Patel - Spacecraft Power Systems PDF

SPACECRAFT
POWER SYSTEMS
SPACECRAFT
POWER SYSTEMS
Mukund R. Patel
CRC PR E S S
Boca Raton London New York Washington, D.C.
Cover photo source: Dornier Satellitensysteme GmbH, Daimler-Benz
Aerospace, Germany. With permission.
Library of Congress Cataloging-in-Publication Data
Patel, Mukund R., 1942-

Spacecraft power systems / Mukund R. Patel
p.cm.
Includes bibliographical references and index.
ISBN 0-8493-2786-5 (alk. paper)
1. Space vehicles – Auxiliary power supply. I. Title
TL1100.P38 2004
629.47 0 44–dc22 2004054509
This book contains information obtained from authentic and highly regarded sources. Reprinted
material is quoted with permission, and sources are indicated. A wide variety of references are
listed. Reasonable efforts have been made to publish reliable data and information, but the author
and the publisher cannot assume responsibility for the validity of all materials or for the
consequences of their use.
Neither this book nor any part may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, microfilming, and recording, or by any
information storage or retrieval system, without prior permission in writing from the publisher.
The consent of CRC Press does not extend to copying for general distribution, for promotion, for
creating new works, or for resale. Specific permission must be obtained in writing from CRC Press
for such copying.
Direct all inquiries to CRC Press, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation, without intent to infringe.
Visit the CRC Press Web site at www.crcpress.com
# 2005 by CRC Press

No claim to original U.S. Government works
International Standard Book Number 0-8493-2786-5
Library of Congress Card Number 2004054509
Printed in the United States of America 1 2 3 4 5 6 7 8 9 0
Printed on acid-free paper
Although reasonable care has been taken in preparing this book, the author
or the publisher assumes no responsibility for any consequence of using the
information contained herein. The diagrams disclosed here are for
illustration purpose only, and may be covered by patents.
. . . dedicated
to my late sister, Bhanumati Amin,
who left loving memories for me to cherish;
and
to my parents, Shakariben and Ranchhodbhai Patel,
who still inspire me in their nineties.
Preface
The first man-made satellite was Sputnik I, launched and operated by Russia
in the low Earth orbit in 1957. In the decade that followed, the American
space exploration programs launched many Earth orbiting satellites. The
first commercial geosynchronous satellite, Intelsat-1, was placed in orbit in
1965, and, in 1969, NASA’s Apollo-11 became the first manned spacecraft to
land on the moon. Since then, many countries have successfully undertaken
small and large space programs. In 2003, a civilian became the first space
tourist and China became the third nation after Russia and the United States
to put a human in space. In 2004, U.S. President Bush announced a new
space initiative for manned landing on the moon by 2015, and then onward
to Mars. In the same year, both China and India announced their own plans
to send unmanned spacecraft to the moon by 2010. On the commercial side,
the recent worldwide communications technology boom has established the
satellite as an integral part of the national infrastructure. Today, many
countries have the capability of placing and operating satellites in orbit.
In the past, the U.S. government has provided a large share of the space
industry revenue, and telecommunications satellites dominated the non-
government market. The 1990s saw milestones in the development of
personal communications systems, expansion of the remote sensing
industry and the approval of sale of spy-quality data. The 2000 world
market for commercial communications satellites was 28 satellites at a
procurement cost of 12 billion dollars, about 50% of which went to U.S.
companies. Added to this was the insurance cost for launch and 1-year on-
orbit operation ranging from 7 to 15% of the launch vehicle and satellite
costs. According to Aerospace America, the total number of launches in
2003 was 90, and is projected to be 150 in 2012. These endeavors symbolize
the increasing expansion of a commercial market. Many changes occurring
in the industry continue the evolution of this dynamic sector of the
technology. New commercial ventures are rapidly expanding. It is on these
ventures that the future expansion of the space industry rests.
The GPS III of the U.S. Air Force with the first launch targeted in the
2012–2015 time frame may be among the largest defense satellite programs
during the next 20 years. It will have some 30 satellites at an estimated cost
of $5 billion. New large-scale commercial satellites, such as Europe’s Galileo
vii
viii Preface
navigation program, will involve many countries, possibly including China.

Science missions also continue to provide new knowledge of our universe.
In the spring of 2001 alone, our horizon was broadened by the discovery of
11 new planets by Keck telescopes in Hawaii, the Haute-Provence
Observatory in France, and the European Southern Observatory in Chile,
bringing the total to about 100 planets in other solar systems. And, recently,
scientists have discovered a solar system just like ours. In our own solar
system, 39 moons of Jupiter, 30 moons of Saturn and a massive reservoir of
water ice just below the Mars surface have been discovered.
Now that commercial and science satellites have established their strong
foothold in the modern world, it is important to optimize their technical
performance and ultimately the rate of return for the investors. The satellite
has certain resources — instruments in science satellites and bandwidth in
communication satellites — all driven by electrical power. This book
focuses on the design, performance, and operation of electrical power
systems for all types of spacecraft, Earth-orbiting, interplanetary, and deep
space for science, commercial, and defense applications.
The design of spacecraft power systems has been advancing at a rapid
rate, along with improvements in the system components. The book
provides the most modern in-depth data and procedures for designing
the electrical power systems to meet the mission requirements at a
minimum cost and weight. The launch cost in 2002 U.S. dollars averaged
about $10,000 per kilogram for placing a satellite in low Earth orbit, and
about $50,000 per kilogram for placing one in a geosynchronous orbit. The
power system weight is therefore a significant design driver. The design
must be totally optimized to the limit such that even a small change in any
one component would result in a penalty of some sort at the satellite level.
Moreover, the engineer must incorporate new technologies as they become
available. However, the satellite is like a spider’s web — touch one thing
and everything moves. It is for this reason that the power system engineer
must be cognizant about more than the solar array and the battery to
optimally design even a conventional satellite. This becomes even more
important in view of the fact that one out of four commercial GEO satellites
launched in the 5-year period 1998 to 2002 has experienced electrical power
problems. Together, they make 60% of the industry-wide insurance claims
for in-orbit operational defects.
The author, a recognized expert in the industry, has written this book to
provide reliable information on a variety of subjects to engineers and
managers involved in spacecraft power systems design, development,
testing, and operation. The details covered in the book would help industry
engineers develop and maintain a leading edge in this rapidly growing and
evolving industry.
Yardley, Pennsylvania
Acknowledgment
The emerging technologies in the spacecraft power systems cannot

possibly be documented in a book without help from many sources.
I have been extremely fortunate to receive full support from many
organizations and individuals in the field. The book is a direct result of
my teaching a continuing education course to engineers working at many
prime contractors and NASA centers. I am grateful to Marshall Kaplan,
president of Launchspace Inc., for inviting me to develop and teach
the course since 1995 under the initial sponsorship of the U.S. Space
Foundation.
Several organizations provided current data and reports on the space
power technologies. They are NASA, the European Space Agency, the U.S.
Department of Energy, and spacecraft manufacturers worldwide. Many
expert individuals at those organizations gladly provided all the help I
asked for. My special gratitude goes to Jim Haines, head of the Power
Systems Section at the European Space Agency, and Bob Beaman, senior
staff engineer at NASA Goddard Space Flight Center, for reviewing the
book proposal and providing valuable comments.
Professional engineers attending my courses often knew more in their
specialties than I did. Their stimulating discussions have continuously
sharpened my knowledge on space power. Several of them not only
encouraged me to write the book on this timely subject, but also provided
valuable suggestions and comments during the development of the book.
I am also grateful to Professor Jose Femenia, Engineering Department
Head; Dr. Warren Mazek, Dean; and Vice Admiral Joseph Stewart,
Superintendent; of the U.S. Merchant Marine Academy, Kings Point, NY,
for supporting my research and publications, including a sabbatical leave
for writing this book.
I heartfully acknowledge the valuable support from all.
Mukund R. Patel
ix
About the Book
This book has evolved from a three-day continuing education course taught
by the author on spacecraft electrical power systems over the last ten years.
The material covered is based on the author’s deep involvement in many
design and development programs at the General Electric Space Division
and the Lockheed Martin Space Systems.
Spacecraft power systems have undergone significant new developments
in the last decade and will continue to do so even at a faster rate in the
current decade. The design and analyses methods used by the engineers
have evolved as needed in every corporation. Except for limited publica-
tions in bits and pieces at various conferences, and a few books having short
sections with sketchy coverage on this vast subject, there is no single book
available as of today that covers the entire scope of spacecraft power
systems.
All aspects of the electrical power system design, analyses, and operation
are covered in this book in details not found in any other book. The wide
coverage of the fundamentals of energy conversion, energy storage, power
conditioning, energy management, and the operational aspects will help
engineers maintain a leading edge in designing and operating a variety of
spacecraft power systems.
The book is written to benefit the spacecraft power system design and
development engineers, mechanical and aeronautical engineers, chief
engineers and program managers. It can serve as a ready reference book.
It assumes a college level background in physics and mathematics, and can
also be used as a textbook for senior or graduate level university courses.
Although the designs are always mission specific, the design numbers or
curves are given at many places as examples or typical values for that class
of satellites only.
The book covers all aspects of spacecraft power systems in 27 chapters
divided in four major parts:
Part A: Satellite overview, space environment and its effects, power
system options, PV-battery system, power system requirements, and design
trades.
x
About the Book xi
Part B: Solar array, battery, power electronics and magnetics, distribution

harness, protection, and ancillary components.
Part C: Energy management, dynamic performance and stability, EMI/
EMC, ESD, reliability, derating, and integration and testing.
Part D: Special systems such as interplanetary and deep space, nuclear
RTG, thermodynamic system with alternator, high-power and high-voltage
systems, electric propulsion, flywheel and superconducting energy storage,
fuel cell, and microwave beam power satellite.
Both systems of units — International and British — are used in the book
to present data, as they came from various sources. An extensive conversion
table is therefore given in the front, and so is a list of acronyms commonly
used in the spacecraft industry and also in the book.
About the Author
Mukund R. Patel, Ph.D., PE, is a research engineer, educator and

consultant in electrical power systems. He has served as Senior Staff
Engineer at Lockheed Martin Space Systems, Principal Engineer at General
Electric Space Division, Fellow Engineer at Westinghouse Research Center,
Space Products Manager at Induction General Inc. and 3M Distinguished
Visiting Professor at the University of Minnesota, Duluth. Presently, he is a
Professor at the U.S. Merchant Marine Academy, Kings Point, NY.
Dr. Patel earned his M.S. and Ph.D. degrees from the Rensselaer
Polytechnic Institute, Troy, NY, in 1967 and 1972, respectively. He is a
Fellow of the Institution of Mechanical Engineers (UK), Associate Fellow of
the American Institute of Aeronautics and Astronautics (AIAA), Senior
Member of the Institute of Electrical and Electronic Engineers (IEEE),
Professional Electrical Engineer in Pennsylvania, Chartered Mechanical
Engineer in the UK, and Member of Eta Kappa Nu, Tau Beta Pi, Sigma Xi
and Omega Rho. He has published over 50 papers at national and
international conferences, holds several patents, and has earned NASA
recognition for exceptional contribution to the power system design for
UARS. He has taught IEEE workshops and in-house courses at GE. Since
1995, he has been teaching the Spacecraft Power System Design Course to
professional aerospace engineers across the United States.
He lives in Yardley, PA, with his wife, Sarla. They have three grown up
children, Ketan, Bina, and Vijal; and two grandchildren, Rayna and Dhruv.
He can be reached at patelm@usmma.edu.
xii
Acronyms and Abbreviations
ACS attitude control system

ADE array drive electronics
AE autumnal equinox
AIAA American Institute of Aeronautics and Astronautics
AMTEC alkaline metal thermoelectric converter
ANSI American National Standards Institute
AO atomic oxygen
APL Applied Physics Laboratory (Johns Hopkins University)
APS auxiliary power supply
a-Si amorphous silicon (PV cell)
AU astronomical unit
AWG American wire gage
BCC battery charge controller

BDCU bi-directional battery charge–discharge unit
BCN battery charge network
BCU battery charge unit
BCR battery charge regulator
BCVM battery cell voltage monitor
BDN battery discharge network
BDR battery discharge regulator
BDU bus data unit
BIU battery interface unit
BJT bipolar junction transistor
BLU battery letdown unit
BOL beginning of life
BPC battery power converter
BPM Battery pressure monitor
BSR back surface reflector
BU backup unit
BVR bus voltage regulator
CASI Canadian Aeronautics and Space Institute

CD command directory
xiii
xiv Acronyms and Abbreviations
CDR critical design review

C/D charge/discharge
CDU command decoder unit
CIGS copper indium gallium diselenide
CMD command
C&DH command and data handling
CNES Center National d’Etudes Spatiales (France)
COTS commercial off-the-shelf (equipment)
CPV common pressure vessel (with two battery cells)
c-Si crystalline silicon (PV cell)
DAR double anti-reflective (coating)

DARPA Defense Advanced Research Projects Agency
DCA Defense Communication Agency
DDCU d.c. to d.c. converter unit
DET direct energy transfer
DEU deployment electronics unit
DIPS dynamic isotopes power system
DMS defense meteorological satellite
DOD depth of discharge (of battery)
DoD Department of Defense (U.S.)
DoE Department of Energy (U.S.)
DPV dependent pressure vessel (battery)
DPU deployment power unit
DRL Deuche (German) Aerospace Center
DSCS defense system communications satellite
(< 1 kW load, 28 0.28 V, GEO bus)
EED electro-explosive device

EHD electro-hydrodynamic
EHT electro-thermal hydrazine thruster
ELF extra-low frequency
EMC electromagnetic compatibility
EMF electromotive force
EMI electromagnetic interference
EMP electromagnetic pulse
EMW energy-momentum wheel
EOL end of life
EOS-AM Earth observing system (3-kW load, 120 V 5 V, LEO bus)
EPS electrical power system
EQX equinox
ESA European Space Agency
ESD electrostatic discharge
ESR equivalent series resistance
Acronyms and Abbreviations xv
FBA fuse board assembly

FESS flywheel energy storage system
FIT failure rate in time (106 hours)
FMECA failure mode and effect analysis
FPGA field programmable gate array
GN&C guidance, navigation and control

GEO geosynchronous Earth orbit
GPS Global Positioning System (of U.S. Air Force, NAVSTAR)
(800-W load, 28-V Regulated, 10900 nm, 638 inclined, MEO
bus)
GRC Glenn Research Center (of NASA at Lewis Field)
GSE ground support equipment
GSFC Goddard Space Flight Center (of NASA)
HAJ hydrazine arcjet

HEO highly elliptical orbit (Molniya)
HTSC high-temperature superconductor
HV high voltage
IECEC Intersociety Energy Conversion Engineering Conference

IEEE Institute of Electrical and Electronic Engineers
IEMS integrated energy momentum wheel system
IGBT insulated gate bipolar transistor
INM Inmarsat satellites (3500-W load, 28-V sun-regulated, GEO
bus)
IMC inter-module connector
I/O input/output
IPS ion propulsion system
IPV individual pressure vessel (for one battery cell)
IR infrared
Isp specific impulse (per kg of fuel)
ISRO Indian Space Research Organization
ISS International Space Station (105 kW, 160/120 V regulated,
LEO bus)
ISO International Standards Organization (Geneva, Switzerland)
I–t current versus time
I&T integration and testing
ITO indium tin oxide
I–V current versus voltage
JPL Jet Propulsion Laboratory
KOH potassium hydroxide (electrolyte)

xvi Acronyms and Abbreviations
LCP light-concentrating panel

LEO low Earth orbit
Li-Ion lithium ion
Li-Poly lithium polymer (solid electrolyte)
LN liquid nitrogen (1938C boiling point)
LISN line impedance stabilization network
LV low voltage
LPC load power conditioner
MEO mid Earth orbit

MBSU main bus switching unit
MEP main enabling plug
M-G motor–generator
MHD magneto-hydrodynamic
MIL-SPEC military specifications (U.S.)
MIL-STD military standard (U.S.)
MMW multimegawatts
MOSFET metal-oxide semiconducting field effect transistor
MPG multiple point ground
MTBF mean time between failures
MWA momentum wheel assembly
NASA National Aeronautics and Space Administration

NASDA National Space Development Agency (of Japan)
NiCd nickel cadmium battery
NiH2 nickel hydrogen battery
NiMH nickel metal hydride
N-S north-south
NSSK north-south station keeping
NOAA National Oceanic and Atmospheric Administration
NSTS National Space Transportation System
OBC on-board computer

OC ordnance controller, also open circuit
PCDU power control and distribution unit

PCE payload control electronics
PDP power distribution point
PCU power control unit
PDU power distribution unit
PEM proton exchange membrane (fuel cell)
PEMS power and energy management software
PM permanent magnet
PMAD power management and distribution
PMCS power management and control software
Acronyms and Abbreviations xvii
PME propulsion module electronics

P-P peak-to-peak (ripple)
PPT peak power tracking
PRI primary unit
PRP power return point
PRU power regulator unit
PPR prime power return
PRA pyrotechnic relay assembly
PV photovoltaic
P–V power versus voltage
PWM pulse width modulation
PYRO pyrotechnic (ordnance)
RBI remote bus isolator

R-DET regulated-direct energy transfer
RFI radio frequency interference
RFC regenerative fuel cell
rms root mean squared
rpm revolution per minute
RPC remote power controller
rss root sum squared
RTG radioisotope thermoelectric generator
RTN return
RUL remaining useful life (of battery)
RUE remaining useful energy (of battery)
RWA reaction wheel assembly
SA solar array
S&A safe and arm
SAD solar array drive
SAPS solar array position sensor
SAS solar array switch
SBA shunt boom assembly
SBR space-based radar
SC short circuit
SD shunt dissipator
SDI Strategic Defense Initiative
SDV shunt drive voltage
SESB single-event secondary breakdown
SEU single event upset
SEL single event latch-up
SLA stretched lens array
SMC safe mode controller
SOC state of charge (of battery)
SPA solar panel assembly
xviii Acronyms and Abbreviations
SPE solid polymer electrolyte (fuel cell)

SPG single point ground
SPS solar power satellite
SPT stationary plasma thruster
SPV single pressure vessel (for all battery cells)
SP-100 Space Power-100 (DoD Program)
SR-DET sunlight regulated direct energy transfer
SS summer solstice
SSA sun sensor assembly
SSU sequential shunt unit
SSM switched shunt module
SSPA solid state power amplifier
SSPC solid state power controller
SSS solid state switch
TCS thermal control system

TDMA time division multiple access
TDRS tracking and data relay satellite (1700-W load, 28-V
regulated, GEO bus, 08 inclination)
TE thermoelectric
TEC thermoelectric converter
TF transfer function
TLM telemetry
TPR thrust to power ratio
TPV thermo-photovoltaic
TRDF total ripple distortion factor
TSS tethered satellite system
TT&C tracking, telemetry and command
TWTA traveling wave tube amplifier
USRA Universities Space Research Association

USSF U.S. Space Foundation (Colorado Springs, CO)
VE vernal equinox
V=f voltage to frequency ratio
V–t voltage versus time
V–T voltage versus temperature
XIPS xenon ion propulsion system
WCBV worst case bus voltage in normal operation

WS winter solstice
Acronyms and Abbreviations xix
Suffixes
mp maximum power point
oc open circuit
op operating point
sc short circuit
Trade names
A2100 Lockheed Martin Space Systems, Sunnyvale, CA
BSS-702 and BSS-601 Boeing Satellite Systems, El Segundo, CA
Bussmann Cooper Bussmann Inc., St. Louis, MO
Littelfuse Littelfuse Corporation, Des Plaines, IL
Microsheet Corning Inc., Corning, NY
Nomex, Mylar, Kapton E.I. Dupont Company, Wilmington, DE
Kevlar, KelF, Teflon
SuperNiCd Eagle Picher Technologies, LLC, Joplin, MO
P600L Fuse Mepcopal Corporation, San Diego, CA
PADS A/D Mentor Graphics, Wilsonville, OR
PSPICE Intusoft Corporation, Gardena, CA
Systems of Units and Conversion Factors
SI unit = English unit Factor below

Length meter foot 0.304 8
mm inch 25.4
micron mil 25.4
km mile 1.609 3
km nautical mile 1.852
m2 foot squared 0.092 9
Area mm2 circular mil 506.7
liter (dm3) cubic foot 28.316 8
Volume liter cubic inch 0.016 39
cm3 cubic inch 16.387 1
m3/s cubic foot/hr 0.028 31
liter gallon (US) 3.785 3
liter/s gallon/minute 0.063 09
Mass kg pound mass 0.453 59
kg Slug 14.593 9
Density kg/m3 pound/mass/ft3 16.020
kg/cm3 pound mass/in3 0.027 68
Force N pound force 4.448 2
Pressure kPa pound/in2 (psi) 6.894 8
kPa Bar 100.0
kPa std atm (760 torr) 101.325
kPa 1mm Hg at 208C 0.132 84
Torque Nm pound-force foot 1.355 8
Power W foot pound/sec 1.355 8
W Horsepower 745.7
Energy J foot pound-force 1.355 8
kJ Btu International 1.055 1
MJ horsepower hour 2.684 5
MJ Therm 105.506
Temperature 8C 8F (8F 32)5/9
K R (8F + 459.67)5/9
Heat W Btu(Int.)/hour 0.293 1
kW Ton Refrigeration 3.517
xx
Systems of Units and Conversion Factors xxi
W/m2 Btu/(ft2 hr) 3.154 6

W/(m28C) Btu/(ft2 hr) 8F) 5.678 3
MJ/(m38C) Btu/(ft3 8F) 0.067 1
W/(m8C) Btu inch/(ft2 hr 8F) 0.144 2
W/(m8C) Btu ft/(ft2 hr 8F) 1.730 4
J/kg Btu/pound 2.326
MJ/m3 Btu/ft3 0.037 3
J/(kg8C) Btu/(pound 8F) 4.186 8
Velocity m/s foot/sec 0.304 8
m/s mile/hour 0.447 04
Magnetics Weber line 108
wb/m2 (tesla) kiloline/inch2 0.0155
Other Units and Conversions

1 e ¼ 0.1592 1018 coulomb
1 eV ¼ 0.1592 1018 J
1 MeV ¼ 0.1592 1012 J
1 calorie ¼ 4.1868 J
1 kg cal(Int.) ¼ 4.1868 kJ
1 erg ¼ 0.1 mJ
1 dyne ¼ 10 mN
1 tesla ¼ 10,000 gauss (lines/cm2)
1 micron ¼ 1 mm
1 nautical mile ¼ 1.15081 mile
1 rad ¼ 100 erg/g
1 rad(Si) ¼ 1 krad ¼ 10 J/kg
1 rad(Si) generates 1.7 1013 electron hole pairs in 1 g of silicon
Absolute zero temperature ¼ 273.168C ¼ 459.67 8F
Boltzmann constant ¼ 1.38 10 23 J/K
Stefan–Boltzmann constant ¼ 5.67 108 W/m2 K4
Space Flight Constants
Acceleration due to Earth gravity 9.806 7 m/s2 (32.173 5 ft/s2)

Earth equatorial radius (mean) 6378.140 km
Geosynchronous orbit altitude 35,786 km
Obliquity of ecliptic at Epoch 2000 23.439 2918
Precession of the equinox 1.396 971 38/century
Earth-sun distance (AU) 149.597 870 million km
Mean lunar distance 0.384 401 million km
Sidereal year a,b,c,d 3.158 814 955 107 seconds
Solar year 365 days, 6 hours, 9 minutes,
9.54 s
Solar flare maxima year 1990 + 11.0 n (n ¼ 1,2, . . .)
Solar flux radiation 1358 5 W/m2 at 1 AU
Earth annual mean albedo 0.30 (fraction of the solar flux
reflected by the Earth)
Earth’s long wavelength radiatione,f,g 237 7 W/m2
Earth’s effective black body temperature 206 to 262 K
Permittivity of free space, o 4 107 henry/m
Permittivity of free space, "o 8.851012 farad/m
1
Velocity of light in free space Vo ¼ pffiffiffiffiffiffiffiffiffi ¼ 299:863 106 m/s
o " o
a. Solar year is measured with respect to the sun. It is the time the
Earth takes to complete one revolution around the sun, which is
356 d, 6 h, 9 m, 9.54 s.
b. Sidereal year is the time interval between two successive passages of
the vernal equinox over the meridian.
c. Solar day is the time the Earth takes to complete a rotation, relative
to the sun, measured from noon to noon at the prime meridian. It is
divided into 24 solar hours.
d. Sidereal day is the time the Earth takes to complete one revolution
around the imaginary axis through the North and South Poles. It is
measured with respect to any fixed start in the sky, and is equal to
23 h, 56 m, 4.09 s, about 4 m shorter than the solar day. It is divided
into 24 sidereal hours.
xxii
Space Flight Constants xxiii
e. Long wavelength radiation is a minimum when albedo is a

maximum
f. Due to orbital orientation, peak solar flux, albedo or long wave-
length radiation may not be incident on a given surface
g. Diurnal (daily) variation can be 36 W/m2.
List of Chapters
Part A: Power System Fundamentals

1 Satellite Overview 3
2 Near-Earth Space Environment 26
3 Power System Options 40
4 Photovoltaic-Battery System 58
5 Environmental Effects 92
6 Power System Requirements 105
7 Design Process and Trades 111
Part B: PV–Battery System

8 Solar Array 135
9 Battery 195
10 Power Electronics and Magnetics 255
11 Distribution Harness and Protection 293
12 Ancillary Components 319
Part C: Power System Performance

13 Energy Balance and Power Management 347
14 Dynamic Performance and Stability 368
15 Electromagnetic Interference and Compatibility 394
16 Electrostatic Discharge 421
17 Reliability and Derating 431
18 Integration and Testing 451
Part D: Special Power Systems

19 Interplanetary and Deep Space Missions 473
20 Radioisotope Thermoelectric Generator 489
21 Dynamic System with Alternator 505
xxiv
List of Chapters xxv
22 High-Power High-Voltage Systems 518

23 Electric Propulsion 566
24 Fuel Cell Power 591
25 Flywheel Energy Storage 604
26 Superconductors in Space 640
27 Microwave Beam Power Satellite 653
Contents
Preface vii
Acknowledgment ix
About the Book x
About the Author xii
Acronyms and Abbreviations xiii
Systems of Units and Conversion Factors xx
Space Flight Constants xxii
List of Chapters xxiv
Part A: Power System Fundamentals
Chapter 1 Satellite Overview 3

1.1 Introduction 3
1.2 Satellite systems 4
1.2.1 Communications and data handling 4
1.2.2 Attitude and orbit control system 4
1.2.3 Tracking, telemetry, and command system 5
1.2.4 Electrical power system 5
1.2.5 Thermal control system 5
1.2.6 Structure and mechanisms system 6
1.2.7 Propulsion system 6
1.3 Earth orbit classification 6
1.3.1 Geostationary orbit 8
1.3.2 Geosynchronous orbit 9
1.3.3 Highly elliptical orbit 11
1.3.4 Low Earth orbit 11
1.3.5 Sunsynchronous orbit 12
1.4 Orbit mechanics 12
1.5 Satellite stabilization methods 13
1.5.1 Gravity gradient method 13
1.5.2 Magnetic damping 13
1.5.3 Spin stabilization 13
xxvi
Contents xxvii
1.5.4 Three-axis stabilization 14

1.6 Launch and transfer orbits 15
1.7 Operational orbit 16
1.8 Eclipse due to Earth 16
1.8.1 Example 20
1.9 Eclipse due to moon 21
1.10 Solar flux 22
1.11 Beta angle 23
1.12 Spacecraft mass 24
Reference 25
Chapter 2 Near-Earth Space Environment 26

2.1 Introduction 26
2.2 Launch and transfer orbit environment 26
2.3 On-orbit environment 27
2.3.1 Lack of gravity and atmosphere 27
2.3.2 Magnetic field 27
2.3.3 Meteoroids and debris 28
2.3.4 Atomic oxygen 29
2.3.5 Charged particles 30
2.4 Van Allen belts 31
2.5 Solar wind and solar flare 34
2.6 Geomagnetic storm 36
2.7 Nuclear threat 37
2.8 Total radiation fluence 37
References 39
Chapter 3 Power System Options 40

3.1 Introduction 40
3.2 Primary battery 41
3.3 Fuel cell 42
3.4 Solar PV–battery 43
3.5 Solar concentrator–dynamic power system 45
3.6 Nuclear–thermoelectric 48
3.7 Nuclear or chemical–dynamic 49
3.8 Other systems 50
3.8.1 Thermo-photovoltaic 50
3.8.2 Solar–thermoelectric 51
3.8.3 Thermionic 51
3.8.4 Alkaline metal thermal to electric
converter 52
3.9 Technology options compared 52
3.10 System voltage options 53
3.11 Scaling for power level 56
References 57
xxviii Contents
Chapter 4 Photovoltaic–Battery System 58

4.1 Introduction 58
4.1.1 Solar array 58
4.1.2 Battery 59
4.1.3 Power regulation 60
4.2 Power system architectures 62
4.2.1 Direct energy transfer 62
4.2.2 Peak power tracker 63
4.3 Fully regulated bus 63
4.3.2 Solar array drive 64
4.3.3 Shunt dissipator 64
4.3.4 Battery 64
4.3.5 Power regulator unit 65
4.3.6 Power distribution circuit 65
4.3.7 Bus voltage controller 65
4.3.8 Mode controller 65
4.3.9 Battery bus 66
4.3.10 Power and energy management
software 67
4.3.11 Loads 67
4.3.12 Ground power cord 67
4.4 Bus voltage control 67
4.5 Control circuit 69
4.5.1 Analog control 69
4.5.2 Digital control 69
4.5.3 Analog–digital hybrid control 70
4.6 Sun-regulated bus 70
4.7 Fully regulated vesus sun-regulated bus 71
4.8 Peak power tracking bus 73
4.9 Architecture trades 78
4.10 The International Space Station 160- to 120-V bus 79
4.11 Large communications satellite buses 83
4.11.1 100-V bus 83
4.11.2 70-V bus 84
4.11.3 Under-50-V buses 86
4.12 Small satellite bus 86
4.13 Micro-satellite bus 89
References 91
Chapter 5 Environmental Effects 92

5.1 Introduction 92
5.2 Solar array degradation 92
5.3 Electrostatic discharge in the solar array 93
5.4 Damage to the power electronics 95
Contents xxix
5.5 Effects on other components 98

5.6 Mass erosion under atomic oxygen 99
5.7 Micrometeoroid and debris impacts 102
5.8 Predicting damage 102
References 104
Chapter 6 Power System Requirements 105

6.1 Introduction 105
6.2 Self-derived requirements 106
6.3 System specifications 107
Chapter 7 Design Process and Trades 111

7.2 Spacecraft level trades 111
7.3 Power system level trades 113
7.3.1 PV cell trades 114
7.3.2 Solar array trades 116
7.3.3 Battery trades 117
7.3.4 Bus voltage trades 117
7.3.5 Pyro power trades 118
7.4 Load power profile 119
7.5 Solar array sizing 122
7.6 Battery sizing 124
7.7 Power flow analysis 125
7.8 Design analyses list 127
7.9 Worst-case error margin 127
7.10 Design process phases 129
7.11 Factory-to-orbit events 130
7.12 Power system functions over life 131
Part B: PV–Battery Power System
Chapter 8 SolarArray 135

8.2 Photovoltaic cell 135
8.3 PV technologies 137
8.3.1 Single-crystal silicon 138
8.3.2 Gallium arsenide 140
8.3.3 Semicrystalline and polycrystalline 140
8.3.4 Thin film 140
8.3.5 Amorphous 140
8.3.6 Multi-junction 141
8.4 Equivalent electrical circuit 141
xxx Contents
8.5 I–V and P–V characteristics 143

8.6 Power from albedo 147
8.7 Back-surface illumination 147
8.8 Array construction 148
8.8.1 Rigid panels 148
8.8.2 Body mounted array 150
8.8.3 Three or more wings 151
8.8.4 Flexible array 151
8.8.5 Inflatable arrays 153
8.9 Array performance 154
8.9.1 Sun intensity 154
8.9.2 Sun angle 154
8.9.3 Temperature effect 155
8.9.4 Sun acquisition 158
8.9.5 Sun tracking 161
8.9.6 Peak power extraction 162
8.9.7 Shadow effect 165
8.10 Array design 166
8.10.1 Degradation factors 169
8.10.2 Radiation damage 169
8.10.3 Fluence calculation 175
8.10.4 Offsetting from sun 176
8.10.5 Magnetic moment 177
8.11 Advances in PV technology 178
8.12 Concentrator array 183
8.12.1 Stretched lens array 186
8.12.2 Light concentrating panel 188
8.12.3 Parabolic concentrator 191
References 193
Chapter 9 Battery 195

9.2 Electrochemical cell 195
9.3 Types of batteries 197
9.3.1 Nickel cadmium 198
9.3.2 Nickel hydrogen 200
9.3.3 Nickel metal hydride 208
9.3.4 Lithium-ion 209
9.3.5 Lithium polymer 212
9.3.6 Silver zinc 213
9.4 Electrical circuit model 213
9.5 Properties and performance 214
9.5.1 Charge/discharge 214
9.5.2 Internal impedance 217
9.5.3 Charge efficiency 217
Contents xxxi
9.5.4 Energy efficiency 218

9.5.5 V–T characteristics 219
9.5.6 Self-discharge 219
9.5.7 Self-heating 221
9.6 Cycle life 224
9.7 Burst power capability 227
9.8 Performance comparison 228
9.9 Battery design 228
9.10 Launch and ascent power 234
9.11 Thermal design 235
9.11.1 Flat pack 237
9.11.2 Vertical stack 238
9.11.3 Wine rack 238
9.12 Safety considerations 238
9.13 Charge regulation 240
9.13.1 Multiple charge rates 242
9.13.2 Single charge rate 242
9.13.3 Unregulated charging 243
9.14 Battery management 243
9.15 Dynamic model 245
9.16 Cycle life model 248
9.17 Primary battery 251
9.18 Advances in battery technologies 251
References 253
Chapter 10 Power Electronics and Magnetics 255

10.2 Switching devices 255
10.3 Shunt regulator 257
10.3.1 Full and partial shunts 258
10.3.2 Linear and PWM shunts 259
10.3.3 Sequential linear shunt 261
10.3.4 Multistage PWM shunt 261
10.3.5 Polyphase PWM shunt 262
10.3.6 Calibrated step shunt 263
10.4 Shunt circuit design 263
10.4.1 Full versus partial shunts 263
10.4.2 Number of shunt circuits 264
10.4.3 Typical shunt circuit design 264
10.5 Bus ripple filter design 265
10.6 Power converters 266
10.6.1 Battery charge (buck) converter 266
10.6.2 Battery discharge (boost) converter 270
10.6.3 Buck–boost converter 271
10.6.4 Flyback converter (buck or boost) 272
xxxii Contents
10.6.5 Transformer coupled forward converter 273

10.6.6 Push–pull converter 274
10.6.7 Inductor coupled buck converter 274
10.6.8 Resonant converter 274
10.6.9 Multiple output converter 275
10.6.10 Load power converter 276
10.6.11 Voltage and current regulators 277
10.7 Magnetics 279
10.7.1 Power losses in core and coil 282
10.7.2 Inductor design 284
10.7.3 Transformer design 286
10.7.4 Manufacturing and testing of magnetics 288
10.8 Design for maximum efficiency 288
10.9 Trends in power electronics 291
References 291
Chapter 11 Distribution Harness and Protection 293

11.2 Ampacity of wires 293
11.3 R-C-L parameters 294
11.4 Conductor materials 297
11.5 Wire insulation and cable shield 299
11.6 Connectors 299
11.7 Harness mass minimization 301
11.7.1 Example 303
11.8 Harness design process 304
11.9 Flexible harness 306
11.10 Fuse protection 306
11.10.1 Fuse rating 309
11.10.2 Types of fuses 311
11.10.3 Fuse characteristics 312
11.10.4 Fuse derating 314
11.10.5 Fuse selection criteria 315
11.10.6 Redundant and parallel fuses 315
11.10.7 Fuse testing 316
11.10.8 Unfused loads 317
11.11 Remote power controllers 317
11.12 Early fault detection 318
Chapter 12 Ancillary Components 319

12.2 Solar array drive 319
12.3 Deployment devices 322
12.3.1 Electro-explosive device 322
12.3.2 Laser-initiated deployment 323
Contents xxxiii
12.3.3 Shape memory metal 324

12.4 Deployment controller 324
12.5 Thermal controller 326
12.5.1 Thermocouple 326
12.5.2 Metallic resistance wire 326
12.5.3 Thermistor 326
12.6 Relays 327
12.7 BPM and BCVM 328
12.8 Battery let-down unit 330
12.9 Current meters 330
12.9.1 Shunt 330
12.9.2 Hall-effect transducer 331
12.9.3 Magnetic amplifier 331
12.10 Capacitors 332
12.10.1 Tantalum capacitor 332
12.10.2 Metalized polypropylene capacitor 332
12.10.3 Ceramic capacitor 333
12.11 Filters 333
12.12 Telemetry and commands 334
12.13 Electronic packaging 336
12.14 Radiation shield 337
12.14.1 Example of radiation shield design 339
12.15 EMI shield 340
Part C: Power System Performance

Chapter 13 Energy Balance and Power Management 347
13.2 Energy balance analysis 349
13.3 Computer program structure 350
13.3.1 Battery module 350
13.3.2 Solar array module 354
13.3.3 Bus module 355
13.3.4 Input variables 355
13.3.5 Output format 355
13.3.6 Program verification 356
13.4 Energy balance simulation runs 356
13.4.1 On-orbit simulation 357
13.4.2 Transfer orbit simulation 358
13.4.3 Transfer orbit load budget 361
13.4.4 Launch and ascent simulation 362
13.5 Battery state of health monitoring 363
13.6 Battery latch-up in sun-regulated bus 365
References 367
xxxiv Contents
Chapter 14 Dynamic Performance and Stability 368

14.2 Bus impedance and system stiffness 368
14.3 Voltage regulation and transients 371
14.4 High frequency ripples 374
14.5 Ripple measurement 376
14.6 Quality of power 378
14.7 Minor fuse blow cross-talk 378
14.7.1 Example 378
14.8 Major fuse blow transient 379
14.9 Stability and bus impedance 382
14.10 Dynamic stability of control system 386
14.11 Dynamic simulation model 386
14.11.1 Solar array model 389
14.11.2 Other component models 391
14.12 Simulation runs 392
References 393
Chapter 15 Electromagnetic Interference and Compatibility 394

15.2 Sources of EMI 394
15.2.1 Conduced EMI 395
15.2.2 Radiated EMI 395
15.3 Modes of coupling 397
15.4 EMI/EMC specifications 403
15.5 EMI suppression methods 407
15.5.1 Twisting wires 408
15.5.2 Grounding 408
15.5.3 Cable shielding 409
15.5.4 Bonding 410
15.6 Common mode EMI 411
15.7 Broadband EMI 412
15.8 Commercial off-the-shelf equipment 413
15.8.1 EMI requirements in commercial
equipment 414
15.8.2 NASA versus commmercial test
methods 415
15.8.3 Testing COTS for space compliance 416
15.9 Electromagnetic pulse/nuclear threat 417
Chapter 16 Electrostatic Discharge 421

16.2 ESD in GEO 421
16.3 ESD in LEO 422
16.4 Dielectric breakdown 423
Contents xxxv
16.5 ESD entry points 424

16.5.1 Strike through gaps 425
16.5.2 Puncture of solid insulation 425
16.5.3 Indirect interaction 425
16.6 Effect of ESD 425
16.7 ESD mitigation 426
16.8 ESD control in solar array 427
16.9 Part sensitivity 429
16.10 Safety during tests 430
Reference 430
Chapter 17 Reliability and Derating 431

17.2 Random failures 433
17.3 Wear-out failure 434
17.4 Fundamental theorems of reliability 434
17.5 Series–parallel reliability 435
17.5.1 Examples 436
17.6 Redundancies 439
17.6.1 Active n for (n1) units 440
17.6.2 Active n for m units 441
17.6.3 Dormant n and active m units 441
17.6.4 Shared versus separate battery chargers 442
17.6.5 Diode o-ring for active redundancy 443
17.7 Failure rate statistics 444
17.8 MIL-HDBK-217 445
17.9 Part-count method of reliability estimate 446
17.10 Derating for reliability 447
17.11 Quick estimate of failure rate 448
17.11.1 Example 448
17.12 FMECA 449
17.13 Non-redundant components 449
17.14 Example of reliability calculation 450
Chapter 18 Integration and Testing 451

18.2 Test limits 451
18.2.1 Survival (non-operating mode only) 455
18.2.2 Qualification (operating mode only) 455
18.2.3 Acceptance (operating mode only) 455
18.2.4 Flight design analysis (all modes) 455
18.2.5 Temperature location 455
18.3 Test sequence 456
18.3.1 Ambient functional performance test 456
18.3.2 Leak 456
xxxvi Contents
18.3.3 Pressure 457

18.3.4 Random vibration 457
18.3.5 Sine vibration 457
18.3.6 Post-vibration performance 457
18.3.7 Acoustic 457
18.3.8 Thermal vacuum cycles 458
18.3.9 EMI/EMC 458
18.3.10 ESD 458
18.3.11 Life test 458
18.4 Solar cell tests 458
18.5 Tests on wires 459
18.6 Power system tests 459
18.6.2 Solar array drive 462
18.6.3 Battery assembly 463
18.6.4 Power electronics 464
18.6.5 Bus impedance and stability 464
18.7 Spacecraft level tests 465
18.7.1 Initial power turn-on test 465
18.7.2 Electrical power system test 465
18.8 Test points 466
18.9 Ground test equipment 467
18.10 Launch site testing 467
Part D: Special Power Systems

Chapter 19 Interplanetary and Deep Space Missions 473
19.2 Temperature in deep space 475
19.3 Lunar mission 478
19.4 Mercury mission 478
19.5 Near-sun mission 480
19.6 Mars mission 481
19.7 Missions to Jupiter and Saturn 484
19.8 Deep space missions 486
References 488
Chapter 20 Radioisotope Thermoelectric Generator 489

20.2 Thermoelectric fundamentals 490
20.2.1 Single-stage unicouple 492
20.2.2 Single-stage multicouple 493
20.2.3 Multistage multicouple 494
20.2.4 RTG assembly 494
Contents xxxvii
20.3 Electrical model of RTG 495

20.4 Maximum power transfer 496
20.5 Effect of temperature and aging 497
20.6 Flight history of RTGs 500
20.7 Segmented TECs 500
20.8 Advanced RTGs 501
20.9 Thermoelectric cooler 503
References 504
Chapter 21 Dynamic System with Alternator 505

21.2 Thermodynamics 508
21.2.1 Stirling engine 508
21.2.2 Brayton cycle 510
21.3 Electromechanical energy conversion 510
21.4 Electrical generator technologies 512
21.4.1 Brief description of each machine 512
21.4.2 Parallel operations 515
21.4.3 Overload capability 515
21.4.4 Voltage regulation and field excitation 515
21.4.5 Speed and frequency limitations 516
References 517
Chapter 22 High-Power High-Voltage Systems 518

22.2 High-voltage PV array 519
22.3 High-power nuclear TEC 520
22.4 High-voltage design issues 524
22.4.1 Paschen breakdown voltage 524
22.4.2 Dielectric stress concentration 525
22.4.3 Corona degradation 526
22.4.4 Atomic oxygen 529
22.4.5 Plasma and charged particles 529
22.4.6 Temperature extremes 534
22.4.7 Design guidelines 535
22.5 High-voltage direct current 536
22.6 Alternating current versus direct current 537
22.7 High-frequency alternating current 539
22.7.1 20-kHz cable 541
22.7.2 EMI in 20-kHz alternating current 542
22.8 High-power components 543
22.8.1 Rotary power transfer 544
22.8.2 Switching and protection 545
22.8.3 Semiconducting devices 547
22.8.4 Capacitor 548
xxxviii Contents
22.8.5 Power converters 548

22.8.6 Transformer and inductor 549
22.9 Very-high-voltage system 550
22.10 Repetitive pulse power 552
22.11 Multi-megawatt burst-power 555
22.12 High-temperature components 561
References 565
Chapter 23 Electric Propulsion 566

23.2 Specific impulse 568
23.3 Types of electric propulsion 570
23.3.1 Electrothermal hydrazine thruster 570
23.3.2 Hydrazine arcjet 570
23.3.3 Ion thruster 571
23.3.4 Stationary plasma thruster 574
23.3.5 Magneto-hydrodynamic 575
23.3.6 Pulsed plasma thruster 576
23.4 Performance comparison 576
23.5 Solar PV propulsion 578
23.6 Solar thermal propulsion 580
23.7 Nuclear power propulsion 581
23.8 Microwave beam propulsion 582
23.9 Tether power propulsion 583
23.9.1 Space debris cleanup 589
23.9.2 Orbit transfer vehicle 589
23.9.3 Keeping the Space Station afloat 589
23.9.4 Exploring the outer planets 590
References 590
Chapter 24 Fuel Cell Power 591

24.2 Electrochemistry of a fuel cell 592
24.3 Electrical performance 593
24.4 Types of fuel cells 596
24.5 Regenerative fuel cells 598
24.6 RFCs for space colonies 599
24.7 RFCs for satellites 600
24.8 Commercial fuel cells 602
References 603
Chapter 25 Flywheel Energy Storage 604

25.2 Photovoltaic–flywheel power system 606
25.3 Flywheel system components 607
Contents xxxix
25.3.1 Rotor rim 608

25.3.2 Hub 608
25.3.3 Magnetic bearings 608
25.3.4 Touchdown mechanical bearings 610
25.3.5 Electrical machine 610
25.3.6 Sensors 610
25.4 Power and momentum management electronics 612
25.5 Energy and stress relations 613
25.6 Flywheel application example 615
25.7 Integrated energy and momentum storage 617
25.7.1 Energy and momentum relations 619
25.7.2 Energy depletion constraints 620
25.8 Energy–momentum wheel example 622
25.8.1 Providing launch and ascent power 623
25.8.2 Replacing batteries and reaction wheels
with IEMS 624
25.8.3 Summary of point design example 625
25.9 Electrical machine options 625
25.9.1 Synchronous machines 626
25.9.2 Induction machine 627
25.9.3 Direct current machine 628
25.10 Motor–generator design issues 629
25.10.1 Constant torque versus constant power
design 630
25.11 Magnetic bearing design issues 631
25.11.1 Low-speed bearings 632
25.11.2 High-speed bearings 632
25.12 Flywheel system controller 633
25.13 NASA flywheel program 635
References 638
Chapter 26 Superconductors in Space 640

26.2 Magnetic energy storage 640
26.3 Critical J, B, and T 643
26.4 Magnet coil design 646
26.5 System configuration 646
26.6 Cryogenic temperature 648
26.7 Cryogenic behavior of components 649
26.7.1 Semiconductors 649
26.7.2 Magnetics 650
26.7.3 Capacitors 651
26.7.4 Resistors 651
26.7.5 Structural materials 652
References 652
xl Contents
Chapter 27 Microwave Beam Power Satellite 653

27.2 Microwave beam 654
27.3 Laser beam 655
27.4 Space-to-ground power 655
27.5 Solar dynamic power generation 658
27.6 Powersat developments 660
References 661
Index 663
Part A
Power System Fundamentals
Chapter 1
Satellite Overview
1.1 Introduction
A satellite consists of various systems designed to meet the mission specific
requirements. All but the simplest satellites require a common set of
systems shown by the solid lines in Figure 1.1. Complex satellites require
additional systems shown by the dotted lines. The systems are classified
into two groups, the payload and the bus. The payload consists of the
communications equipment in commercial satellites or science instruments
in research satellites. The bus consists of all remaining equipment grouped
into several functional systems that support the payload. The power system
is one of the bus systems that consist of the solar array, battery, power
electronics, distribution harness, and controls. Other essential bus systems
are the communications and data handling system to receive commands
and return information, telemetry sensors to gage the satellite state, and a
FIGURE 1.1 Satellite systems.
3
4 Spacecraft Power Systems
central computer to coordinate and control activities of all the systems.

Satellites with complex missions also require systems to determine the
spacecraft attitude and orbit orientation, and propulsion to control both.
Satellite design is generally optimized to the fullest extent such that any
change would result in a higher cost. The total mission cost, however, is a
complex function of many variables, and so is the power system cost. The
cost, C, in dollars per watt of power generated can be expressed as a
function of four major variables as follows:
C ¼ fðX1 ; X2 ; X3 ; X4 Þ ð1:1Þ
where X1 ¼ cost per kg of the power system mass launched,

X2 ¼ cost per liter of the power system volume launched,
X3 ¼ cost per watt of the power system generation capacity, and
X4 ¼ cost of altitude control related with power system components.
1.2 Satellite Systems

The typical communications satellite bus consists of the following systems.
1.2.1 Communications and Data Handling

The communications and data handling system performs three indepen-
dent functions:
It receives and demodulates information transmitted to the satellite from

the ground station via command links.
It transmits data, both recorded (remote) and real-time, from the satellite
to the ground station via data links.
It transmits bus equipment and other telemetry data from the satellite to
the ground station via telemetry links.
1.2.2 Attitude and Orbit Control System

The attitude and orbit control system determines the exact position of the
satellite with respect to the local vertical, thus providing precise pointing of
the communications antennas, imaging sensors, and any other mission
sensors. The attitude control function accepts error signals from which the
basic or the precision attitude determination function provides 3-axis
attitude control using three reaction wheels. The basic attitude deter-
mination function obtains the pitch and roll data from the Earth sensor and
yaw data from the gyroscopes with updates from the sun sensor to
provide a basic 3-axis pointing within 0.1 accuracy. The precise attitude
Satellite Overview 5
determination function uses three gyroscopes with updates from a star

sensor to provide a 3-axis pointing within 0.01 accuracy.
1.2.3 Tracking, Telemetry, and Command System

The tracking, telemetry, and command (TT&C) system accepts analog,
discrete, and digital data from various systems of the spacecraft, and
processes them into a continuous data stream for direct transmission to the
ground or for on-board storage for later transmission. These data are
analyzed and evaluated on the ground to determine the spacecraft state of
health and the operational configuration. The command and control
function is all digital. It provides ascent guidance from the booster
separation through transfer orbit, and controls the satellite attitude and
operating modes on orbit. The control is exercised in accordance with
commands and data received from the ground station, supplemented by
signals and data supplied by other systems of the satellite. This system also
provides the error correction coding, a key function of the system.
1.2.4 Electrical Power System

The electrical power system generates, stores, conditions, controls, and
distributes power within the specified voltage band to all bus and payload
equipment. The protection of the power system components in case of all
credible faults is also included. The basic components of the power system
are the solar array, solar array drive, battery, battery charge and discharge
regulators, bus voltage regulator, load switching, fuses, and the distribution
harness. The harness consists of conducting wires and connectors that
connect various components together.
In the Earth orbiting satellite, the solar array is rotated once per orbit by
the solar array drive to track the sun at or near normal angle. The rotation is
rate-servo controlled. The body information and position errors are
computed by the satellite computer to derive rate control signals. The
nominal rate of rotation is 0.06 per s. Using slip rings and carbon brushes is
one way of providing the rotary joint between the rotating array and the
satellite body. The control signals for required rotation rate come from the
TT&C system, which also selects the rotation direction.
1.2.5 Thermal Control System

The thermal control system maintains the temperature of all equipment
within the specified limits during normal and abnormal operations. It
provides both passive and active cooling as needed. Typical components of
this system are fixed radiators, louvers, multi-layer blankets, coatings,
tapes, heaters, thermostats, temperature sensors, and control electronics.
Thermistors are widely used as temperature sensors. The components are
sized for the average electrical power dissipation, the external heat input
from the sun, the Earth’s reflected sunlight (albedo), and the long
wavelength (infrared) heat radiated from the Earth.
1.2.6 Structure and Mechanisms System

The structure and mechanisms system primarily provides a frame for
mounting and linking various mechanical components together.
Mechanisms for deploying the booms, the solar array and other compo-
nents after orbit injection are often included in this system. The deployment
power circuits and devices are well shielded for electromagnetic inter-
ference (EMI) to prevent unspecified deployment. The deployment motion
is derived by spring-loaded rotary mechanisms. A rotary vane damper
filled with viscous silicone fluid governs the deployment rate. The solar
array deployment generally involves cable cutting and/or rod cutting.
Structures are often made of magnesium and aluminum. Composites are
also common. Some steel and occasionally beryllium are used where
needed.
1.2.7 Propulsion System

The propulsion system provides propulsion torque for 3-axis control during
ascent and for maintaining the satellite momentum below a specified
maximum level during the mission. It often uses a mixed system of high-
pressure regulated helium or nitrogen and liquid hydrazine. The gaseous
helium or nitrogen and hydrazine are carried in high-pressure cylindrical
titanium alloy tanks. The propulsion systems also provides the translational
delta-V for orbit changes and orbit trims
Figure 1.2 depicts the anatomy of the Global Positioning Satellite (GPS), a
mid-Earth orbit communications satellite fleet of the U.S. Air Force. The
payload antenna is facing the Earth. The solar array panels are oriented
north–south on two booms running through the solar array drive. Mounted
inside the north and south panels of the spacecraft body are the batteries,
power regulators, and the control electronic boxes (not shown).
1.3 Earth Orbit Classification

The Earth is a sphere with a slight flatness at the top. Its diameter is
12,713.54 km at the poles and 12,756.32 km at the equator, the difference
being 42.78 km. Air surrounds the Earth and extends up to 160 km above
the surface, beyond which the atmosphere gradually fades into space. The
satellites orbiting the Earth are classified into several groups with their
typical parameters listed in Table 1.1. The orbits are often described by the
following abbreviations:
FIGURE 1.2 Configuration and evolution of GPS, mid-Earth orbit navigation satellites of the
U.S. Air Force.
GEO: geosynchronous Earth orbit, circular at 35,786-km altitude

MEO: mid Earth orbit, circular at 2000 to 20,000-km altitude
LEO: low Earth orbit, generally circular at 200 to 2000-km altitude
HEO: highly elliptical orbit, such as Molniya
A communications satellite provides interconnectivity over a large area

for point-to-point, point-to-multi-point, and broadcast communications. It
can serve fixed as well as mobile terminals anywhere — on land, on the sea,
in air or in space. A typical satellite transponder receives an uplink signal
from a ground station, frequency converts, amplifies and retransmits it to
the ground.
Table 1.1 Classification of the Earth’s orbits
Apogee Perigee Inclinationd

Orbit type (km)a (km)b Eccentricityc (degrees) Periode
Geostationary 35,786 35,786 0 0 1 sidereal

day
Geosynchronous 35,786 35,786 Near 0 0-90 1 sidereal
day
Elliptical 39,400 1000 High 62.9 1=2 sidereal
Molniya day
Low Earth Various Various 0 to high 0 to 90 >90 min
a
Closest distance from the Earth surface.
b
Farthest distance from the Earth surface.
c
Ratio of difference to sum of apogee and perigee radii.
d
Angle between orbit plane and equatorial plane.
e
One sidereal day is 23 h, 56 min, 4.09 s.
In the case of LEO and MEO, the orbit parameters are chosen to avoid the
radiation belts that surround the Earth at altitudes of 1.3 to 1.7 and 3.1 to 4.1
Earth radii. A typical LEO satellite has an altitude of 500 to 1500 km, an orbit
period of 1.5 to 2 h, and is visible to a given Earth station for only a few
minutes in every orbit period. A typical MEO satellite is between 5000 and
12000 km altitude with orbit period of several hours. In a highly elliptical
inclined orbit, it can see the polar regions for a large fraction of its orbit
period.
A GEO satellite moving west to east at an altitude of 35,786 km (22,237
miles) results in a nominal orbit period of 24 h, and remains stationary with
respect to the Earth. Three such satellites spaced 120 apart in the equatorial
plane can provide continuous coverage of the globe except near the poles.
The launch vehicle booster and its upper stages deliver the satellite in the
transfer orbit, which is an elliptical orbit with the Earth at one of its foci and
the apogee at the geosynchronous orbit. An apogee kick motor is then fired
to circularize the orbit at the geosynchronous height. The primary features
of various orbits are described below.
1.3.1 Geostationary Orbit

The geostationary orbit is a very special geosynchronous orbit, and, in fact,
is unique. It is exactly circular with a radius of 42,164 km in the Earth’s
equatorial plane with zero degree inclination and zero eccentricity. A
satellite placed in this orbit is synchronized with the Earth’s rotation rate
and direction (eastward). It does not move with respect to the Earth, and
sees the same object on the Earth steadily. The orbit period is the same as
that of the Earth’s rotation, i.e., 23 h 56 m 4.09 s. As a result, the satellite’s
beam-to-Earth and the ground station’s beam-to-satellite are steady in
position. This simplifies the design and operating requirements of both the
satellite and the ground station. However, it takes more fuel to reach and
maintain the geostationary orbit than any other orbit around the Earth at
that altitude. Numerous satellites already placed there make it difficult to
get a desirable location in this orbit that would avoid radio frequency
interference from neighboring satellites. The Tracking and Data Relay
Satellite (TDRS) of the U.S. Department of Defense is an example of a
geostationary satellite. A satellite placed in this orbit tends to drift away
from its assigned station. Hence, a periodic station-keeping operation is
required.
The time in space is kept in sidereal time, which measures the rotation of
the Earth in relation to a fixed star. Solar time is used on Earth to measure
the Earth’s rotation in relation to the sun. The same star is not in the same
place at the same solar time, but is at the same place at the same sidereal
time from day to day. A sidereal day consisting of 24 sidereal hours is the
time the Earth takes to rotate once on its axis past an imaginary line from
the Earth’s center to any star. Thus, the sidereal time is measured from a
point in the sky called the vernal equinox, although no bright start marks
this point.
The geostationary orbit period is exactly 1 sidereal day. It is slightly
shorter than the mean solar day of 24 h because of the sun’s apparent
motion resulting from the Earth’s rotation around the sun, which is 360 in
365.24 days, i.e., 0.9856 per day. By the time the Earth has rotated once in
relation to a distant star, it has moved westward along its orbit, as depicted
in Figure 1.3. The sun is then 0.9856 east of its position at the start of the
Earth’s rotation. The Earth needs additional time to rotate eastward to come
back in line with the sun. The Earth must thus rotate a total of 360.9856 in 1
mean solar day so that the meridian will align itself with the sun from one
noon to the next in exactly 24 h (86,400 s). The time for the Earth to rotate
0.9856 past one rotation is 86,400/(0.9856/360.9856) ¼ 235.91 s. The side-
real period of rotation is therefore 86,400 235:91 ¼ 86,164.09 s, or 23 h,
56 m, 4.09 s, which is shorter by 3 m and 55.91 s than the mean solar day.
1.3.2 Geosynchronous Orbit

Most commercial communications satellites operate in numerous geosyn-
chronous orbits. The geostationary orbit described above is one unique
orbit in the entire class of the geosynchronous orbit. The distinction
between the two is minor and fine, but is important. The geosynchronous
orbit is similar to the geostationary orbit, except that its inclination can be
any value between 0 and 90 . Inclinations other than 0 requires ground
station tracking antennas. Sometimes, that may not be a disadvantage
because the ground stations require tracking antennas for other reasons.
Mobil platforms, such as planes and ships, also require tracking antennas.
The geosynchronous orbit is chosen for fuel-efficient launch and orbit
FIGURE 1.3 Sidereal day and mean solar day for Earth.
maintenance. If a satellite is placed in i inclination orbit, the point directly

below the satellite oscillates between i north and i south every day, and
appears to drift to the north and south in a figure eight as shown in
Figure 1.4. The angular height of this figure is just the magnitude of i /
180 radians. This motion away from the equator induces a longitudinal
difference between the ideal and actual satellite points. The difference
FIGURE 1.4 Satellite motion in geosynchronous orbit with i inclination.

appears as the satellite moves towards the equator. The maximum

longitudinal deviation is (i /180)2/4. The variation in distance is
(i /180)Ro, where Ro is the orbit radius. The ground stations must be
aligned with the north–south (N–S) motion of the satellite in an inclined
orbit.
When the sun and the moon are not in the equatorial plane, the N–S
components of their combined gravity force change the orbit inclination of
the geostationary and geosynchronous satellites at a rate about 0.85 per
year. Station-keeping maneuvers are required to compensate for such orbit
drift. This consumes some fuel, which must be carried on board to last for
the mission duration. The use of arc jets minimizes the fuel requirement by
increasing the propulsion efficiency. The arc jets and station-keeping fuel
requirement can be eliminated if 0.85 drift per year is acceptable, or can be
accommodated in the mission design, as in TDRS. In exchange, the satellite
would require several degrees of yaw maneuvering on daily basis to remain
pointed at the ground station.
1.3.3 Highly Elliptical Orbit

Among highly elliptical orbits, Molniya is one specific orbit named after a
Russian communications satellite with a 1000-km perigee and a 39,400-km
apogee. Having the period of 1=2 sidereal day in this orbit, the satellite comes
to the same longitude on every other apogee. The advantage of the Molniya
orbit is good coverage of the entire northern hemisphere. The disadvantage
is no coverage over the southern hemisphere. Moreover, it requires more
satellites and needs two tracking antennas at each ground station. GPS,
although not in a Molniya orbit, has an orbit period of 1=2 sidereal day
because of its selected MEO location in a circular orbit. Some US military
satellites have used the elliptical Molniya orbit with 63.4 inclination to
cover Russia for 10 h out of a 12-h period.
1.3.4 Low Earth Orbit

This is approximately a circular orbit at low altitude. The International
Space Station (ISS) and NASA’s space shuttle orbiter operate in low Earth
orbit. Most communications satellites operate in GEO, but some newer
constellations are planned and/or placed in LEO between 500 and 2000-km
altitudes and 30 to 90 (polar) inclinations. Being closer to the Earth, smaller
and simpler satellites can be used in this orbit. Also, two-way communica-
tions introduces a time delay of only 0.02 s versus 0.5 s in geosynchronous
orbits. On the negative side, LEO communications satellites require tracking
of omni-directional antennas, and many birds are needed for wide
coverage.
1.3.5 Sunsynchronous Orbit

A satellite in this orbit maintains a constant angle between the sun’s
direction and the orbit plane, and always sees the sun at the same angle. It is
used for special applications.
1.4 Orbit Mechanics

Kepler’s three laws of planetary motion, based on Newton’s laws, apply to a
satellite orbiting a planet. They are as follows:
First law: The satellite orbit is an ellipse with the planet at one focal point.
Second law: The line joining the planet and the satellite sweeps equal
areas in equal times. If the time intervals it1 and it2 in Figure 1.5
are equal, then the swept areas A1 and A2 are also equal.
Third law: The square of the orbit period is proportional to the cube of
the semi-major axis,
42 a3
i:e:; To2 ¼ ð1:2Þ

where a is the semimajor axis of the orbit, and is the gravity constant of the
planet. For the Earth, is 3.986 1014 m3/s2 or 3.986 105 km3/s2.
FIGURE 1.5 Kepler’s second law of planetary motion.

For circular orbits, the third law gives the orbit period To in terms of the
orbit radius Ro,
2R1:5 o
To ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi s; or To ¼ 2:7644 106 R1:5
o h ð1:3Þ
398600
For the orbit period to be 1 sidereal day of 86,164.09 s, the orbit radius must
be 42,164 km. Deducting the mean radius of the Earth surface 6378 km, we
get the geosynchronous satellite altitude of 35,786 km above the Earth
surface. This altitude is about six times the Earth’s radius.
The satellite velocity in circular orbit is given by
2Ro
v¼ ð1:4Þ
To
which is 3.075 km/s in GEO orbit. In comparison, the Earth travels in its
orbit around the sun at speed of 30 km/s, about ten times faster.
1.5 Satellite Stabilization Methods

The satellite stabilization in orbit is achieved by either an active or passive
method described below.
1.5.1 Gravity Gradient Method

Gravity gradient is a passive method, sometimes used in small LEO
satellites. The difference in the attractive force of gravity on the parts closest
and farthest from the Earth creates a moment that maintains the satellite
aligned with the local vertical. This method requires long booms in order to
have adequate moment, and it does not work in geosynchronous orbits
because of near-zero gravity there.
1.5.2 Magnetic Damping

Magnetic damping is another passive method. It uses long booms with
magnets that interact with the Earth’s magnetic dipole field to produce the
stabilization moment.
1.5.3 Spin Stabilization

Spin stabilization is an active method that has been commonly used in most
satellites until recently. It is still used for small satellites. The spinning
bicycle wheel and the spinning top are stable above certain minimum spin
rate. The satellite stability can be similarly maintained by storing angular
momentum in a spinning body on board the satellite. For spin stabilization,

the moment of inertia about the desired spin axis must be greater than that
about any orthogonal axis. Small satellites are spun in entirety. Large
satellites using complex antennas are split in two sections, the despun
antenna section and a spinning cylindrical body. The solar cells are
mounted on the spinning body as shown in Figure 1.6. A typical spin
rate is 30 to 60 revolutions per minute. Heavier satellites require higher spin
rates for stability. Spin-stabilized satellites are also called dual-spin or
gyrostat satellites.
1.5.4 Three-axis Stabilization

Three-axis or body stabilization is another active method commonly used in
modern satellites. In this method, the satellite uses several spinning
momentum wheels located inside the body as shown in Figure 1.7. The
orientation is automatically maintained by servo-control that adds or
subtracts momentum from the spinning momentum wheels. Thrusters are
used periodically to maintain the orientation as necessary. The satellite
usually has a box-shaped body with flat solar panels (wings) extending
from the north and south faces. Table 1.2 compares key features of the spin-
stabilized and 3-axis stabilized satellites. The 3-axis stabilization generally
results in lower dry mass for satellites with solar array power exceeding a
few hundred watts. For this reason, it is widely used in modern large high-
power satellites.
All spacecraft, once disturbed from their stable position, may oscillate for
a long time in characteristic modes which may be close to being unstable. It
is important that these modes are identified and suitable damping is
FIGURE 1.6 Spin-stabilized gyrostat satellite.

FIGURE 1.7 Three-axis body-stabilized satellite with definition of attitude axes.
introduced by the attitude and/or orbit control system to restore the vehicle
to the stable position. Fuel movement in the tanks may also add in to the
oscillations, but it is normally controlled by baffles. There are five points
within the reference frame in space at which a stationary body will be in
equilibrium. All these points are in the plane in which the dominant masses
rotate. They are referred to as the Lagrangian or Libration points, and are of
potential use for the spacecraft in the Earth–Moon type systems.
1.6 Launch and Transfer Orbits

The communications satellite is placed in a geosynchronous orbit in two
main steps. The launch vehicle places the satellite first into a low Earth
Table 1.2 Key features of spin-stabilized and 3-axis stabilized satellites
Spin-stabilized 3-axis stabilized
Inherently stiff due to rotational inertia Bias or zero momentum maintains the
stability
Simple mechanical structure Complex attitude control
Only 1/3rd of the solar array generate power at Full solar array generates power all the
any time time
Power limited by body size that fits the launch Can have high power by adding solar
vehicle panels
Less flexibility in design Great flexibility in design
Suitable for small satellites Suitable for large satellites
circular orbit, called the parking orbit. Then, the so-called Hohmann
transfer takes the satellite to the final orbit using minimum fuel. The first
velocity increment changes the low circular orbit into a highly elliptical
transfer orbit with perigee that of the final circular orbit. The second
velocity increment at the apogee of transfer orbit places the satellite in the
final circular orbit. When the perigee and the apogee kick motors are fired,
some sort of stabilization is needed because the thrust would tumble the
satellite and cause incorrect orbit injection.
The fully deployed satellite, which is 3-axis stabilized in the operational
orbit, can use spin stabilization in the transfer orbit when the solar panels
are stowed into a box shaped body. The satellite is despun by applying
reaction wheel torque to bring to a non-spinning state at the end of transfer
orbit. The de-spinning operation takes about 10 min. Until the solar array is
fully deployed, the sunlit panel radiates heat from the front face only, as
opposed to both the front and back faces after the deployment. Moreover,
the exposed panel is oriented normal to the sun for maximizing the power
generation except during maneuvering. To keep the temperature of the sun
side panel from rising above the tolerance limit, the satellite is spun at a low
rate, such as 1/10th to 1 revolution per minute. Spinning at such a slow
barbecue rate is merely for thermal reasons even when the spinning is not
required for stability. The spin rate is gyro controlled. One can deploy the
array in the transfer orbit, but it adds a mechanism and structural
complexity, resulting in added mass, low reliability, and difficult transfer
orbit maneuvers.
1.7 Operational Orbit

As the satellite revolves around the Earth in operational orbit at inclination
measured from the equatorial plane, it changes its orientation with respect
to the sun with seasons as depicted in Figure 1.8. The north of the Earth
rotation axis is inclined 23.45 towards the sun on the summer solstice day,
and 23.45 away on the winter solstice day. On the autumnal and vernal
equinox days, the axis inclination is zero, resulting in equal day and night.
1.8 Eclipse due to Earth

Since the ecliptic and equatorial planes are inclined to each other by 23.45 ,
the angle of incidence of the sunlight on the solar arrays varies from 66.55 to
90 . The corresponding incident solar flux varies from 91.75% on a solstice
day to 100% on an equinox day. However, the satellite on equinox days
encounters longest eclipse once per day when the Earth blocks the sunlight
from illuminating the satellite.
When the satellite is in shadow of the Earth, the solar array power
generation ceases and its temperature drops sharply. Predicting the eclipse
FIGURE 1.8 Satellite in Earth’s orbit with seasonal variations.
duration is, therefore, important for the spacecraft power system design.
For the geosynchronous satellite, the longest eclipse occurs on the vernal
and autumnal equinoxes when the sun is in the equatorial plane as shown
in Figure 1.9. The duration for which the entire sun is blocked is called the
umbra (total eclipse marked by dotted arc). The total arc when the sun is
fully or partially blocked is called the penumbra (arc a–b). It is proportional
to the mean solar day accounting for the Earth’s orbital motion during the
eclipse. The umbra duration varies with the seasons, the longest being
69.4 min occurring around March 21 and September 21. From the
geometrical considerations of the geosynchronous orbit in Figure 1.9, the
penumbra duration is 73.7 min (1.228 h) and the umbra is 4.3 min shorter
than the penumbra. Since the solar array output voltage and current during
this 4.3 min would not meet the requirement for the power system
operation, penumbra is taken as the eclipse duration for the power system
design.
As the sun moves above or below the equator after an equinox, the
eclipse duration becomes shorter and shorter, and finally becomes zero
when the inclination of the sun becomes high enough (Figure 1.10). The
number of days the geosynchronous satellite sees an eclipse, and the eclipse
FIGURE 1.9 Eclipse in geosynchronous orbit — umbra and penumbra.
duration on that day, are shown in Figure 1.11. The eclipse onset time on a
particular day is of interest to the satellite design engineer, because it
determines the required services and the battery requirement onboard the
satellite.
In near-equatorial, circular, low Earth orbits, eclipses of approximately
equal duration occur once every orbit period. The eclipse duration is
FIGURE 1.10 GEO eclipse, once per orbit in spring and autumn seasons only.
FIGURE 1.11 GEO eclipse duration longest on vernal and autumnal equinox days.
dependent on the orbit altitude, inclination, and the sunlight incidence

angle on the orbit plane (Figure 1.12). It can vary by a factor of two in LEO.
For a circular orbit, the eclipse duration (in h) is given by
8 !1=2 9
>
> REarth 2 >
>
>
> 1 >
>
>
< Rorbit >
=
1 1
Te ¼ þ sin1 ð1:5Þ
2 >
> cos >
>
>
> >
>
>
: >
;
where ¼ angle of the sunlight incidence on the orbit plane, i.e., the angle
between the Sun–Earth line and the local normal of the orbit plane.
The angle varies seasonally between (i þ ), where i ¼ orbit inclina-
tion with respect to the equator and ¼ angle between the sun line and the
ecliptic plane (23.45 ). As increases, the eclipse duration decrease, which
improves the load capability of the electrical power system. At certain value
of high , no eclipse occurs. There are polar and near-polar low Earth orbits
FIGURE 1.12 Eclipse in near-equatorial LEO, once per orbit in all seasons.
FIGURE 1.13 Eclipse duration and orbit period vs circular orbit altitude.
that never have an eclipse of the sun. On the other hand, the longest eclipse
occurs at ¼ 0.
1.8.1 Example
For a satellite in a 6343-mile radius and 20 inclination orbit, the above
equation gives Te ¼ 0.63 h or 38 min long eclipse (note that the argument of
the sin1 must be in radians).
For circular orbits, the eclipse duration and the number of eclipses per
year are plotted in Figures 1.13 and 1.14, respectively. The ratio of the
maximum eclipse to minimum sunlight duration is an indicator of a
FIGURE 1.14 Maximum number of eclipse per year vs orbit altitude.

FIGURE 1.15 Minimum sun time and maximum eclipse duration in circular Earth orbits.
challenge posed to the power system design engineer. The greater the ratio,
the heavier the battery requirement to power the load during an eclipse. It
also requires a larger solar array to capture the energy required during
shorter periods of sunlight, and to divert a higher fraction of it to charge the
battery while simultaneously supplying full load to the payload. Such a
demand is greater on a low Earth orbit satellite, as seen in Figures 1.15 and
1.16.
1.9 Eclipse due to Moon

In addition to eclipses due to the Earth’s shadow on the satellite, the moon
can also obstruct the sun as seen by the satellite. Such eclipses are irregular
in occurrence, varying from 0 to 4 per year, with an average of two. Usually
they are spaced far apart, but in the worst case two eclipses can occur in
24 h. The duration can vary from several minutes to more than 2 h, with an
FIGURE 1.16 Ratio of maximum eclipse to minimum sun time in circular Earth orbits.
average of 40 min. The satellite may experience additional battery depth of

discharge and fall in temperature in the case where an eclipse due to the
moon occurs adjacent to an eclipse due to the Earth. In most missions,
however, eclipses due to the moon impose no additional design require-
ment, but that has to be ascertained. Otherwise, shedding a noncritical load
temporarily during the worst moon eclipse often circumvents the situation.
1.10 Solar Flux

The energy received from the sun in space varies with the distance squared.
The Earth’s orbit around the sun is approximately circular with a slight
eccentricity of 0.01672. The distance, therefore, varies within 0.01672 times
the average distance between the sun and the Earth, which is 149.6
million km, defined as one astronomical unit (AU) of distance. Thus, the
solar flux varies over (1 0.01672)2 or 1 0.034 of the yearly average.
Within these small variations, the Earth is closest to the sun around January
2 (perihelion), and farthest around July 2 (aphelion). For many years, the
average solar radiation in the Earth orbit has been taken as 1358 5 W/m2
on the surface normal to the sun. Measurements reported by Frohlich1 show
a higher average value of 1377 5 W/m2, which is now generally accepted.
However, the conservative number of 1358 5 ¼ 1353 W/m2 continues in
wide use.
Table 1.3 shows the ratio of the seasonal flux over the yearly average flux
on the equinox and solstice days. It also shows the seasonal variations in the
orbit inclination. Satellites using solar arrays with single-axis tracking
gimbals generate less power in accordance with the cosine of the sun angle.
The last column gives the combined effect of the cosine and the flux
variation due to eccentricity on the solar array output power. The output on
the summer solstice day is 11% lower than that on the vernal equinox day.
Table 1.3 Variation of solar flux and angle of incidence with the seasons
Angle of incidence Flux on array/yearly

Seasonal flux/ with single-axis average on sun-
Day of the year yearly average tracking gimbals tracking surface
Vernal equinox, 1.001 0 1.001

March 21
Summer solstice, 0.967 23.45 0.887
June 22
Autumn equinox, 0.995 0 0.995
September 21
Winter solstice, 1.034 23.45 0.949
December 22
FIGURE 1.17 Seasonal variations in solar parameters over 1 year in geosynchronous orbit.
The seasonal variations in solar parameters in geosynchronous orbit are

collectively depicted in Figure 1.17.
1.11 Beta Angle

The beta angle, , is defined as the angle between the Earth–Sun line and
the orbit plane when the spacecraft is closets to the sun (orbit noon). It
varies seasonally between 0 and (i þ io) , where i ¼ orbit inclination, and
io ¼ angle between equator and ecliptic plane, which is 23.45 .
The value of ¼ 90 results in the greatest solar flux on the satellite body,
and ¼ 0 gives zero flux. In most satellites with the array always pointed
to the sun by the solar array drive motor, the angle has an insignificant
effect on the generation of electric power. However, the thermal control
system is impacted by the angle. Low may require additional heaters,
while high may require additional cooling. The angle impacts the solar
array temperature, which in turn has a small secondary effect on the power
generation. The most significant effect of the angle on the power system
design comes from the eclipse duration as given by Equation 1.5. As
increases, the eclipse duration decreases, which consequently requires a

smaller battery and less charging power during sunlight.
The angle is not to be confused with the sun angle, , often used to
define the sunlight incidence angle on the solar array, which could be
canted to catch the sun normally in case is not 90 . The sun angle is
defined as the angle between the solar array plane and the sun-pointing
vector. The power generation is proportional to cos , so of 90 results in
the maximum power generation, and 0 gives zero power.
The power system engineer starts with the orbit parameters specified by
the customer, which primarily set the orbit period, eclipse duration and the
angle. These parameters in turn have the greatest impact on the power
system design.
1.12 Spacecraft Mass

Hundreds of small spin-stabilized and large 3-axis stabilized satellites have
been built and deployed in various orbits. The spacecraft dry mass and the
corresponding electrical power requirements for GEO satellites have been
steadily rising as seen in Figure 1.18. The new technological developments
in solar cells and evolutionary design improvements have jointly con-
tributed to the higher solar array output power per kilogram for modern
communications satellites. For large 3-axis body stabilized GEO satellites
with total eclipse load of 5 to 10 kW, the electrical power system (EPS) mass
is about 30% of the total spacecraft dry mass, the payload mass is also about
30%, and the remaining 40% is in the structure and all other spacecraft
systems.
The average mass of 16 commercial satellites contracted in 2002 was
about 3600 kg each, which is considered the mass of medium size satellites.
On the large side, Thuraya for the United Arab Emirates will have lift off
mass of 5200 kg.
FIGURE 1.18 Power vs dry mass of GEO communications satellites.

Reference
1. Frohlich, R.C., Contemporary Measures of the Solar Constant: The Solar Output
and its Variations, Colorado Associated University Press, Boulder, CO,
1977, pp. 93–109.
Chapter 2
Near-Earth Space Environment
2.1 Introduction
The near-Earth space environment, i.e., thermosphere and ionosphere,
presents a substantial risk to space systems. The spacecraft power system
must withstand the space environment and meet full performance
specifications over the entire mission life, starting from the launch phase
to disposal at the end of life. General design criteria required for this
purpose are studied and documented by space research organizations such
as NASA.1 As the solar array is directly exposed to the space environment,
it is particularly vulnerable to damage. In fact, the rate of damage on a solar
array determines the spacecraft life in many cases. The environmental
factors also affect the overall design of all other components of the power
system.
2.2 Launch and Transfer Orbit Environment

High levels of acceleration, shock, and vibration are present during lift-off
and during pyro-driven deployments at various stages. The resulting stress
levels, varying with the launch vehicle, have effects on the power system
design, particularly on the solar panels. For example, the solar panels are
required to withstand the launch acceleration of about 3g on the space
shuttle Orbiter, and about 10g on the launch vehicle Saturn, where g is the
acceleration due to gravity on the Earth’s surface. The pyro shock levels
during deployments can be extremely high for a few milliseconds. The
shock spectrum is generally characterized by high-frequency components.
The peak shock level in Ariane during payload separation, for example,
could be about 2000g at frequencies above 1.5 kHz.
In transfer orbit, the solar panel is still folded, but must withstand the
accelerating force at perigee and the braking force at apogee. Thermally, the
outer solar panels must withstand the Earth’s heat radiation and albedo, in
addition to the solar radiation, without exceeding the specified temperature
limit.
26
Near-Earth Space Environment 27
2.3 On-orbit Environment

The power system design is influenced primarily by the following on-orbit
environmental factors.
2.3.1 Lack of Gravity and Atmosphere

The lack of gravity and atmosphere in space has significant effects on the
power system design. Zero gravity and vacuum jointly deprive all
spacecraft equipment from the natural convection cooling which is normal
on Earth. The internal thermal design in space, therefore, depends primarily
on conduction cooling and, to some extent, on radiation cooling. The heat
rejection to the outer space, however, must depend only on radiation
cooling.
The vacuum causes sublimation and outgassing, more in some materials
than others. The subsequent condensation of the sublimed vapor on cold
surfaces can cause short circuits in electrical parts. Since zinc has a high
sublimation rate, and some polymers have high outgassing rates, their use
in space is restricted. The solar array current collecting slip rings and
brushes can experience cold welding under high contact pressure in a
vacuum. The use of lubricants having low sublimation rates, or keeping
them in sealed pressurized enclosures, is therefore important.
2.3.2 Magnetic Field

The interaction of the Earth’s magnetic field, B, and the magnetic dipole
moment, M, created by an electrical current loop produces torque, T, on the
spacecraft. The cross-vector product of M and B gives this torque (in Nm),
i.e.,
T ¼MB ð2:1Þ
where M is expressed in Am2 and B in tesla. The magnetic moment is

produced by current loops in the spacecraft power circuits. The moment (in
A m2) is defined as the vector product of the current and the loop area, i.e.,
M ¼ I A. Its direction is perpendicular to the loop area as shown in Figure
2.1.
In the geostationary orbit, the normal component of B is constant around
0.104 mT, and the radial component varies between 0.042 mT as the satellite
revolves around the Earth. In orbits at other altitudes, the magnetic field
varies inversely with the orbit radius cubed. The normal component of B
produces a torque in the equatorial plane, averaging to zero over the orbit
period. Both components of the torque affect the satellite’s attitude. The
power system contribution to the magnetic torque comes from the solar
panel, battery, and wires connecting various system components.
FIGURE 2.1 Definition of magnetic moment.
The magnetic moment of the spacecraft is minimized by design and

compensation in the following ways:
Lay out the current carrying conductors such that they form the
smallest possible loops.
Make two adjacent loops in opposite directions, so that they compen-
sate each other.
Twist wires wherever possible to neutralize the moments of adjacent
twists.
After using one or more of the above design means, the net residual
magnetic moment is determined by tests after the satellite is assembled. A
moment that would limit the resulting torque below 100 mNm in the
operational orbit is generally considered acceptable.
2.3.3 Meteoroids and Debris

Impact of solid objects can cause damage to the solar array. Small particles
not large enough to cause immediate damage can cause gradual degrada-
tion in output power over time. The particle mass and the hit rate (flux)
depends on the orbit. Several NASA studies conducted for predicting the
meteoroid flux in the Earth’s orbit have given consistent results. They show
that the average number varies inversely with the meteoroid mass as given
by the following simple expression:
1
meteoroid flux ¼ ð2:2Þ
m
where the flux is measured in number of particles of mass greater than m

grams per square kilometer per year. The most common meteoroids are
small, having masses between 0.1 and 100 mg, generally called micro-
meteoroids. The probability of having n impacts with particles of mass
between m1 and m2 on a given surface area during y years is estimated
using Poisson’s probability distribution function. In the absence of accurate
information, their mass density is assumed to be 0.5 g/cm3 and a mean
FIGURE 2.2 TDRS meteoroid environment fluence over 15 years.
impact speed of 20 km/s. The actual fluence varies with the orbit. For
example, in the geostationary orbit of TDRS, the estimated meteoroid
fluence for a 15-year mission is shown in Figure 2.2. The yearly fluence rate
in other orbits is shown in Figure 2.3.
In addition to the natural micrometeoroids, the man-made debris can
vary from 1 mm to 10 mm in diameter. They have an average density equal
to that of aluminum, the most widely used material for satellite and launch
vehicle components. Their relative velocity with respect to the spacecraft
varies from zero to twice the orbital velocity, with an average around
10 km/s. The impact energy of large meteoroids can instantly damage the
solar array protective glass and cells. Micrometeoroids, on the other hand,
erode the glass surface gradually with subsequent degradation in the power
output over time.
2.3.4 Atomic Oxygen

Atomic oxygen is present in low Earth orbit. It severely erodes some
materials, such as silver, which is extensively used in the solar array
FIGURE 2.3 Debris flux rate estimate in various orbits.

construction. Similar erosion is also seen in some electrical insulation such

as KaptonTM and silicon rubber. The erosion comes not only from the
chemical activity, but also in large part by the atoms travelling at
several km/s velocity relative to the spacecraft. In addition to the erosion
(surface recession), atomic oxygen can also form stable oxides on the metal
surface. For these reasons, exposed bare silver in solar arrays and silicon or
Kapton insulation on electrical wires are undesirable. This topic is dealt
with in Chapter 5.
2.3.5 Charged Particles

The sun continuously radiates energy in several forms — visible light,
invisible infrared, ultraviolet, X-rays, gamma rays, radio waves, electrons,
protons, and plasma (electrically charged hot gas). The abundance of
charged particles that come with the sun’s radiation makes space a hostile
environment. The energy in these particles impinging on a surface causes
damage, which accumulates over a period of time. The following terms are
widely used in discussing the radiation of charged particles in the space
environment.
One million electron volts (1 MeV) is the unit of the equivalent energy
level of various charged particles. It is defined as the energy released
by the total charge of one million electrons falling through one volt
in the electric potential field. With the electron charge of
0:1592 1018 C, 1 MeV equals 0:1592 1012 J.
Flux density expresses the number of charged particles present per unit
volume.
Flux is the term that expresses the rate of flux hitting a surface. It is given
as the number of particles impinging upon a unit surface area per
unit time, i.e., in number/m2 s or in MeV/m2 s to account for their
total energy level. The flux varies as the spacecraft moves around the
orbit. It is greater on the sun-side of the Earth than on the farther side.
Often the orbit average flux is quoted.
Integrated flux or fluence expresses the accumulated number of particles
impinging upon a unit surface area over some time duration in orbit.
Its unit is number/m2 year or often MeV/m2 over the mission life to
account for the total energy level the surface must withstand.
Dose expresses the amount of energy absorbed per unit mass of the
subject matter. It is measured in rad or rad (Si). Since silicon makes
up the bulk of micro-electronic devices, it is taken as the reference
material for comparing the radiation energy. One rad is equal to 100
ergs of absorbed energy in 1 g of target material. One rad (Si) is the
unit of energy that will generate 1.7 1013 electron–hole pairs in 1 g

of silicon.
The common sources of charged particles in the space environment are as

follows:
Solar radiation, consisting of mostly electrons and protons coming from

the sun.
Solar wind brings bursts of charged particles in addition to the normal
radiation from the sun. It is primarily made of protons and some
electrons radiating from the sun. Their flux and energy level around
the Earth’s orbit are low during normal solar wind. The mean proton
flux is about 200 million protons/cm2 with a mean energy level of
several keV. However, the energy levels can reach 100 MeV during
solar flares, and even 1 GeV during intense flares.
Cosmic radiation comes mostly from the outer space and some probably
from the sun. It consists of about 85% proton, 12% alpha particles
and 3% electrons. The proton energy level is high in GeV, but the flux
rate is low: in the order of a few particles/cm2s.
2.4 Van Allen Belts

The Earth’s magnetic field covers the magnetosphere, a doughnut shaped
region of space influenced by the Earth’s magnetic field. It acts on the
electrons and protons coming from the space. The Van Allen radiation belts
are parts of the magnetosphere that contain a large number of particles. The
magnetosphere normally shields the Earth from these particles. However,
when a disturbance on the sun radiates an abundance of particles, some
reach the Earth’s atmosphere near the magnetic poles and cause a glow
known as an aurora, popularly known as the northern lights (or the
southern lights, in the southern hemisphere).
The Earth’s magnetic field varying with the radius cubed has a radial
gradient, and it converges near the magnetic poles a shown in Figure 2.4.
The solar wind brings charged particles to the Earth’s magnetic field. These
particles are deflected by the Lorentz force, F, given by the following cross-
vector product:
F ¼ qV B ð2:3Þ
where q is the particle charge, V is the particle velocity, and B is the Earth’s
magnetic field intensity. The particle moving radially towards the Earth at
velocity V generates a circumferential force that will impart velocity V . The
resultant velocity is then VR, as shown in the plan view. The radial gradient
FIGURE 2.4 Lorentz force on a charged particle moving in Earth’s magnetic field.
in the field imparts a spiral motion to the particle. The spiraling particle
comes to rest once all the energy is absorbed, and then springs back and
forth as explained below.
A particle moving in a converging magnetic field generates a force that
pushes the particle into a weak magnetic field (Figure 2.5), causing it to drift
at velocity Vz. The Vz decreases as the particle moves towards the equatorial
plane of weak magnetic field; and V increases to keep the kinetic energy
constant. The particle eventually comes to a stop in the z-direction and is
reflected back into the weaker field. Since the field is again converging on
the other side, the particle is reflected back again. Thus, the converging field
confines the charge particle between two magnetic mirrors. This way, most
FIGURE 2.5 Charged particle moving in a convergent magnetic field.

FIGURE 2.6 Van Allen radiation belts.
particles remain trapped in a certain belt all the time. However, particles
having sufficiently high axial kinetic energy do not get reflected, and escape
the confinement belt.
Most particles become trapped in two doughnut-shaped radiation belts
known as the Van Allen belts, which are a part of the magnetosphere. The
belts are stronger on the sun-side than on the side away from the sun, as
shown in Figure 2.6. The trapping is concentrated in two belts shown in
Figure 2.7. The electrons get trapped in a belt spanning 2 to 5 Earth radii
with a heavy concentration between 3 and 4 Earth radii. The high-energy
protons penetrate closer to the Earth and get trapped in another belt
spanning around 1 to 2 Earth radii with a heavy concentration around 1.5
Earth radius. Since the charged particle flux varies over the satellite orbit,
the integrated radiation flux (fluence) is used in the power system design.
FIGURE 2.7 Extent of Van Allen radiation belts around Earth per NASA’s AP-8 Environment
Model.
The solar array, in particular, requires protection against radiation

damage. The geosynchronous satellites orbiting at 6.6 times the Earth
radius are way outside the inner proton belt and at the outer edge of the
outer electron belt. Therefore, they are exposed to only a small part of the
outer electron belt. However, they are more vulnerable to high-energy
proton generated by solar flares.
2.5 Solar Wind and Solar Flare

A continuous stream of protons, electrons, and ionized gas leaving the sun
and sweeping the solar system is called the solar wind. It is very thin,
containing about 5 charged particles per cubic centimeter of space. The
solar wind streams constantly, but not evenly. These particles travel past
the Earth at speeds of about 480 km/s. In comparison, the Earth travels in
its orbit at 30 km/s. When these particles flow past the Earth, some get
trapped in the Earth’s magnetosphere.
A solar flare is a gush of ionized gas with charged particles spewed out
by the corona, the sun’s glowing outer shell. It is associated with dark
sunspots bounded by intense magnetic fields that come and go in 11-year
cycles. The flare can cause auroras and disturb radio signals on Earth. The
ionized gas is extremely hot (>1 million C), such that atoms of hydrogen
and helium are homogenized into dilute plasma, composed mainly of
negatively charged electrons and positively charged protons. Although this
constitutes a million tons of matter moving at a million miles per hour, its
density is so low that it is still considered a vacuum. However, its effects
can be devastating as depicted in Figure 2.8.
Solar flares occur in 11-year cycles. The last major period was November
and December 2000. Several flares occur during each solar cycle in which 7
years are active and 4 inactive. The intensity levels of the flares vary over a
wide range. Therefore, their energy levels and the corresponding prob-
abilities that must be considered in designing the power system are
mutually agreed upon by the satellite customer and the manufacture. The
relatively strong 1989–1990 flares were extensively analyzed by spacecraft
engineers because of the quantity and quality of data obtained and their
impact on many spacecraft in orbit at the time. The newly developed model
suggests that the proton fluences below 10 MeV were significantly
underestimated in the past. This causes more degradation of the photo-
voltaic (PV) cells in the Isc, although not much in Voc. From this study, it is
estimated that a solar array design based on the old King model2 of 1974
would have about 20% probability of missing the power output require-
ments by a few percent compared to the new Feynman model3 of 1989–1990
flares.
FIGURE 2.8 Solar flare and effects on Earth.
Table 2.1 gives the total annual fluence in a typical geosynchronous orbit.
The integrated flux of solar flare protons per 11-year cycle based on NASA/
JPL model with 80% confidence level is given in Table 2.2.
Spacecraft going beyond the solar system, such as Pioneer and Voyager,
experience the so-called termination shock wave at the edge of the solar
system. The termination shock wave is a zone around the solar system
where the solar wind crashes into the tenuous gas and dust that fills
interstellar space.
Table 2.1 Total annual fluence in a typical geosynchronous orbit

(number/cm2)a
Particle type Normal solar flares Intense solar flares
Trapped electrons 1014 1014

Trapped protons Negligible Negligible
High energy solar flare protons 107 1010
a
The energy level of the electrons varies from a few keV to several MeV,
and that of protons in several hundreds MeV range.
Table 2.2 Integrated

flux of solar flare
protons per 11-year
solar cyclea
Energy Solar protons

(>MeV) (number/cm2)
1 5.3 1011
5 2.0 1011
10 1.0 1011
30 2.3 1010
50 9.7 109
70 6.6 109
100 2.5 109
a
Based on the NASA/
JPL model with 80%
confidence level.
2.6 Geomagnetic Storm

There is a well-recognized link between solar activity, disturbances in the
geomagnetic field, and disruptions to man-made systems such as satellites,
communications networks, defense systems, and even the sturdy power
grids on Earth. Analysis suggests that these vulnerabilities to magnetic
disturbances will continue and perhaps even increase as these systems
continue to evolve and grow.
What could happen to the spacecraft power system under the geomag-
netic storm can perhaps be appreciated by examining what has happened to
the power grids on Earth in the past. Under such storms, the magnetic field
of the Earth is disturbed, causing the induced currents in overhead
transmission lines. The high-voltage lines, due to their long length and
large loop area with the Earth’s surface, can induce a high current. This low
frequency d.c.-like geomagnetically induced current then enters the high
voltage transformers, driving them in deep magnetic saturation, causing the
voltage collapse and severe overheating. The great storm of March 13, 1989
affected large areas of North America with a power blackout for 6 million
Hydro Quebec customers in Canada for a long time. Voltage regulation was
affected, undesired relay operations occurred on system equipment, and
new areas of vulnerability were exposed due to unanticipated common
mode currents saturating the transformer cores, drawing high currents and
tripping off the transmission lines. Less severe storms in September 1989,
March 1991, and October 1991 reinforced the fact that geomagnetic
disturbances can hamper utility operation. The October 2003 storm knocked
out satellites (e.g., Japan’s Earth observation satellite ADEOS-11), hampered
radio communication and caused an electrical blackout in Sweden. To
protect airline passengers against the charged particles, the U.S. Federal
Aviation Administration warned pilots to stay below 25,000 feet when

flying near the poles in order to avoid excess radiation. The storm arrived
within a few hours of NOAA’s prediction time.
The space research models of plasma population and circulation,
monitoring the interplanetary real-time solar wind and the geomagnetically
induced currents, and tracking the auroral zone current can provide timely
information for the next solar maxima, which begins in 2010–2011.
Major effects of solar flares and geomagnetic storms on the spacecraft
power system are:
Rapid degradation of the solar array power generation. This is a life-

shortening, not a life-threatening, effect.
Common-mode EMI noise entering the electrical power system
components through exposed cables. This can upset the power system
possibly causing a permanent damage. In the 1990s, the power systems
of a few communications satellites (e.g., Anik-E1 of Canada and Intelsat-
K) were damaged, and a few were suspected to have failed under
intense solar flares. An in-depth investigation indicated that a spark
caused by geomagnetic storm shorted out connections between the
satellite solar panels and several dozens of relays.
2.7 Nuclear Threat

Some defense satellites are required to withstand a man-made threat of
nuclear detonation of a specified intensity. The energetic particles released
during a nuclear event are fission electrons, neutrons, gamma rays and
X-rays. Their energy levels depend solely on the destructive power of
the nuclear device deployed. The threat level specified for the design is
determined from the probability and the consequence considerations.
The level of nuclear threat, however, is always a classified number under
the U.S. Department of Defense (DoD) secret classification system.
2.8 Total Radiation Fluence

The energy levels of various sources of the natural radiation in the Earth’s
orbit are as follows:
Trapped electrons 0.1 to 7 MeV

Trapped protons 0.1 to 100 MeV
Solar flare protons 1 to 200 MeV
Solar flare alphas 1 to 300 MeV
Galactic cosmic rays >1 GeV
FIGURE 2.9 Trapped proton environment in 852-km, 99 -inclined circular orbit.
(Source: NASA/GSFC.)
The total fluence is calculated for a given mission using the environmental
model specified by the customer. As an example, the orbit integrated flux of
trapped proton and electron environment in a 852-km, 99 -inclined circular
orbit is shown in Figures 2.9 and 2.10, where the y-axis is the fluence with
energy levels on the x-axis.4
The total fluence in equivalent MeV/cm2 day is the area under such a
number (n) versus energy level (E) curves for the electrons, protons, solar
flares, and any other source of radiation, natural or man-made. That is,
FIGURE 2.10 Trapped electron environment in 852-km, 99 -inclined circular orbit.
(Source: NASA/GSFC.)
Z1 Z1 Z1 Z1
fluence ¼ n dE electrons þ protons þ flares þ nuclear
o o o o
Z1 ð2:4Þ
þ any other
o
The integrated fluence determined by Equation 2.4 is then used to

determine the damage on the solar array and other components during
the mission life.
Several recently completed cross-disciplinary scientific and engineering
research projects are significantly improving space environment character-
ization. Operational users of space systems are the direct beneficiaries of
these research advances.
References
1. NASA, ‘‘Space Vehicle Design Criteria (Environment),’’ Technical Report No.
SP-8005, 1980.
2. King, J.H., ‘‘Solar proton fluence for 1977–1983 space missions,’’ Journal of
Spacecraft and Rockets, 11, 401–407, 1974.
3. Feynman, J., ‘‘New interplanetary proton fluence model,’’ Journal of
Spacecraft, 27(4), 403–408, 1990.
4. Stassinopoulos, J.M., Barth, J.M., and Smith, R.L., ‘‘METSAT Charged
Particle Environment Study, Revised Edition, Method 2,’’ NASA Report No.
GSFC X-600-87-11, 1987.
Chapter 3
Power System Options
3.1 Introduction
Power requirements in very early satellites were several watts. In today’s
communications satellites, it is several kilowatts and is growing. Some
strategic defense spacecraft power requirements are estimated to be in
hundreds of kilowatts and some concepts require hundreds of megawatts
of burst power. Solar radiation is the only external source of energy
available in space. Any power system not using solar energy must carry its
own source of energy on board, such as the primary battery, fuel cell,
nuclear or chemical fuel.
The basic components of the spacecraft power system are shown in
Figure 3.1. They are the primary energy source, energy conversion, power
regulator, rechargeable energy storage, power distribution and protection,
and power utilization by the user’s equipment (loads). Candidates for the
primary energy source include solar radiation, radioisotopes, nuclear
reactors, and electrochemical and/or chemical fuel. The energy conversion
may be photovoltaic, thermoelectric, dynamic alternator, fuel cell, or
thermionic. The energy storage has been primarily electrochemical,
although flywheel technology is under development at NASA Glenn
Research Center (GRC).
From the available options that are compatible with a given mission and
its environment, the satellite level optimization study is conducted to select
the best combination of energy source, energy conversion, and energy
storage technologies. Final selection must meet multiple criteria, but the
FIGURE 3.1 Basic components of a spacecraft power system.
40
Power System Options 41
FIGURE 3.2 Optimum energy sources for various power levels and mission durations.
primary criteria are always low mass and low life-cycle cost. Such selection
is largely influenced by the product of the power level and the mission
duration as shown in Figure 3.2. The dividing lines among various options
are only approximate and have large overlaps. The following sections
briefly describe these options with their optimum application ranges. The
detailed description and performance of often-used options are covered in
separate chapters.
3.2 Primary Battery

A primary battery can economically power a small elementary spacecraft
requiring only several watts over several days. Early short mission space-
craft flew with primary batteries such as AgZn and NaS. Even today, low
power short life satellites carrying instruments with low duty ratio may be
designed using a primary battery such as LiCFx as the only power source,
thus eliminating the solar panel and battery charge electronics.
The battery cell consists of two electrode plates submersed in an
electrolyte as shown in Figure 3.3(a). The electrochemistry of the cell
generates an electrical potential difference between the electrodes, which
can drive electrical current through an external load circuit. Thus, the
battery converts the stored chemical energy between the electrode plates
into direct current electricity. The cell can deliver only a certain amount of
charge, measured in ampere-hours (Ah), before all of its energy is depleted.
The cell voltage decays with the Ah discharged as shown in Figure 3.3(b).
The primary battery has nonreversible electrochemistry. It cannot be
FIGURE 3.3 Battery cell construction and voltage characteristics.
recharged once its Ah capacity has been drained. It is then discarded, often
jettisoned from the spacecraft to shed mass.
3.3 Fuel Cell

Powering loads greater than several watts for more than a few days or a few
weeks is not practical using a battery, but is easily done with a fuel cell. The
fuel cell, developed as an intermediate-term power source for space
applications, was first used in a moon buggy and continues being used to
power NASA’s space shuttles. It also finds other niche applications at
present. A fuel cell converts chemical energy in the fuel, such as hydrogen
and oxygen, into electricity (Figure 3.4(a)). Since the fuel continuously
FIGURE 3.4 Fuel cell construction and voltage characteristics.

refurbishes the energy, the cell does not run out of energy. Hence, the fuel
cell is not rated in terms of the Ah capacity, but in terms of the power
generation rate. The cell voltage remains constant (Figure 3.4(b)) as long as
the fuel is supplied at the required rate. Therefore, the fuel cell can be an
optimum choice for supplying hundreds or thousands of watts over a few
weeks using the on-board fuel.
A fuel cell is a static electrochemical device that generates direct current
electricity by a chemical reaction without altering the electrodes or the
electrolyte materials. In concept, hydrogen and oxygen are combined to
produce electricity and water, which is the reverse of the electrolysis of
water. The crew in manned missions can use this water. The fuel does not
burn as in an internal combustion (IC) engine. Thus, the fuel cell operates
differently from both the electrochemical battery and the IC engine. The
conversion efficiency of the fuel cell is not limited to that of the Carnot cycle,
because the fuel cell bypasses the thermal-to-mechanical conversion and its
operation is isothermal. That is why its efficiency can be, in principle, much
greater than that of the IC engine. The efficiencies of some commercial fuel
cells approach 70 to 80%, about twice the combustion engine efficiency. The
space qualified fuel cell efficiency, however, is around 10% at present, but
has a potential for a significant increase.
3.4 Solar PV–Battery

One of the most valuable breakthroughs in the space industry was probably
the photovoltaic (PV) cell used to convert sunlight into electricity for Earth-
orbiting satellites. Today, it is the most widely used energy conversion
technology in the industry that has fueled the information revolution using
high-power communications satellites. Power requirements in tens of watts
to several kilowatts over a life ranging from a few months to 15 to 20 years
can be met with an array of photovoltaic cells. Satellites requiring
continuous load power even during an eclipse must use a rechargeable
battery along with the PV array. The battery is charged during sunlight and
discharged to power the load during an eclipse. A power regulator and
control circuits are used as required for the mission.
The general layout of the PV–battery power system is shown in Figure
3.5. All components other than the solar array are generally located inside
the satellite body. The orientations of the core body and the solar array are
maintained relative to the sun and the Earth. The core body is normally
maintained in a near constant orientation relative to the Earth, while the
drive and the gimbals orient the solar array to the sun. The drive rotates
360 once per orbit as the satellite revolves around the Earth. The gimbals
rotate to compensate for the variation in the solar angle and also to
prevent array shadowing if applicable. Not all satellites have gimbals, but
almost all using the solar energy for power generation have an drive. The
FIGURE 3.5 Solar photovoltaic–battery power system configuration.
most common form of drive is a slip ring assembly with a solar array
drive in 3-axis stabilized satellites, and a rotary power transfer assembly in
gyrostats. Angular errors induced by the structural distortions are often
compensated by the and/or drive settings.
The seasonal variations of the angle and the eclipse duration over 1 year
for the International Space Station in 400-km (220-n.m.) altitude and 51.6
inclination orbit are shown in Figure 3.6.1 For a given system design, the
FIGURE 3.6 Beta angle and eclipse duration variation with season for the International Space
Station. (Source: NASA Glenn SPACE Team/J. Hojnicki.)
FIGURE 3.7 Degradation in I–V characteristics of a typical PV cell under radiation.
power available to the load varies over the year due to seasonal variation in
the angle. At high when the eclipse duration is zero, the load
capability of the electrical power system would be the greatest, as no battery
charge power is required. For the ISS, there would be no eclipse at all for
> 71 , making the orbit sun-synchronous.
The PV cell has been a building block of space power systems since the
beginning. The cell is a diode-type junction of two crystalline semiconduc-
tors, which generates electricity under sunlight. Its performance at the
beginning of life (BOL) is characterized by the output voltage and current at
its terminals as shown by the heavy line in Figure 3.7. The two extreme
points on this curve, namely the open circuit voltage, Voc, and the short
circuit current, Isc, are often used as the performance indicators. The
maximum power a cell can generate is the product of Voc, Isc, and a factor
that is approximately constant for a given junction. The I–V characteristic of
the PV cell degrades as shown by thin lines with the increasing fluence of
charged particles on the solar array in the space environment. Such
degradation results in decreasing power generation with time. With the
combination of seasonal variations of angle and yearly degradation of
charged particles, the power generation of the solar array over the mission
life varies as shown in Figure 3.8.
3.5 Solar Concentrator–Dynamic Power System

Solar energy can be used in systems other than photovoltaic cells. For
example, the sun’s energy is collected in the form of heat using a
concentrator. The heat, in turn, is used to generate steam and drive a
rotating turbo-generator or a reciprocating alternator: either way uses a
thermodynamic energy converter.
FIGURE 3.8 Degradation of solar array output power versus service years.
The dynamic power system was a primary candidate for early space
station design, having an estimated power requirement of 300 kW. The
system configuration is shown in Figure 3.9. A parabolic concentrator
focuses the sun’s heat on to a receiver, which boils a fluid. The fluid can be a
suitable liquid or even a liquid metal, such as potassium chloride. High-
pressure steam produced in the receiver drives a steam turbine based on the
Rankine cycle. The fluid can also be a gas, such as a mixture of helium and
FIGURE 3.9 Solar concentrator–dynamic system. (Source: NASA Glenn Research Center.)
xenon, having a molecular weight around 40. The heated gas drives a
turbine working on the Brayton cycle. The gas-based system, however,
minimizes erosion and the problem of sloshing when transporting a liquid.
In either case, the high-pressure high-temperature fluid drives the turbine,
which in turn drives an electrical generator. The energy conversion
efficiency is about twice that of the photovoltaic system. This minimizes
the deployed collector area and the aerodynamic drag in the low Earth
orbit. An indirect advantage is that the energy storage is interwoven in the
system at no extra cost. It primarily resides in the form of latent heat with
the phase change at high temperature around 1000 K.
The usable energy extracted during a thermal cycle depends on the
working temperatures. The maximum thermodynamic conversion effi-
ciency that can be theoretically achieved with the hot side temperature, Thot,
and the cold side temperature, Tcold, is given by the Carnot cycle efficiency,
which is
Thot Tcold
Carnot ¼ ð3:1Þ
Thot
where the temperatures are in degrees absolute. The higher the hot side
working temperature and lower the cold side exhaust temperature, the
higher the efficiency of converting the captured solar energy into electricity.
The hot side temperature, Thot, however, is limited by properties of the
working medium. The cold side temperature, Tcold, is largely determined by
the cooling method and the environment available to dissipate the exhaust
heat.
The dynamic power system incorporates the thermal energy storage for
hours with no degradation in performance, or for longer duration with
some degradation. This feature makes the technology capable of producing
high-value electricity for meeting peak demands. Moreover, compared to
the solar–photovoltaic system, the solar–thermal system is economical, as it
eliminates the costly PV cells and battery. The solar concentrator–dynamic
system with a turbo-alternator also offers significant advantage in efficiency
and weight, and hence the overall cost over solar PV technology. The
efficiency advantage comes from the higher efficiency of the engine (about
30%) as compared to silicon solar cells (about 15%), and higher efficiency of
thermal energy storage of the receiver (about 90%) as compared to the
battery efficiency (about 75%).
The concept is sufficiently developed for use in the future, particularly in
high-power LEO missions. It may also find applications in high-power
defense spacecraft where large solar arrays can make the mission
nonmaneuverable and vulnerable to enemy detection and attacks. The
higher efficiency requiring less solar collection area results in reduced drag
and less concern regarding station dynamics, approach corridors, and
experimental viewing angles. The reduced drag is particularly important
because it allows lower flight altitudes within given constraints of drag-

makeup fuel and orbit decay time. At high power approaching the 100-kW
range for space-based radar (SBR), the solar array collector area becomes
prohibitive. The solar dynamic system can be extremely cost effective over a
wide range of power between a few kilowatts and hundreds of kilowatts. It
was considered for the dynamic isotopes power system (DIPS) in the 5 to
10 kW power range, and the space station in the 200 to 500 kW power range.
3.6 Nuclear–Thermoelectric
Interplanetary and deep space missions far away from the sun cannot be
effectively designed using photovoltaic power generation because of weak
solar flux. The spacecraft must therefore carry on board a primary energy
source, such as a radioactive isotope or a nuclear reactor. In the first
alternative, the radioisotope heats a thermoelectric (TE) material such as
lead telluride, which generates electrical potential (Figure 3.10). The
principle is similar to that in a thermocouple, only with higher conversion
efficiency. The radioisotope thermoelectric generator (RTG) is being
routinely used for interplanetary missions requiring power levels of several
hundred watts. The reactor, on the other hand, has been considered for high
power in the 30 to 300 kW range. Both power sources have the advantage of
supplying power all the time, thus eliminating the need for a battery in base
load systems having no peak power requirement. The obvious disadvan-
tage is the heavy radiation shielding required around the electronic
components. Also the nuclear fuels that are safe and easy to handle with
little shielding, such as curium-244 and plutonium, are expensive.
Inexpensive easily available fuels, such as strontium-90, are unsafe.
High-energy particles emitted from the radioactive isotope material are
the primary energy source, which heats the absorbing material. The thermal
power radiation decreases proportionally with the remaining mass. The
mass of the isotope material decay exponentially at a rate characterized by
the half-life, T1=2 , which is long. Therefore, the power generation essentially
remains constant for decades. This makes the nuclear energy source ideal
FIGURE 3.10 Thermoelectric converter.

for long-life interplanetary missions; many of them require several hundred

watts power. However, it has also been used for some defense missions
requiring a high degree of radiation hardness under nuclear threat, such as
in the Space Power-100 (SP-100) program having power requirements
ranging from 30 to 300 kW. Fairly advanced development has been carried
out under the SP-100 funding, which is described in detail in Chapter 22.
3.7 Nuclear or Chemical–Dynamic

Hundreds of kilowatts or multi-megawatts of electrical power require a
chemical or nuclear reactor as the primary energy source and rotating a.c.
generator. A short-life burst power mission may consider carrying on-board
chemical fuel. On the other hand, a long-life mission may require a nuclear
reactor. The primary source of power in such a reactor is nuclear fission,
just like that in a ground-based nuclear power plant, only much smaller in
scale. A fissile material such as uranium-235 works as the heat source to
vaporize a fluid, typically liquid metal such as mercury or sodium
potassium. The vapor then drives a turbine-generator using the Brayton
or the Rankine cycle. The Stirling cycle is suitable in the 50 W to 50 kW
power range for long life space missions, while the Brayton cycle is suitable
in the 50 kW to 10 MW power range for advanced electric propulsion. A
variety of concepts using a nuclear–dynamic system was considered for the
strategic defense initiative (SDI) programs in the late 1980’s and the early
1990s. The SDI conceptual design power levels were in the multi-megawatt
range between 30 and 300 MW. The basic requirement for such missions is
to generate high power in the pulse or burst mode for a short duration of
time, such as engagement at full power for about 15 min for a few times
over the mission life. For the remaining time during the entire mission, the
system must coast consuming minimum quiescent power. The design
driver for such a power system is large-scale energy storage with a slow
charge rate, and energy discharge at a high rate. For energy storage, both
the flywheel and the superconducting magnet have been considered along
with other options. The multi-megawatt nuclear dynamic systems are
further discussed in Chapter 22. A nuclear–dynamic system for a few
hundreds watts of power for interplanetary and space defense missions has
been developed by NASA GRC, and is being space qualified. It uses a
radioisotope as the heat source, a Stirling thermodynamic engine, a
reciprocating electrical alternator, and a.c.–d.c. power converters if needed
for d.c. loads. It is intended to replace the presently used RTGs. It works at
temperatures of 1000 C on the hot side and 500 C on the cold side of the
fluid. Compared to RTGs, it has a higher system efficiency (20%), higher
specific power, and is scalable to higher power levels. The space-worthy
technology has been demonstrated on a NASA test bed for applications up
to 10 kW by building and testing a 2 kW unit. The technology is similar to
that used for balloon flights where a parabolic solar concentrator at the top
of the balloon collects the energy, boils a liquid, and drives a Stirling engine
at the base of the balloon.
3.8 Other Systems

There are other power system options under various stages of development.
They may find applications in some special niche missions. In evaluating
potential applications of any new technology, the prime considerations are
always the mass, cost, reliability, and the technology risk. For example, a
system using any fluid has a potential for leakage and sloshing. On the
other hand, a system having no moving parts have three inherent
advantages: high reliability, no vibration, and no torque on the platform.
3.8.1 Thermo-Photovoltaic
In the thermo-photovoltaic (TPV) scheme, the radioisotope or solar heat is
directed at the PV cells as opposed to the present use of RTGs for generating
electricity. It has a significantly higher conversion efficiency. Therefore, it
may find applications as the mission-enabling technology where its high
cost may be acceptable. The system can have a cylindrical or a flat
configuration as shown in Figure 3.11. A heated surface radiates infrared
heat onto an array of photovoltaic cells sensitive in the infrared range. A
part of the energy is converted into d.c. electricity; and some is reflected
back and dissipated as heat. The energy conversion process is different to
that in the conventional PV cell. The efficiency varies with the radiator
temperature. It ranges from 10% at 800 C to 12% at 1100 C based on the
absorbed heat, but is only 3 to 4% based on the total incident energy.
Most current TPVs use low bandgap PV cells such as 0.55 eV InGaAs or
0.73 eV GaSb to optimize the cell response to energy sources in the 1 to 2 mm
range. The low bandgap of these cells leads to low open circuit voltage
(about 0.25 to 0.45 V) and poor fill factor caused by high intrinsic carrier
FIGURE 3.11 Thermo-photovoltaic converter.

concentration. Thus, the cell must be operated at low temperatures,

generally below 60 C to have adequate output voltage. As a result, a
large radiator area is required. The temperature has considerable influence
on the PV cell performance. The bandgap decreases with increasing
temperature and the photocurrent increases with temperature by absorbing
extra longer wavelength photons from the emitter. However, the open
circuit voltage of the cell decreases linearly with increasing cell temperature
due to the exponential increase in the saturation current.
The TPV concept has recently seen renewed interest with new develop-
ments in semiconductor technology. Recent advances have produced low
bandgap (0.50 to 0.55 eV) material that is lattice-matched to GaSb
substrates. These developments enable TPV systems to achieve reasonable
efficiency and power density with radiators operating at about 1000 C. At
relatively low radiator temperatures, there are many viable options for the
heat source and a number of applications become attractive. A conversion
efficiency of 11.7% has been reported2 at a radiator temperature of 1076 C
and a module temperature of 30 C. This is the highest achieved efficiency
for an integrated TPV system. It is the ratio of the peak (load-matched)
electrical power output and the heat absorption rate. The TPV cell degrades
like the GaAs cell, hence it is suitable for application in Earth orbits passing
through high radiation belts.
3.8.2 Solar–Thermoelectric
The power system for a near-sun probe operating at high temperature
cannot effectively use PV cells because of severe temperature degradation
in performance. In such a mission, the collected solar heat can be directed at
a pile of thermo-electric converters (TECs). The energy conversion details in
such systems are common with the RTGs covered in Chapter 20. The only
difference between the RTG and the solar–TE is the source of heat. The RTG
heat source is the nuclear reaction in a suitable radioisotope, while heat in
the solar–TE scheme comes from the sun.
3.8.3 Thermionic
In this conversion process, the thermal energy is converted into electricity
by using the electrons released from a hot body, known as thermionic
emission, or the Edison effect. The electrons released from the cathode are
collected at the anode, and a closed path through a load is established to
complete the circuit back to the cathode. It has no moving parts. The
concept is an old one, but is attracting new interest due to advances made in
high-temperature materials. The thermionic converter is basically a heat
engine with electrons as the working fluid, and is subject to the Carnot
efficiency limitation. For this reason, it operates at much higher tempera-
tures with a hot side around 1800 to 2000 K and heat rejection around 800 to
1000 K. This imposes limits on candidate materials and life. Also, the
thermionic converter produce a very low voltage requiring extensive power
conversion and large power conditioning loss. Further developments are
under way with the DoD funding at present. A typical example of
development is the thermionic fuel element that integrates the converter
and nuclear fuel for space power in the kilowatts to megawatts range for
long missions. Converters filled with ionized gas, such as cesium vapor, in
the inter-electrode space, yield higher specific power due to space charge
neutralization.
A 100-kW thermionic power system with a nuclear reactor has been built
and tested in Russia, which is believed to be scalable to 300 kW and perhaps
to a megawatts power level. A solar concentrator–thermionic power system
may be practical up to 100 kW. This technology is not suitable for power
levels lower than several kilowatts. General Atomic has designed a 50 kW
unit in which solar heat is concentrated directly on the cathode, thus
eliminating a working fluid. The conversion efficiency is around 10% using
a high-side temperature of 1800 K, and 5% using the high-side temperature
just below 1000 K. The specific power can approach 700 to 1000 W/kg.
3.8.4 Alkaline Metal Thermal to Electric Converter

The alkaline metal thermal to electric converter (AMTEC) is a direct thermal
to electric energy conversion system. It operates at temperatures around
1000 K on the high side and 600 K on the low side, and yield efficiency of 18
to 20%. It is a static system suitable for power levels below 100 W. It uses a
unique characteristic of the alkali metal conducting ceramic -alumina.3 It is
a solid electrolyte that conducts sodium ions, but is an electron insulator.
An electrochemical potential is generated when sodium is present at two
different pressures separated by an electrolyte. The sodium is circulated
through the converter using a porous stainless steel wick that uses capillary
forces to transport the sodium from the low-pressure region to the high-
pressure region. It has no moving parts and develops low voltage requiring
a high level of power conditioning. A commercial version of this technology
with titanium nitride electrodes is used for small down-hole instruments in
the oil industry. Developments are under way for space-worthy AMTEC
using a radioisotope as the heat source for interplanetary and deep space
missions requiring power in the few hundred watts range.
3.9 Technology Options Compared

The practical limit and performance characteristics of major power system
options are summarized in Table 3.1. Although the PV–battery power
system is the most widely used system for satellites, a variety of alternative
power system technologies have been flown, developed, and are being
Table 3.1 Practical limit and performance comparison of various power system
options
Specific
Power system Practical power limit Net system efficiency power
option (kW) (%) (W/kg)
Solar–PV 20 15–30 5–10

Isotope–TEC 1 7–15 7–15
Nuclear–TEC 100 7–15 –
developed for various missions. Table 3.2 summarizes the present state of
the spacecraft technology options4 along with current development work
that has promising applications in the near future. It also includes some
concepts that potentially may find beneficial applications in the future.
Development engineers continuously evaluate new technologies for
possible incorporation into new designs as they become available from
research laboratories. Many times, these changes bring incremental benefits.
Quantum benefits can be achieved by incorporating several changes
simultaneously into a new design. For example, DoD and NASA have
funded development and testing of protoflight solar array designs which
could yield specific power over 100 W/kg, a factor of 3 greater than the state
of the art, and a factor of 5 greater than the state of the practice. The design
under consideration integrates three promising technologies: a flexible
copper indium diselenide thin-film PV cell, smart mechanisms using shape
memory metal, and a multifunctional lightweight structure.5 An important
criterion in the application of a new technology is the development and
flight qualification status. New components must be subjected to time-
consuming and expensive testing to prove their ability to withstand launch
and space environments.
3.10 System Voltage Options

Early spacecraft with loads of a few hundred watts used 28 V d.c., primarily
based on the product specifications readily available for the aircraft power
system at the time. Since then, the power levels have increased significantly.
As power is the product of voltage and current, high power requires a high-
voltage bus in order to keep the current at a reasonable level. Otherwise, the
excessive power loss in switching devices and the I2R loss in conductors
reduce the system efficiency considerably. Today’s spacecraft bus voltages,
somewhat standardized by the product lines of various manufacturers and
government agencies, are 28 V, 50 V, 70 V, 100 V, 120 V, and 160 V, as shown
in Figure 3.12. The 160-V limit comes primarily from the bare conductor
interaction with space plasma, particularly in the low Earth orbit. Above
Table 3.2 Technology options and status of various power system components
Technology option Technology status
PV cell: converts sunlight into electricity Used in most long-mission spacecraft for power
generation. The mission design life is generally
limited to about 15 years due to degradation in
the Van Allen radiation belts.
TE cell: converts radioisotope heat into Well developed and flight-proven for a few
electricity hundred watts. It is expensive. Solar heated TE
may be considered in regions of intense Van
Allen radiation belts and/or missions with
nuclear threats.
AMTEC: converts isotope heat into Under development. Suitable up to a few
electricity using an alkali metal TE hundred watts systems for interplanetary and
converter deep-space missions, rovers, etc.
Thermionic: converts heat into electricity Solar and nuclear heated thermionic converters
have been built and tested in the laboratory.
Suitable for hundreds of kilowatts. Have high
specific power.
Nuclear reactor: converts heat into US prototype designs have been developed, but
electricity not proven in space. Russian space programs
have flown many such units. Scalable to the
megawatt power range.
Fuel cell: converts a fuel’s chemical Used routinely for the space shuttles. Was
energy into electricity considered for the ISS. May find more
applications in the future.
Battery energy storage Most spacecraft use secondary batteries, and
some short mission spacecraft use primary
battery. Fully developed.
Flywheel energy storage Under active development for installation on
ISS in 2006–2007 time frame. Targeted to
replace the battery.
Thermo–PV: converts heat into electricity Under development. Low specific power, but
using PV type cells may find niche application.
160 V, the solar array current leakage to plasma increases exponentially

with potential sparking above 180 to 200 V.
The plasma current is collected at exposed conductors having a bias
voltage. With an insulated conductor, parasitic power loss due to plasma
current at pinholes or coating defects have been shown to be small in
ground plasma chamber testing at NASA GRC. Tests on several cable-
insulating materials have shown that the collected plasma current remains
negligible below a few hundred volts positive. However, at voltages of
200 V and above, insulation pinholes showed snap-over effects. If con-
ductors are at high positive potentials relative to the plasma, snap-over may
greatly increase the electron current and the resulting power drain. Still,
regardless of the spacecraft grounding scheme, it is unlikely that more
spacecraft will be at potentials greater than þ100 V above the plasma
potential, so snap-over is unlikely to occur. An exception is the 160-V ISS
FIGURE 3.12 Optimum voltage for various power levels.
solar array, where the plasma contactor keeps the structure close to the
plasma potential. However, the most positive end of the solar array being
more than 100 V positive, the solar cell edges would collect more current
than otherwise, essentially increasing the power drain and demanding a
high current capacity for the plasma contactor. For other high-voltage LEO
spacecraft, such as the 200-V SP-100, analysis has shown that the power loss
from currents to other surfaces is small compared to the total delivered
current, and thus the percentage efficiency loss is also small.
A rule of thumb is that for every square meter of exposed conductor in
LEO at 100 V positive, a parasitic structure current of about 1 mA may be
expected. Thus, for a payload of about 100 m2 surface area on a 100-V bus,
for instance, only 100 mA of structure current may drain from the power
system capacity. This is negligible compared to 100 A that a 10-kW power
system would deliver at 100 V.
Voltages higher than 160 V can be used in low Earth orbit with insulated
cables covered in a shielded enclosure as shown in Figure 3.13, and by
FIGURE 3.13 Covered cable tray protection against voltage breakdown.

encapsulating all connectors and circuit boards. NASA has selected a 120-V
distribution system for the ISS with necessary step-down converters for
existing 28-V hardware. Early in the ISS design, 270 V d.c. and 440 V 20 kHz
a.c. were considered, but, finally, a 160-V solar array voltage and 120-V
distribution voltage were selected. Alternating current was seriously
considered for the space station design in the 1980s for multiple benefits,
but was dropped as the power requirement was significantly scaled down
and high development costs were projected.
High-voltage design, however, impacts upon the component selection.
Moreover, the space environment considerations also limit the voltage to a
certain level for a given mission. The factors that influence the voltage
selections are:
Power level being the primary driver

Space environment and space plasma
Paschen minimum breakdown voltage between bare conductors
Human safety
Availability of components, such as semiconductor devices, power
distribution and protection devices, tantalum capacitors, etc.
Design issues at much higher voltages up to several hundred kilovolts are

discussed in Chapter 22.
3.11 Scaling for Power Level

Selecting the bus voltage from available options is a matter of the technical
factors discussed above and in Chapter 22, taking into consideration the
corporation’s in-house design heritage. In the case of a new development
characterized by an absence of the heritage design, the following guideline
may be used:
optimum voltage ¼ 0.025 power requirement (3.1)
For example, a 5000-W payload power system would have a mass-

optimized design at 0.025 5000 ¼ 125 V. This voltage is then shifted to
the nearest standard voltage. If the voltage falls in the bands given in Table
3.3, the design engineer would have a good availability of parts which have
been widely used in other spacecraft design, both commercial and
government.
For top level screening of various power system options for a given
mission, the mass of a previously built similar design can be scaled to
estimate the mass of a new system design with a different power level. The
following is a rough empirical scaling law:
Table 3.3 Voltage bands
High voltage Low voltage Medium voltage

(V) (V) (V)
Distribution voltage 28 70 120

Solar array output 38 80 160

new power requirement 0:7
mass of new design ¼ mass of similar design
power in similar design
ð3:2Þ
For example, the mass of a new system with 10 times the power
requirement is 5 times the similarly built system using the same technology.
This equation suggests that specific power doubles for every 10-fold
increase in the power level.
For a.c. systems, the power system mass depends both on the power level
and also on the frequency. The design and fabrication experience of
commercial and aircraft industry suggest the following rule of thumb.

kW
mass of a:c: system ¼ ð3:3Þ
f
where the exponent is equal to 0.5 for small systems in hundreds of watts,
and 0.75 for large systems in hundreds of kilowatts.
References
1. Hojnicki, J.S. et al. ‘‘Space Station Freedom Electrical Performance Model,’’
NASA Glenn Research Center, Report No. TM-106395, 1993.
2. Brown, E.J. et al. ‘‘Measurements of conversion efficiency for a flat plate
thermo-photovoltaic system using a photonic cavity test system,’’ in
Proceedings of the 35th Intersociety Energy Conversion Engineering Conference,
AIAA, 2000, Paper No. 3029.
3. Pantalin, J.E. et al. ‘‘Advanced AMTEC converter development,’’
Proceedings of the 35th Intersociety Energy Conversion Engineering
Conference, ASME, 2001, pp. 519–524.
4. Hyder, A.K. et al. ‘‘Spacecraft Power Technologies,’’ Imperial College Press/
World Scientific Publishing Co, London, 2003.
5. Marshall, C.G. et al. ‘‘Example of a prototype lightweight solar array and
the three promising technologies it incorporates,’’ Proceedings of the 35th
Intersociety Energy Conversion Engineering Conference, SAE, 1999, Paper No.
01-2550.
Chapter 4
Photovoltaic–Battery System
4.1 Introduction
Photovoltaic conversion of the sun’s energy is the most common source of
electrical power in space. An array of photovoltaic cells powers the load
and charges a battery during sunlight. The battery powers the load during
an eclipse. If the solar array, the battery and the load were operated at the
same constant voltage, no voltage regulator would be needed. All the
equipment could be wired to the same bus. However, the solar array output
voltage is higher at the beginning of life, and when the array is cold for
several minutes after each eclipse. Also, the battery has a lower voltage
during discharge than during charge. Since the system is required to
provide power to the load at a voltage regulated within specified limits, a
voltage regulator is always needed to match voltages of various power
components during the entire orbit period.
The photovoltaic power system, therefore, primarily consists of a solar
array, a rechargeable battery, and a power regulator which regulates power
flow between various components to control the bus voltage. Other
components such as various sensors are also required to make the array
and the battery work together. The total power system is thus coordinated
internally as well as externally through interfaces with other systems of the
spacecraft. The top-level performance characteristics of the basic PV–battery
system are described in this chapter, leaving the component level details for
later chapters.
4.1.1 Solar Array

The solar array is made of numerous PV cells stacked in series–parallel
connections to obtain the desired voltage and current from the assembly. It
converts the incident photon energy into d.c. voltage, which drives current
through the external load circuit. The solar array works more like a constant
current source over its normal operating range. Its terminal voltage versus
current characteristic, referred to as the I–V curve, is shown in Figure 4.1.
The end of life (EOL) curve must meet the performance requirement. The
characteristic curve changes significantly with temperature and the
radiation dose of charged particles as shown in the figure. The power,
being the product of the voltage and current, varies as in Figure 4.2. The
58
Photovoltaic–Battery System 59
FIGURE 4.1 Current versus voltage characteristics of a photovoltaic cell.
power output of an array is maximum at the knee-point voltage. The system

produces less power when operating at any other voltage. Also, the power
output gradually degrades with accumulated radiation dose. At the end of
life, the array generates less power than at the beginning of life.
4.1.2 Battery
The battery is made of rechargeable electrochemical cells connected in a
series–parallel combination to obtain the desired voltage and current. Its
terminal voltage depends primarily on the state of charge (SOC), and to
some extent on the operating temperature. The battery charge is measured
in terms of the ampere-hours stored between the positive and negative
plates. The voltage is highest when the battery is fully charged, and the
lowest when it is fully discharged. Since the battery works more like a
constant voltage source over the normal operating range, its terminal
characteristic is generally expressed in terms of the battery voltage versus
the state of charge. Figure 4.3 depicts the voltage of one fully charged cell as
it discharges and then gets recharged. The voltage scale in the figure
FIGURE 4.2 Power versus voltage characteristics of a photovoltaic cell.

FIGURE 4.3 Battery cell voltage versus state of charge in one complete cycle.
represents both the NiCd and the NiH2 cells. The battery voltage drops
significantly with increasing discharge, and then rises during charge. The
average voltage during charge is higher than that during discharge.
4.1.3 Power Regulation

Power regulation is primarily accomplished by battery charge and
discharge converters, a shunt dissipator, and a mode controller that
responds to the bus voltage error signal. The shunt dissipator is necessary
to control the bus voltage during sunlight as described hereafter. The solar
array (source) and the constant power load have their own I–V character-
istics as shown in Figure 4.4. The system can operate at either of the two
intersection points A or B. However, point A is inherently unstable because
the load slope is less than the source slope. Point B, on the other hand, is
inherently stable. Without a shunt control, the system would operate at
point B, producing a lower power. With shunts designed to regulate the
FIGURE 4.4 Stability of operating point and shunt control during sunlight.
sunlight voltage at say 35 V; the system will pull back from point B to point
C by shunting the excess current to the ground. The shunt current in this
operating mode would be Ishunt, the difference between the source current at
D and the load current at C.
The mode controller sets the system’s operating mode in response to the
error signal, which is the difference between the actual bus voltage and the
reference voltage representing the required bus voltage. Depending on the
error signal value and its polarity (positive or negative), the mode controller
sends a control signal to either the shunt regulator, or the battery charge
regulator, or to the battery discharge regulator (Figure 4.5). One of these
three sub-regulators, in turn, maintains the bus voltage within the specified
limits. Details of the next level of the mode controller during the battery
discharge regulation are shown in Figure 4.6.
External interfaces of the power system with other systems of the
spacecraft are mission specific, and difficult to describe in general.
However, Figure 4.7 depicts key external interfaces in a typical commu-
nications satellite.
FIGURE 4.5 Mode controller error signal routing concept.

FIGURE 4.6 Mode controller error signal to battery discharge regulator.
4.2 Power System Architectures

The solar array, battery, and the shunt characteristics described above,
along with the load voltage requirement, are extremely important in
selecting the power system architecture that is most suitable for the mission.
The mission-imposed or self-derived requirements on the power system
also have bearings on the architecture selection. The PV–battery system is
configured in one of the following architectures that would optimize the
system performance for a given mission.
4.2.1 Direct Energy Transfer

Direct energy transfer (DET), in which the solar power is transferred to the
loads with no series component in between. The necessary exceptions are:
(1) the slip rings to provide a rotary joint between the Earth-facing
FIGURE 4.7 External electrical interfaces in a typical power system.

spacecraft body and the sun-facing solar array, and (2) the power
distribution unit consisting of load switching relays and fuses to protect
the power system from faults in the load circuits. The DET can be further
subdivided into two classes: (1) the fully regulated bus, and (2) the sunlight
regulated bus. The components and operation of these two busses are
similar, except that the latter has no battery discharge converter in the
power regulator unit (PRU).
4.2.2 Peak Power Tracker

Peak power tracker (PPT), in which the solar array output voltage is always
set at the value which results in the maximum power transfer from the
array to the load. A series power converter between the array and the load
matches the load voltage requirement and the array output voltage. For this
architecture to be cost effective, the power loss in the PPT converter must be
less than the gain in operating the system at the peak power point all the
time.
4.3 Fully Regulated Bus

In a fully regulated direct energy transfer bus, commonly known as the
regulated bus, the bus voltage is controlled within a few percent during the
entire orbit period. The typical bus voltage variation is 2 to 5% of the
nominal voltage. The architecture of this bus is shown in Figure 4.8. It has
the following components.
FIGURE 4.8 Fully regulated direct energy transfer architecture.

4.3.1 Solar Array

The solar array (SA) is an array of PV cells that convert sunlight into
electricity. The large array is generally divided into many parallel circuits
(strings), each with isolation diodes so that a failed string would not draw
power from the healthy ones. The ungrounded end of each section and
individual string is isolated from the main bus and battery by diodes. These
diodes may be located inside the PRU for protection, or on the back of the
solar array exposed to the environment. In one case, the slip rings are
involved, but not in the other. The electrostatic discharge induced transient
voltage suppressing snubber capacitors may be located inside or outside.
The array negative terminal is usually grounded. If the positive were
grounded, the power system would work just as well, except that we must
use p–n–p semiconductors instead of traditional n–p–n devices in the PRU.
The power converter design would change accordingly. One advantage of
the positive ground is that the structure becomes negatively charged, and
attracts atomic oxygen ions at lower speed, causing less sputtering of the
anodized aluminum commonly used in the spacecraft structure. The bus
diodes pair d(2), typically Schottky diodes, prevents loss of the bus voltage
in case the battery charger gets an internal fault, which may clear after some
time. Such diodes — appropriately called the DET diodes — have been
used in many defense meteorological satellites (DMSs). They are not
absolutely needed, and would reduce efficiency. In its absence, adequate
fusing should be used to protect the bus from faults in a battery charger.
The current required to clear a charger fault would come from the solar
array whether or not d(2) is present.
4.3.2 Solar Array Drive

The solar array drive (SAD) consists of slip rings, a motor, and motor drive
electronics. It continuously orients the solar array to face the sun for
generating the maximum possible power during the entire sunlight period
of the orbit.
4.3.3 Shunt Dissipator

During sunlight, particularly in the beginning of life, this component
dissipates power that is unwanted after meeting the load power and the
battery charge power requirements.
4.3.4 Battery
The battery stores energy in an electrochemical form to supply power to the
loads during eclipse periods over the entire mission life. NiCd secondary
batteries were extensively used in satellites until the mid 1980’s. In the
newer designs, however, they have been replaced by NiH2 for better
performance and lower mass.
4.3.5 Power Regulator Unit

The power regulator unit (PRU) provides an interface between the solar
array bus and the battery. The battery voltage varies widely with the cell
voltage varying from 1.0 V when fully discharged to 1.55 V when fully
charged. The discharge converter in the PRU boosts the battery voltage to
the bus voltage during an eclipse and the charge converter bucks the array
voltage to the battery voltage during sunlight. For GEO missions, the charge
converter rating is much smaller than the discharge converter rating. In
such situations, designing the charge and discharge converters separately
would be beneficial, as their individual designs can be better tuned to their
own ratings and requirements. For LEO missions, on the other hand, a bi-
directional converter may be beneficial, as the charge converter and the
discharge converter ratings are comparable. The following approximations
may help trading various options in conceptual designs. A 1-kW battery
discharge converter may weigh 2 to 3 kg. A 1-kW PRU designed to operate
between –25 and 60 C with a reliability of 0.99975 may weigh 6 to 7 kg
including the battery charge converter and dedicated support electronics.
4.3.6 Power Distribution Unit

The power distribution unit (PDU) ensures that all loads, except critical and
essential loads, are powered through switches and fuses. The fuses are not
to protect the loads as much as to protect the power system from faults in
the user equipment.
4.3.7 Bus Voltage Controller

The bus voltage controller consists of the bus voltage sensor, the reference
voltage and the error signal amplifier. The amplified error signal output of
the bus controller enters the mode controller, which in turn sends command
signals needed to regulate the bus voltage within required limits.
4.3.8 Mode Controller

The mode controller automatically changes the EPS mode in response to the
error signal as follows. The mode selector switch is typically a magnetic
latching relay.
4.3.8.1 Shunt mode

During sunlight, if the solar power exceeds the load and battery charge
requirements, the mode controller turns on the shunts to dissipate excess
power, else the bus voltage will rise above the allowable limit. During this
mode, the battery is charged at full, partial or trickle charge rate, as
required.
4.3.8.2 Charge cut-back mode

When the battery is approaching full charge, the charge rate is cut back to
control the battery temperature. This mode includes trickle charge rate
when the battery reaches a 100% state of charge. The charge cut-back mode
is also activated in case the solar power exceeds the load requirement, but
not enough to supply the required charge current to the battery. In such a
case, the battery charge power is cut back to the level that will maintain the
bus voltage within the allowable limits.
4.3.8.3 Discharge mode

In the absence of solar power during an eclipse, the battery is discharged to
maintain the bus voltage. The battery voltage falls with the decreasing state
of charge. The discharge converter, therefore, must increase the boost ratio
accordingly. The PRU does this automatically by increasing the duty ratio
of the discharge converter as the battery voltage decays.
4.3.8.4 PRU bypass mode

In case a fault occurs in any of the loads, the fuse must be blown as quickly
as possible in order to minimize the bus voltage decay. The delay in PRU
control loop response and the additional impedance introduced by the PRU
would significantly extend the fuse blow time. To circumvent this, the
battery is instantly connected to the bus by the bypass diode quad d(4),
which quickly delivers the battery energy to the fuse in case of a fault, and
blocks power flow to the battery at all other times.
4.3.9 Battery Bus

The battery bus is essentially a tap point directly off the battery. During the
launch and ascent phases of the mission, the PV array is not deployed, and
the battery meets all the energy needs. For example, all electro-explosive
devices (EEDs) for deployment are powered directly from the battery bus
for high reliability and low bus impedance (as there is nothing between the
power source and the load).
4.3.10 Power and Energy Management Software

Although the power and energy management software (PEM) is part of the
software system, it is dedicated to the EPS performance, health monitoring,
control, and protection. In case of emergencies, or during planned mission
operations, PEMS sheds loads in a preset sequence if and when the battery
state of charge cannot support all loads. The battery telemetry consists of
the battery voltage, current, temperature, individual cell voltages and the
internal pressures of selected cells. Some of these telemetry readings also go
to PEMS.
4.3.11 Loads
The term load includes all loads, i.e., the payloads (transmitters, receivers,
science instruments, etc.) as well as the bus system loads. Most loads in
satellites are constant power loads. Some loads may have a low duty ratio.
For all loads, the orbit average power requirement is taken into account in
sizing the PV array and the battery.
4.3.12 Ground Power Cord

To preserve the battery power during pre-launch testing and final checks
before lift-off, the on-board system uses external ground power via an
umbilical cord. To further preserve the battery, power transfer is scheduled
as late as possible in the countdown.
4.4 Bus Voltage Control

The regulated bus in normal operation maintains the bus voltage between
the specified upper limit and the lower limit. The mode control scheme for
GPS IIR type 28-V regulated MEO bus with about 1000 W load, may be set
up as follows:
Shunt mode regulates the bus voltage by using a shunt dissipator. In this
mode, the batteries are fully charged and the solar array power
generation exceeds the spacecraft’s needs, so the excess current from
the solar array is shunted. This mode is established when the shunt
drive voltage signal is greater than 2.5 V d.c. and the battery charge
currents are not responsive to the shunt drive voltage.
Battery charge control mode regulates the bus voltage using the battery
chargers as linear shunting circuits. In this mode, the batteries are
charged and the spacecraft loads are met by the solar array. If the
solar array power capability is only slightly above the system load,
the battery charge current must be limited. This mode is established
when the shunt drive voltage is 1.0 to 2.5 V d.c. and the battery charge
current is linearly responsive to the shunt drive voltage and is
between 0 and 3.6 A.
Discharge mode regulates the bus by the battery discharge converter. In
this mode, solar array current is not adequate for spacecraft bus
loads, so the batteries are in discharge to provide the power needed
for the bus. The solar array power capability is less than the bus load
demand. This mode is established when the shunt drive voltage is
less than 1.0 V d.c.
Reference and error amplifiers measure the difference between the 28-V
volt bus and reference voltage. Depending on the magnitude and sign of the
error signal, signals are sent to activate the shunt dissipators, the battery
charge controllers or the battery discharge controllers. These modes are
mutually exclusive and easily identified by monitoring the shunt drive
voltage telemetry signal. The PRU performance is significantly different for
each mode but varies only slightly for various input conditions within each
mode.
In another example, in an EOS-AM type fully regulated 120-V LEO bus
with a 3000-W load, the bus voltage is regulated within 5 V to remain
between 115 and 125 V as shown in Figure 4.9. This regulation band is
further subdivided into five sub-bands. The voltage control is implemented
using a mode controller, which places the power system in one of the
following operating modes:
FIGURE 4.9 Bus voltage versus operating mode bands in a 120 5-V regulated bus.
Discharge mode when the bus voltage falls below the specified limit
(point a)
Dead band (do nothing band) when the voltage is between the dead-
band limits (a and b)
Charge mode when the bus voltage rises above the dead-band limit
(point b)
Shunt mode when the battery is fully charged and solar array output
power exceeds the load power requirement (point c)
As a refinement, the control scheme may have two dead bands between
the charge mode and the shunt mode, one for placing the battery in high
charge rate, and the other for a low charge rate, as shown between the 122
and 124-V bands in Figure 4.9.
The bus voltage is regulated at a specified bus sense point, generally
defined inside the power regulator unit. For dynamic stability of the
feedback control loop, the mode controller amplifier provides a minimum
phase margin of 45 and gain margin of 10 dB. The bus is generally
designed to accommodate payload shedding in case the bus voltage drops
to 90–95%, and is capable of powering the essential spacecraft functions to
75–80% of the nominal bus voltage.
4.5 Control Circuit

The control circuit may be analog, digital or a hybrid, as described below.
4.5.1 Analog Control

Analog control has been used in spacecraft for decades. A fully analog
control loop for the main bus voltage regulator gives much better dynamic
response to the power bus transients. The bandwidth of this control loop is
kept high to optimize the bus regulation while reducing the noise.
4.5.2 Digital Control

Digital control is an alternative to the traditional analog control. A digital
shunt regulator uses relatively small shunt dissipators to provide small
signal control of the bus. As the analog shunt approaches its limit, it would
clock up/down counter, which controls the number of shunt switches in
use. While it may present challenges with respect to dynamic response of
the system controller, it offers flexibility in tailoring the system to multiple
missions. For example, by simply changing the gain constant in the
software table, one can adjust the system’s transient response. One can
also incorporate a number of different battery charge regimes, and then
adjust the charge rate in orbit with single command from the ground. This
arrangement is adaptable to a variable number of solar array segments and

to a number of array configurations. The result is more flexibility in using
standard modules of the solar array and the battery, which translates into
cost reduction.
4.5.3 Analog–Digital Hybrid Control

In a hybrid control scheme, a high bandwidth analog control loop for the
main bus voltage regulator assures better dynamic response to the power
bus transients while keeping the EMI low. The digital control loop is used
for managing the lower functions where a low bandwidth is acceptable. A
field programmable gate array (FPGA) may be used to command solar
array shunts on and off the bus and for commanding the battery charge
states. The FPGA samples the power system parameters, calculates the
averages and monitors large rate changes in order to make adjustments
needed in the solar array shunts and battery charge state. Such a hybrid
control system results in good bus transient response with flexibility,
reprogramming capability, bandwidth, and design simplicity.
4.6 Sun-Regulated Bus

If the design objective is to minimize complexity, an obvious approach is to
distribute power from both sources — the solar array and the battery —
directly to the load. Although such a direct energy transfer bus is sometimes
known as unregulated bus, the bus voltage is regulated by shunt control
during sunlight, and is unregulated only during an eclipse. For this reason,
such a bus is also known as the partially regulated bus, sunlight regulated
bus, or sun-regulated bus. Typical sun-regulated bus architecture is depicted
in Figure 4.10. The difference in the sun-regulated bus and the fully
regulated bus is only in the PRU. The sun-regulated bus has the usual
battery charge regulator to regulate the battery charge rate during sunlight
as commanded, but does not have the battery discharge converter. Instead,
the battery discharges directly to the bus during eclipse through diode ‘d’,
which is not called the ‘bypass diode’ any more, but is now called the
‘battery discharge diode’. It only allows discharge from the battery, but
blocks any uncontrolled charge current coming directly to the battery,
leaving the charging function to the charge regulator. Thus, the battery is
disconnected from the bus during sunlight when the shunt controller is
regulating the bus voltage. This architecture would be most beneficial in
multiple battery systems in GEO where the sunlight duration is long and
the eclipse is short.
In this architecture, the voltage regulation during sunlight is achieved by
shunt control circuits. Since the battery is directly connected to the bus
without a discharge converter in between, the bus voltage is the same as the
FIGURE 4.10 Sun-regulated direct energy transfer architecture.
battery voltage. Therefore, the bus voltage falls as the battery discharges
during eclipse, and rises as it gets recharged during sunlight. The bus
voltage variations are the same as the battery voltage. A nominal 28-V bus
voltage typically varies from 22 to 35 V during the orbit period. In the
absence of shunt regulation, the solar array output voltage would settle at
the naturally stable operating point B in Figure 4.4, which would be too
high particularly in the beginning of life and for several minutes after each
eclipse when the array is cold. The maximum to minimum bus voltage ratio
without the shunt control may approach 3 in some cases during mission
life, versus about half as much with the shunt control.
The mode controller controls the battery charger and the shunts as
needed. The power and energy management software maintains the energy
balance. Non-critical loads are powered through switches and fuses.
4.7 Fully Regulated Versus Sun-Regulated Bus

Power system design must give equal importance to conditioning and
distributing power from both the solar array during sunlight and from the
battery during eclipse. This poses a difficult design problem since both
power sources have fundamentally different characteristics. The solar array
is inherently a constant current source whose output voltage is limited by
the forward junction voltage of the PV cell. On the other hand, the battery is
inherently a constant voltage source of low internal resistance giving a
discharge voltage regulation of about 10%. It has high current and power
capability and its size is determined by the energy requirement during
eclipse. However, the fully regulated bus finds application in spacecraft
having the load power requirements above 3 kW, typically in GEO. The
sun-regulated bus, on the other hand, finds applications mostly in satellites

having the load power requirement of less than a couple of kilowatts,
typically in LEO.
The sun-regulated bus is simple and hence more reliable. However, the
equipment connected to the bus must be designed to operate over wide
variations in voltage up to 25% around the nominal value. Some loads,
such as traveling wave tube amplifiers (TWTAs) may accept such
variations, but most others require a voltage regulator at the point where
the power is distributed to the equipment. The sun-regulated bus saves the
cost and the power loss in the battery discharge converter. However, the
power loss in a battery discharge converter in the regulated bus and the
battery bypass diode in the sun-regulated bus compensate each other.
The fully regulated bus offers certain advantages at the cost of complexity
and potentially lower reliability. Low system impedance can cause voltage
variations due to cross-coupling between items of equipment. This may
increase the electromagnetic interference particularly in the time division
multiple access (TDMA) mode. However, the advantages of a fully
regulated bus are as follows:
Low battery mass and cost, as it allows use of fewer higher capacity
battery cells.
Low solar array mass and cost due to not having a battery latch-up
problem, which requires over-sizing the array as described in Chapter
13. The saving is typically 7 to 10%, and can be as high as 15 to 20% in
some cases.
Lighter power distribution harness mass, since the main bus works at a
constant voltage. In contrast, the sun-regulated bus requires heavier
wires to carry high current at the lowest voltage level at the end of
eclipse.
Lightweight load power converters, since they are supplied with
constant voltage input.
In the sun-regulated bus, if the battery latch-up occurs at the eclipse exit, the
resulting bus current can be in theory 60% higher than that in the fully
regulated bus. However, due to the reduced thermal load in eclipses and
the availability of charge power, latch-up in equinox is not a design driver.
Summer solstice is more critical period, when the bus latches onto a fully
charged battery, limiting the array over-sizing to 10 to 15%. The over-sizing
can be further reduced by temporarily reducing the heater load.
The harness connecting two pieces of equipment is usually designed to
limit the voltage drop under maximum load current. The specified
allowable voltage drop primarily depends on the bus voltage, and can be
from 100 to 500 mV in a 28-V bus, and 1 to 2 V in a 120-V bus. The sun-
regulated bus requires thicker conductors to limit the voltage drop at high
current at the end of eclipse. For a 5-kW load power capability, the power
distribution harness, including the metallic cable shield and various
connectors, could weigh approximately 20 to 30 kg in a sun-regulated bus
and 15 to 20 kg in a fully regulated bus.
The load power conditioner (LPC) provides interface between the EPS
bus and the radio frequency (RF) amplifier of the communication loads. The
output ratings of LPC vary in the 50 to 500-W range. Generally, it has a pre-
regulator and a switch mode converter if the bus is sun-regulated, or a
linear regulator with fully regulated bus. The LPC efficiency ranges from 90
to 98% depending on the design. However, for comparable power ratings,
its efficiency is about 2% higher for a fully regulated bus than that for a sun-
regulated bus, and the LPC mass is lower by 2 to 4 g/W of output.
A fully regulated bus in general provides straightforward power system
specifications, simple design interface, in-orbit operating flexibility, auton-
omous overload and eclipse entry control, and simple testing at the user
end. At power levels above a few kilowatts, it also provides high efficiency,
low mass and low overall cost at the power system level and/or the satellite
level. On the other hand, it needs a battery discharge converter. This can be
expensive, as each item of equipment has some base cost associated with
the design, documentation, manufacture, quality control, and testing.
The fully regulated bus also offers great flexibility in battery cell selection.
Fewer high capacity cells can be selected, since the battery voltage can be
any value within the charge and discharge converter duty ratio limit, This
generally reduces the battery cost and footprint by selecting one of the
standard ampere-hour capacity cells to match the exact requirement of the
bus. On the other hand, the required cell ampere-hour rating in a sun-
regulated bus may fall between two standard capacity cells, forcing the
design engineer to select the higher capacity cell, costing more in mass and
money.
4.8 Peak Power Tracking Bus

The solar array generates more power at higher voltage at the beginning of
life and when cold coming out of an eclipse. The maximum power can be
extracted only if the bus voltage is varied with years in service and with
temperature. However, the load must be supplied power at the same
voltage, generally lower than the maximum power producing voltage of the
solar array. A suitable switching regulator between the solar array and the
load, as shown in Figure 4.11, remedies this disparity between the
maximum power producing voltage and the constant load voltage. The
series regulator input voltage is then maintained at the maximum power
producing level by the peak power tracker, and the output voltage is
stepped down to the constant load voltage by varying the duty ratio as
required. The peak power tracking is activated only when the battery needs
FIGURE 4.11 Peak power tracking architecture for a mission with wide variations in solar
flux and temperature.
charging or the load demand exceeds the solar array output. Otherwise, the
excess power is left on the array raising the array temperature. The battery
relay is opened up when the battery is fully charged.
The peak power tracking (PPT) electronic controller senses the maximum
power point in one of the following ways:
The solar array output power — the product of voltage and current —
is continually computed and fed to the peak power tracker. The array
operating voltage is changed until the peak is detected.
As seen in Chapter 8, the bus dynamic and static (a.c. and d.c.)
impedances are equal in magnitude at the peak power point. A ripple is
injected into the solar array bus, and the dynamic impedance dV/dI
and the static impedance V/I are continuously measured. The bus
voltage is adjusted such that both impedances are equal.
The ratio of the Vmp to the Voc for any solar array is approximately
constant, say K (typically 0.70 to 0.75). The Voc of a solar cell coupon
maintained in the same environment as the main array is continuously
monitored. The operating voltage of the main array is then adjusted to
KVoc to extract the maximum power.
The inner voltage control loop regulates the solar array output voltage
to the reference value from the PPT controller. By changing this
reference value at regular intervals, the PPT controller moves the
operating point of the solar array. In each time interval, the PPT
controller calculates the solar array power slope by multiplying the
sensed solar array voltage and current. If this power slope is positive,
the PPT controller increases the reference value until the sensed power
slope is negative, and vice versa. Thus, the operating point of the solar
array is located near the peak power point where the power slope is
zero. The algorithm can be written as
P
Vref ðn þ 1Þ ¼ Vref ðnÞ þ K ð4:1Þ
V
where K is a suitable constant. The peak power voltage thus derived is

fed to the series switching regulator, which then converts the array
voltage to the load voltage. Figure 4.12 shows a complete PPT
assembly.
FIGURE 4.12 Power regulator unit with peak power tracker.

(Source: Dornier Satellitensysteme Gmbh, Daimler-Benz Aerospace. With permission.)
The peak power tracking architecture is particularly useful in the

following applications, where the additional weight, power loss, and cost
of adding such an assembly can be justified:
Small satellites having no pointing gimbals, such that the solar array is
not always oriented towards the sun
Satellites having the solar radiation and array temperature varying over
a wide range, indirectly varying the array voltage
In low Earth orbit, where the battery must be charged in a short period.
The PPT allows maximum power to be captured for several minutes after
each eclipse when the array is cold. Architecture without the PPT feature,
such as a DET bus, would waste a significant amount of power as shown in
Figure 4.13. If a DET system were designed to deliver the required power at
one-half the illumination at EOL, the power waste would be CD watts at
EOL full sun, BD watts at BOL full sun, and AD watts at EOL full sun on a
cold array. The PPT design eliminates this waste by utilizing all the power
that can be generated.
FIGURE 4.13 Power wasted in direct energy transfer architecture in certain conditions.
The main advantages of the peak power tracking are that it maximizes
the solar array output power all the time, and it does not require the shunt
regulator and the battery charge regulator. On the other hand, it results in
poor system efficiency due to power loss in the peak power tracking
converter. Moreover, since this loss is dissipated inside the spacecraft body,
it negatively impacts the thermal system.
The PPT can have three configurations: series, parallel, and series–
parallel, as shown in Figure 4.14.1 The series–parallel configuration yields
better system efficiency because the input and the output power conversion
is processed by a single converter for all operating modes, as seen in Table
4.1.
A PPT algorithm has been developed2 and tested using only the solar
array voltage information, giving the tracking control without using current
sensors. This results in low ripple current, hence lighter bus filter
capacitors.
FIGURE 4.14 Battery charge and discharge options in peak power tracking architecture.
Table 4.1 Operating modes of series–parallel battery discharge

regulators in the PPT system
Mode Series regulator Battery charger Battery discharger
Eclipse Off Off Regulated bus

PPT discharge PPT Off Regulates bus
(partial sun)
PPT charge Regulates bus PPT Off
(full sun)
Trickle charge Regulated bus Trickle charge Off
4.9 Architecture Trades

A primary criterion of choosing between alternative architectures is the
overall mass, efficiency and cost of the system. Table 4.2 summarizes the
trades between various architectures with their best applications. The final
selection of the architecture depends on the mission specific details.
However, past experience indicates that peak power tracking architecture
Table 4.2 Pros and cons of various architectures and their best applications
System Sun regulated Fully regulated Peak power tracking
Pros High power transfer Well regulated No need for shunt

efficiency from solar input voltage to regulator and battery
array and battery to all loads. charge regulator.a
load.
Simpler, lighter Makes the maximum use
Fewer power system and more of the incident solar
components. efficiency load energy.
converters.
Cons More complex load Needs more Lower efficiency than DET
converters. power at EOL in many cases.
converters.
Battery latch-up More heat dissipation
concern. Series power loss inside the spacecraft body.
between battery
Larger solar array. and load.
Best Small load variations. Loads requiring Large variation in solar
application in close regulation. array input energy
missions with Small variations in (illumination) throughout
these features illumination for most Large solar array the mission.
of the sun period. output voltage
variations.
a
True for a single battery bus with the battery connected directly to the bus. For a multiple
battery system, or a fully regulated bus, each battery must have its own charger for effective
battery charge management.
is generally advantageous for small satellites in low or irregular orbits

having a power requirement of less than 500 W. Between 1000 and 3000 W,
the sun-regulated direct energy transfer architecture would most probably
be advantageous. For power levels exceeding 5000 W, the regulated direct
energy transfer architecture is generally found advantageous. Although the
regulated bus requires additional equipment, the added mass is compen-
sated by the elimination of converters and regulators in the loads.
Moreover, decoupling the battery design from the solar array voltage,
and operating the system at a constant voltage optimizes the operating
point of the solar array. The battery voltage is then chosen to optimize the
number of battery cells in the most economical capacity.
The PPT architecture is more suitable for LEO satellites having relatively
short periods of sunlight. It is also attractive for missions with large
variations in solar flux, solar array temperature, and sun angle in satellites
with no sun tracking gimbals. It makes the best use of the solar array all the
time in such missions.
In addition to the DET and PPT architectures described above, some
spacecraft use hybrid architecture, such as two power busses, one regulated
and the other sun-regulated. Charging the battery from a dedicated solar
array section rather than from the main bus is another possibility.
The architectures of some of the spacecraft busses flying at present are
described in the next sections.
4.10 The International Space Station 160- to 120-V Bus

The International Space Station is truly the largest and most complex space
structure ever built, with 16 international partners. The complete assembly
shown in Figure 4.15 weighs one million pounds and the total interior space
of its six laboratories equals two Boeing 747 aircraft. It is taller than a 30-
story building and wider than the length of a football field. The attitude is
controlled within 1 stabilized at a rate less than 0.1 /s. It is in 335 to 500-km
low Earth, 51.6 inclined orbit with a 90-min orbit period and 35-min
eclipse. The experiments planned for its on-board laboratories are targeted
to enhance our understanding in:
The Earth, by viewing 75% of its surface

Long term exposure of micro-gravity on humans
Producing new medicines and materials
Like any other large space programs, such as the manned landing on Moon,
the engineering and technology spin-offs on our everyday lives on Earth are
the side benefits.
The ISS power system generates 105 kW using a solar array with an area
of nearly 1 acre. The loads are powered during an eclipse by thirty-eight
FIGURE 4.15 International Space Station with solar array and other modules in view.
(Source: NASA.)
81-Ah NiH2 batteries3 via bi-directional battery charge–discharge units

(BDCUs), autonomously controlled by the bus voltage set point. It has two
interconnected power systems; the 160/120-V US built system, and 120/
28-V dual voltage Russian built system to power the US, European, and
Japanese modules. The two systems are generally independent, but are
interconnected via d.c. converters to allow bi-directional transfer of power.
Figure 4.16 is the functional block diagram of the US system, including the
converter at the interface with the Russian power source. The US system
is described below.3
The solar array is made of four modules (wings) for a total of 76 kW
power under nominal conditions, and more during favorable conditions.
Each wing consists of two thin blankets held under tension on each side of a
central collapsible mast. The entire assembly turns on a gimbal, which
provides one axis of rotation for sun pointing. A second orthogonal axis
rotation is provided at the gimbal, where the entire solar array connects to
the rest of the truss structure of the station. The 76 kW from the US system
added to the 29 kW from the Russian system makes the 105 kW capability of
the complete station assembly. Each of the four solar modules is made in 82
strings of eighty 8 8 cm PV cells and coverglass for protection against
space charged particles. The PV cells are crystalline silicon with 14.5% EOL
efficiency over a 15-year life. The module printed circuit is Kapton/copper/
Kapton laminate welded to each cell to provide a series interconnection for
FIGURE 4.16 International Space Station single channel power flow diagram.
(From E.B. Gietl et al. NASA Report No. 210209, 2000.)
the assembly. Parallel bypass diodes are used every eight cells for reliability
in case of cell damage and to avoid reverse current heating during
prolonged shadows. Sequential shunt units (SSUs) on the solar arrays
operate at 20-kHz switching frequency.
The seasonal sun-pointing is done by gimbals and the orbit sun-
following by drive and roll rings. For currents of the space station
magnitudes, the roll rings provide superior power transfer performance
over the slip rings with rubbing contacts, as described further in Chapter
22.
The solar array output voltage is 160 V, which is the highest voltage that
can be practically used in a low Earth (high plasma) orbit in view of
potential plasma arcing and/or leakage current concerns. The 160 V is
stepped down to 120 V using d.c.–d.c. converter units (DDCUs), each rated
at 4.25 kW, for utilization inside the user modules. The DDCUs provide
150% current limiting capability and 20 dB isolation between the generation
point and the distribution points for personnel safety.
The solar array area and the operating voltage are greater than in any
other spacecraft flown before. Therefore, the nature of the single point
ground on the ISS in high plasma in LEO poses an arcing problem. To
preclude such arcing, a device called the plasma contactor located on the
truss creates a plume of ionized xenon gas, which acts as a conductive

bridge between the station and the plasma. It protects the array and other
conductive surfaces of the station from arcing, pitting, and erosion.
The battery is made of 48 battery packs, each with thirty-eight 81-Ah
IPV NiH2 cells, designed for 5 years (40,000 charge/discharge cycles) at
35% depth of discharge (DOD), although the actual operational DOD is
around 15%. With two packs connected in series, the battery voltage
varies between 95 and 115 V. Battery replacements are planned every 5
years to assure the power margin. With 15 years life of the station, the
battery replacements every 5 years constitute significant recurring cost.
The maximum charge rate is 50 A, which tapers down to 40, 27, 10 and
5 A, and finally to 1 A trickle charge rate. No bypass diodes are used in
either direction, but the cells are closely matched. An open cell will lose
the entire battery string. The operating temperature is in the range of 0 to
10 C.
The switching and fault protection is achieved by solid-state remote
power controllers (RPCs) in six ratings from 3.5 to 65 A in both current
limiting and non-current limiting designs. The RPCs trip at different set
points of over-current, over-voltage, and under-voltage to isolate faults as
close as possible to the faulted equipment. The solid state power controllers
(SSPCs) provide switching and protection. They reset automatically for
critical loads and remain tripped for non-critical loads. The upstream
coordinated fault protection in the 160-V segment is achieved by larger
main bus switching units (MBSUs) containing a remote bus isolator (RBI).
The RBIs are essentially large relays capable of interrupting up to 350 A of
d.c. fault current.
The power system stability is also a serious concern because the loads on
the station are constantly changing as new scientific experiments are
brought on board. As a result, the output impedance of the DDCUs and the
input characteristic of the loads have been specified to ensure stability with
any combination of the expected loads. The stability criteria under such
diverse load combinations are discussed in Chapter 14.
A hierarchy of redundant computers linked via the MIL-STD-1553 bus
controls the ISS power system. The computers autonomously control many
functions such as sun tracking, battery energy storage, and thermal control.
The power flow balance between major segments of the station is
coordinated by the on-board command and control systems, which also
provide interface control between the segments.
Future plans for new technology developments on the ISS that are related
to the electrical power systems include: (1) replacing the silicon solar cells
with GaAs panels to increase conversion efficiency, (2) using high discharge
cycle NiH2 batteries, and (3) using flywheel energy storage to significantly
increase the specific energy.
4.11 Large Communications Satellite Buses

In view of reducing cost and delivery time, all prime manufacturers have
developed standards busses, which can meet various power requirements
without making many changes. The standard bus, however, has a degree of
flexibility to tweek a little if and when needed to meet a customer’s
specifications. A few such standard busses presently available for commu-
nication satellites are briefly described below.
4.11.1 100-V Bus

A standard fully regulated DET bus operating at 100 V d.c. is offered by
Boeing Satellite Systems under the trade name BSS-702TM. Its power system
architecture is shown in Figure 4.17.4 The BSS-702 bus is a dual voltage bus,
primary 100 V d.c. for high power equipment and secondary 30 V d.c. for
low power component. It uses a NiH2 battery and silicon triple junction
solar cells. A xenon electric ion-propulsion system is used for N–S station
keeping, which is powered from the 100-V bus.
After each discharge of the battery, the basic function of any battery
charge management is to return the battery to the highest achievable SOC
with the minimum amount of overcharge. In the BSS-702 bus, this is
achieved by performing the following health monitoring and maintenance
functions:
FIGURE 4.17 Fully regulated 100-V bus architecture.

(From R. Hill, Proc. 36th IECEC, ASME, p. AT-59, 2001. With permission.)
Monitor and process the battery voltage, current, temperature, and

pressure sensor readings.
Control over-temperature to prevent inadvertent overcharge.
Autonomously control the charge rate by switching between high, low,
and trickle charge rates as needed in different seasons.
Adjust battery bias to account for voltage degradation in NiCd and
pressure growth in NiH2 cells.
Detect and report low SOC conditions.
Each battery is built in multiple packs to distribute the mass and heat as
needed to balance the spacecraft layout. Each pack has redundant heaters,
reconditioning circuit health sensors, and cell bypass diodes in case of cell
failure. The sensor monitors the cell voltage, pressure, temperature, and
bypass and reconditioning circuit status telemetry. The cells are 5.5-inch
diameter NiH2 and are stacked vertically with the electrodes on two
separate sides.
The solar array panel tracks the sun in one axis only, so the seasonal
variations in power generation can be more than 10% over the year. The
original solar array design was a concentrator type with channeled solar
reflector panels along both sides of the wing with GaInP2/GaAs/Ge PV
cells mounted on a graphite face sheet on an aluminum honeycomb
substrate. After suspecting some plume related problems, the array was
replaced with a traditional flat panel array.
Loads are either unfused, fuse protected, or switched by relay or
transistor, depending on the current magnitude and the nature of the load.
4.11.2 70-V Bus

A standard fully regulated DET bus operating at 70 V is offered by
Lockheed Martin Space Systems under trade name A2100TM. Of the 16
commercial satellites contracted in 2002 and the first half of 2003, five were
for A2100. It is available up to 15-kW power level, scalable up to 20 kW
using multi-junction GaAs PV cells. The power system architecture is
shown in Figure 4.18.5 The solar array is made of flat panels with Si, GaAs
or multi-junction PV cells as needed for the required power level. The PV
cell coverglass is coated with indium tin oxide (ITO) to eliminate arcing due
to electrostatic discharge (ESD). The honeycomb substrate with a graphite
epoxy composite face sheet saves mass over that with aluminum. The
booms and yokes are made from aluminum or graphite. One solar array
drive (SAD) per solar array wing with redundant motor winding provides a
large torque margin in case jerking is needed. Power slip rings and signal
slip rings are in separate sections.
The batteries are made of NiH2 cells assembled in two batteries for the
required energy storage. The battery power converters in the PRU can be
FIGURE 4.18 Fully regulated 70-V bus architecture.

(From A.A. Salim, Proc. 35th IECEC, AIAA, p. 2809, 2000. With permission.)
directional (buck and boost separately), or bi-directional if it is beneficial.

The charge converters have multiple charge rates. The batteries have
multiple heater control zones for maintaining desired temperature with
redundant temperature and pressure sensors. For open cell protection, a
thermally activated switch is preferred over the bypass diode, as the former
results in lower loss of power. Reconditioning is considered unnecessary for
the NiH2 batteries. However, if required by the specifications, resistances
are connected across the entire battery, or across a battery section, or across
each cell. In the first option, which is the most mass effective, two parallel
resistors are first commanded across the battery. As the battery voltage
approaches 1.0 per cell, the discharge rate is cut back by commanding one
of the resistors off to avoid cell voltage reversal. The charge rate is also cut
back in case of any peaking power on the bus sensed by a voltage dip.
The battery nominally operates between –5 and 0 C during the charge
period. However, its temperature rises to about 20 C at the end of the
longest eclipse. The allowable cell-to-cell and battery-to-battery gradients
are 5 to 10 C. Locating the battery outside the spacecraft body can be
advantageous in maintaining the 0 to 5 C temperatures during the charge
periods. The MEP is the module-enabling plug for the battery. The electro-
explosive devices (EEDs) for deployment are powered directly from the
battery at the 22nd cell tapping, and are protected by fuses in the EED
enable plugs.
The main power bus assembly provides the load distribution, current
sensors, voltage sensors, and bus filter. The fuse box houses protective fuses
and also load power switches for controlling the bus load when needed.
The transient over-voltage and under-voltage following a large load step,
such as turning on an arc jet (>4 kW total), is individually analyzed to
ensure meeting the bus ripples requirement.
In the A2100 bus, the energy balance is managed by controlling two
batteries independently in the following steps:
Using the current telemetry, determine the total battery charge and
discharge current
Determine the cell voltage and pressure
Compute the state of charge
Control the battery charge rate
If the battery temperature exceeds a set limit, reduce the charge rate
Shed load if the battery discharge falls below the set level
Provide for parallel or sequential battery charge
Prevent the disconnection of all battery power converters
Provide ground override capability
The bus under-voltage protection under a set limit is provided by

commanding and latching the payload power converter off. The shaded
loads can be turned on only by a ground command.
A shunt is connected across each solar circuit. The shunt switch is either
completely turned on or off. The on-switch shunts the power to the ground,
while the off-switch feeds the power to the bus. The desired bus regulation
with fine resolution is maintained by using one active shunt. Such
configuration also minimizes the bus voltage ripple.
4.11.3 Under-50-V Buses

British Aerospace’s standard 42.5-V fully regulated and 28 to 42.5-V sun-
regulated busses have been used on many small satellites. US manufac-
turers, on the other hand, offer 28-V fully regulated and 22 to 35-V
sun-regulated busses for small satellites requiring load power under a
couple of kilowatts. One such bus offered by Daimler-Benz Aerospace is
depicted in Figure 4.19.
4.12 Small Satellite Bus

The following power system features are typical in small satellites:
A solar array covering a wide range of solar flux and temperature with
wide swings in the I–V characteristics.
Photovoltaic–Battery System
87
FIGURE 4.19 Sun-regulated low voltage bus architecture. (Source: Dornier Satellitensysteme Gmbh, Daimler-Benz Aerospace. With permission.)
Peak power tracking architecture and no sun-tracking gimbals.

Body mounted solar array or three to four flat panels on 1 to 2 mm
aluminum substrate with insulation and optimized coverglass. Or,
flexible solar array with cloth-type substrate weighting about 5% of the
rigid substrate.
One battery, if permissible; otherwise two; NiCd or NiH2; both to have
a reliable flight history.
Early small satellites were spin stabilized, with their drum-like body
covered with PV cells all around. Such configuration used the cells
ineffectively; because it uses less than one third of the PV cells for power
generation at any given time. Today’s spacecraft have 3-axis active attitude
control with solar cells installed on flat panels pointed normally to the sun.
The PPT is more suitable for being one common architecture for a
number of small satellites for a variety of missions. The design can be
modular and adaptable without component redesign, thus being cost
effective. The power control unit maximizes the energy delivered to the bus
by driving the solar arrays to operate at their maximum power point over a
wide range of temperature, sun inclination, and illumination intensity. The
tracker system, in addition to its flexibility, does not dissipate power
associated with the shunt regulation inside the spacecraft. It adjusts the
operating point of the solar array as needed, thereby leaving the excess
power on the array.
A PPT bus that has flown successfully had the following design features,
which can be easily changed to accommodate other missions.
The solar array continuously tracked the sun using a solar array drive.
The PV cell could be either 8 mil or 2 mil thick silicon or 5.5 mil thick GaAs.
The NiH2 battery used 22 cells. The maximum battery DOD for LEO was
nominally 30% and 35% with one cell failed. For GEO, it was nominally 70%
and 75% with one cell that failed. The charge/discharge ratio was
maintained at 1.05 for LEO and 1.2 for GEO.
The peak power tracker received telemetry from the battery charge
control loop, which varied the modulation of the control signal operating
the tracker in each PRU.
The PRU worked as the interface between the solar array and the bus,
designed to deliver the total power generated by the solar array to the
spacecraft bus. It contained a peak power tracker comprised of a buck
regulator with pulse width modulation (PWM) control and the required
battery voltage–temperature (V–T) control. After receiving the battery
telemetry and comparing with the selected V–T charge curve, the PRU
varied the duty ratio of the buck converter. This changes the solar array
operating point thus reducing the power to the bus. All power regulation
functions were monitored and controlled within the PRU, including the
battery charge control. Separate PRUs were used for each wing if more than
one solar array wing under different operating conditions were needed for
some satellites in the standard product line. This ensures that the EPS
design remained modular to accommodate all possible mission scenarios
with little or no design changes.
4.13 Micro-Satellite Bus

Very small satellites with load power requirements in watts can use simple
and lean architecture shown in Figure 4.20. The solar array, the battery, and
the loads are permanently connected in parallel. The battery feeds to the
loads automatically during an eclipse, and recharges itself during sunlight
by imposing its drained voltage on the bus. Once fully charged, the battery
voltage is relatively constant, and the battery works as the buffer. Any
excess current is absorbed by the shunt resistance connected in parallel with
the battery by means of a switch.
Two simple architectures for satellites that require power only during
sunlight are shown in Figure 4.21(a) and (b). The shunt voltage regulator
makes the power source behave like a current source. On the other hand,
the series regulator makes the source behave like a voltage source. The
disadvantage of the series approach is that, without the shunt regulator, the
solar array output voltage in a typical GEO satellite would exceed 150% of
its EOL voltage. For a few minutes after exiting the eclipse, the cold array
can produce as much as 300% of the nominal bus voltage.
Low voltages (7 to 15 V) have been considered for satellites with load in
tens of watts. However, such low voltage is likely to end up in high cost
because it requires components other than standard 28-V class components.
For example, ST5 spacecraft designed by NASA GSFC6 for a few watts load
used a battery dominated low voltage bus depicted in Figure 4.22. To
minimize the number of components for high reliability and low mass, it
used a body mounted solar array with triple-junction cells, one lithium-ion
battery, and simple power electronics. The electronics that traditionally
FIGURE 4.20 Simple power system for a micro-satellite.

FIGURE 4.21 Micro-satellite power system with shunt regulator and series regulator.
FIGURE 4.22 Micro-satellite power system with no regulator.

provide the solar array regulation and battery charge control are
eliminated. Low loss diodes are used in series with a solar array for fault
isolation. The bus voltage is the same as the battery voltage, which may
vary from 2.7 to 4.0 V per cell. The design can be modular in that for a
higher power microsat, two modules can be connected in series as shown in
the figure.
References
1. Cho, Y.J. and Cho, B.H., ‘‘A novel battery charge–discharge of the
regulated peak power tracking systems,’’ in Proceedings of the 34th
Intersociety Energy Conversion Engineering Conference, SAE, 1999, Paper
No. 01-2445.
2. Veerachary, M., Senjyu, T., and Uezato, K., ‘‘Voltage-based maximum
power point tracking control of PV system,’’ IEEE Transactions on Aerospace
and Electronics Systems, 38(1), 262–267, 2002.
3. Gietl, E.B., Gholdston, E.W., Manners, B.A., and Delventhal R.A.,
‘‘Electrical power systems of the international space station — A platform
for power technology development,’’ in Proceedings of the 35th Intersociety
Energy Conversion Engineering Conference, AIAA, 2000, Paper No. 35-AP-SS-
1. Also NASA Report No. TM-210209, June 2000.
4. Hill, R., ‘‘Boeing Satellite Systems 702 Electrical Power System,’’ in
Conference, ASME, 2001, Paper No. AT-59.
5. Salim, A.A., ‘‘In-orbit performance of Lockheed Martin’s Electrical Power
System for A2100 Communication Satellite,’’ Proceedings of the 35th
Intersociety Energy Conversion Engineering Conference, AIAA, 2000, Paper
No. 1-AP-SP-1
6. Castell, K. and Wingard, R., ‘‘Recent advances in power system design at
GSFC,’’ Proceedings of the 34th Intersociety Energy Conversion Engineering
Conference, SAE, 1999, Paper No. 01-2534.
Chapter 5
Environmental Effects
5.1 Introduction
The radiation and other aspects of the space environment that have
damaging effects on various power system components are discussed in
this chapter. The charged particles impinging on the spacecraft surface slow
down by collision and lose their energy. The deposited energy in the
material causes damage by atomic excitation in metals and semiconductors,
and ionization in plastic materials. Various particles lose energy at different
rates as they penetrate the surface and cause damage at different rates. The
higher the rate, the more damage they produce. The damage is most severe
in the Van Allen belts because of the high level of radiation. Although the
belt region, which is between the low Earth orbit and the crowded
geosynchronous orbit, is potentially a valuable piece of orbital real estate,
the region has remained relatively unexplored because of the practical and
economic difficulties of shielding conventional satellites from charged
particles. NASA has measured charged particles in many orbits to develop
and validate a computer model. In 2002, nine astronauts aboard the ISS each
wore a 0.04-inch square silicon chip dosimeter that measured the levels of
radiation they were exposed to during space walks. On Earth, similar
devices are used in hospitals worldwide to measure the radiation given to
cancer patients.
5.2 Solar Array Degradation

The solar array is the most affected component in the power system under
space radiation. The degradations are listed in Table 5.1 in order of their
importance. The array performance degrades with the increasing level of
fluence, but not in a linear relationship. The damage is more in the initial
years than in the latter years. The impinging particles produce defects in the
crystalline structure of the PV cells. The resulting defects degrade the
voltage and current outputs of the cell. Low-energy particles create damage
close to the surface, and therefore lower the open circuit voltage. On the
other hand, high-energy particles penetrate deeper in the base and lower
the lifetime of electron hole pairs, thus decreasing the short circuit current.
The power generation capability continues to degrade as the radiation dose
accumulates over the years. When it falls below the specified requirement,
92
Environmental Effects 93
Table 5.1 Environmental causes and effects on solar array performance
Environment Effect
Ionized particles Damage to the p–n junction

Micrometeoroids and debris Damage to cell active area and interconnects
Increased cell shunt resistance
Contamination from out-gassing Degradation in optical properties of the coverglass
and propulsion plume
Space plasma Increased leakage current
Thermo-elastic cycling Cracks in solder joints of the interconnects
Bypass diode failure Shorts in cell blocks and reduced string current
Ultraviolet radiation Degradation in optical properties of the coverglass
the solar array is considered to have reached the end of life. Figure 3.7 and
Figure 3.8 showed the gradual degradation of the I–V curves under
increasing fluence levels, which result in continuously decreasing power
output with years in service. With traditional silicon cells, the average
power degradation is about 1% per year in LEO and less in GEO. Because
the radiation damage to the n-type crystalline lattice is less than that on the
p-type, the n–p type PV cells are more radiation resistant than the p–n type
used in early satellites. Today, almost all cells used in the industry are n–p
type. A coverglass, such as fused silica or ceria doped MicrosheetTM, over
the PV cell is used for protection against radiation. The degree of radiation
protection provided by the coverglass comes from both the density and
thickness of the material. For example, 12-mil fused silica coverglass
provides the same protection as 10-mil ceria Microsheet. Normally, the
coverglass is coated with anti-reflective material, such as silicon monoxide.
The coating minimizes reflection of the sunlight, thus enhancing the solar
energy absorption and the power output of the cell. Susceptibility to
radiation can be further enhanced using radiation-hard solar cells. Higher
resistivity of the cell substrate (base), while initially less efficient, may
provide higher EOL power in a long mission spacecraft.
5.3 Electrostatic Discharge in the Solar Array

The arcing in a solar array due to electrostatic discharge (ESD) is known to
have caused power loss in both LEO in GEO satellites. Small ESD
discharges do not have enough energy to cause damage. However, the
risk is increased in a high-voltage array, particularly in the low Earth orbits.
Bus voltages greater than 120 V inherently increase such a risk because of
current leakage and arc discharges. The mechanism of damage is as follows.
An anamolously large charge collection from the space environment can
occur on positively biased array components, and the negatively biased
portions may be involved in arc discharges. These discharges may damage
solar cells and produce electromagnetic interference. The vast majority of

arcing would be of a microsecond duration. The arcs are generally localized
near the coverglass edges and are believed to involve electron currents from
the conductor. ESD arcing between adjacent PV cells produces intense low
impedance plasma, leading to an array-driven current coupling between
adjacent cells as shown in the electrical circuit model in Figure 5.1. The
resulting ohmic heating damages the conducting path, possibly causing a
permanent short in the affected string. The plasma charge eventually
discharges in the space and reaches a balanced voltage. In a high plasma
orbit, this voltage is high.
Flight experiments have shown strong dependence of the arc rate on bias
voltage and plasma density and the existence of a threshold bias voltage.
Dielectric materials such as the coverglass or adhesive must be near a
highly negative conductor or semiconductor, and both components should
be in contact with space plasma. The charging rate in GEO is believed to be
determined primarily by the flux of electrons with energy in the order of 30
keV. The arcing rate is sensitive to temperature, and most GEO arcing
occurs just after coming out of an eclipse. The primary mechanism of arcing
in high voltage solar arrays is believed to involve differential charging. It
triggers an arc followed by a more energetic discharge between cells that is
sustained by the array power itself. If Kapton insulation on the array
substrate is carbonized, a permanent short and a loss in power generation
can occur. This risk can be mitigated by grouting adjacent cells to increase
the trigger threshold of the sustain discharge.
Figure 5.2 from Crofton and Francis1 shows data from a simulated test on
a solar array under a 20-keV electron gun and sunlight. As expected, the
substrate and coverglass could become charged to kilovolt levels. The data
FIGURE 5.1 Electrical circuit model of solar array string in plasma.

FIGURE 5.2 Aluminum substrate and coverglass charging under 20-keV electron gun in
sunlight.
were obtained by starting at the highest electron beam current, and

reducing it in steps at intervals of several minutes each. The conducting
substrate voltage decayed quickly with time, but the coverglass potential
was sustained due to the long time constant of the resistive coverglass once
the charging source was removed. Such differential voltage can arise
because the flux of electrons and protons in GEO can vary minute by
minute.
The spacecraft charging environment in LEO is primarily due to ions. It
consists of low energy plasma of less than 1 eV with a density ranging from
105 to 106 particles per cm3. The floating potential of the spacecraft is
affected by the bus voltage and the area of the solar array. The large area of
the coverglass floats near the plasma potential of a few volts. The spacecraft
structure, on the other hand, is driven negative to approximately the bus
voltage. Under this condition, the cells near the structure ground have high
voltage gradient across the coverglass and the cell, leading to arcing across
a small clearance (<1 mm) between the coverglass and the cell inter-
connects. The resulting relatively dense and highly conductive plasma
plume expands over the array and sweeps past the cells. The charge held in
the coverglass-to-cell capacitance is released in the conductive plume,
causing current coupling directly between the cells. The cell-to-cell coupling
is driven by current and voltage produced by the array strings, and can last
much longer than the initiating arc between the coverglass and the cell
interconnects. If sufficient string current is available, such sustained arc can
cause heating damage, leading to a permanent short circuit and power loss.
NASA tests have shown a sustained cell-to-cell arcing threshold of about
60 V bus voltage and 1 A string current. The solar array arc discharge
mitigation options are discussed in Chapter 8.
5.4 Damage to the Power Electronics

The charged particles can interact with sensitive regions of semicon-
ductor devices and degrade the electrical performance and even trigger
catastrophic failures. The photons, protons, and electrons cause gradual

degradation in performance with increasing radiation dose. The space
environment also contains high-energy protons, neutrons, and heavy ions
that can cause single-event displacement damage by introducing high-
energy transients in the sensitive region of the power MOSFET. The effect
can be a single-event burnout, single-event gate rupture or single-event
latch-up. The burnout causes permanent short circuit between the source
and the drain, rendering the MOSFET useless. The gate rupture causes a
permanent short between the gate and the drain. The latch-up can also be
caused by electrical transients on input or output lines, high temperature or
improper sequencing of power supply biases. Various mechanisms of such
damage are as follows:
Ionization damage: The electron–hole pairs are produced by ionization

in the device insulation or protective coatings. The charge thus
trapped degrades the electrical properties of the device.
Dose-rate damage: The electron–hole pairs produced by ionization over
a short duration produces a current transient. The resulting voltage
transients can upset logic or memory devices and can induce latch-
up.
Displacement damage: High-energy particles collide with silicon atoms
and displace atoms from their crystal lattice, producing defects
leading to degradation in electrical performance of the device.
The above three types of damage do not occur one after another in steps as
the radiation dose increases in a device. All three types may occur
simultaneously in a single device. The results of the damage can be
classified into two categories:
Failure under single-event radiation from cosmic rays

Gradual degradation under the accumulated dose of radiation
The worst-case cosmic ray radiation can cause the following types of failure
in semiconducting devices:
Single-event upset (SEU). This causes a change in the state of the stored
memory (bit-flip) in the semiconductor device. However, it does not
result in a damaged device.
Single-event latch-up (SEL). This causes a high current condition with
loss of stored data, inability to function, and even burnout. The device
must be powered off to restore the function.
Table 5.2 Types of damage in semiconducting devices due to charged particles
Device Ionization damages Displacement damage
Diodes Decrease in forward voltage drop Decrease in forward voltage drop

Increase in leakage current Increase in leakage current
Decrease in Zener voltage Increase in breakdown voltage
Bipolar transistors Decrease in gain Decrease in gain
Increase in leakage current Increase in leakage current
Increase in saturation voltage
Single-event secondary breakdown (SESB). This is caused by a very

high current due to secondary breakdown of parasitic transistors in
power MOSFETs.
The types of damage on various semiconducting devices under a dose of

charged particles are summarized in Table 5.2. Under radiation effects, the
diode forward voltage drop can decrease by 50 mV maximum, and the
leakage current can increase by 10 mA maximum. Generally, diodes used in
a metal enclosure do not pose radiation damage risk. However, they may be
damaged when placed unprotected outside the spacecraft, such as on a
solar array for cell bypass or string isolation purposes.
Many d.c.–d.c. converters use a power MOSFET because of its higher
switching speed relative to the bipolar transistor. The MOSFET also has
a lower potential of thermal runaway and has a simpler input drive
requirement. However, the power MOSFET is more susceptible to
degradation and catastrophic failure under radiation.
The radiation studies performed by Attia et al.2 are summarized in Table
5.3 and Table 5.4. The study showed that the effect of total dose radiation on
the Buck, Boost and Cuk converters were insignificant. However, all
converters lose their performance function under the worst-case single-
event failure. For these reasons, space qualified converters must use
radiation hardened power MOSFETs, which cost much more than their
Table 5.3 Reduction in converter output voltage at

various radiation doses
Radiation
dose (krad) Buck Boost Buck–Boost Cuk
0 0 0 0 0
4 negligible negligible 0 negligible
8 negligible negligible 69% negligible
12 negligible negligible 100% negligible
Table 5.4 Converter output voltage

after a single-event failure
Converter Output voltage
Buck Same as the supply voltage

Boost 0
Buck–Boost 0
Cuk 0
commercial varieties. However, improved circuit designs that can reliably

use commercial MOSFETs in space can yield significant cost benefits.
Given that the SEU bit-flips in computer memory, it induces data error
that propagates through the run-time operational flight program, causing
erroneous outputs from the flight computer system.
5.5 Effects on other Components

Ultraviolet radiation modifies the surface resistivity, opacity, and thermal
characteristics of the coverglass and its adhesive used in the solar array
construction. The result is reduced illumination and increased temperature,
both of which degrade the solar array output power. Since charged particles
also darken the coverglass, it is a usual design practice to allow a couple of
percent power degradation due to coverglass darkening.
Space missions to Jupiter involve radiation exposure exceeding 1 Mrad,
which corresponds to the intense radiation in the Van Allen belts. The RTG
power source used for Jupiter missions cannot deliver the peak power
needed during certain science-acquisition events, so batteries are used in 2
to 10 Ah ratings. In the NiH2 battery, the hydrogen electrode polymeric
compression seal contains TeflonTM, which has poor radiation tolerance,
and hence is not suitable for such a mission. Super NiCdTM batteries have
high radiation tolerance, and have been tested up to 20 Mrad.
The glass dielectric capacitors are high-reliability components suited for
any environment where radiation or high temperature (200 C) could alter
or destroy the circuit performance. They have been tested under neutron
and gamma radiation environments and have excellent transient and long-
term radiation performance characteristics. They are radiation hardened
and qualified to MIL-C-23269 and MIL-C-11272 for withstanding nuclear
radiation and high-voltage transients with little or no degradation. Their
performance relative to other types of capacitors in fast neutron flux fields
is shown in Figure 5.3. As the bar graph illustrates, glass capacitors can
operate in fields ten times more intense than any other capacitor type tested.
Furthermore, when exposed to nearly 10 108 rad, they exhibit minor
parametric variations immediately following radiation. They have low
FIGURE 5.3 Radiation hardness of various capacitors under integrated fast neutron flux.
dielectric absorption, zero aging rate, and a temperature coefficient

retraceability.
Outgassing of materials in a space vacuum can cause dielectric break-
down at voltages above the Paschen minimum. Proper material selection
and venting is needed to avoid such breakdown, particularly in high-
voltage systems. Any material can be used in space as long as the volatile
content is less than 0.1% and the total mass loss in space is less than 1%.
Permanent magnets do not degrade under radiation.
5.6 Mass Erosion under Atomic Oxygen

Atomic oxygen (AO) is a significant environmental constituent for space-
craft in a low Earth orbit. It is formed by photo-dissociation of diatomic
oxygen by ultraviolet radiation from the sun, more in LEO, where
recombination or the formation of ozone is improbable. It reacts with
many commonly used spacecraft materials resulting in degradation and
mass erosion of exposed spacecraft surfaces. The AO can oxidize and
damage surfaces, especially in the ram direction with the spacecraft ram
velocity simulating a high-energy beam. Volatile oxidation products may be
lost, leaving a surface without its protective coverings and this can lead to
electrostatic discharge to or from exposed conductors. Moreover, chemi-
cally active ions may be attracted by charged spacecraft surfaces, and the
reaction rates increase with energy. Such considerations are important
mainly for low planetary orbits, such as in LEO, and also in low Mars’s
orbit.
The AO has significant impact on the design and durability of the
materials. The following are significant characteristics of AO in LEOs:
Most abundant species in LEO atmosphere (80% AO, 20% nitrogen)

Average velocity is approximately 8 km/s
Average energy is about 5 eV
Fluence (atoms/cm2) is dependent on solar activity, velocity vector, and
altitude
Mechanism of material reaction is not well understood
The reaction rate on a material depends on the AO density, velocity vector,

and the material reactivity. Materials in order of high to low degradation
resistance are metal oxides, metals, inorganic polymers, and organic
plastics. Thus, shielding with metal oxides, such as oxidized aluminum,
can offer adequate protection against AO damage, provided that the
shielding is strong enough to resist micrometeoroid damage.
Sputtering may also be a concern for long duration LEO missions
carrying surfaces at high negative voltage. A complicating factor is that little
information is available on the nature of chemical effects on sputtering rates
under AO. However, some work has been done for developing a sputtering
model to help evaluate the surface material loss rates in LEO.
Measurements of low-energy sputtering by atomic oxygen support a
theoretical model. One distinguishing characteristic of the model is that
there is no absolute threshold for sputtering, but the rate at low energy is
determined by the thermal tail of the surface material atom velocity
distribution. The sputtering rate is a strong function of the surface voltage.
Because the spacecraft voltage is usually well under 1 kV, sputtering is only
important for mission durations in years in high-density plasma, such as
that in LEO or low Mars orbit. Sputtering may also be a long-term problem
in the lunar environment if the locally produced environmental plasma has
a high enough density. For long life missions in a high-density plasma
environment, surfaces should be insulated to prevent direct contact of high
potential conductors with the plasma. Earlier in the ISS designs, engineers
showed that the sputtering rate greatly increases by ions focusing onto the
pinhole where micrometeoroids, debris or manufacturing defects produce
small pinholes in insulators. Underlying conductors must be of sufficient
thickness to withstand puncture at accelerated sputtering rates. Materials of
low sputtering rate also may be used, provided that they satisfy thermal,
atomic oxygen, and other requirements. Optical, thermal or other specia-
lized coatings in the line of sight to the sputtering pinhole will become
coated with the sputtered material over a long period of time. It is desirable
to place surfaces such that they have no direct line of sight to surfaces that
might undergo sputtering.
Sputtering by ions in LEO is also possible. However, such a sputtering
rate per incident particle may be 0.1 and the flux on the order of 1012
per square centimeter per second. In contrast, reaction rates for atomic
oxygen may be 1.0 and the ram flux 1014 per square centimeter per second.
Thus, instead of a time scale of 10 years for sputtering, we might expect a

degradation time scale of a few days under atomic oxygen. Indeed, only a
few weeks of time in LEO is required to destroy 1-mil layer of a highly
reactive material like Kapton. However, not all materials are reactive to
atomic oxygen. For LEO missions longer than a week, surface materials that
are non-reactive with high-energy atomic oxygen must be used.
The spacecraft in LEO runs into the atomic oxygen with an impact energy
of about 5 eV. Both the impact energy and the high reactivity of the atomic
oxygen result in oxidation and erosion of most hydrocarbon polymers. The
erosion is commonly measured in terms of the volume of the material
removed per incident oxygen atom. Such measurements have been
performed on board many spacecraft and have been reported by many
investigators.3
The erosion rates of most polymers, except silicones and materials
containing metal atoms which produce nonvolatile oxides, vary widely
from 0.337 1024 to 6.3 1025. Such wide variations make it difficult for
the design engineer to know more accurately the durability of candidate
materials at the design stage. Moreover, new materials continuously emerge
which do not have such information due to expensive in-space testing.
Ground-based testing has been used with limited success in predicting the
in-space erosion rate based on the content and structure of hydrocarbon
polymers.
The erosion rate (e) is often stated in terms of the surface thickness loss
per unit time (i.e., mm/year). That is,
M
e¼ ð5:1Þ
A
where iM ¼ mass loss, A ¼ surface area, and ¼ mass density. The

erosion rate is also influenced by the direction of atomic oxygen impinge-
ment. Most spacecraft have surfaces that are exposed to atomic oxygen at
various angles during each orbit. Therefore, it is important to measure the
erosion rates under impingement from various directions. Banks et al.3 have
reported a study for some 40 materials commonly used in space.
The silicone insulation on wires and the solar array could be adversely
impacted by atomic oxygen. When exposed to atomic oxygen, silicon tends
to shrink, harden, and experience surface conversion of silicon to silica. This
causes surface erosion, which is more on the front surface than on the tail
surface. Long duration spacecraft, such as the Russian Mir space station,
have indicated significant deposits of silica on nonsilica containing surfaces.
Such depositions cause UV darkening, which changes the solar absorption
properties and degrades the solar array output power. Silicones intensively
investigated are DC-93-500 used to attach the coverglass to solar cells and
CV-1147 used between solar cells and the substrate on the ISS. Teflon shows
the worst radiation damage, whereas Kapton (polyamide) is much better.
Silver Teflon erosion rate has been recently reported to be 1 mil/year in the
ISS orbit. This rate is 10 times higher than some old estimates. In coating
other solids or films with Kapton, the coefficient of thermal expansion must
match; otherwise it may eventually crack. For LEO missions, it is
recommended that: (1) sulfuric acid anodization be used as an insulating
surface in regions where thermal control is not an overriding concern, and
(2) AO protected aluminized Kapton blankets (Kapton surface on the
outside) of 1000-V dielectric strength be used in all other outside surfaces of
a high-voltage spacecraft.
5.7 Micrometeoroid and Debris Impacts

The solar array gradually becomes damaged by micrometeoroids punching
small holes or shattering large surface areas of the cell. In the case of small
holes of a couple of millimeters diameter, the impacted cell would still work
without losing the whole string, but with a degraded performance. In
the case of a shatter, the cell fails open, losing the whole string. In the past a
few spacecraft have incurred such damage under micrometeoroid showers.
The damage is indicated by a sudden decrease in power output
immediately following a wobbling of the spacecraft.
Micrometeoroid and debris impacts may puncture insulators, fluid lines,
pressure vessels, and manned compartments. Such damage potential must
be accounted for in the spacecraft design. Studies of the expected flux of
debris and micrometeoroids of a certain size are uncertain. This is due to
the uncertainties in the amount of debris produced by abandoned space-
craft, its collision and breakup into smaller fragments, and models of the
atmospheric drag that eventually remove it from orbit. The SP-100 had
enough built-in redundancy to survive collisions with 1 cm size debris
particles. On the other hand, the ISS required a shield to protect its
pressurized manned modules against debris particles up to 1 cm in size.
In late 1998, the Earth passed through a Leonid meteor storm that
originated from the dusty passage of the parent comet Tempel-Tuttle. It
lasted about 10 h, and was one of the most intense meteoroid showers in
decades. To guard against possible damage, TDRSs and other satellites
were rotated to have the solar panels face the storm edge-on to minimize
the target areas to speeding meteoroids. The satellites were also powered
down for at least a part of the 10-h passage through the Leonid.
5.8 Predicting Damage

Radiation damage to spacecraft components can be predicted fairly
accurately now. For example, the damage found on the solar array of
MIR, when examined in 2000, was as predicted for the effects of radiation,
micrometeoroids, thermal absorptivity, and emissivity, for example. The
MIR solar array did not have internal bypass diodes, only string isolation
diodes. PV cells were silicon with 14% efficiency and 76% fill factor. No
arcing or soldering damage or open circuits were found. Some cells were
shorted due to micrometeoroid damage through a ceria microsheet cover-
glass. The array showed a high level of contamination, which was believed
to be there when built due to poor bake-out procedures.
The most damaging effects of the environment on the power system
occur during solar storms. The damage can be minimized if the intensity of
such storms can be predicted in advance. A new scale for rating the severity
of solar storms has been developed by NOAA. The storms are rated on a
scale of 1 to 5, with 5 being the most severe. The scale categories and their
effects on power system are given in Table 5.5. Satellite owners can take
protective measures as a storm approaches, including reducing power
levels on certain components that are vulnerable. For example, the solar
array can be rotated on the edge to reduce the surface area normal to the
radiation flux. Such precautionary measures are taken when required in
view of the predicted risk.
A new sensor under the trade name CEASE is available on the market to
monitor local radiation in the environmental and to provide an autonomous
real-time warning of the following:
Total radiation dose and dose rate

Surface charging
Single-event effects
Solar cell damage
The sensor can alert satellite operators to hazardous conditions before the
mission is affected. Under such an alert the operators can, in turn, re-
prioritize the operation, inhibit any anomalous sensitive tasks, such as
attitude control adjustments, or initiate other prudent actions as indicated
by the warning flag.
Table 5.5 Scale categories of space environment based on effects on spacecraft
Category Effects
S-5 Permanent damage to solar panels, loss of controls, serious EMI including
common mode noise
S-4 Accelerated degradation of solar panels, orientation problems
S-3 Damage to exposed components, decrease in soar array current
S-2 Isolated problems possible
S-1 None
References
1. Crofton, M.W. and Francis, R.W., ‘‘ESD measurements on solar cell
coupons in a simulated GEO environment,’’ Proceedings of the 35th
Intersociety Energy Conversion Engineering Conference, IEEE, 1999, Paper
No. 2634.
2. Attia, J.O. et al. ‘‘Radiation effects on dc–dc converters,’’ Proceedings of the
35th Intersociety Energy Conversion Engineering Conference, IEEE, 1999, Paper
No. 2696.
3. Banks, B.A. et al. ‘‘A space experiment to measure the atomic oxygen
erosion of polymers,’’ Proceedings of the 35th Intersociety Energy Conversion
Engineering Conference, IEEE, 1999, Paper No. 2695.
Chapter 6
Power System Requirements
6.1 Introduction
The overall mission requirements are set by the customer and given to the
satellite contractor in the form of mission specifications and statement of
work. The customer performance requirement specifies only the launch,
transfer orbit, and operational orbit constraints, and the payload power
demand over the mission life with specified power margin. This leads to the
spacecraft requirements. All systems of the bus derive their requirements
from the spacecraft requirements, and all components of a system derive
their requirements from the respective system requirements. The flow-
down of the requirements along with the design, manufacturing, assembly,
and testing is shown in Figure 6.1.
The spacecraft requirement specifications generally do not specify the
power system requirements per se. It merely includes a general statement
such as ‘‘ the power system shall generate, condition, distribute and store
electrical power to meet the requirements of the spacecraft payload and bus
systems during all phases of the mission.’’ To meet such a broad statement
of work, almost all requirements of the power system are self-derived based
on the design analyses and approach taken by the satellite manufacturer.
The basic satellite level parameters impacting the EPS design are the orbit
altitude, orbit inclination, and the mission duration. These parameters are
used to determine the orbit period, sunlight and eclipse durations, and the
solar angle between the orbit plane and the Earth–Sun line. The power
system design is then driven by the load power requirement in all phases of
the mission, which are as follows:
Launch and ascent. This phase extends from the lift-off to the sun
acquisition, during which the battery supplies the needed power.
Transfer orbit. During this phase, the geosynchronous spacecraft is
transferred from LEO to HEO and then to GEO with the solar array still
stowed. Only the outer panels are exposed to the sun and generate a
fraction of the rated power. The load power is, therefore, rationed. The
energy balance in this phase is desirable, but is not a rigid requirement.
A small energy deficit may be drawn from the battery up to the
allowable DOD as long as it is fully recharged before commencing the
full load operation.
105
FIGURE 6.1 Satellite and power system requirements flow down.
Parking orbit: Often this orbit is specified for storage until the satellite
is placed in service in the operational orbit.
Operational orbit: All solar panels are now deployed and generate full
power. In this phase, all required loads are on. The battery charge and
discharge energy balance must be maintained in each orbit.
Disposal orbit: A disposal method at the end of the life may be
specified, and the satellite may be required to function at a specified
power level in the disposal orbit.
Thus, the power system is required to perform during various phases of

the mission in a variety of orbits — LEO, HEO, and GEO. However, the
orbit period, the maximum eclipse duration, and the load power require-
ment in the operational orbit determine the ratings of the solar array,
battery, and other components. The power generation diminishes with
increasing radiation dose, more so in low and highly elliptical Earth orbits
which pass through the Van Allen radiation belts. For a given load power,
therefore, the solar array area requirement increases as the mission duration
becomes longer. The overall power system architecture is selected based on
the prior experience on similar satellites that have met similar requirements
in the most economical and mass effective manner.
6.2 Self-Derived Requirements

As mentioned earlier, the electrical power system requirements are seldom
found in the customer specifications for the spacecraft under procurement.
Power System Requirements 107
They are derived from the spacecraft level requirements and in-house trade
studies. The EPS self-derived requirements are based on various analyses
performed on the power system under design. The final requirements
generally come from the operational orbit analyses, but other orbits must
also be analyzed to ascertain that the proposed requirement is met in the
worst case. The power system design team commonly performs the
following worst-case analyses:
Power flow to determine the component ratings and heat dissipations

Energy balance to determine the battery rating
Voltages at the terminals of various equipment
EOL and BOL solar array power generation capability
Load switching and fault response, including major fuse blow event
Bus stability under various feedback control loops
Energy balance in transfer orbit
dV/dt specifications, which come from three sources:
Voltage fall before the fuse blow
Voltage rise after the fuse blow
Sudden loss of power during integration and testing
The following are some major self-derived requirements of the power

system:
Solar array EOL power level

Solar array pointing and rotation for sun orientation
Battery capacity, in ampere-hours
Battery DOD and charge control
Bus voltage regulation
EMI/EMC and ESD
6.3 System Specifications

The power system requirements vary over a wide range from low power
(few hundred watts) to high power (several kilowatts) in LEO, MEO or
GEO altitudes. The power system specification document is developed
based on the performance analyses with the mission parameters as input.
This is done well before the detail design starts. It is a well-drawn
document containing a level of detail matching the complexity of the
mission. A typical outline of such specifications is given in Table 6.1. The
complete requirement specification is typically a 100 to 200-page document,
Table 6.1 Power systems requirement

specifications — topical outline
1. Scope of the spacecraft

2. Applicable documents
3. Power system requirements
a. Interface with other systems
i. Functional interface
ii. Logical interface
iii. Physical interface
b. EPS characteristics and performance
i. Load capability
ii. Operating phases and modes
iii. Battery charge and discharge management
iv. Main bus voltage, normal and transients
v. Battery bus characteristics
vi. Station-keeping power
vii. External power input
viii. Command and telemetry
ix. Ground interface
x. Reliability
xi. Software
c. Major component characteristics and performance
d. Design and construction
4. Quality assurance
5. Documentation
which is generally unique to the particular spacecraft. This section lists

some of the most basic requirements for a PV–battery power system for an
Earth orbiting satellite. Such a mini-specification is generally used
informally or with some formality at an early stage. The blanks below
represent a number to be inserted, which is derived from the mission
specific analysis performed by the system engineer. Some blanks are filled
in italics, which are given merely as representative examples of a 50-V or
70-V GEO bus. They would change with the bus voltage and mission
requirements.
The electrical power system shall provide for the generation, storage,
control, protection and distribution of electrical power to the
spacecraft payloads and housekeeping loads during the entire
mission life of ______ years.
Operational orbit load: during sunlight _____ watts, during eclipse ____
watts.
Peak load of additional _____ watts for ____ minutes in every ________
minute period.
Pulse load of additional _____ watts for ____sec with duty ratio of 5
percent, with 1 ms turn on time and 1 ms turn off time.
Power System Requirements 109
Transfer orbit load: during sunlight_____ watts, during eclipse _____

watts, with solar array stowed and only the outer panel exposed.
The EPS shall provide sufficient capability to support the spacecraft
transfer orbit loads with the solar array stowed for (a) 9 hours when
the incident sun angle is in the range of 15 to 165 as measured from
the spacecraft spin axis, or (b) 30 hours under incident sun angle in
the range of 40 to 140 , or (c) indefinitely when the sun angle is in the
range of 60 to 120 .
Load of ____ watts during launch, ____ watts during separation and
spin, and ____ watts during spin and sun acquisition. Mission power
profile up to entering the transfer orbit as per Figure____.
Bus voltage_____ Vdc _____V, regulated direct energy transfer bus.
Single fault tolerant system where faults shall not propagate to other
components, subsystems or systems.
Autonomous redundancy changeover in case of a fault.
EPS Reliability of 0.95 over the entire mission life, including the ground
and space storage, if required.
Telemetry and command to monitor the power system operation as per
Table____.
Autonomous operation in all normal functions, unless overridden by
ground command.
Power management software via 1773 data bus to control the battery state
of charge/discharge, temperature, pressure, V-T curves, charge/
discharge balance between batteries and load shedding if and when
necessary.
Each non-critical load fused for the EPS protection.
Bus transients to comply with ____________ document.
EMC to comply with ___________ document.
Orbit environment to comply with ____________ document.
Launch environment to comply with ____________ document.
Survivable to additional threat as per _______ document.
Piece parts derated as per ________________ document.
Umbilical cord for ________ watts of ground power to the EPS bus and
for _______ amperes battery charging current.
Two batteries, each with ________ NiH2 cells with ____ Ah ratings.
Maximum worst-case battery DOD 70 percent during operational orbit
and 80 percent during transfer orbit.
Battery charge rates of 3 A and 5 A, and the trickle charge rate of 0.5 A.
Battery temperature to be maintained between 10 C and 25 C, with

5 C maximum cell-to-cell gradient and 5 C maximum inter-cell
gradient.
Ordnance controller powered from both batteries independently at
voltages between 18 and 34 volts.
The deployment ordnance controller shall contain relays (enable, arm
and fire) required to switch power from the batteries via independent
paths to the EEDs used for separation of deployable structures and
ignition of the apogee kick motor.
Bus dynamic impedance < 200 m from 500 Hz to 20 kHz and > 0.60 m/
Hz frequency for frequencies below 500 Hz.
EPS-induced ripples < 50 mV peak-to-peak from dc to 400 MHz, except
250 mV p–p between 30 kHz and 45 kHz. Total RMS ripple <120 mV.
Bus ripples at housekeeping loads due to effect of payload current
ripples < 500 mV p–p for frequencies up to 50 MHz.
Harness wires no thinner than 24 AWG. Aluminum wires shall not be
used.
Maximum voltage drop to users: 0.50 V for loads < 75 watts, 0.75 V for
loads >75 watts, 1.0 V to heaters and mechanism loads.
All load power conditioners must supply power to their loads with 10 ms
power interruption from the bus under a fuse blow transient or other
similar events.
The spacecraft power return bus to be hard grounded to the spacecraft
chassis at the power regulator unit.
Testability shall be provided at the EPS level and the spacecraft level.
Chapter 7
Design Process and Trades
7.1 Introduction
The power system design process starts with analyzing the orbit para-
meters, the load power requirement, and heritage data on similar satellites
built and flown earlier. The design that meets the mission requirements at
minimum mass, volume, and cost at present generally incorporates new
technologies as they become available and space qualified. The design
process starts with a top-level trade analyses — formal or informal — in
order to select the power system architecture that would lead to an
optimized design. The detailed design then follows and continues in several
iterations until it is fully optimized. The engineer would know that the
design is fully optimized when a small change in any one component
would result in a penalty of some sort at the power system level or at the
spacecraft level.
7.2 Spacecraft Level Trades

As expected, the power system interfaces with other systems of the
spacecraft as shown in Figure 7.1. The power system engineer works with
the chief engineer in doing the spacecraft level design trades. The ultimate
goal of the program is always the minimum mass and cost at the spacecraft
level. For that reason, while focusing on the power system, the power
system engineer keeps a peripheral vision on the overall spacecraft mass
and cost with applicable input from the spacecraft system engineer and/or
the chief engineer. Reducing 1 kg mass in one component that adds less
than 1 kg in another component is a winning trade. Such trades are
continuously revisited as new spacecraft technologies become available.
The following factors have a large influence on the architecture selection
of the PV–battery power system. They are driven primarily by the top-level
mission requirements and do not involve the power system level trades.
However, understanding their influence on the power system can quickly
lead to the selection of an optimum EPS architecture.
1. As the satellite altitude increases, the orbit period increases and the
eclipse time, as a fraction of the sunlight time, decreases. The demand
on the battery discharge and the charge power therefore decreases.
111
FIGURE 7.1 Power system interfaces with other systems of spacecraft.
2. The micrometeoroids and orbital debris are worse in the 50–55

inclination LEO orbits. The charged particle radiation environment is
also worst in this orbit over the South Atlantic Ocean where the Van
Allen belt dips to a low altitude. At higher inclinations in LEO, the
Earth’s infrared radiation decreases while the albedo (solar radiation
reflected from the Earth) increases.
3. The angle between the Earth–Sun line and the orbit plane varies
seasonally. It reaches maximum values on the summer and winter
solstice days, which is max ¼ i þ io, where i ¼ orbit inclination, and
io ¼ 23.45 (the angle between the equator and the ecliptic plane). As
increases, the eclipse time decreases, and the solar array temperature
also decreases. At higher , therefore, the EPS is capable of supplying
greater load, both during eclipse and also during sunlight. The eclipse
duration is the longest when ¼ 0 and the shortest or even zero when
is maximum.
4. The array pointed away from the normal sun-line reduces the solar flux,
increases the coverglass reflectivity and changes the view factor to
Earth affecting the array temperature. Typically, a 3 error in the
gimbals pointing and 2 in the array flatness, thus a total of about 5
error, is allowed in the sun pointing. The cosine law up to 30 governs
the power loss. In the 30–80 range, the loss is greater than that given by
the cosine law, and the array generates no power beyond 80 . This has
an impact on the transfer orbit maneuvering when the array could be
way off-normal.
5. A planner solar array with dual-axis gimbals requires tracking of the
sun-line to the array normal no more than . With single-axis gimbals,
the array must have a cant angle equal to the annual average angle. A
concentrator array requires two-axis sun tracking to maintain the sun-
line normal to the array surface.
Design Process and Trades 113
6. Two-wing design is effective in maintaining the balance of solar

pressure on the wings. One-wing design must compensate for the
offset between the center of mass and the center of solar pressure.
Otherwise, additional fuel is needed to maintain the spacecraft attitude.
7. Due to 1.5% eccentricity in the Earth’s orbit around the sun, the solar
flux in LEO and GEO varies seasonally about 3% with an annual
average of 1358 5 W/m2, where 5 W/m2 is an allowance for
measurement errors. The flux is near the average value on both
equinox days, maximum near the winter solstice and minimum near
the summer solstice. Thus, the solar array receives about 6% more solar
flux at the winter solstice than on the summer solstice day. However, it
generates only about 4% more power because of the increased
temperature under the increased flux. The 11-year solar cycle of ionized
particles has no effect on the solar flux.
7.3 Power System Level Trades

Once the spacecraft level trades are settled, the power system engineer
focuses on the internal EPS level trades that may reduce mass and/or cost.
The power system mass as a percentage of the satellite dry mass can range
from 25% in LEO satellites to 45% in GEO satellites. This mass must be
supported during launch and then transferred to the final orbit at high cost.
Saving of even a few percent of the power system mass can result in
appreciable savings at the spacecraft level. For example, the impact of the
electrical power system on the structural system, although not obvious and
direct, is strong.
The first task of the power system design engineer is to select the
optimum architecture. As established in Chapter 4, a direct energy transfer
sun-regulated low-voltage bus is generally optimum for satellites having a
load under a couple of kilowatts. For large high-power satellites, the fully
regulated high-voltage bus would generally lead to the optimized design.
For a shortl mission operating over wide orbit parameters, peak power
tracking architecture may be suitable. For an interplanetary or deep space
mission, nuclear power sources may be optimum. All these options must be
considered for the mission under design in order to arrive at the optimum
power system architecture.
Next, the bus voltage level and the power generation and energy storage
technologies must be jointly selected to optimize the total power. The major
driving factors are the payload power level, operating orbit parameters,
mission life, and number of satellites in the program procurement. For the
self-derived satellite level load requirement, the trade study is done to select
various key component for the power system, such as the PV cell,
coverglass thickness for radiation protection, array substrate, battery cell,
power converter topologies, etc. An example of such trades is displayed in

Figure 7.2. As the trade study proceeds from left to right, the selections
made are shown by continuing lines. Since the solar array and the battery
are two components that primarily contribute to the power system mass
and cost, they receive more attention and see more rapid technological
changes than other components. They impact not only on the EPS design,
but also on other spacecraft systems.
7.3.1 PV Cell Trades

The fundamental design requirement is to generate the required electrical
power at the end of life. The degradation of the solar array power output
under charged particles is a trade within the power system design that often
does not cross the system boundary. The PV cell selection depends on the
mission environment and the design life. Some cells, such as low substrate
resistivity cells, have high BOL power, but degrade fast, while others do the
opposite. The orbit environment and the mission duration strongly
FIGURE 7.2 Power system design trades flow-chart.

influence the cell selection. A cell that gives high BOL power may be
beneficial for a short life mission, whereas a cell that is radiation hard and
gives high EOL power may be beneficial for a long mission. Figure 7.3 is an
example of such a trade. The cell designated as A (10 cm base resistivity)
would be better for a mission which would accumulate more than 10 1014
MeV fluence over 15 years lifetime. It gives lower BOL power but offers
high radiation resistance. For a shorter mission, the cell designated B (2
cm base resistivity, such as used in EOS-AM spacecraft) would require
smaller array area since it has high BOL power but is not as radiation hard.
With the same PV technology, it is generally true that the cell that gives
high BOL power degrades faster, giving less power at the EOL. The mission
life thus plays a definitive role in cell selection.
For a LEO or GPS type MEO satellite, the radiation degradation in power
generation over 10 years life is about 40–50% for silicon cells and 25–35% for
GaAs/Ge cells. The additional cost of high radiation resistance cells may be
more than recovered in long duration missions in such high radiation
orbits.
For minimizing the mass of a long life satellite, attitude control reaction
jets that use chemical propellant are often replaced with torque rods and
reaction wheels. However, the torque rod performance is limited by the
Earth’s magnetic field and the reaction wheel performance is limited by the
angular momentum storage capability. Subsequently, reaction control jets
may be required to supplement the torque rod system in large disturbance
torque environment. This is especially true for low park orbits where drag
is high. In such situations, the cost of a high efficiency array with a small
area may be more than compensated by not requiring reaction jets to
supplement the torque rod.
FIGURE 7.3 PV cell selection trade for radiation resistance.

The power system must also provide reliable power to ascertain a

targeted disposal, especially if electric thrusters are used for disposal
maneuvers. As for the end of life disposal of the satellite, international
organizations may establish stricter regulation in terms of the timelines,
disposal orbits, reliability, and liability for disposing of the satellite. The
array area having high drag may directly impact the disposal time and
reliability, and may cause the satellite to lose attitude control.
7.3.2 Solar Array Trades

A mass reduction in the solar array has multiple benefits at the satellite
level, such as reduced drag, inertia, and propellant mass. Such win–win
trades must be thoroughly investigated at an early design stage. The solar
array area is linearly proportional to the required power generation. At the
same time, it also has significant influence on the propulsion and attitude
control systems. A large array increases the drag and moment on the
satellite, more so in LEO satellites and in low Earth park orbits of
geosynchronous satellites. A small array of high efficiency cells might be
worth the added cost, considering that it would allow lower park orbits for
the same drag. A lower park orbit allows the launch vehicle to insert more
mass into the orbit, resulting in a reduced launch cost per kilogram of the
payload. A small array may also result in a shorter transfer time from the
park orbit to the geosynchronous operational orbit as a result of lower drag
in the early part of the transfer orbit. This may yield greater benefit,
considering that the transfer from LEO to GEO may take hundreds of days
with large arrays. Aligning the center of pressure and center of gravity of
the satellite body can minimize the drag.
If a narrow orbit control is required in LEO, a large array may add
significant propellant mass in a long duration mission, costing probably
more than the savings in using low efficiency PV cells. In GEO, a large array
causes orbit drift due to solar pressure. Therefore, the satellite requires
more propellant fuel for N–S station keeping in operational orbit during the
entire mission life.
If a precision attitude control is required, it is necessary to have a low
mechanical natural frequency of the solar array in the first mode of
vibration. Small arrays inherently have high frequency. A large array, on
the other hand, would require adding structural rigidity and decoupling at
an additional cost. The torque compensation hardware must be sized for
drag and solar pressure. The torque is generally canceled by changing the
reaction wheel speed, and then by de-spinning the wheels using torque
rods over the duration of the orbit. Thus, a large array can increase the size
of the torque rods and the reaction wheels.
7.3.3 Battery Trades

LEO and GEO satellites differ significantly in their power system
requirements. GEO satellites have a high launch cost and require fewer
battery charge/discharge cycles. LEO satellites, on the other hand, have a
low launch cost, high ionized radiation, and a much greater number of
battery charge/discharge cycles. The battery chemistry selection may,
therefore, differ significantly for these two classes of satellite. It would also
influence the thermal control system. The NiCd battery was the workhorse
of the space industry until the mid-1980s, when the NiH2 battery replaced it
for both LEO and GEO missions. The NiH2 battery, however, requires a
narrow operating temperature range of –10 C to þ10 C. This adds some
mass to the thermal control system. There are various derivatives in which
the basic NiH2 technology is available, commonly known as the IPV, CPV,
SPV, and DPV cells as described in Chapter 9. There is only a small mass
difference between each type. The lithium-ion battery has recently made
inroads into the space industry for its significantly higher specific energy
and is expected to replace the NiH2 battery in the near future. However, it is
sensitive to over-charge and over-discharge. This often adds some mass in
the charge control electronics.
7.3.4 Bus Voltage Trades

For communications satellites, selecting one of the standard busses offered
by prime contractors may be the most economical option. The bus voltages
are standardized for various ranges of the load power. However, the
voltage selection may be open for designing a special science or defense
spacecraft. Even in such cases, the bus voltage is sometimes set by the
heritage and the preferred database. Otherwise, a trade study is needed to
select the optimum voltage for the mission on hand. Developing pre-
liminary designs for various available voltages is the only way to arrive at
the optimum. The mass of each component is estimated under various bus
voltage options and added to the total power system mass and volume. As
for the cost, it may be assumed that the cost is proportional to the mass, at
least in the first approximation. Figure 7.4 is an example of such trade for a
2000-W TDRS-type GEO satellite with a 15-year mission life. Notice that the
solar array mass is not shown in the chart, but is included in the total mass.
As it is a series–parallel assembly of numerous PV cells, it is fairly
independent of the bus voltage. In a high-voltage bus, more PV cells are
connected in series and less in parallel, but still the same number of cells is
needed to generate the required power. For this example mission, the figure
indicates that a 50-V or 70-V bus is the optimum. The 100-V bus reduces the
harness mass, but it also increases the battery mass and volume
significantly. This is because the battery ampere-hour rating in a 100-V
bus becomes so small that its energy storage per kilogram mass and liter of
FIGURE 7.4 Bus voltage trade example for a 2000-W GEO satellite.
volume fall significantly. As a result, a 100-V bus does not reduce the total
mass compared to a 50-V or 70-V fully regulated bus.
7.3.5 Pyro Power Trades

In the case where a fully regulated direct energy transfer bus has been
selected, another top level trade is whether to power the EED pyro circuits
directly from the regulated bus or from the battery, bypassing the battery
discharge converter. This trade is depicted in Figure 7.5. It is not for cost or
mass reduction, but for the EMI and reliability improvements as described
hereafter. Both the primary and backup converters for each battery are
normally active during operational orbit. During launch, the battery
discharge converter can be off if the pyro power is supplied directly from
the battery bus. However, it must be on if the pyro power is supplied from
the main bus. In this case, the satellite is launched with the backup
converter on and the primary converter off. The reason for launching in
such backup mode is design related; that is, switching from the primary to
backup converter is automatic, whereas switching from the backup to
primary converter happens only on ground command.
The pyro firing during launch is generally a source of high EMI that may
cause switching transients, which may disturb some critical components
and sensors. The primary unit under severe transient glitches may get
turned off, and the systems may automatically switch to the backup unit. If
this happens, it is not possible to switch back to the primary unit. This
amounts to losing the redundancy in this critical function. If the satellite is
launched in the backup mode, an automatic switching to the primary mode
under a transient voltage glitch does not amount to losing the redundancy.
FIGURE 7.5 Powering EEDs by battery bus versus main bus.
If launched in the backup mode, it is switched to the primary mode by

ground command shortly after the orbit acquisition, on the third or fourth
revolution.
7.4 Load Power Profile

Once the satellite level trades are settled, and the EPS architecture has been
selected, the next step in the design process is to compile a detailed load
power profile of each component to be powered. It is also called the power
budget or the power allocation in an early phase of the design. The input
power for the on and off durations of each item of equipment ’peak power
when on, quiescent power when off, and the thermal dissipation are
compiled. The component dissipation may vary significantly during
sunlight and eclipse, and during the beginning of life and end of life.
Such variations are properly accounted for.
The power versus time curve for each item of load equipment may be
condensed into two parameters: the peak power and the duty ratio. The
duty ratio is defined as the ratio of the average power to the peak power
over a certain time period. For a well-defined load, which is on or off, the
duty ratio is merely the ratio of on-time to the period of repetition, that is
D ¼ Ton/T. For a continuously varying load at the other extreme, it is
defined as
energy consumed in one repetition period

duty ratio ¼ ð7:1Þ
peak power repetition period
i.e.,
RT
PðtÞ dt
o
D¼ ð7:2Þ
Ppk T
The average power required by any equipment is then Pavg ¼ DPpk. Figure
7.6 is a typical communication load profile for a geosynchronous satellite.
The load reaches high peaks of short duration when the transponders are
communicating with the ground.
The average power is used to determine the solar array and the battery
ratings. On the other hand, a load averaged over five thermal time constants
of the equipment is used to determine the equipment power rating, the wire
gage, and the fuse size. The peak power and duty ratio of most equipment
are significantly different during the launch, the transfer orbit, and the
FIGURE 7.6 Typical communications satellite load profile.

Table 7.1 Seasonal variations in power generation and heater power

requirement
Seasonal day Solar fluxa Sun angle Power generationb Heater load

Vernal equinox 1.001 0 1.001 High
Summer solstice 0.967 23.5 0.887 Low
Autumn equinox 0.995 0 0.995 Low
Winter solstice 1.034 23.5 0.948 Medium
a
Normalized with the annual average.
b
Accounting for seasonal variations in both solar flux and sun angle.
operating orbit. They may also vary seasonally during the operating orbit.
The worst-case seasonal power situation is sometimes known from a similar
satellite designed and flown in the past. When there is no such heritage to
draw from, considering 4 key days in the year, as listed in the first column
of Table 7.1, may be adequate for a geosynchronous satellite. On these 4
days, both the power generation and the heater power requirements vary.
Since the energy balance must be maintained all the time in the operating
orbit, any one of these 4 days may most probably be the worst case for
setting the power system capability. Because the heater power requirement
is low when the power generation is low, it may not be obvious which day
is the worst for the power system design. In the absence of some heritage
data, it may be necessary to conduct a preliminary design for all 4 days to
determine the worst case, and then to focus on the worst case for detail
design. The summer solstice is usually the worst day for the load power
capability of the GEO communications satellite. After establishing the worst
day, the design process focuses on meeting the load power requirement on
that day.
The load requirement of each of the bus systems and the payload
components are then complied during the transfer orbit and the worst-case
operating orbit. Table 7.2 and Table 7.3 are examples of loads and their
characteristics in one satellite during the transfer orbit and operational
orbit, respectively. The load power values in these tables are input powers
to the components that must be supplied by the power system. In new
Table 7.2 Transfer orbit load power profile for System A
Component 1 Component 2 Component n System total
Quantity X1 X2 Xn XTO
Peak power Y1 Y2 Yn YTO
Peak heat dissipation Z1 Z2 Zn ZTO
Duty ratio D1 D2 Dn DTO
Average power P1 P2 Pn PTO
Average loss (dissipation) L1 L2 Ln LTO
Table 7.3 Operation orbit load power profile for System A
Component 1 Component 2 Component n System total
Quantity X1 X2 Xn XOP
Peak power Y1 Y2 Yn YOP
Peak heat dissipation Z1 Z2 Zn ZOP
Duty ratio D1 D2 Dn DOP
Average power P1 P2 Pn POP
Average loss (dissipation) L1 L2 Ln LOP
designs when the loads are somewhat uncertain, appropriate margins are
incorporated in the power estimates. For example, it may be acceptable to
the design team to have no margin for heritage components requiring no
change at all, 10% margin for components modified from a heritage design,
and 30% margin for new component yet to be developed.
Various system loads obtained from Table 7.2 and Table 7.3 are then
condensed into the total power system load budget as shown in Table 7.4.
All bus systems are to be included in this table, including the payload and
the housekeeping load for the electrical power system itself.
7.5 Solar Array Sizing

The load power profile developed above is then used to determine the solar
array output power requirement. For example, for a communications
satellite having a 2500-W communication payload power, the solar array
power generation requirement is shown in Table 7.5 in two seasons and
during an eclipse. The autumnal equinox day, having the longest eclipse, in
this case determines the solar array output requirement of 3940 W, which
includes a 5% power margin and a 5% reliability margin. Allowance must
also be made for array shadowing and failed bypass diode, if applicable.
The solar array must generate this power at the end of life. The beginning of
Table 7.4 Total average load and heat dissipation budget (allocation) for power
system
Transfer Operational Transfer orbit heat Operational orbit

System orbit power orbit power dissipation heat dissipation
System A PTO POP LTO LOP

System B .. .. .. ..
System Z .. .. .. ..
Payload .. .. .. ..
Spacecraft PTO POP LTO LOP
total load
Table 7.5 Example of sizing solar array output power for a communications
satellite (average watts)
Autumn Summer During longest

equinoxa solsticeb eclipse
Communication payload system 2500 2500 2500

Communication system load margin (10%) 250 250 250
Total communication system load 2750 2750 2750
Bus systems
TT & C 50 50 50
Attitude control 50 50 50
Thermal control 150 100 50
Electrical power 20 20 20
I2 R conduction loss 50 50 60
Battery charging 500 100 0 (disch.)
Total EPS load 3570 3150 2980
Power margin (5%) 180 160 150
Reliability margin (5%)c 190 170 165
Total solar array power 3940d 3480 3295
generation required at EOL
a
Non-eclipse season near September 21.
b
Eclipse season near June 22.
c
Allows one string failure (in a typical 20 circuit solar array).
d
Solar array output power required ¼ 3940 W minimum.
life power would be greater depending on various degradation factors,

which are:
Environmental degradation due to ionized radiation

Cell-to-cell current mismatch (assembly loss)
Sun angle off-pointing angle (due to array flatness, structural misalign-
ments, and sun-tracking errors)
All other, such as micrometeoroids, space debris, contamination,
plasma, ultraviolet radiation, yellowing of coverglass, and thermal
cycling
The values of individual degradation factors come from the considera-

tions detailed in Chapter 8, and the total degradation is obtained by
multiplying all those factors. If the total degradation comes to 33%, as an
example, then the solar array must generate a BOL power of 3940/
(1 0.33) ¼ 5880 W in order to meet the 3940 W load requirement at EOL.
This is the Pmax it must deliver at its knee point voltage Vmp with
appropriate margins both in power and also in voltage. The Pmax is
temperature dependent, so we need to know the array operating
temperature. The temperature is calculated by equating the heat input
and output for the front and back of the array. All cells on an array are
assumed to operate at the same temperature. The heat input is the solar
radiation, albedo, Earth infrared radiation, and from the adjacent solar
array. The heat output is the power converted into electricity and the power
radiated back into space. If multiple wings connected in parallel operate at
different temperatures, the operating voltage is selected to be a few volts
less than the Vmp of the higher temperature wing to avoid the colder array
operating above Vmp where the power output drops rather quickly.
Assuming that the solar array output power calculated above is at the
shunt dissipater interface, where the calculated voltage is, say, 110 V in a
100-V bus, the BOL design power of 5880 W would require the solar array
output current of 5880/110 ¼ 53.5 A. Once the array output voltage and
current are determined, the array design proceeds as described in Section
8.9. The number of strings in the array and number of cells in each string are
then determined as follows:
Number of series cells per string ¼ array output voltage (110 V in the
above example)/cell voltage at the operating point
Number of strings in the array ¼ array output current (53.5 A in the
above example)/cell current at the operating point
7.6 Battery Sizing

The detailed battery design procedure is described in Chapter 9. At a
preliminary design stage, the round trip energy efficiency approach can be
used to derive the approximate rating of the battery. Both NiCd and NiH2
batteries have a round trip energy efficiency of about 70%. For the life
considerations, the maximum DOD must be below a certain value that
depends on the number of charge/discharge cycles the battery is required
to deliver during the mission life. With these considerations, the ampere-
hour rating is derived from the expressions below:
Battery cells in series ¼ minimum permissible battery voltage for pyro

or power electronics/minimum cell voltage at the end of discharge at
EOL
Wh discharged ¼ eclipse load in watts eclipse duration in hours
Ah discharged ¼ Wh discharged/1.25 average cell voltage cells in
series
Ah capacity ¼ Ah discharged/permissible DOD for required cycle life
Wh charge ¼ Wh discharged/0.70 round trip energy efficiency
Charge power ¼ Wh charge/charge time during sunlight (7.4)
7.7 Power Flow Analysis

The energy collected during sunlight must match the energy required by
the loads, plus the energy used to charge the battery and the power losses in
various components. Such a balance must be maintained over the orbit
period with the worst-case eclipse duration. In the simplest case with
constant load during the orbit (Figure 7.7), the two shaded areas A1 and A2
must be equal in each orbit period during the entire mission life. A small
deviation from such a stringent requirement in an occasional orbit could be
manageable for a short time, but must be thoroughly examined for
acceptability from the battery DOD considerations.
The design optimization is done to maximize the power output to the
load while maintaining the energy balance. The start point of the energy
balance analysis is usually the beginning of the eclipse. The battery SOC is
determined at the end of eclipse based on the load and the power losses.
The design engineer then determines the required solar array output to
bring the battery back to full SOC while meeting the load power during
sunlight. During the entire orbit period, the engineer also determines the
maximum power, voltage or current handled by each component, which
then sets the component power, voltage, and current ratings. The best
design is obviously the one that leaves minimum margins in all
components, and also is acceptable to the program team for the level of
uncertainty in the design or the system operation. At the minimum, the
system should not be battery power limited or array power limited.
Typically, a few percent margin in both the battery and the solar array is
considered adequate.
The spreadsheet shown in Figure 7.8 illustrates the power flow necessary
to maintain the energy balance for an EOS-AM satellite requiring 3000 W
constant load during sunlight and also during an eclipse. It may use
flywheel energy storage instead of a battery, but that would not change the
analysis method. In addition to the power generation (solar array) and the
energy storage (battery), it includes other components, such as the power
FIGURE 7.7 Energy balance over one orbit period with constant load.
FIGURE 7.8 Power flow and energy balance analysis example for a 3000-W LEO satellite.
electronics and distribution harness. Various power electronic converters

are used to change the voltage as needed at various points in the power
system. Their efficiencies are accounted for in the systems model. The
power consumed in the control electronics, such as the bus voltage
controller; bus voltage error amplifier, are included in the efficiency
numbers.
The power flow analysis determines the voltage and current throughout
the power system, and establishes the power handling capability and heat
dissipation of various components. It computes the power through various
components, based on the component efficiencies marked on the spread-
sheet. The converter efficiency, harness resistance and diode voltage drops
are factored in the computations. The current, voltage, and power loss in
the distribution harness are also determined in each segment of the system.
There is always an excess power generation, more so at the beginning of
life. The shunts, therefore, are required to bypass the excess power to the
ground, or leave on the solar array itself. In the design calculations at the
end of life, the shunt power should be equal to the power margin plus the
reliability margin.
The result of an energy balance analysis example for a 3000-W load LEO
satellite on a spreadsheet is depicted in Figure 7.8. As seen in the lower half
of the figure, the solar array must generate 5839 W and the battery must
have an energy storage rating of 6804 Wh. An 80-cell NiH2 battery would
require a 69-Ah rating to support a 3000-W load during an eclipse. The
assumptions used in this example are listed on the spreadsheet. The
spreadsheet algorithm is derived from the design methods developed in
various chapters of this book.
For a given EPS design, such a spreadsheet is also typically used to
perform the following two types of analyses:
Forward analysis to establish the EPS load capability, assuming that the
battery must be fully charged at the end of the sunlight in each orbit.
Backward analysis to verify whether a given load profile can be
supported without exceeding the component ratings, assuming that the
power can be drawn form the solar array and the batteries within the
permissible DOD during that time period. This is often done to assess
the peak power capability during a planned or contingency operation,
or when the load profile is not flat during the loading period.
7.8 Design Analyses List

Various design analyses are conducted side by side as the design matures.
They may confirm or change the design in order to meet the required
specifications under all normal and abnormal conditions. Table 7.6 is a list
of such analyses that are typically conducted as applicable for a given
mission.
7.9 Worst-Case Error Margin

Various uncertainties in input parameters may introduce probable errors in
the design estimates. Such errors are managed by providing appropriate
margins. If no guidance is available from the management, a conservative
engineer may stack all probable errors to derive the total error that is the
sum of all individual errors. The probability of occurrence of such a total
error is incredibly slim, of course. In situation where life safety in not
Table 7.6 Typical design analyses required for

the power system design
Design analyses performed:

Launch and ascent power
Transfer orbit energy balance
Operational orbit energy balance
Component dissipations
Stability and transient analysis
Transient due to step load changes (on and off)
Transient during fault and recovery (fuse blow)
Ripple analysis due to repetitive load pulses
Stability margin and sensitivity analyses
Battery latch-up analysis for sunlight regulated bus
Each of the above analyses is performed at:
Beginning of life with
Minimum load
Maximum load
Nominal and failure conditions
End of life with
Minimum load
Maximum loads
Nominal and failure conditions
involved, it may be acceptable to use the root sum squared (RSS) error
calculated as follows.
If errors in individual performance parameter are E1, E2, . . . with various
input parameter uncertainties, then the total error stacked in series is
Estacked ¼ E1 þ E2 þ . . . . However, the most probable error is given by the
RSS of the individual errors, i.e.,
sffiffiffiffiffiffiffiffiffiffiffiffiffi
X n
ERSS ¼ E2i ð7:4Þ
1
As a simple example, consider a design having four probable sources of

error, each 2%. The total stacked error is 8%, whereas the RSS error
calculated as above is only 4%, or one half as much. The difference
between the stacked and the RSS errors widens with a greater number of
error sources. In another example, say we wish to estimated the most
probable variance from the expected performance of an integrated circuit
under three damaging effects, namely the temperature, aging, and radiation
of charged particle. The procedure of estimating the most probable RSS
error is as follows:
Estimate the component performance with the normally expected

values of the operating stresses.
Then determine the deviation in performance under possible sources of

uncertainties at their worst-case levels individually.
Now, determine the total worst-case deviation in performances due to
all uncertainties combined using the RSS variance method, i.e.,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
total ¼ 2radiation þ 2temp þ 2aging ð7:5Þ
If the most probable worst case deviation thus calculated is unacceptable

with some margin, appropriate design changes must be made to bring it
under the acceptable level by either decreasing the temperature, or
increasing the device hardening, or both.
7.10 Design Process Phases

In today’s competitive environment it is not enough to ensure that the
system and the equipment are designed to be capable of meeting high
standards of performance and reliability. A consideration of manufacturing
processes and expected tolerances is part of good and complete design. A
thorough and correct design the first time is of paramount importance for
maintaining the program schedule and cost. The following Rule of Ten can
be a strong motivator to the design engineer: Not doing something right the
first time costs ten times as much to find and fix each time it escapes to the
subsequent stage from the design to the delivery. This rule results in the
following string of cost multipliers of fixing an error at various phases of the
total design and delivery cycle:
Design Production Test Launch pad
1 10 100 1000
The engineer must also detect and correct the design early enough for
potential manufacturing and test problems. This would minimize, pre-
ferably reduce to zero as a goal, the design changes, test failures,
manufacturing disruptions, trouble shooting, and re-work. For such cost-
effective implementation of the design, all engineers — design, manufac-
turing, test, and field engineers — work in the total team on the total satellite.
Only such a concurrent design team can ensure that the design of each
system is compatible with all other systems, and also with the manufactur-
ing and test procedures. For this purpose, the design process incorporates
various reviews as it proceeds from the contract award to the delivery. A
typical design phase sequence is listed in Table 7.7.
Table 7.7 Power system design, analyses and review

phases
1. Requirements allocation and conceptual design phase

Requirements definition/allocations
Review of system requirements
Top level design trades and conceptual design
2. Preliminary design and analysis phase
Component specifications
Preliminary design and analysis
Preliminary design review
3. Detail design and analysis phase
Detail design and analysis
Critical design review
4. Support phase
Support component and system tests
Support outside suppliers
Support satellite integration and test
Prepare operations manuals
Support launch
5. Continuing tasks
Cost and schedule control through the spacecraft delivery
7.11 Factory-to-Orbit Events

In the entire design process, the engineer stays focused on what the power
system is required to do not only in the operational orbit, but also during
the factory tests, on the launch pad, and during the orbit acquisition. The
following are some typical functions and requirements in various phases of
the satellite, which the power system engineer must incorporate in the
design:
Storage: Unforeseen circumstances may require a partially built space-

craft to be placed in storage, typically 3 years maximum in a
controlled environment.
Ground tests: Solar array and battery may not be installed, requiring
external power for testing via ground enable plug. Workhorse battery
may be used during testing, which may not be an exact duplicate of
the flight battery in performance.
Prelaunch: External or internal power is used, flight batteries may be
installed, and solar array is stowed.
Launch: Battery power is used at high discharge rate, and the system
operates through the launch environment of shocks and vibration.
Transfer orbit: Solar array generates limited power. The battery powers
the pyro bus loads only. Redundancy management is enabled.
Deployment: Partial or full solar array power is available after

deployment. Only the bus and the pyro loads are powered.
Pre-operational drift orbit: Full power generation, but still low loads.
Non-operational on-orbit: Full power generation, but less then full load.
Supports eclipse and battery charging.
Normal operational orbit: Full power generation and full load. The
eclipse and battery energy balance is maintained in every orbit.
Station keeping: Full power generation and full load. In addition, the
battery supports arcjets during N–S station keeping.
Reacquisition: Loads shed by ground commands, arrays slewed back-to-
back if this mode persists for more than 30 min or so.
Deactivation or de-orbiting: Power system supports this maneuver as
the last function. Then the power system role is over.
7.12 Power System Functions Over Life

The power system engineer must also be cognizant of the major satellite
level events from the pre-launch to the customer hand-over. A typical
sequence used for DMSP satellites follows:
Prelaunch: The spacecraft is on the launch pad when the decision is

made to launch or abort. The system is typically single-string, so
potential single point failures are identified. A launch decision
proceeds as follows:
The power system is on and the satellite goes on the internal power
about 10 min before lift-off, when the solar simulator is turned off.
Both batteries are on to supply power during the final 10 min before
lift-off and during ascent. No charging can occur until the satellite
is deployed and faces the sun.
The power converters are on and in discharge mode. The hardware
detection and switches are enabled. Software-controlled redun-
dancy is typically off.
The array drive electronics is on, but the array is not driven. All
components drawing power directly from the battery are on, except
turned off by an internal switch.
Ascent mode: The power system is on, but has no change from the
prelaunch phase.
Pre-handover: Two redundant cutters for each of the two bands are cut
to deploy the solar array. Both bands must cut for the deployment to
occur. Two cutters for the boom hinges and dampers are activated.
Redundant pin pullers activate the solar array cant angle.
Handover: The spacecraft control computer initiates handover based on
time as adjusted during the flight.
Orbit operation: Power system maintains the energy balance in each
orbit. The satellite is fully operational.
Part B
PV–Battery System
Chapter 8
Solar Array
8.1 Introduction
The solar array is made of numerous photovoltaic (PV) cells connected in a
series–parallel combination to obtain the required voltage and current. The
fundamental characteristics of the PV cell — the building block of the array
— is covered first in this chapter before moving on to the array construction,
performance, and design.
8.2 Photovoltaic Cell

A photovoltaic cell converts sunlight into direct current electricity. The PV
effect was discovered in 1839 by the French physicist Antoine Henri
Becquerel. It remained in the laboratory until 1954, when Bell Laboratories
produced the first silicon solar cell. It powered the first Russian satellite in
1958, followed by applications in the U.S. space programs. Since then, it has
been an important source of power for satellites because of its high power
output per unit mass. Having developed maturity in space applications, PV
technology is now spreading into terrestrial applications ranging from
powering remote villages around the world to feeding power grids. The
installed capacity of terrestrial PV systems worldwide in 2004 was over
3000 MW. Almost 40% of the PV panels installed in the world are produced
in the U. S. Unlike the dynamic systems using turbine or engine driven
generators, the PV array is static, produces no vibration or noise, and needs
no active cooling.
The photovoltaic effect is the electrical potential developed between two
dissimilar materials when their common junction is illuminated with
photons (Figure 8.1). When the junction absorbs light, the energy of the
absorbed photons is transferred to the electron system of the material,
resulting in the creation of charge carriers that are separated at the junction.
The charge carriers may be electron–ion pairs in a liquid electrolyte, or
electron–hole pairs in a solid semiconducting material. In either case, they
create a potential gradient (voltage), are accelerated under the electric field,
and circulate as electrical current if the external circuit is closed. The
product of the voltage and current in the circuit is the photon power
converted into electricity. The power that is not converted into electricity
remains at the junction and elevates the cell temperature.
135
FIGURE 8.1 Photovoltaic effect converts sunlight into electricity.
The PV cell in practical use consists of a p–n or n–p junction of two

semiconductors with a terminal for each. The physics of the PV cell is very
similar to the classical junction diode. The origin of the photovoltaic
potential is the difference in the chemical potential — called the Fermi level
— of the electrons in the two isolated materials. When they are joined, the
junction approaches a new thermodynamic equilibrium. Such equilibrium
can be achieved only when the Fermi level is equal in the two materials.
This occurs by the flow of electrons from one material to the other until a
voltage difference is established between the two materials, which is just
equal to the initial difference of the Fermi level. Absorption of photon
energy in the silicon cell knocks out an electron from the atom, leaving a
positively charged hole. Free electrons return through an external circuit
under the potential difference delivering power to the load.
From the solid-state physics point of view, the cell is basically a large area
diode with the junction positioned close to the top surface. Early cells were
p–n type with p layer facing the sun. They degraded at a rapid rate in the
Van Allen radiation belts. The cell design was changed to n-on-p to improve
the radiation resistance. The new n–p cell extended the mission life over a
decade.
The basic cell construction is shown in Figure 8.2. The metallic contacts
are provided on both sides of the junction to collect electrical current
induced by the impinging photons on one side. A conducting foil or solder
contact is provided all over the bottom dark surface and on one edge of the
top illuminated surface. Thin conducting mesh on the remaining top
surface collects the current while allowing the light through. The spacing of
the conducting fibers in the mesh is a matter of compromise between
maximizing the electrical conductance and minimizing the blockage of the
light. In addition to these basic features, several enhancements are also
included in the construction. For example, the front face of the cell has an
anti-reflective coating to absorb as much light as possible by minimizing the
reflection. The protection against damage due to energetic particles hitting
the surface is provided by the coverglass applied with a transparent
adhesive.
Solar Array 137
FIGURE 8.2 Basic construction of PV cell with performance enhancing features.
The series resistance is one of the limiting factors in determining the

conversion efficiency of PV cell. It is influenced by the current collecting
grid thickness (usually less than 1 mm), and the semiconductor structure.
The cell manufacturer undertakes a multi-dimensional optimization to
achieve the best design. The space qualified cell efficiency is quoted at air
mass zero (AM0), which represents the condition in outer space. The
voltage developed and the fraction of the incident energy converted into
electricity depends on the characteristics of the junction material and the
wavelength of the radiation.
8.3 PV Technologies
The most important measure of the PV cell performance is the photoelectric
energy conversion efficiency and the cost per watt capacity. Together, these
two parameters indicate the economic competitiveness of the photovoltaic
with alternative power generation technologies. The conversion efficiency,
, of the photovoltaic cell is defined as
electrical power output

¼ ð8:1Þ
solar power impinging the cell
Theoretical maximum energy conversion efficiency for three common

materials is 16% for germanium (Ge), 24% for silicon (Si) and 29% for
gallium arsenide (GaAs). The energy conversion is different from the
energy absorption. Not all energy that is absorbed by the cell is converted
into electricity. The sunlight wavelength spectrum is very important.
Certain semiconductors have a cut-off wavelength, such as 1.1 mm in the
silicon cell. Photons with longer wavelengths are unable to generate hole–
electron pairs. They pass through the p–n junction and are absorbed at the
bottom of the cell in the form of heat. The effect of changing the wavelength
of illumination is shown in Figure 8.3. It may be noted that about two thirds
of the solar radiation energy lies between 0.4 and 1.1 mm wavelengths. A
photon of blue light having 3 eV energy generates around 0.5 eV electricity
and the remaining 2.5 eV is absorbed as heat. This heat must be dissipated
back into the space; otherwise it would decrease the electrical conversion
efficiency. A high-emissivity coating on the back surface helps to dissipate
the heat effectively, thus improving the electrical conversion efficiency.
The continuing development efforts to produce more efficient low cost
cells have resulted in various types of PV technologies available on the
market today in terms of conversion efficiency and cost. The major types are
discussed in the following sections.1
8.3.1 Single-Crystal Silicon

Single-crystal silicon is a widely available cell material that has been the
workhorse of the space industry. In the most common method of
production, known as the Czochralski process, the silicon raw material is
first melted and purified in a crucible. A seed crystal is then placed in the
liquid silicon and drawn at a slow constant rate. This results in a solid
single-crystal cylindrical ingot (Figure 8.4).2 The manufacturing process is
slow and energy intensive, resulting in high raw material cost. The ingot is
sliced using a diamond saw into 200 to 400 mm (5 to 10 mil) thick wafers.
The wafers are further cut into rectangular cells to maximize the number of
cells that can be mounted together on a rectangular panel. The cell is
generally in several centimeter squares in size. Unfortunately, almost half of
the expensive silicon ingot is wasted in slicing it and forming square cells.
The material waste can be minimized by making the full size round cells
from round ingots. Using round cells is uneconomical in satellites where the
FIGURE 8.3 Power versus wavelength spectral response of silicon cell.

Solar Array 139
FIGURE 8.4 Single-crystal ingot-making by Czochralski process. (Source: G. Cook et al. DOE
Report No. DE91015001, 1995.)
panel space is at high premium. Another way to minimize waste is to grow

crystals on ribbons. Some U.S. companies have set up plants to draw PV
ribbons, which are then cut by a laser to reduce waste.
The high level of cleanliness during the manufacturing process, followed
by stringent screening for various possible defects, makes the space-
qualified cells expensive. A flexure test eliminates failure-prone wafers at
an early stage. The probability of cell cracking can be substantially reduced
by mechanical proof testing of silicon wafers before further processing. This
is based on a study3 that demonstrated that weak wafers could be
eliminated by subjecting all the wafers in a manufacturing batch to a
biaxial-flexure test. The test is performed after wafers have been chemically
polished but before junctions are formed in them. Wafers that are likely to
crack during further processing or in service break as soon as they are
flexed in a test jig. The wafers having no internal flaws that could eventually
cause mechanical failure withstand the flexure test. The study also showed
that the biaxial flexure test does not induce additional damage in otherwise
good wafers. If a wafer is sound, the testing does not create flaws that
would later weaken the specimen. The experimental results were analyzed
statistically. The biaxial strength of silicon wafers that failed from surface
flaws was found to correspond to a Weibull distribution. The critical flaw
sizes were found to be 4.6, 3.1, and 2.1 mm in proof testing at stress levels of
215, 260, and 310 MPa, respectively.
8.3.2 Gallium Arsenide

GaAs has much higher conversion efficiency than silicon cells, but is
expensive. It has a 1.43-eV direct band gap. It is relatively insensitive to
wide temperature cycling, and has high radiation resistant. Modern GaAs is
generally produced by growing GaAs film on a Ge substrate.
8.3.3 Semicrystalline and Polycrystalline

Instead of drawing single crystals using seeds, the molten silicon is poured
into a rectangular crucible. This process under a controlled cooling rate
forms partial and/or multiple crystals. The rectangular ingot thus produced
eliminates the squaring-up process and the associated waste of materials.
The remaining fabrication process follows the steps of the single crystal
silicon cell. This is a relatively fast and inexpensive process for manufactur-
ing thick crystalline cells. The conversion efficiency is low, but the cost is
much lower, giving a net reduction in cost per watt of power. Such cells are
developed for terrestrial applications, but are uneconomical at present for
space applications.
8.3.4 Thin Film

Copper indium gallium diselenide (CIGS) and cadmium telluride (CdTe)
are thin film materials, typically a few micrometers or less in thickness,
directly deposited on glass, stainless steel, ceramic or other compatible
substrate materials. While the basic copper indium diselenide cell has a
band gap of 1.0 eV, the addition of gallium increases the band gap to 1.4 eV,
resulting in more efficient collection of photons near the peak of the solar
spectrum. The thin film technology uses much less material per square area
of the cell, hence is less expensive per watt of power generated. These are
new types of photovoltaic cells entering the market that has drawn some
interest in the space industry.
8.3.5 Amorphous
In the amorphous technology, silicon vapor is deposited in a couple of mm
thick amorphous (glassy) film on glass or stainless steel rolls, typically 2000
feet long and 13 inch wide. This process forms a random bond structure, as
opposed to a crystalline structure. It significantly reduces the carrier
mobility and results in poor conversion efficiency, which is about 10% at
present. However, it uses only 1% of the material. Moreover, the sheet
manufacturing, instead of single cells from ingots, is inexpensive. The cost
per watt generated is significantly lower. With the stabilized efficiency
about one half of that in the crystalline silicon (c-Si) cells, the amorphous
silicon (a-Si) is not a serious candidate for space power at present. However,
Solar Array 141
crystalline technology has been with us for several decades and has entered
a plateau of a slow learning curve. On the other hand, amorphous
technology is new and hence a steep learning curve is anticipated. The
expected result is a rapid decline in the amorphous silicon price per watt by
increasing the efficiency and decreasing the manufacturing cost. On this
premise, in 1997, two plants in the U.S. started manufacturing a-Si sheets in
large quantity for terrestrial applications. It has been tested in space and
used in protoflights also.
8.3.6 Multi-Junction
Ever increasing demands on spacecraft power systems has produced
double and triple junction cells with improved performance, radiation
resistances, and lower cost. Recent advances have produced the tandem
multi-junction GaAs cell similar to the a-Si tandem cell, where one cell
consists of two tandem GaAs cells separated by thin tunnel junction of
GaInP, followed by a third tandem GaInP cell, separated by an AlInP tunnel
junction. The tunnel junctions mitigate voltage drop of the otherwise
forward-biased p–n junction that would appear between any two tandem
p–n junctions in opposition to the photon-induced cell voltage.
An n-on-p silicon junction converts only red and infrared light into
electricity, not blue and ultraviolet. The GaInP/GaAs cell captures infrared
photons as well, thus improving the efficiency. Efficiencies up to 35% in
double junction and 40% in triple junction space-qualified cells have been
measured by Spectrolab Inc. for DoE. The triple junction efficiency has a
theoretical limit of 50%. That is like capturing one half the sun. Today, the
GaInP/GaAs on Ge substrate multi-junction cells are being increasingly
used by many spacecraft builders seeking high efficiency and good
manufacturability.
8.4 Equivalent Electrical Circuit

The steady state equivalent electrical circuit of the cell is shown in Figure
8.5, which represents the complex physics of the PV cell. The cell acts as a
FIGURE 8.5 Equivalent electrical circuit of PV cell.

constant current source shunted by a perfect diode. The circuit parameters

are as follows. The series resistance Rs represents the internal resistance to
the current flow. It is primarily due to the resistivity of the material. The
shunt resistance represents the leakage current across the junction. It
depends on the n–p junction depth, the impurities, and the contact
resistance. The value of Rsh is inversely related with the leakage current
to ground. In an ideal PV cell, Rs ¼ 0 (no series loss), and Rsh ¼ 1 (no
leakage to ground). In a typical high quality 2.5 2.5-cm silicon cell, Rs ¼
0.05 to 0.10 and Rsh ¼ 200 to 300 . The PV conversion efficiency is
sensitive to small variations in Rs, but is insensitive to variations in Rsh. A
small increase in Rs can decrease the PV output significantly. Since
magnitudes of the diode current and the resistances Rs and Rsh vary with
temperature, the cell output and the conversion efficiency decrease with
increasing temperature.
In the equivalent circuit, the current delivered to the external load equals
the source current (photo current), Is, proportional to the illumination, less
the diode current, Id, and the ground shunt current, Ish. The open circuit
voltage, Voc, of the cell is obtained when the load current is zero, i.e., when
I ¼ 0, and is given by
Voc ¼ V þ IRs ð8:2Þ
The diode current is given by the classical diode current expression

qV
oc
Id ¼ Io eAKT 1 ð8:3Þ
where Io ¼ the diode saturation (dark) current

q ¼ electron charge ¼ 0.1592 1018 coulomb
A ¼ curve fitting constant
K ¼ Boltzmann constant ¼ 1.38 1023 J/K
T ¼ absolute temperature, in kelvin
The load current is therefore given by the expression
qV
oc V
I ¼ Is Io e AKT 1 oc ð8:4Þ
Rsh
The last term in Equation 8.4 is the ground leakage current. In practical
cells, it can be ignored since it is negligible compared to Is and Id. The
terminal current versus voltage (I–V) characteristics under dark, partial
light, and full light are shown in Figure 8.6. The voltage developed and the
fraction of incident flux converted into electrical power depend on the
junction material and wavelength of the radiation. The maximum power
Solar Array 143
FIGURE 8.6 Terminal I–V characteristics under full, partial, and zero illuminations.
that can be extracted from the cell under full light is equal to the area of the
dotted rectangle. The maximum power is approximately proportional to the
short circuit current. In the forward direction, the power generation is
maximum at the knee point of the curve, where the voltage is typically 0.50
V, the current about 40 mA/cm2 and power about 20 mW/cm2 of the cell
area in crystalline silicon cells. For GaAs/Ge cells, these values are 1.0 V,
30 mA/cm2, and 27 mW/cm2, respectively.
The diode saturation current can be determined experimentally by
applying voltage Voc to the cell in the dark and measuring the current
going into the cell. This current is often called the dark current or the
reverse diode saturation current. If the voltage is externally applied in the
reverse direction, say during a system fault transient, the current remains
flat and power is absorbed by the cell. However, beyond a certain negative
voltage, the junction breaks down as in diode, and the current rises to a high
value. In the dark, the current is zero for voltage up to the breakdown
voltage, which is the same as in the illuminated condition.
8.5 I–V and P–V Characteristics

The electrical characteristic of PV cell in sunlight in the useful working
range is represented by the I–V curve in the first quadrant only, as shown in
Figure 8.7(a). Two most important parameters widely used for describing
the cell electrical performance are the open circuit voltage, Voc, and the
short circuit current, Isc. The short circuit current is measured by short-
ing the output terminals, and measuring the terminal current under full
FIGURE 8.7 Cell characteristics in sunlight and the maximum power point.
illumination. Ignoring the small diode and the ground leakage currents
under zero terminal voltage, Isc is the photocurrent, Is. The current under this
condition is the maximum current the cell can deliver. The I–V characteristic
is developed from the test data at various illuminations, temperature, and
ionized radiation doses. The bottom right of the curve at zero current is the
open circuit voltage measured with the output terminals open. The
maximum photo voltage is produced under the open circuit voltage.
Again, by ignoring the ground leakage current, Equation 8.4 with I ¼ 0
gives the open circuit voltage as

AKT Is
Voc ¼ logn þ1 ð8:5Þ
q Io
The constant KT/q is the absolute temperature expressed in volts (300 K ¼

0.026 V). In practical photocells, the photocurrent is several orders of
magnitude greater than the reverse saturation current. Therefore, the open
circuit voltage is many times the KT/q value. Under constant illumination,
Is/Io is a sufficiently strong function of the cell temperature, and the solar
cell ordinarily shows a negative temperature coefficient of the open circuit
voltage.
The output power of the panel is the product of the voltage and the
current outputs. In Figure 8.7(b), the power is plotted against the voltage.
Notice that the cell produces no power at zero voltage or zero current, and
produces the maximum power at voltage corresponding to the knee point
of the I–V curve. This is why the PV power circuits are designed such that
the panel operates closed to the knee point, slightly on the left-hand side.
Therefore, the PV panel is modeled approximately as constant current
source in the electrical analysis of the system.
Solar Array 145
On the left hand side of the knee point, the cell works like a constant
current source, generating voltage to match with the load resistance. On the
right hand side, the current drops rapidly with small rise in voltage, and the
cell works like a constant voltage source with an internal resistance.
Figure 8.8 and Figure 8.9 are typical data sheets of two representative
space qualified cells. The basic requirement of the solar cell modules for
space applications is generally described in MIL-STD-83576 for the specific
FIGURE 8.8 Silicon PV cell performance data: Typical 8-cm, 8-mil thick on 10-cm
substrate with back surface field reflector, 14% efficiency in AM0.
FIGURE 8.9 GaAs/Ge PV cell performance data: Typical 4 4-cm, 3.6-mil thick, 19%
efficiency in AM0.
needs of the U.S. Air Force. However, the specifications of solar cells
intended for general use in space are not well defined. Even though the
PV cell performance standard is defined for specific procurement, the
performance measurement standard also needs to be defined. The ISO
Technical Committee C20 has issued one such standard ISO-15387 for the
aircraft and space vehicles that is endorsed by ANSI. NASA/Jet Propulsion
laboratory (JPL) performs such test in aircraft and balloons at 35 km height,
and plans to offer such tests in the space shuttle at about 500 km altitude at
a cost of around $5000 for a 2 4 cm cell. Comparing such tests conducted
by one organization with another is difficult. For example, in a round robin
test by six organizations to ISO-15387, including tests conducted at JPL and
in the PV engineering test bed at NASA GRC, the results showed deviation
Solar Array 147
of up to 3% on the same cell. The cell temperature coefficient is extremely

sensitive to the solar spectrum, and can vary by a few-fold, making the
spectrum duplication in tests extremely important. For this reason, power
engineers allow a 3% margin for standard cells, and perhaps more for
multi-junction cells, for possible error in power generation estimates using
test results.
8.6 Power from Albedo

Albedo is the ratio of the sunlight reflected by a planet to that received by it.
The Earth’s albedo impinging on the front and back sides of the solar array
can generate power at low voltage, which can be used if the solar array is
designed accordingly. NASA test data in LEO show that the PV conversion
efficiency under the Earth’s albedo on cells mounted on the back side is
about 50% of that on the sunlit front side. This can increase the array current
by 10 to 15% and about 5% increase in the voltage due to lower back side
temperature. However, if cells were mounted on the back side of the array
also, the cost per watt generated would be high, making it uneconomical in
most cases. As for the front side albedo superimposed on the full sun, it
does not significantly increase the power generation. Usually, the albedo
effects are ignored for the sake of simplicity, particularly in GEO satellites.
8.7 Back-Surface Illumination

The PV cells are sometimes mounted on flexible cloth rather than on rigid
panels. The flexible cloth array under the back side illumination would
transmit significant energy to the front, which the front cell would convert
into useful power. The power generation under back surface illumination
on a coupon of 10 series connected cells representing the flexible cloth solar
array of the ISS was tested under side and back illuminations. The cells
were 8 cm 8 cm 200 mm, n-on-p, crystalline silicon cell with 125-mm
borosilicate coverglass. They were mounted on 125-mm thick composite
polyimide–glass screen cloth substrate with silicon adhesive. The cells were
interconnected by 37-mm thick flat copper conductors welded to contacts on
the back of the solar cells. The copper interconnects covered 31% of the solar
cell area. The coupon was subjected to standard flash tests on the front side,
sideways and on the back side. As the illumination angle of incidence
varied, the cell I–V characteristic changed primarily due to the projected
area loss up to about 60 . Above that, other mechanisms also become
significant, such as Fresnel reflection, edge effect on light collection,
scattering and refraction, shadowing and other effects covered under the
Kelly cosine. The noteworthy conclusions of these tests were as follows:4
1. The back-illuminated cells produced 41% of the front-illuminated

power at essentially the same voltage.
2. Under edge illumination, the cells produced about 70% of the voltage
attained with normal front side illumination.
These data are valid for the BOL. However, it has been shown that under
the equivalent radiation dose of 1.5 1013 MeV after 11.6 years of the ISS
operating life, the back side response degraded by only 10%. This is a
significant power generation from the back side illumination of the flexible
cloth mounted cells. It can be factored in the analysis if needed in certain
mission situations to avoid over-sizing the arrays on the extreme
conservative side.
Analysis has shown that back side illumination from sources other than
direct sun, such as albedo from the Earth, the moon, and the spacecraft
itself, produce negligible power.
8.8 Array Construction

The solar array consists of numerous cells mounted on a base substrate. The
construction can take various forms described below. With a variety of
array constructions possible, the bulk of the arrays are still made with
crystalline silicon cells on rigid flat panels.
8.8.1 Rigid Panels

In this traditional construction method, the cells are mounted on a rigid
substrate made of an aluminum honeycomb with face sheets. Traditionally,
the face sheets have been made from aluminum, but fiber composite sheets
such as graphite, NomexTM and KevlarTM have also been successfully used
for mass reduction. Honeycomb substrates are available in 1=4 , 1=2 , 3=4 and 1
inch thicknesses. Silicon adhesive is used to fix the PV cells onto the panels.
Extreme care is exercised in applying the adhesive, since even a trace of
silicon in the cell production facility can degrade the cell output voltage
considerably.
The array is protected from the space environment primarily by a
coverglass such as fused silica MicrosheetTM. An antireflective coating
minimizes the light reflection. It enhances the light absorption by the cell
and produces more power. A suitable coating is also applied to the back
surface for temperature control. The ESD concern is addressed by ITO
coating on the front surface. However, it reduces the power generation,
which must be compensated by larger panel area or thicker coverglass.
Figure 8.10 shows a satellite with one wing of the solar array with key
panel construction features. The series strings of cells on the panels are
routed such that the magnetic moment is minimized. If the propellant
plume is a concern, the strings are oriented perpendicular to the plume flow
Solar Array 149
FIGURE 8.10 Satellite showing one wing of solar array fully deployed.
such that the plume affect only one whole strings versus some parts of all
strings. This way, if one string were damaged due to a heavy plume, the
required power would still be produced with the built-in redundancy in the
number of string circuits. Exposing all strings to the plumes may pose
relatively higher risk.
Figure 8.11 is a representative cross section of the panel depicting cell and
various other layers with their typical thickness.
The rigid panels are initially stowed with the spacecraft body during
launch. They are deployed by cutting various spring-loaded cables and
bolts with electro-explosive devices. Figure 8.12 displays the sequence of
one wing stowed, being deployed, and after full deployment.
FIGURE 8.11 Panel cross section showing cells mounted on honeycomb substrate with face
sheets.
FIGURE 8.12 Solar array wing: (a) stowed, (b) being deployed, and (c) after full deployment.
8.8.2 Body Mounted Array

The body mounted array finds applications in small satellites for science
missions, such as the one shown in Figure 8.13. The cells are mounted
directly on the spacecraft body without using a honeycomb substrate. It
saves substrate mass and also gimbals. The obvious disadvantage is that it
needs of 57% more cells for the same power. However, it is generally found
advantageous in the following situations:
The spacecraft distance from the sun, the sun angle or the operating
temperature vary over a wide range, or
Flat deployed panels interfere with an instrument’s line of view.
When the body mounted array rotates, the illuminated strings are at rated
voltage and the dark strings are at zero voltage. This can cause ESD
discharge along the circumference of the body. The remedy is to use bleed
resistance across the isolation diode as shown in the figure. It equalizes the
potential of the sunny side and the dark side through a bleed current.
Solar Array 151
FIGURE 8.13 Body mounted array with bleed resistors in solar strings.
8.8.3 Three or More Wings

The three-wing array shown in Figure 8.14 may give essentially the same
benefits as the body mounted array, provided that the deployed array
wings do not interfere with the instruments. The wings can be canted at
different angle as required for powering the mission. This architecture is
most likely to be used for small science mission satellites with peak power
tracking system design. With multiple wings, a separate PPT converter and
control are required for each wing as shown in Figure 8.15. Otherwise, the
PPT controls would conflict with each other.
8.8.4 Flexible Array

The flexible array made by installing PV cells on Kevlar cloth — open weave
or close knit — can reduce the array mass. The array is folded when stowed,
FIGURE 8.14 Three-wing solar array for small science mission satellites.
FIGURE 8.15 EPS architecture for 3-wing array with PPT for each wing separately.
and is deployed like an accordion panel or rolled out (Figure 8.16). Such
arrays have been used on the first Hubble Space Telescope, EOS-AM,
Olympus, and ISS. They must be designed to withstand the thermal snap on
exiting the eclipse. The snap is caused by the cell side facing the sun heating
up quickly, while the back side is still cold due to insignificant heat transfer
from front to back. Hence, a large thermal gradient builds up across the cloth
thickness. This and the mismatch in the coefficients of thermal expansion
between the cell area and the metal frame jointly cause the front to bend
FIGURE 8.16 Accordion type flexible cloth solar array for EOS-AM satellite. (Source: NASA.)
Solar Array 153
(buckle) since the flexible cloth cannot support bending stress. This produces
bowing in the array as shown in Figure 8.17. The electrical power is reduced,
since the array is not completely flat now. It comes back to a flat shape in
about 30 min when the temperature on the two sides is equalized across the
thickness.
As an example of estimating the power reduction, consider a 2-m wide
cloth panel. Such a width may bow by 20 cm or so. In that case, the angle
in the figure is 11.3 , which is roughly the average off-angle across the
width. Cos ¼ 0.98 gives 2% loss of power until the cloth regains its flat
shape. This is about 30 min in every orbit. In LEO satellites with 1-h
sunlight, this loss may need at least 1% larger array area.
The rigid array does not experience such bowing, since the aluminum
honeycomb takes the front heat to the back quickly, and the front and the
back face sheets spreads the heat evenly over the entire front and back
surfaces.
8.8.5 Inflatable Arrays

Flexible arrays come in two types:
The flat accordion type described above with conventional cells

Inflatable, often in balloon shape, with a few micrometers thick
amorphous PV sheets on 1 or 2 mils thick substrate.
The inflatable array is a new technology that has already been developed
and flight tested. It significantly reduces the mass and can be packaged in
small volume. Where usable, it reduces the overall program cost by
reducing the launch vehicle size, or by allowing more satellites per launch.
The inflatable array utilizes a flexible thin-film PV on a flexible composite
laminate structure densely packed for launch, and deployed in space using
inflation gas. It is finally cured and rigidized in situ in orbit without further
need of inflation pressure to maintain structural rigidity.
FIGURE 8.17 Bowing of flexible array under thermal snap.

8.9 Array Performance

The major factors influencing the electrical performance of the solar array
are the sun intensity, sun angle, and the operating temperature. These
factors are discussed below.
8.9.1 Sun Intensity

The I–V characteristic of PV array shifts down at lower sun intensity with a
small reduction in voltage as shown in Figure 8.18. However, the
photoconversion efficiency of the cell is insensitive to the solar radiation
in the practical working range. Figure 8.19 shows that the efficiency is
practically the same at full sun (1353 W/m2) and 1=2 sun, and starts falling
off rapidly only below 1=4 sun (340 W/m2). This means that the conversion
efficiency is high in penumbra. We obtain a lower power output in
penumbra only because of the less solar energy impinging on the cell.
However, the power extractable during penumbra cannot be utilized in
practical designs because of the low voltage. Hence, it is ignored.
8.9.2 Sun Angle

The cell output current is given by
Is ¼ Io cos ð8:6Þ
where Io is the photocurrent with normal sun ( ¼ 0). The cosine law holds
well for sun angles ranging from 0 to about 50 . Beyond 50 , the electrical
output deviates significantly from the cosine value, and the cell generates
no power beyond 85 , although the mathematical prediction would give
FIGURE 8.18 Current versus voltage characteristics at various illumination levels.

Solar Array 155
FIGURE 8.19 Photo-conversion efficiency over wide range of solar flux.
7.5% power generation. The actual power–angle curve of the PV cell is

called Kelly cosine, and is given in Table 8.1. The Kelly cosine presented in
the table is useful to assess accurately the power available from sun at low
angles during the transfer orbit.
8.9.3 Temperature Effect

With increasing temperature, the short circuit current of the cell increases,
whereas the open circuit voltage decreases as shown in Figure 8.20. Since
the increase in current is much less than the decrease in voltage, the net
effect is the decrease in power, which is quantitatively evaluated by
examining the effects on current and voltage separately. Say Io and Vo are
the short circuit current and the open circuit voltage at reference
temperature T, and and are their temperature coefficients in units of
A/ C and V/ C, respectively. If the operating temperature is increased by
iT, then the new current and voltage are given by
Table 8.1 The Kelly cosine values of

the photocurrent in silicon cells
Sun angle Mathematical Kelly cosine

(degrees) cosine value value
30 0.866 0.866
50 0.643 0.635
60 0.500 0.450
80 0.174 0.100
85 0.087 0
FIGURE 8.20 Temperature effect on I–V characteristic.
Isc ¼ Io þ T; and Voc ¼ Vo T ð8:7Þ
Since the operating current and voltage change approximately in the same
proportion as the short circuit current and open circuit voltage, respec-
tively, the new power is
P ¼ VI ¼ ðIo þ TÞðVo TÞ
This can be simplified in the following expression by ignoring the small

term containing the products of and ,
P ¼ Vo Io þ TVo Io T;
which reduces to
P ¼ Po ½ðIo Vo ÞT ð8:8Þ
For a typical 2 4 cm single crystal silicon cell, is 250 A/ C and is

2.25 mV/ C. Therefore, power varies approximately as
P ¼ Po ½1 0:005T ð8:9Þ
This expression indicates that for every degree centigrade rise in the
operating temperature, the silicon cell power output decreases by 0.50%.
Figure 8.21 depicts the power output versus voltage characteristics at two
operating temperatures. It shows that the maximum power available at the
lower temperature is higher than that at the higher temperature. Thus, a
cold temperature is better for the PV cell, as it generates more power.
However, the two Pmax points are not at the same voltage. In order to extract
Solar Array 157
FIGURE 8.21 Temperature effects on P–V characteristic.
maximum power at all temperatures, the PV system must be designed such

that the array output voltage can increase to V2 for capturing Pmax2 at lower
temperature, and can decrease to V1 for capturing Pmax1 at higher
temperature. If the array is operating at a fixed regulated voltage, the
higher power generation capability at a cold temperature cannot be utilized
by the loads, and the excess power generation by the cell must be wasted in
the shunt circuits. One can capture and utilize the peak power all the time
with varying temperature only by using the PPT system architecture.
Otherwise, the power versus temperature variation would be as shown in
Figure 8.22.
The array undergoes a wide temperature cycle in each orbit. During
sunlight, it operates around 50 to 60 C on the front face, 40 to 50 C on the
FIGURE 8.22 Power generation versus operating temperature of silicon cell.

back face, and with a 10 C gradient. The temperature is calculated from the
thermal model with the heat received primarily from the sun. The solar
array temperature is determined by the following thermal equation:
solar flux þ Earth’s albedo þ Earth’s thermal radiation þ heat coming from
adjacent components of the spacecraft ¼ electrical power output þ heat
radiated back into space (8.10)
The incidence and radiated heat from the spacecraft body varies with
location in the orbit as shown in Figure 8.23. The array temperature
variations over one GEO and LEO orbits are shown in Figure 8.24. During
an eclipse in GEO, the temperature drops exponentially to as low as
175 C. The time constant depends on the mass composition of the array
component, and is typically in the 30 to 60 min range. The front to back face
temperature gradient for rigid array with aluminum face sheet can be 5 to
10 C under steady sun and up to 20 C on sunlight snap after eclipse.
Various techniques are used to control the temperatures of spacecraft parts.
Figure 8.25 shows some passive techniques used in spacecraft thermal
design.
8.9.4 Sun Acquisition

The array wings after deployment must be oriented to acquire the sun
normal to the array surface. The sun sensor used for this purpose has two
PV cells mounted on two 45 wedges and connected differentially in series
opposition through an actuator motor as shown in Figure 8.26. When the
sun is perfectly normal as in (a), the currents from both cells are equal to
Io cos 458. Since they are in series opposition, the net current in the motor
is zero, and the array stays put. On the other hand, if the array is not normal
to the sun, with an error angle as in (b), the sun angles on the two cells are
different, giving two different currents, namely
FIGURE 8.23 Heat input and output of satellite in Earth orbit.

Solar Array 159
FIGURE 8.24 Solar array temperature variations over one LEO and GEO orbits.
I1 ¼ Io cosð45 þ Þ; and I2 ¼ Io cosð45 Þ ð8:11Þ
The motor current is therefore,
Im ¼ I1 I2 ¼ Io cosð45 þ Þ Io cosð45 Þ ð8:12Þ
Using the Taylor series expansion
h2 00
fðx þ hÞ ¼ fðxÞ þ hf 0 ðxÞ þ f ðxÞ þ
2!
We can express the two currents as
I1 ¼ Io cos 45 Io sin 45;

FIGURE 8.25 Thermal control on satellite components by passive techniques.
and
I2 ¼ Io cos 45 þ Io sin 45 ð8:13Þ
The motor current in Figure 8.26(c) is then
p
Im ¼ I1 I2 ¼ 2Io sin 458 ¼ 2Io ð8:14Þ
if is in radians.
FIGURE 8.26 Sun pointing actuator principle.

Solar Array 161
The motor turns towards the sun until the current is zero, and then it
remains there. At that position, the error angle, , is zero, i.e., the array is
normal to the sun. The two current sensors shown in (c) work as the
sunlight/eclipse binary detectors also.
As in the ISS, sun sensors are generally mounted on the main solar array
body itself. They may indicate that the solar array is off pointed when in
fact the array is shadowed from the sun, or the sun sensors themselves are
shadowed, or when the solar array is feathered in order to avoid the rocket
exhaust plumes during rendezvous with the space shuttle or any other
vehicle.
8.9.5 Sun Tracking

Sun tracking is required for the solar array to face the sun continuously as
the spacecraft orbits the Earth. This is done by an actuator that follows the
sun like a sunflower. There are two types of sun trackers:
One-axis gimbals, which follows the sun from east to west during the
day.
Two-axis gimbals which tracks the sun from east to west during the
day, and from north to south during the seasons of the year. The dual-
axis tracking is done by two linear actuator motors, which aim the sun
within 1 of accuracy (Figure 8.27).
The EPS provides a means of independently orienting and rotating the

deployed north and south solar wings about the pitch axis of the spacecraft.
After the deployment, the ground control is involved in aligning the solar
array to the sun. Once aligned, the array rotates by the clock time without
ground intervention to maintain the sun orientation.
FIGURE 8.27 Dual-axis gimbals.

Some sun pointing error cannot be avoided even after acquiring the
normal sun by the sun sensor and then tracking by the SAD. The error
generally comes from the cell flatness error of about 2 , and the gimbal
tolerance error of about 3 . The total 5 error must be accounted for in the
array design. Cos 5 ¼ 0.996, which means 5 sun pointing error will reduce
the power generation by 0.4%.
8.9.6 Peak Power Extraction

The gimbal motor drives the module to face the sun for collecting the
maximum solar flux. However, that in itself does not guarantee the
maximum power output from the module. As was seen earlier, the module
must electrically operate at the voltage that corresponds to the peak power
point under the given operating conditions. First we examine the electrical
principle of the peak power tracking operation.
Some loads such as heaters have constant resistance, with power varying
with the voltage squared. On the other hand, most communications loads in
satellites are constant power loads, drawing more current at lower voltage.
The operating point of any power system is the intersection of the source
line and the load line. If the source having the I–V and P–V characteristics
shown in Figure 8.28 is supplying power to the resistive load R1, it will
operate at point A1. If the load resistance increases to R2 or R3, the operating
point moves to A2 or A3, respectively. The maximum power is extracted
from the module when the load resistance is R2. Such load matching with
the source is always necessary for maximum power extraction from the PV
module.
Some loads such as heaters have constant resistance, with power varying
with the voltage squared. On the other hand, most communications loads in
satellites are constant power loads, drawing more current at lower voltage.
FIGURE 8.28 Operation and load matching of PV source with constant resistance load.
Solar Array 163
The operation with constant power loads P1 and P2 is shown in Figure

8.29. The load line P2 has two points of intersection with the source line,
denoted by B1 and B2. Only point B2 is stable, as any perturbation from it
will generate a restoring power to take the operation back to B2. Therefore,
the system will operate at B2.
The necessary condition for the electrically stable operation of the solar
array is

dP dP
> ð8:15Þ
dV load dV source
If the array is operating at voltage V and current I on the I–V curve, the
power generation is P ¼ VI watts. If the operation moves away from the
above point, such that the current is now I þ iI, and the voltage is V þ iV,
the new power is
P þ P ¼ ðV þ VÞðI þ IÞ ð8:16Þ
After ignoring a small term, Equation 8.16 simplifies to
P ¼ VI þ IV ð8:17Þ
The iP should be zero at the peak power point, which necessarily lies on a
locally flat neighborhood. Therefore, at the peak power point, the above
expression in the limit becomes
dV V
¼ ð8:18Þ
dI I
FIGURE 8.29 Operation and load matching of PV source with constant power load.
We take note here that dV/dI is the dynamic impedance of the source, and
V/I is the static impedance. Various electrical methods of extracting the
peak power from the module are now described below.
8.9.6.1 Impedance method

In the first method, a small signal current is periodically injected into the
array bus and the dynamic bus impedance Zd ¼ dV=dI and the static bus
impedance Zs ¼ V=I are measured. The operating voltage is then increased
or decreased until Zd ¼ Zs .
8.9.6.2 Power gradient method

In another method, the operating voltage is increased as long as dP/dV is
positive. That is, the voltage is increased as long as we get more power. If
dP/dV is sensed negative, the operating voltage is decreased. The voltage is
retained if the dP/dV is near zero within a preset dead band.
8.9.6.3 Voltage ratio method

The third method makes use of the fact that for most PV cells, the ratio of
the voltage at the maximum power point to the open circuit voltage (i.e.,
Vmp =Voc ) is approximately constant, say K. For example, K equals 0.72 for
high quality crystalline silicon cells. An unloaded cell is installed on the
array and kept in the same environment as the power-producing module,
and its open circuit voltage is measured. The operating voltage of the
power-producing array is continuously adjusted at KVoc to produce the
maximum power.
8.9.6.4 Current regression method

Yet another method uses statistical least-squared error regression curve
fitting of numerous test data on a given cell. For example, in extensive
testing of GaAs cells under various illumination (50 to 2500 W/m2) over a
rather wide temperature range (120 C to 140 C), JPL found that Isc varies
linearly with illumination Io according to the following expression:
Io ¼ ao þ a1 T þ a2 Isc ð8:19Þ
where T is the temperature in degrees centigrade. Furthermore, Voc varies

linearly with log Io.
8.9.6.5 Power regression method

The JPL tests also found that the PPT algorithm could also be set up as
Pmax ¼ bo þ b1 T þ b2 Io þ b3 TIo W=m2 ð8:20Þ

Solar Array 165
This method can be used at least as a back-up for the software routine.
Gluck5 has published test results with values of all a and b constants in the
above equations for numerous GaAs cells tested at JPL.
8.9.7 Shadow Effect

The array consists of many parallel strings of series connected cells. One
such string is shown in Figure 8.30. A large array may be partially
shadowed due to a structure interfering with the sun line. If a cell in a long
string becomes completely shadowed, it loses the photovoltage but still
must carry the current by virtue of its being in series with other fully
operational cells in the string. Without internally generated voltage, it
cannot produce power. Instead, it acts as a load, producing the local I2R loss
and heat. The remaining cells in the string must work at higher voltage to
make up the loss of the shadowed cell voltage. Higher voltage in healthy
cells means lower string current as per the I–V characteristic of the string.
This is shown in the middle dotted I–V curve in Figure 8.30(b). The current
loss is not proportional to the shadowed area, and may go unnoticed for
mild shadow on a small area. However, if more cells are shadowed beyond
the critical limit, the I–V curve is reduced below the operating voltage of the
string, making the string current fall to zero, losing all power of the string.
FIGURE 8.30 Shadow effect on a long PV string of solar array.

FIGURE 8.31 Bypass diodes in PV string to minimize power loss under heavy shadow.
The commonly used method of eliminating the loss of a string due to the
shadow effect is to subdivide the string length into several segments with
bypass diodes (Figure 8.31). The diode across the shadowed segment
bypasses only that segment of the string. This causes a proportionate loss of
the string voltage and current without losing the whole string power. Some
modern PV cells come with such internally embedded bypass diodes to
eliminate the need of separate diodes on the back of the panel. Cross-
strapping cells in adjacent strings, also shown in the figure, is often used to
improve the reliability against open cell, but is not related with the shadow
effect mitigation.
If part of the array is shadowed for part of the orbit, the decrease in
incident energy during the sunlight portion of the worst-case orbit may
need to be accounted for in the energy balance. It is given by
1 X
Esun ¼ ð1 shadowi Þ cos i t ð8:21Þ
Tsun i
where Tsun ¼ sunlight duration, i ¼ angle between the solar vector and the
array normal at time step i, shadowi ¼ fraction of the solar array surface
shadowed at time step i, and it ¼ time step interval of the analysis.
8.10 Array Design

At the end of life, the array must generate required power equal to the sum
of all load, the power losses and the battery charge power with some
margin. The usual end of life solar power margin in the worst case with any
one failure in the power system is typically 3 to 7% depending on the
Solar Array 167
maturity of the load estimate. The solar array in LEO must recharge the
battery in a short time during the sunlight. As a result, the ratio of the
needed array power to the load power is approximately 2.0 in LEO
compared to about 1.1 in GEO satellites.
The PV cells on the array are segmented in many strings connected in
parallel as shown in Figure 8.32. The following relations approximately
hold true with a suitable margin in the array segmentation. The margin
varies as the design progresses from the conceptual stage to the final design.
array voltage
Nseries cells per string ¼ ð8:22Þ
Vmp
Icircuit
Nstrings per circuit ¼ ð8:23Þ
Imp
array current
Nparallel circuits ¼ ð8:24Þ
Icircuit
where N denotes a number, and Vmp and Imp are the cell voltage and cell
current at the maximum power point, respectively. Each circuit delivers the
output current to the bus via dedicated isolation diode and sliprings. One
extra circuit is provided for single point failure redundancy. The number of
power slip rings is thus equal to the number of parallel circuits. The
segmentation is done such that the circuit current is usually below 5 A,
which is the derated current capability of some standard SAD slip rings.
FIGURE 8.32 Array segmentation in strings with isolation diodes for each slip ring.
The solar array substrate is grounded using a separate slip ring and the
signal slip rings are kept separate from the power slip rings.
Substrate materials are selected to meet the space environment over the
mission life. Differential coefficient of thermal expansion between various
array materials is important to avoid thermal cycling of the array under
thermo-elastic stresses. Otherwise, the cell mounting on the substrate may
delaminate, leading to premature failure. Mechanically, large solar array
wings having output power of a few kilowatts are often required to have
the natural frequency exceeding 20 Hz in the stowed configuration. In the
deployed configuration, the typical requirement is 0.10 Hz in the bending
mode and 0.5 Hz in the torsion mode. This may have some bearing on the
cell and substrate selection.
The following factors directly or indirectly influence the total number of
cells required in the array design. Some of them are discussed in this
section, while others are covered elsewhere in the book:
Environmental radiation
Coverglass protection
Degradation factors
Operating temperature
Temperature coefficients for Isc and Voc
Cell size suitable for manufacturing
Magnetic dipole moment minimization
Power shunting method
As for the PV cell type, the silicon cell has met the industry need for
decades. GaAs/Ge has been used in the past where needed for high power
or high radiation resistance, although mainly in defense spacecraft.
Compared to the silicon cell, the GaAs cell gives approximately 40%
more power per square meter and 20% more power per kilogram with rigid
panels on an aluminum honeycomb. The material cost of the GaAs cell is,
however, an order of magnitude higher than the silicon cell. The space-
worthy rigid array cost in 2002 was $400 to $500 per watt using Si cells and
about three times as much using GaAs. In volume and/or weight limited
power systems, GaAs may find applications where high cost can be justified
from other considerations. Recently, however, various other PV cells, such
as the double and triple junction cells, have made inroads into spacecraft
power systems design. And the technology change will continue. The
responsibility for selecting the most optimum PV cell at the time of the
design remains with the power systems engineer.
Solar Array 169
8.10.1 Degradation Factors

Degradations in the solar array performance come from the following two
distinct groups of factors: (a) initial degradations at BOL due to
manufacturing imperfections and tolerances, and (b) accumulated degrada-
tion over time up to the EOL due to various environment effects.
(1) Initial degradations at BOL

Cell mismatch in the assembly
Cell-to-cell wiring loss
Array wiring loss over panels and boom
Plasma effect, causes leakage current from the array to space
(2) Accumulated degradation over time to EOL
Ionized particles, cause damage due to cumulative radiation dose
Ultraviolet radiation, changes the optical properties of the cover-
glass and the cell
Thermo-elastic stress cycles, cause soldered joints to crack over
time
Micrometeoroids and debris impacts, reduce power generating
area of cells and damage conductors, particularly at the solder
joints
Flue gases, change optical properties of coverglass and cell
Bypass diode failure, loses a sub-string reducing the string current
Table 8.2 gives representative values of major degradation factors. The
effect can be represented by the equation:
total degradation ¼ BOL degradation þ in-service degradation to EOL

(8.25)
8.10.2 Radiation Damage

Different charged particles in the space environment have different mass
and velocity, resulting in significantly different energy levels. Heavy
particles confine damage to the vicinity of the particle track, hence doing
more damage for a given energy. Different particles having the same energy
have different damaging effect on Isc and Voc. Therefore, their damaging
effect is converted into the equivalent of 1 MeV of energy dose. The sum is
expressed as, say, 2 1015 e/cm2, where the letter e is called electron fluence
but it really represents 1 MeV energy. Therefore, the fluence levels are
different for Isc and Voc, and the higher of the two applies to the Pmax point.
One way to approximately account for the Pmax degradation is to take the
product of the Isc and Voc degradation. However, such approximation in
Table 8.2 Degradation factors for a

900 W satellite in MEO 10,900 nautical
miles, 69 inclination orbit, 15 years
Isc factors
Natural radiation Note a
Assembly mismatch loss bol 0.98
Coverglass charge particles 0.99
Coverglass coating (ITO) bol 0.98
Ultraviolet rays 0.97
Propellant contamination 0.98
Micrometeoroid damage 0.98
Voc factors
Natural radiation Note a
Coverglass charge particles 0.99
Pmax factors
Natural radiation Note b
Wiring loss (cell-to-cell) bol 0.98
Panel wiring voltage drop 0.05pPbol 1.50 V
a. To be determined from the radiation

fluence over the mission life.
b. Approximate value equal to the product
of the Isc and Voc degradation factors can be
used in top-level calculations only. In
detailed calculations, it is accounted
through the Isc and Voc factors.
some situations may not be accurate. For the silicon cell, the equivalent
fluence for Voc and Pmax is same, i.e., Voc and Pmax degradations are
identical. But for GaAs cells, the equivalent fluence for Voc and Pmax are
different, i.e., different particles degrade the GaAs cells differently with
respect to Voc and Pmax.
The degradation curves are established by testing numerous cells at
various fluence levels. The most common source of degradation factors is
NASA/JPL Handbooks, such as PUB 82-69 were originally published with
data on silicon cells. The GaAs data were added in the early 1990s. The
triple junction cell degradation curves have been published by organiza-
tions such as Aerospace Corporation. Some representative degradation
curves for silicon cells are shown in Figures 8.33 through 8.37. The radiation
not only affects the Isc and Voc, but also their temperature coefficients as
shown in Figure 8.38 and Figure 8.39. The Isc coefficient values in Figure
8.38 are for a 4 6 cell. Divide them by 3 for a 2 4 cell.
The coverglass is used to protect the cell from radiation damage. Its
thickness is increased as long as the saving in the array mass is more than
the added mass in the coverglass. This gives an optimum coverglass
thickness for a minimum array mass including the coverglass. For example,
estimated values for natural radiation in GPS orbit with various coverglass
thicknesses on GaAs/Ge cells are given in Table 8.3. The parametric curves
are generated at various coverglass thicknesses for candidate PV cells
Solar Array 171
FIGURE 8.33 Normalized Isc versus equivalent 1-MeV/cm2 fluence for 2-cm silicon cell.
(Source: NASA/JPL.)
FIGURE 8.34 Normalized Voc versus equivalent 1-MeV fluence for 2-cm silicon cell.
(Source: NASA/JPL.)
FIGURE 8.35 Normalized Pmax versus equivalent 1-MeV fluence for 2-cm silicon cell.
(Source: NASA/JPL.)
FIGURE 8.36 Normalized Imp versus equivalent 1-MeV fluence for 2-cm silicon cell.
(Source: NASA/JPL.)
Solar Array 173
FIGURE 8.37 Normalized Vmp versus equivalent 1-MeV fluence for 2-cm silicon cell.
(Source: NASA/JPL.)
FIGURE 8.38 Temperature coefficient of Isc versus equivalent 1-MeV fluence for 2-cm
4 6-cm silicon cell: divide by 3 for a 2 4 cell. (Source: NASA/JPL.)
FIGURE 8.39 Temperature coefficient of Voc versus equivalent 1-MeV fluence for 2-cm
silicon cell. (Source: NASA/JPL.)
(silicon or any other) as displayed in Figure 8.40. As the coverglass

thickness is increased, the coverglass mass increases but the array mass and
coverglass decrease due to reduced fluence with better protection. The total
array mass therefore takes the parabolic shape with a minimum for a certain
coverglass thickness, which is then selected for the design.
In a typical GEO communications satellite with typical coverglass, the
natural radiation fluence on silicon cells with coverglass is about 3 1013
MeV per year. Under such radiation, the power generation capability of the
array degrades with the service life as shown in Figure 8.41. Natural
radiation varies with orbit altitude and inclination. Figure 8.42 shows the
variation in radiation and EOL/BOL power ratio required with two types of
Table 8.3 Natural radiation in GPS orbita (1 MeV

for GaAs/Ge cells of 0.075 g/cm2 thickness)
Coverglass (mm) Isc Voc Pmax
0.10 9.0 1014 9.7 1014 9.7 1014

0.20 7.2 1014 7.3 1014 7.3 1014
0.30 6.2 1014 6.2 1014 6.2 1014
0.50 4.8 1014 4.8 1014 4.8 1014
1.00 3.0 1014 3.0 1014 3.0 1014
a
Nuclear radiation on Isc, Voc and Pmax are to be added.
Solar Array 175
FIGURE 8.40 Optimum coverglass thickness for radiation protection results in minimum
array mass.
traditional silicon cells designated by K4 and K6 in a 45 inclined orbit at

various altitudes.
8.10.3 Fluence Calculation

The following parameters generally enter the total radiation fluence
calculations:
Orbit altitude, inclination, and duration of the mission

Trapped electron model, such as NASA’s AE7
Trapped proton model, such as NASA AP8
FIGURE 8.41 Solar array power capability degradation versus years in GEO service. (Si cell
with 6 mil coverglass.)
FIGURE 8.42 Radiation fluence and power degradation versus orbit altitude for 3-year
mission in 45 inclination orbit.
Solar flare model as specified by the customer

Fission electrons and neutrons from nuclear threat, if specified
Solar array cross section with material thickness and densities in the
front and in the back of the cell junction. The surface mass density of all
the front and back shielding is then converted to the total surface mass
density in g/cm2. The face sheets are included, but the honeycomb
contribution is neglected.
The radiation damage equivalent to 1 MeV is then calculated for each

design option, such as the PV cell type and the coverglass thickness, as
shown in Table 8.4 for various options.
8.10.4 Offsetting from Sun

Excess power is always available at the BOL. The full load power is often
not needed for a long time, such as during on-orbit storage. We may expect
that offsetting the array from the normal sun could minimize the ionized
radiation damage and extend the array life. Such expectation is based on the
less than 90 angle of incidence of charge particle on the array. However,
offsetting does not help in this regard, because the direction of the charged
particles does not matter in their degradation effect. The fluence is like a
cloud of homogeneous gas or liquid in which the array is immersed,
particularly in GEO and MEO. The immersed area is what really matters. In
LEO, however, there are some indications that the direction perhaps
matters, but probably not much.
If the inboard panels are kept stowed for several years until they are
needed later in the mission, the radiation damage on the stowed panels
Solar Array 177
Table 8.4 Radiation fluence on Isc with one group of optionsa
Particle Front thickness Back thickness Total fluence
Proton 7.08 1010 0.00 7.08 1010

Electrons 8.30 1014 2.65 1014 1.09 1015
Flare proton 2.49 1014 6.90 1013 3.18 1014
Flare alpha 1.22 1013 8.30 1011 1.30 1013
Nuclear fission electron and neutron TBD TBD TBD
TOTAL 1.09 1015 3.35 1014 1.42 1015
TOTAL summary: equivalent fluence for Isc ¼ 1.42 1015 MeV/cm2.

From similar table, equivalent fluence for Voc ¼ . . .. . .. . ... MeV/cm2.
Equivalent fluence for Pmax ¼ . . .. . .. . . MeV/cm2.
Nuclear fluence of 3.2 1014 to be added, same for Isc, Voc and Pmax.
a
These values are representative of Isc fluence on 10 mil silicon cells (0.061 g/cm2
thickness), 8 mil ceria doped microsheet coverglass, 10 cm substrate, on 2 x 20 mil
aluminum face sheets on 1 inch honeycomb, over 14 years in GPS II orbit.
would be somewhat less. However, this reduction in the damage is

insignificant, may be 1% for every 2 years of preservation. As mentioned
earlier, the radiation particles are like a cloud of gas in which the array is
immersed. Deployed or not deployed, the charged particles enter the array
and do the damage.
8.10.5 Magnetic Moment

The satellite’s magnetic moment is primarily produced by the solar array
current loops and the torque coils. It is a vector defined as M ¼ (current
area of the loop) Am2, regardless of the shape of the loop. The vector
direction follows the right hand rule. For a loop current in one plane, the
moment is perpendicular to the plane. In the Earth’s magnetic field, B, it
produces a torque, T (in Nm), of magnitude given by the cross-product
T ¼ M B ¼ M B sin ð8:26Þ
The potential energy, U (in joules), stored in the loop is given by the dot
product
U ¼ M B ¼ M B cos ð8:27Þ
The moment increases with current. Since current is proportional to the

array area and the array area is proportional to power, the moment
increases roughly as square of the power, i.e.,
M ¼ k P2 ð8:28Þ
where k is the constant of proportionality derived from test values in

similarly built designs.
The torque direction is such that the coil (dipole) tends to orient itself
parallel to the B field such as to minimize the potential energy, which is
always the stable position. In doing so, if the coil turns at an angular speed,
!, the power delivered by depleting the potential energy is T!. When
changing the attitude (position and direction) to maintain the desired orbit,
additional fuel must be spent by the attitude control system to overcome
this torque. The amount of this additional fuel is proportional to the time
integral of T! over the mission life, i.e.,
Z
fuel ¼ T! dt ð8:29Þ
life
The magnetic moment of the solar array is minimized by laying the string
circuits such as to minimize the loop area. This is done not only to minimize
the fuel requirement, but also to minimize its influence on payloads,
particularly the instruments that are sensitive to magnetic field.
8.11 Advances in PV Technology

Advances in PV cell technologies continue in the following areas:
Manufacturing cost reduction

Efficiency improvements
Spectrum response improvement
Using low resistivity base
New semiconducting materials
Radiation resistance enhancements
This film and multi-junction (cascade) cells
Concentration cells, parabolic and linear
Amorphous cells (non-crystalline glass type)
Because much of the present PV technology uses crystalline semiconductors

similar to integrated circuit chips, the production cost has been high. The
PV cell manufacturing process is energy intensive. Every square centimeter
cell area consumes a few kilowatt hours before it faces the sun and
produces the first kilowatt hours of energy. However, the manufacturing
energy consumption is steadily declining with continuous implementation
of new production processes. Newer processes are being developed and
continuously implemented to lower the energy use per square centimeter of
the finished cell. In the current manufacturing process, the front and the
Solar Array 179
back of the cells are diffused separately, taking 1 to 3 h of manufacturing

time. Researchers at Georgia Institute of Technology are considering rapid
thermal processing technology. This process simultaneously forms the front
and the back of the cell in several minutes, and the metal contacts are
applied by screen printing. It not only cuts the processing time, it also
lowers the temperature and energy needed for fabrication.
Recently, double junction and triple junction GaInP2/GaAs/Ge cells have
made inroads in the spacecraft power systems designs.6 The n-on-p design
provides enhanced radiation hardness. This is achieved simply by
deposition on a p-type Ge substrate and forming an active junction in the
substrate by in situ arsenic diffusion. Tunnel junctions are used to link the
individual Ge bottom, GaAs middle and GaInP2 top cells. The I–V
characteristic of a typical large triple junction cell is shown in Figure 8.43.
Triple junction cell efficiency up to 40% has been tested in controlled
laboratory conditions, and 30% in production cells used in the spacecraft,
with a long-term goal of 40%. However, it has been reported that the cells
FIGURE 8.43 Cross section and I–V characteristic of typical large triple junction cell.
degrade faster than initially predicted. The present cost of 30% efficiency
cells is 12 to 15 $/cm2 ($350 to $450 per watt) at the cell level and $800 to
$1200 per watt at the array level.
Developments are under way at JPL for a new 46% efficient cell using
40% efficiency triple junction cell in the front and 10% efficient thermo-
electric cell in the back to use the waste heat. This technology, when
available, may produce a new cell with a maximum theoretically possible
efficiency of 47% in the laboratory and 37% in production cells.
The multi-junction c-Si cell significantly increases the efficiency, while a-
Si cells drastically decrease the material cost. The disadvantage of the a-Si
cell is that the cell efficiency is low to begin with, and degrades further by
about 20% in the first several months under sunlight before stabilizing.
However, afterwards, it suffers less radiation damage because of its very
thin PV junction. Gamma rays do not cause damage while penetrating the
super-thin junction. In fabricating the PV cell, better crystallivity means
better efficiency, but also greater radiation damage. Even slight damage to
the well-aligned crystalline structure degrades the power generation. The
random atomic structure in the amorphous cell cannot become misaligned
and there is no reduction in performance.
Amorphous silicon has an extremely low production cost, is inherently
radiation hard and performs better at high temperature. Such cells in large
volume for terrestrial applications cost $3 per watt, with the annual
production capacity in tens of megawatts at present. For space applica-
tions, their downside is lower efficiency compared to the c-Si cells. Multi-
junction thin-film a-Si deposited on a thin flexible substrate is an emerging
technology that may find space applications. The thin film cell is a few
microns thick and is supplied on aluminum sheets each with a thickness
of a few mils. The triple junction cells are produced by plasma-enhanced
chemical vapor deposited on a thin substrate with a textured silver/zinc
oxide back reflector. The top cell uses a-Si alloy of about 1.8 eV band gap
to capture the blue photons, the middle cell uses an a-SiGe alloy of 1.6 eV
band gap to capture the green photons, and the bottom cell uses a-SiGe
alloy of 1.4 eV band gap to capture the red photons. The sunlight, which
is not absorbed, is scattered back from the back surface at an oblique
angle to facilitate multiple passes. Indium tin oxide is deposited onto the
cells, which serves the dual purpose of ESD control and antireflective
coating. The top silver grid collects the current. Assuming 9% efficiency,
Guha7 estimates that the specific power of the cell with 0.5-mil thick
stainless steel substrate is about 770 W/kg, and about 2750 W/kg with 1
mil thick Kapton substrate. The triple-junction a-Si cell efficiency of 13%
has been achieved in small cells under AM1.5 spectrum prevalent on the
Earth. Since the AM0 spectrum in space is much richer in the blue
wavelength region, the component thickness in the triple-junction cells are
optimized differently. In space optimized cells under AM0 radiation
tested in NASA GRC laboratory, 0.27 cm2 cells have shown 12% efficiency,
Solar Array 181
and large 11 cm2 cells have shown 10.8% efficiency, all at BOL after the
initial stabilization.
In space optimized a-Si cells, Guha also reports the following:
Temperature degradation in power in the 20 to 70 C range is found to

be between 0.2 and 0.3% per C. This is about one half of that in the
conventional crystalline silicon cells. For this reason, a-Si is expected to
perform better at the high temperatures typically found in LEO and
GEO.
Radiation degradation is lower if the cells are annealed at moderate
temperature. At 70 C, the power degradation was 5% after a dose of 1
1015 MeV electrons/cm2 and 16% after a dose of 1 1016 MeV protons/
cm2 after annealing at 110 C. Crystalline Si cells would lose most power
under such doses.
No change in performance was found after more than 33,000 thermal
cycles between þ150 C and 110 C at a rate of 80 C/min, nor after
exposure to atomic oxygen of 1.3 1020 atoms/cm2.
Table 8.5 due to Guha compares the estimated EOL performance of the thin-
film triple junction a-Si cells with the conventional c-Si cells. The most
significant benefit of the a-Si cell is its low cost and low mass.
To drive the cost down further, scientists at the Swiss Federation of
Technology have recently developed a titanium dioxide cell with 33%
efficiency which, when fully developed, may cost about the same as a
comparable thickness of glass.
A new cell on the horizon is the thin film copper–indium–selenide cells
bonded on thin lightweight substrate of flexible sheets that are kept flat
prior to deployment. The shape-memory springs would unfold the array
Table 8.5 Comparison between amorphous and

crystalline silicon cells for space applications
Performance Crystalline Amorphous
BOL efficiency 14% 9%

Temperature effect 20% 8%
Radiation degradation 25% 3%
EOL efficiency 8% 8%
Cost High Low
Mass saving in 7 kW array Reference 50 kg
Interconnect Reference Much less
Stiffness Rigid Flexible
Technology status Proven Being tested
(Source: S. Guha et al. Proc. 34th IECEC, p. 2553, ASE,

1999.)
frame into panels. The specific power of such panels can be up to 100 W/kg
at the array level.
Very large arrays generating 100 kW to 1 MW power for lunar or Mars
colonies may become possible using thin film cells on a monolithic polymer
substrate at a cost of about $100 per watt and specific power of 1000 W/kg.
For conventional spacecraft applications, thin films must have at least 15%
efficiency on monolithic polymer substrates to make them mass and cost
competitive. A thin film means a few micrometers (< 5 mm) thickness using
any semiconducting material, but mostly in amorphous or polycrystalline
single or multi-junction form. Their terrestrial versions use a glass substrate,
but the cells for space may use aluminum, stainless steel, or polyimide
substrates. The thin film cells are replacement cells (not monolithic) that
come in a few few inches rectangular sizes (e.g., 4 6 inches). At present,
the specific power of the thin film cells is about twice that of silicon cells.
The radiation resistance of these cells is inherently high because of both the
thin film and also the randomly aligned non- or poly-crystalline structure,
so the charged particle cannot damage the structure uniformly.
Table 8.6 summarizes major PV technologies on the market today.
The Naval Research Laboratory and NASA Glenn have incorporated a
suitcase-size solar cell experiment on the ISS. It is designed to monitor and
relay the voltage, current and temperature data on a variety of the present
and emerging PV cells for 1 year. The primary emphasis of the experiments
is on evaluating the long-term performance of current and future multi-
junction solar cell technologies.
Table 8.6 Comparison of various PV cell technologies (typical representative

values or relative values)
Efficiency W/ $/Wa Radiation

Cell BOL m2 BOL hardness CICb kg/m2 Maturity
Silicon 12–14% 150 100 Low 1 Heritage

GaAs/Ge 18–19% 225 300 Med 1 Heritage
GaInP2/ 24–26% 300 300 Med 1 New in use
GaAs/Ge
Amorphous 8% 80 << 100 High 0.1 Not ready
Si
CuInSe 10% 135 << 100 High 0.1 In Development
X 100 Suns 32–34% GaAs tandem
X 150 Suns 30–32% GaInP2/GaAs tandem
X 1000 Suns 36–38% GaInP2/GaAs/Ge
triple junction
a
Rough order of magnitude costs in 2004.
b
CIC ¼ Cell þ Interconnect þ Coverglass.
Solar Array 183
8.12 Concentrator Array

In an attempt to improve the conversion efficiency, the sunlight is
concentrated into tens or hundreds of times the normal sun intensity by
focusing on a small area using low cost lenses. The primary advantage is
that such a cell requires a small fraction of active area compared to the
standard cell. This significantly reduces the PV material requirement,
although the total energy collection area remains the same. Besides
increasing the power and reducing the size or number of active cells,
such cells have the additional advantage that the cell efficiency increases
under concentrated light up to a point. For example, Krut reports that a
GaInP/GaAs dual junction cell efficiency of 29% under one sun increased to
30% under two suns and 34% under 100 to 500 suns. The efficiency
improves provided the cells are kept cool by radiating heat into space. Krut
cites the following concerns about the introduction of concentrating solar
panels in spacecraft:
Need for major changes in conventional flat panel stowage and

deployment methods
Increases sun-tracking because of the optics dictating narrow accep-
tance angles
Lack of space heritage of some concentrator material
Higher operating temperature of cells, and the resulting drop in cell
efficiency
The technical problems, cost and schedule risk associated with any new
technology.
Figure 8.44 is the technical data sheet for a GaAs concentrator cell. The
concentrator silicon cell is already being used, whereas use of the GaAs cell
is lagging behind, partly because a higher concentration is needed for peak
efficiency. The conversion efficiency peaks at tens of suns in silicon cells,
whereas it is at hundreds of suns in GaAs.9 For solar power satellites
discussed in Chapter 27, high concentrator GaAs cells would be needed
with a concentrator factor over 1000, perhaps in the range of 5000 to 10,000,
requiring extremely small area cells. It is easier to produce a high efficiency
cell of small area than to produce large area cells with comparable
efficiency. On the other hand, the major disadvantage of the concentrator
cells is that they require focusing optics, which results in added cost.
Two options of concentration cells are depicted in Figure 8.45. In the
dome shaped concentrator, the active cell lays in the center of an aluminum
cup covered with typically a 15-mil transparent silicon dome. Since the
dome lens produces high level of concentration, even slight misalignments
due to a small pointing error can result in significant loss of power. The thin
film inflatable array with dome concentrator is under development for
FIGURE 8.44 GaAs concentrator cell performance data: Typical 22% efficiency at 20 Suns.
space antennas and is being considered for extending into power systems,
particularly for high power electrical propulsion.10 The line concentrator
concept, on the other hand, produce relatively less concentration, but is less
sensitive to the pointing error.
Concentrator solar cells can dramatically reduce the cell cost, which is 35
to 45% in conventional arrays. However, two key issues seriously affect the
cell design at high concentrations. One is the minimization of the series
resistance, and the other is more active cooling of the array to remove the
energy not converted into electricity. The mounting structure can be a
flexible cloth or rigid panel made of a honeycomb structure. The rigid panel
approach takes away some mass benefit of the flexible blanket array, but
Solar Array 185
FIGURE 8.45 Solar concentrator cell construction.
offers well-proven deployment and ability of placing PV cells on the outer

most panel for the transfer orbit power.
The primary advantage of the concentrator technology is low cost per
watt, which is achieved by replacing the active solar cell area with less
expensive optical surface. A secondary advantage come from the fact that
the enclosed cell is naturally protected from radiation and space plasma.
This makes it suitable for high voltage applications. Tests have shown the
dielectric withstand capability of 1000 V with 400 V operational capability.
A specific power of 200 W/kg at the cell level has been achieved.
Concentration ratios of 3 to 10 have been considered for space and around
100 for ground-based applications. In designing such an array, some areas
of concerns have been:
Degradation of optics.
Thermal control of the solar cells under concentrated sunlight,
particularly on exiting the eclipse. However, the thermal gradients
would be relatively small because of the thin structure.
Error in pointing the array to the sun. The greater the concentration, the
more power loss even with a small pointing error.
The concentration cells have limited benefits in traditional space applica-

tions for two reasons. The reduced active area saves some PV material cost,
but the pointing error may offset the benefit. Moreover, the energy
collection area remains the same, hence the same substrate mass. A
concentration ratio above 10 results in a diminishing rate of return for
space use. The potential space application is in a high voltage array
operating around 500 V or higher. Its low leakage current at high voltage
offers significant benefits, such as in electric propulsion.
The solar concentration technology falls into three main groups described
below.
8.12.1 Stretched Lens Array

Stretched lens array (SLA) technology uses a stretched silicon membrane
Fresnel lens to concentrate sunlight onto the PV cell, usually triple junction.
The line-focused silicon lens provides 8- to 10-fold concentration onto a
radiatively cooled PV cell. The lens is made of 200 mm silicon laminated on
75 mm ceria glass for support and UV radiation protection. The laminated
lens is supported as a stretched membrane between end arches as shown in
Figure 8.46. The PV cells are attached to 190 mm composite sheet supported
as a stretched membrane between two edge elements. The composite has
high thermal conductivity to keep the PV cells cool by radiation. The lens
frames are foldable for launch in the same manner as in a planar solar wing.
The entire solar array wing has a mass density of about 1.5 kg/m2, which
includes lenses, PV cells flex circuit, wiring, composite panels, hinges, yoke,
and the deployment mechanisms. NASA Glenn has tested 27.4% power
conversion efficiency based on the total panel area, yielding about 375
W/m2 at 25 C. At GEO operating temperature of 80 C, the yield could be
300 W/m2, or about 200 W/kg. The pointing error tolerance is approxi-
mately 2 about the critical axis.
Among the advantages of SLA technology are:
High efficiency, hence high specific power

The line focus concentrator offers low cost mass-production of lens, as
the active PV cell area is now 1/8th compared to the conventional solar
array
FIGURE 8.46 Stretched lens array prototype panel. (Source: NASA/GRC.)

Solar Array 187
Relaxation of array alignment to only one axis

It provides inherent protection of the active PV cells against charged
particles, hence suitable for defense satellites in high radiation orbits or
with nuclear threats
Ability to withstand high level of space plasma. It is demonstrated to
withstand cell voltage excursion of 500 V relative to the plasmas with
minimal environmental interaction
Minimized plasma interaction when biased to high voltage offers a
potential use in high voltage electric propulsion around 500 V
The GaInP/GaAs on Ge substrate multi-junction cell is being increasingly

used in SLA due to its high efficiency and manufacturability. The
commercial dual and triple junction cells have 28 C AM0 conversion
efficiency of about 30% at one sun, and even higher at concentration ratio of
several suns. Figure 8.47 shows the I–V characteristic of an 8.5 24 cm lens
aperture cell. The efficiency improvements come from the improved
spectrum response. On the down side, keeping the operating temperature
low for power output maximization requires effective heat spreading and
thermal radiation of the concentrated energy into space. Another drawback
is a potential loss of power generation capability under pointing error.
However, it is noteworthy that such cells have soft angular dependence
shown in Figure 8.48 and Figure 8.49 as reported by Krut and Karam.6,8 The
test module delivered over 97% of peak power at 8 in the plane (soft
angle) and 97% at 4 in the plane (hard angle). Thus, the cell is capable
of tolerating such pointing errors without significant loss in power
generation.
The solar concentrator array with refractive linear element technology
(SCARLET) is pursued by the Ballistic Defense Organization and NASA
Glenn. The experiments flew first in Advanced PV and Electronics
Experiment (APEX), and then again in 1995 on Comet/Meteor spacecraft
FIGURE 8.47 SLA module I–V test results at AM0. (Source: NASA/GRC.)
FIGURE 8.48 Off-normal performance of 10 crystalline silicon cell in -plane (easy angle
plane).
with both linear and parabolic concentration array with GaAs/Ge cells. The
results are reported by Jones et al.11 This technology is suitable particularly
for high voltage arrays. It was also tested on NASA/JPL’s Deep Space 1
spacecraft to provide 2500 W power to the spacecraft and the ion propulsion
engine. It demonstrated 200 W/m2 and 45 W/kg specific power12 with a
dome lens.
8.12.2 Light Concentrating Panel

In light concentrating panel (LCP) technology, a pair of optically flat panels
in a channel shape reflect sunlight onto the PV cells as shown in Figure
FIGURE 8.49 Off-normal performance of 10 crystalline silicon cell in -plane (hard angle
plane).
Solar Array 189
8.50.13 The concentration ratio, which can be achieved using such reflectors,
has an upper limit of 2.75. The reflector panels can be rigid aluminum or
fluorinated polyamide film. In the latter, the reflective surface is achieved
by vacuum sputtering aluminum onto a polyamide substrate. The latter is
more advantageous for mass reduction. The reflectors roll up on themselves
and are stowed besides the folded PV panels, leaving one panel exposed to
generate power during the transfer orbit. A system of pulleys and cables
synchronizes the panel deployment and maintains the support booms
parallel to each other. Springs in the support booms keep the reflectors
under tension after the PV panels are deployed. The booms at every other
panel supports the reflectors and tensions them. The ends of the booms can
rotate past the perpendicular to the PV panels, making the reflectors
illuminate the entire length of the PV array even at extreme diurnal motion
of the sun in geosynchronous satellites. There are no shadows on the cells
even when the array angle is not zero. A panel length of at least twice its
width gives the maximum packing factor. The reflector flatness is critically
FIGURE 8.50 Light concentrating panel assembly.

(From M.A. Brown and J. Moore, Proc. 36th IECEC, ASME, p. AT-38, 2001. With permission.)
important to avoid local hot spots. It is estimated that such a channel

concentrator design with polyamide film reflectors can yield specific power
of 180 W/kg using 30% efficient PV cells.
LCP technology was demonstrated on Mightysat II launched in July
2000.14 Its design concentrates sunlight using lightweight composite mirror
strips forming an orthogrid of rectangular mirror troughs. Light with
concentration ratio of about 2.75 to 1 is reflected onto standard solar cells.
The cells are mounted on a high-conductivity flat composite laminate with
an integrated Kapton film. The mirror and cells are mounted on a composite
base sheet having high thermal conductivity to conduct the heat away. The
PV cells interconnect to a flexible wiring layer integrated into Kapton film.
The box-shape array provides the needed structural rigidity, replacing
the honeycomb structure in traditional planar panels. The active area with
PV cells is now reduced to one third. Stern and Bonebright estimate that the
solar panel cost can be reduced by 40% by using this technology.14
The original solar array design in Boeing’s BSS-702 bus was the LCP type.
It was used to generate maximum power from an array stowed in a given
fairing. The design used angled solar reflector panels along both sides of the
wing with GaInP2/GaAs/Ge PV cells mounted on a graphite face sheet on
an aluminum honeycomb substrate. After all the tie-rods are released, the
panels unfold like an accordion and the wings are deployed in a sequence
shown in Figure 8.51. The reflectors then unfold in a coordinated manner.
The BSS-702 concentrators were essentially mirrors fixed on both sides of
the solar panels at an angle giving the panel a distinctive U-shape, reflecting
the sunlight, and generating about 30% more power. However, Boeing
detected premature power loss experienced by several BSS-702 satellites
launched in 2000 and 2001. The average loss was 6% per year. The rate of
decline depended on how the satellite was used. The problem was
suspected in the concentrators’ vulnerability to contamination by gaseous
emissions from the spacecraft, gradually darkening the mirror surface.
While the exact cause of the trouble remained unclear, it was determined
that the xenon-ion propulsion system used on the spacecraft was not the
cause. However, the LPC arrays were replaced with a traditional planar
design. To boost the power without using concentrators, Boeing switched to
solar panels with triple junction, instead of double junction cells, and
adding panels to the arrays.
The root cause of the continuous linear degradation of power in the BSS-
702 channel array was orally presented at the IECEC 2003 conference in
Portsmouth, VA. It has been determined to be contamination with a volatile
condensable material, coupled with the concentrator geometry. The two
factors jointly reduced the reflectance, reducing the transmission of
illumination, ultimately leading to mechanical failure due to additional
shrinkage. The contamination was due to out-gassing accumulated on the
channel panels. Planar arrays do not accumulate and also operate at lower
temperature. Moreover, multi-junction cells are more sensitive to such
Solar Array 191
FIGURE 8.51 Light concentrating panel stowed and being deployment.

(From M.A. Brown and J. Moore, Proc. 36th IECEC, ASME, p. AT-38, 2001. With permission.)
contamination. Tests have eliminated possibilities of ion thruster and UV

darkening, for example.
8.12.3 Parabolic Concentrator

In this technology, which is shown in Figure 8.52, the solar energy is
concentrated onto a small area by using a reflective mirror in a parabola of
revolution. Referring to the geometry shown in Figure 8.53, the solar ray
bundle reflected from the center of the mirror form a circle on the focal
FIGURE 8.52 Parabolic concentrator for dynamic power system.
plane. The diameter of this first order image is di ¼ 2f tan , where is the
solid angle of the sun image, which is 0 0 0 32 00 from the Earth, and f is the
focal length. This gives di ¼ 0.0093f. A ray bundle reflected from the edge of
the mirror will form an ellipse on the focal plane. This distortion is known
as aberration. The intensity of the sunlight is uniform across the first order
image, but falls off to zero at the aberration edge. All the energy collected is
contained within the aberration image. The fraction of the total energy
collected which is concentrated in the first order image is

1 þ cos 2
i ¼ ð8:30Þ
2
where is the angle shown in Figure 8.53.
FIGURE 8.53 Parabolic concentrator geometry.

Solar Array 193
The concentration ratio, Ci, of the paraboloid is defined as
energy density in the first order image

Ci ¼ ð8:31Þ
energy density received from the sun
Obviously, there are two concentration ratios, one for the first order image
and the other for the aberrated image. Their values can be derived as
follows:
Ci ¼ 46100 sin2 ð8:32Þ
Co ¼ 11550 ð sin 2Þ2 ð8:33Þ
where C0 is the concentration ratio for the aberrated image.

Thus, very large concentration ratios of tens of thousand are possible in
the first order image, creating the energy density of tens of megawatts per
square meter in the Earth’s orbit.
Due to tolerances of construction, practical concentrators depart appre-
ciably from the above theoretical values. The deviations could be from
imperfect parabola in the form of flat or over-curved spots, asymmetry or
misalignment with the mirror axis. A practical mirror is often described by
its index of perfection, defined as
actual Ci
i ¼ ð8:34Þ
theoretical Ci
For optical quality mirrors, i is near unity, whereas practical mirrors may
have i around 0.50 or less.
If a misalignment exists between the sun line and the mirror axis, the
image moves with respect to the axis and also becomes distorted. The
amount of the lateral displacement of the center of the image, or its edges,
can be shown in terms of Figure 8.53 to be

$ 2 f
¼ ð8:35Þ
57:3 cos m
where ¼ displacement of the image from the zero error position and i ¼
angular misalignment error in degrees.
References
1. Carlson, D.E., Recent advances in photovoltaics,in Proceedings of the 30th
Intersociety Energy Conversion Engineering Conference, ASME, 1995, pp. 621–
626.
2. Cook, G., Billman, L., and Adcock R., Photovoltaic Fundamental, DOE/Solar
Energy Research Institute, February 1995, Report No. DE91015001.
3. Chern, P.C. and Leipold, M.H., Stress Fiat and Proof Testing of Silicon Wafers,
Jet Propulsion Laboratory, 1987, Report No. NASA TSP-5.
4. Delleur, A.M., Kerslake T.W., and Scheiman D.A., Analysis of direct solar
illumination on the backside of space station cells, in Proceedings of the 34th
Intersociety Energy Conversion Engineering Conference, ASE, 1999, Paper No.
01-2431.
5. Gluck, P.R. and Bahrami, K.A., A simple algorithm to compute the peak
power output of GaAs/Ge solar cells on the Martian surface, in Proceedings
of the 30th Intersociety Energy Conversion Engineering Conference, ASME,
1995, pp 321–325.
6. Karam, N. H. et al. Development and characterization of high efficiency
GaInP/GaAs/Ge dual and triple junction solar cells, IEEE Transactions on
Electron Devices, July 1999, pp. 2116–2125.
7. Guha, S. et al. Low cost and light weight amorphous silicon alloy solar
array for space applications, in Proceedings of the 34th Intersociety Energy
Conversion Engineering Conference, ASE, 1999, Paper No. 2553.
8. Krut, D.D. et al. ‘‘High efficiency space concentrator with soft angular
dependence and related cell technology, in Proceedings of the 36th
Intersociety Energy Conversion Engineering Conference, ASME, 2001, Paper
No. AT-90.
9. Algora, C. and Diaz, V., Influence of series resistance on guidelines for
manufacturing of concentrator p-on-n GaAs solar cells,in Progress in PV
Research and Applications, John Wiley and Sons, 2000, Vol. 8, pp. 211–225.
10. Laug, K.K. and Holmes, M.R., Paraboloidal thin-film concentrators and
their use for power applications, in Proceedings of the 34th Intersociety
Energy Conversion Engineering Conference, SAE, 1999, Paper NO. 2552.
11. Jones, P.A., Murphy, D.M., and Piszczor, M., A linear refractive PV
concentrator solar array flight experiment, in Proceedings of the 30th
Intersociety Energy Conversion Engineering Conference, ASME, 1995, Vol. I,
pp. 309–314.
12. O’Neill, A. et al. The stretched lens array (SLA), an ultra light concentrator
for space power, in Proceedings of the 36th Intersociety Energy Conversion
Engineering Conference, ASME, 2001, pp. 79–84.
13. Brown, M.A. and Moore, J., Solarcon concentrator solar array, in
Conference, ASME, 2001, Paper No. AT-38, pp. 75–78.
14. Stern, T.G. and Bonebright, P., Flight tests of a solar array concentrator on
Mightysat II, in Proceedings of the 35th Intersociety Energy Conversion
Engineering Conference, ASME, 2001, Paper AT-37, pp. 69–74.
Chapter 9
Battery
9.1 Introduction
The energy storage is required to meet the spacecraft load demand not only
during eclipse, but also when the demand exceeds the power generation at
any time. The most widely used energy storage technology is the battery,
which stores energy in an electrochemical form. There are two basic types:
Primary battery, in which the electrochemical reaction is irreversible. It

cannot be reused after full discharge, so it is discarded. It finds
applications in very short duration missions as the primary source of
power. It has a specific energy that is an order of magnitude higher
compared to the secondary battery.
Secondary or rechargeable battery, in which the electrochemical reaction
is reversible. After a discharge, it can be recharged with direct
current from an external source. It finds applications in long duration
missions where another independent power source is available for
recharging the battery periodically. Most satellites use the secondary
battery.
9.2 Electrochemical Cell

The battery is made of numerous electrochemical cells assembled in a
series–parallel circuit combination to obtain the required voltage and
current. The internal construction of a typical electrochemical cell is shown
in Figure 9.1. It has positive and negative electrode plates with insulating
separator and a chemical electrolyte in-between. The two electrode plates
are connected to two external terminals mounted on the casing. The cell
stores electrochemical energy at a low electrical potential. The cell voltage
depends solely on the electrochemistry, and not on the physical size.
Commonly used electrochemistries produces 1.5–3.5 V when fully charged.
The cell capacity, denoted by C, depends on the physical size. It is defined
as the ampere-hours (Ah) charge the cell can deliver at room temperature
until it reaches a cut-off voltage of about 2/3rd of the fully charged voltage.
The battery can deliver C amperes for 1 h or C/n amperes for n hours. The
capacity measures the Ah output at the terminals, not the input or that
195
FIGURE 9.1 Typical electrochemical cell construction.
stored between the plates. A 1.5-V cell discharged to 1.0 V delivers

practically full capacity of the cell. The Ah capacity to 0.1 V discharge is
only a few percent greater than that at 1.0 V discharge.
The battery voltage rating is stated in terms of the average voltage during
discharge. The higher the battery voltage, the higher the number of cells
required in series. The product of the voltage and the Ah rating makes the
energy rating in watt-hours (Wh) the battery can deliver to a load from the
fully charged state.
The battery charge and discharge rates are stated in fraction of the
capacity. For example, charging a 100-Ah battery at 10-A rate is said to be
charging at C/10 rate. Discharging this battery at C/2 rate means drawing
50 A. At this rate the battery will be fully discharged in 2 h. The state of
charge (SOC) of the battery at any instant is defined as
Ah capacity remaining in the battery

SOC ¼ ð9:1Þ
rated Ah capacity
The state of charge affects the cell voltage, specific gravity, and freezing
point of the electrolyte. The electrolyte in a fully charged battery has high
specific gravity and freezes at a much lower temperature. On the other
hand, a fully discharged battery freezes at a higher temperature. This shows
the importance of keeping the battery fully charged when exposed to low
temperatures.
The battery depth of discharge (DOD) is defined as
Ah capacity drained from fully charged battery

DOD ¼ ð9:2Þ
rated Ah capacity
Obviously,
DOD ¼ 1 SOC ð9:3Þ

Battery 197
Major secondary batteries used in the spacecraft industry at present are

nickel cadmium (NiCd), nickel hydrogen (NiH2), and lithium-ion (Li-ion).
New electrochemistry is being continuously researched by NASA for space
applications,1 and by the United States Advance Battery Consortium for
a variety of ground based applications — consumer electronics, electric
vehicles, utility load leveling, and renewable power systems. Lithium–
polymer (Li-poly) and nickel metal hydride (NiMH) are two such examples
in the commercial world.
The selection of electrochemistry for a given application is a matter of
performance and cost optimization. The following figures of merit are often
used in comparing relative performance of various electrochemistries:
Specific energy, defined as the gravimetric energy density, which is the

energy stored per unit mass, Wh/kg.
Energy density, defined as the volumetric energy density, which is the
energy stored per unit volume, Wh/l.
Specific power and power density, defined as the power the battery can
practically deliver per kilogram of mass and liter of volume,
respectively. It is sometimes necessary to think in terms of the
power parameters also, since the internal resistances of the battery
may limit the rate at which the energy can be discharged within
practical design limits.
Cycle life, defined as the number of charge/discharge (C/D) cycles the
battery can deliver while maintaining the cut-off voltage.
9.3 Types of Battery

The NiCd battery had been the workhorse of the spacecraft industry since
the earliest missions, and is still used in some missions. However, since the
mid-1980s, it has been replaced with the NiH2 battery for general use. The
NiH2 provides deeper DOD for comparable cycle life, thus requiring lower
Ah capacity, which translates into lighter weight. Today, the industry
appears to be moving towards the lithium based batteries for potentially 2
to 5 times the specific energy compared to the NiH2 chemistry. No single
electrochemistry available on the market can meet the wide range of space
mission requirements. All chemistries will perhaps continue in use where
they fit the best for a minimum mass and cost design. However, the more
commonly used electrochemistries and their unique construction and
performance features are presented in this chapter.
9.3.1 Nickel Cadmium

The NiCd is a matured electrochemistry that powered almost all satellites
until mid-1980s. It comes in a sealed vented prismatic cell with positive and
negative electrode plates made of a sintered nickel structure. The positive
plate is impregnated with nickel hydroxide active material and the negative
plate with cadmium hydroxide active material. The plates are die-cut with
corner radii and coined edges to prevent shorting. The negative plate
design includes a thin TeflonTM impregnation on the surface. This feature
extends the operational life by enhancing oxygen recombination and
retarding cadmium migration. The positive and negative plates are
separated by nonwoven nylon or asbestos, which has been replaced with
Pellon 2536TM. The electrolyte is potassium hydroxide and the casing is
made of stainless steel.
The NiCd electrochemistry offers relatively lower specific energy, is
temperature sensitive, and has shorter cycle life. Moreover, the cadmium
has been placed under environmental regulatory scrutiny. For these
reasons, NiH2 has replaced the NiCd for most space applications.
9.3.1.1 Memory effect

A major disadvantage of the NiCd battery comes from its peculiar
characteristic known as the memory effect. The NiCd battery remembers its
DOD in most frequent use in the recent past, does not work well beyond
that, and partially loses its unused capacity for subsequent use. For
example, if a NiCd battery is repeatedly charged and discharged 25% of
its capacity to point M in Figure 9.2, it remembers point M. If the battery is
discharged beyond point M in a subsequent use, the cell voltage will drop
much below its original normal value as shown by dotted line in the figure.
FIGURE 9.2 Memory effect in NiCd cell.

Battery 199
The end result is a partial loss of capacity after repeated shallow discharge
cycles. The phenomenon is like losing a muscle in the human body due to
lack of use over a long time.
9.3.1.2 Reconditioning
A remedy to counter the memory effect for restoring the battery to full
capacity is reconditioning, in which the battery is fully discharged to almost
zero voltage and then fully charged to about 1.55 V per cell. In the spacecraft
power system, this is done twice a year at a convenient time in the orbit. In
GEO, it is done outside the two eclipse seasons. In LEO, however, the
spacecraft must carry an extra battery just for reconditioning purposes,
adding significant mass and cost. Other types of batteries have negligible
memory effect, offering a significant advantage over the NiCd battery.
A battery letdown unit is used to recondition the battery in orbit
operation. It is capable of discharging the battery completely at more than
one discharge rate through dedicated resistors, either at the cell level or at
the battery level. In the latter case, each battery has its own battery letdown
resistor. Generally the battery reconditioning starts at high rate around
C/50. Upon the first cell reaching 0.5 V, the battery letdown unit is
commanded to disconnect one of the discharge paths, and continue the
battery discharge at C/100 rate. Upon any battery cell reaching 0.1 V, the
remaining discharge path is disconnected to terminate the discharge. After
the full discharge, the battery is then brought back to full charge in a normal
manner. The battery cell voltage monitor provides information on each cell
voltage during reconditioning and also during flight. The cell voltage data
during flight is not directly useful, as little can be done if anomalies are
found in flight. However, the data can be useful to the customer in the
following ways:
Predict the remaining life

Plan replacement in serviceable spacecraft
Reduce the load if necessary and possible
Redesign the battery for future spacecraft
9.3.1.3 Advanced NiCd

The NiCd electrochemistry has been relatively fixed for several decades
with some incremental improvements in performance in terms of life and
specific energy. For example, reduction in electrode loading levels has been
used in recent years to limit internal stresses that cause cell degradation.
Despite a series of improvements, NiCd does not match other electro-
chemistries at present.
A major improvement program was undertaken under Department of
Defense funding. Under this program, optimized electro-deposited nickel
and cadmium electrodes, advanced separator system, and special charge

control methods were developed to achieve longer life at higher DOD, and
eliminate the memory effect. Such space-qualified cells are available from a
few sources, one under the trade name SuperNiCdTM. These cells compete
well with NiH2 at a power level of 1 to 2 kW, and can reduce the battery
weight by a factor of 2 or more. Such an advanced NiCd battery gives about
twice the life at a given DOD, or twice the DOD for a given life. The final
result is the same — a significantly lighter battery. In the tests simulating
LEO service, Sanyo’s advanced NiCd battery of 35-Ah capacity survived
more than 35,000 cycles at 20 C and 40% DOD.
The advanced NiCd competes with NiH2 for some niche low power
spacecraft, particularly in defense applications. It has negligible memory
effect. Its cycle life for a given DOD is several-fold higher than the
conventional NiCd and a few-fold higher than NiH2.2 It had been used in
over 50 missions by the year 2000 in capacities ranging from 5 Ah to 50 Ah,
typically in small satellites in LEO for NASA, the U.S. Navy, and the U.S.
Air Force.3 These batteries are not being reconditioned. The typical design
DOD has been around 75% in 5- to 12-year GEO missions, and 10 to 20% for
1- to 8-year LEO missions. Advanced NiCd gives better performance than
the old NiCd only below 50-Ah capacities, and matches with NiH2 in higher
capacities. The primary limitation for this capacity restriction is the heat
removal difficulty across the plates. Although the charging is endothermic
during initial charge, it becomes exothermic beyond 75 to 80% SOC.
9.3.2 Nickel Hydrogen

The NiH2 has been the most widely used cell in space during the last 20
years, both in LEO and GEO satellites. Over 100 satellites carrying more
than 10,000 cells have been flown. They have accumulated almost one
billion cell-hours of highly reliable operation. The NiH2 battery takes the
best from two electrochemistries, the nickel oxide electrode from the NiCd
and the hydrogen catalyst electrode from the fuel cell. The NiH2 cells can
withstand some abuse in over-charge and over-discharge. On the other
hand, NiH2 has the following disadvantages:
Low energy density in Wh/l

Pressure vessel rupture, handling, and safety concerns
High self-discharge rate around 0.5%
High loss of capacity on storage
In the most widely used configuration of NiH2, each elementary cell

developing 1.25 V is packed in an individual pressure vessel (IPV). Figure
9.3 depicts a simplified view of the IPV cell components. The positive
electrode is made of aqueous impregnated slurry of sintered nickel. The
Battery 201
FIGURE 9.3 NiH2 cell components — simplified view.
negative electrode is made of platinum. The electrodes are stacked in pairs

with the leads extending through the center core. A separator is placed on
each side of the positive electrode. The electrolyte is 26 or 31% potassium
hydroxide (KOH). Other factors include ceramic to metal seals with integral
terminals, double stack supports (no cantilever) for shock resistance, and
zirconium separators that yield longer life. Life tests approaching 42,000
cycles have been demonstrated at 60% DOD.4
A plastic diffusion screen is placed on the back side of each negative
electrode to diffuse gas to the back side. A number of such electrode pairs
with separators are placed in one pressure vessel to obtain the required Ah
capacity. The internal gas pressure rises to about 1000 psig when the cell is
fully charged. To sustain such high pressure, the vessel is made of age
hardened Inconel-718 with burst pressure of at least 2.5 times the maximum
operating pressure at full SOC. It must comply with MIL-STD-1522
pressure vessel requirements. The vessel is made in two parts, each with
an end-dome and one half of the cylindrical center body. The assembled
vessel has the ring of electron beam weld in the center. Because of the high-
pressure design requirement, the vessel contributes significant portion
(about 40%) of the total cell mass.
Two types of positive electrode plates have been used. Both are based on
a sintered nickel substrate, with one impregnated using aqueous procedure
and the other using an alcohol procedure. Both behave almost similarly in
all aspects. Two commonly used electrodes thicknesses are 30 mils for GEO
and 35 mils for LEO applications. The 35-mil electrode gives longer wear
life and slightly higher discharge voltage. However, it uses heavier
electrodes and double separators, hence longer cell length and higher
weight. Some manufacturers process the electrodes in aqueous solution,
whereas some in alcoholic solution, although both are purged later on. The
alcoholic solution gives a little better performance, but is believed to result

in a capacity fading in wet storage in discharged condition. That is, when
revived, the battery does not restore to the same capacity for hundreds of
cycles. Hence, it needs some trickle charging in wet storage. The cell
processed in aqueous solution does not pose such a problem. It comes to
full life on the first recharge after wet storage for up to 3 years.
Battery technology is rapidly changing. Some of the recent improvements
for increasing the specific energy in NiH2 chemistry are described by
Britton.5 A few examples are the catalyzed wall-wicks incorporated in the
KOH solution, and the replacement of traditional sintered substrate used in
nickel electrodes with lightweight porous nickel felt. A positive plate is the
key to all improvements in cycle life, voltage stability, and extended
storage. The positive plate is typically 0.75 to 1 mm thick with active
material loading of around 1.5 g/cm3.
The cell capacity is a function of the loading level of the positive active
material, KOH concentration, and temperature. The earlier NiH2 construc-
tion used 21% KOH and a double layer zirconium separator, which was
replaced by a single layer for mass reduction in GEO satellites. However,
the cell for LEO applications still uses the double layer zirconium with
thicker electrodes to meet the longer cycle life requirement.
The 3.5-inch diameter cell had been standard for a long time, but is
inefficient for larger than 100 Ah capacity due to increased length. The 4.5-
inch diameter cell is a direct scale-up of the widely used 3.5-inch diameter
cell, while maintaining the stack length. The 3.5 and 4.5-inch diameter cells
in 35- to 150-Ah capacity range is currently flying in many satellites. To
extend the Ah range on both sides, 2.5- and 5.5-inch diameter cells have also
been fully developed and are flying in a number of satellites. As seen in
Figure 9.4, the specific energy varies with the ampere-hour rating and also
FIGURE 9.4 Specific energy of NiH2 cells — industry range.

Battery 203
with design variations in the separator, electrolyte concentration, and gas

management provisions such as a catalyzed wall-wick.
For the International Space Station (ISS), the NiH2 was selected after
evaluating many other options, including the regenerating fuel cell. The ISS
NiH2 cell is made of dry sintered Ni electrode, sintered Teflon-bonded
platinum black H2 electrode with Teflon backing and photo etched nickel
grid, 31% KOH concentration electrolyte, and zirconium oxide clothe
(ZircarTM) separators. The positive electrode is electrochemically impreg-
nated using an alcohol–water process. The cell contains two back-to-back
stacks of many pairs of electrodes, placed in two halves of the cylindrical
stainless steel case welded at the longitudinal center.
Compared to NiCd, the NiH2 chemistry has superior charge/discharge
cycle life and low internal resistances. Another attractive feature of the
NiH2 cell is the linear dependence of the SOC with the internal pressure
that can be measured by strain gages mounted on the outside wall.
Temperature compensated strain gages are accurately matched in bridge
circuits, and processed by signal conditioning amplifiers. A two-stage
regulator provides stable power for the circuit. Adhesive and conformal
coating materials are used for mounting the gages and the flexible circuit
board is made of space-qualified, low out-gassing materials. The strain
gages are calibrated individually on each cell.
Although the NiH2 cell does not exhibit a noticeable memory effect, a
number of battery design engineers believe that reconditioning is
required in both NiCd and NiH2; in fact, in all batteries with a positive
nickel electrode. As a result, reconditioning has been done in the past as
a practice inherited from the NiCd experience. Cumulative flight data to
date suggests no need of reconditioning. Most customers do not require
it anymore, and most power system designs have eliminated this
function. However, the battery letdown unit is often installed on NiH2
battery to discharge before shipping, on the launch pad to measure the
capacity, or before placing in storage. If the letdown unit is not made
part of the battery assembly, a separate unit is needed for any one of
these reasons.
The mass distribution in the cell is shown in Figure 9.5. It is dominated by
the pressure vessel, which requires a thick wall to limit the hoop stress
under high pressure. Moreover, the cylindrical shape makes packaging of a
large battery in a 15 to 20 kW satellite difficult due to the large volume.
Figure 9.6 depicts a NiH2 battery assembly with cells mounted vertically.
Over the past decade, cell manufacturers have introduced the following
design improvements to reduce the battery mass and volume.
9.3.2.1 Common and single pressure vessels

The vessel portion of the individual pressure vessel mass can be reduced by
packing two cells in a common pressure vessel (CPV), or by packing all cells
FIGURE 9.5 Mass composition of NiH2 cell.
FIGURE 9.6 50-Ah 22-cell battery assembly using cylindrical NiH2 cells mounted vertically.
Battery 205
in a single pressure vessel (SPV). The CPV looks just like the IPV, except for
the internal connections (Figure 9.7). Two cells internally connected in
series in CPV doubles the terminal voltage to 2.5 V. The battery is then
assembled as usual, using one half as many CPV cells. Handling and testing
fewer cells save in the manufacturing cost. The CPV cells have been used
extensively on NASA/JPL programs. However, venting one vessel results
in the loss of two cells, partially offsetting the gain in mass and volume, or
compromising the reliability.
The SPV design places all cells in a single vessel. About 100 SPVs have
successfully flown in various NASA/JPL/Navy/Air Force programs.
Iridium satellites are flying with 50-Ah SPV batteries. The single pressure
vessel requires a thicker wall to limit the hoop stress, partially
compensating the mass savings. Overall, the SPV does save mass and
volume, but the reduced reliability is a significant concern. A pressure
leak or an internal arcing would result in a total loss of the battery.
Moreover, it does not have the means of monitoring the individual cell
voltages. The failed-cell bypass circuit generally used in the IPV battery
design cannot be used in this case.
The following are commonly available NiH2 cells from various vendors:
2.5 inch diameter IPV for small satellites 2 to 20 Ah

3.5 inch diameter IPV 20 to 100 Ah
4.5 inch diameter IPV 100 to 330 Ah
5.5 inch diameter IPV for large satellites 200 to 400 Ah
4.5 inch diameter CPV 2.5 V 60 to 100 Ah
FIGURE 9.7 Common pressure vessel NiH2 cell.

(From D.K. Coates et al. Proc. IECEC, ASME, p. AP-90, 1995. With permission.)
Table 9.1 NiH2 cells for space applications
Rated Design Diameter Tip–tip Weight BOL 10 C 72 h

20 C Ah typea (inches) length (kg) pressureb Cap. Ah OC Ahc
25 BB 3.45 9.2 0.830 500 33 24

50 BB 3.45 10.8 1.450 625 53 37
110 BB 4.67 11.4 2.760 600 120 88
160 BB 4.67 13.3 3.625 850 167 112
200 BB 4.67 14.4 4.700 1000 208 134
250 BB 4.67 17.5 5.900 1000 260 170
25 Re 3.45 9.30 1.000 500 25 18
55 Re 3.45 11.1 1.700 700 55 39
110 Re 4.67 12.1 2.975 600 120 88
160 Re 4.67 14.4 4.100 850 167 112
a
Back to back has single separator suitable for GEO applications. Recirculating has double
separator suitable for LEO applications.
b
Internal cell pressure in pounds per square inch.
c
Ampere-hour capacity at 20 C after 72 h of self-discharge at open circuit.
Typical for all ratings: Discharge rate C/2 to C, charge rate C/20 to C/2, trickle charge rate
C/100 to C/200, 26–31% KOH, operating temperature 10 C to 30 C, Optimum operation at
5 C, Non-operating temperature 20 C to 40 C.
5.5 inch diameter CPV 2.5 V 100 to 200 Ah

10 inch diameter SPV for small satellites 15 to 50 Ah
Table 9.1 gives data on commonly available IPV cells.
9.3.2.2 Dependent pressure vessel

The dependent pressure vessel (DPV) uses hockey-puck shape cells.6 Each
cell has its own pressure vessel, but with reduced wall thickness and flat
sides. The casing is made of SS-304, a laser closure weld and the usual nylon
compression seals. Nonwoven polymeric separators, 26% KOH, a micro-
fiber nickel electrode and a binary supported catalyst in a hydrogen
electrode are common. The maximum operating pressure is about two
thirds that in IPV. The cells are packed in a canteen-rack configuration,
clamped between heavy end plates with four tie-rods (Figure 9.8). The
pressurized fully charged cells depend on adjacent cells for structural
support, offering all the reliability features of the IPV battery with reduced
volume. An internal failure in one cell can be accommodated by having one
spare cell as in the IPV battery. The electrical performance of the DPV is
identical to that of the IPV under the same loading conditions. The DPV cell
results in somewhat higher specific energy and significantly higher energy
density. In conceptual designs using DPV, 30% mass reduction and 50%
volume reduction have been estimated. However, some of the concerns of
the DPV are:
Battery 207
FIGURE 9.8 Dependent pressure vessel NiH2 battery.

A pressure leak in one cell may potentially crush all cells in a zipper
mode. This concern is addresses by the cell designed with an internal
bracket to support the failed cell and prevent crushing under the
pressure of adjacent cells (Figure 9.9).
Complex thermal control due to concentrated heat dissipation
Must buy the whole battery from the vendor, which is uncommon at
most satellite builders.
FIGURE 9.9 Dependent pressure vessel internal support bracket.

Calibrating the strain gage reading with pressure on a hockey-puck cell

is difficult. Without a reliable pressure data, the DPV has no advantage
over the NiCd for interpreting the battery SOC.
For these reasons, the DPV does not offer significant advantage over the IPV
cell, and is unlikely to find applications in space in the foreseeable future.
9.3.3 Nickel Metal Hydride

Concerns in working with cadmium in the NiCd battery and disposing of it
helped to spark interest in using the NiMH battery for a variety of
applications. In general, NiMH electrochemistry presents few health and
safety risks, primarily because the electrode materials are nontoxic. Its
anode is made of a metal hydride, which eliminates the environmental
concerns of cadmium. The NiMH offers an improvement in specific energy
over that in NiCd. Another performance improvement is that it has
negligible memory effect. The NiMH, however, is less capable of delivering
high peak power, has high self-discharge rate, is adversely sensitive to high
temperature, and therefore is susceptible to damage due to over-charging.
Compared to NiCd, NiMH is expensive at present, although the future
price is expected to drop significantly. This expectation is based on current
development programs targeted for its large-scale application in electric
vehicles. For space applications, it has been flown on a few Japanese,
European, and American flights.
Electrochemically, the NiMH is similar to the NiH2, except that the
working hydrogen is stored as solid metallic hydride rather than gas.
Having no internal pressure to withstand, these cells are made prismatic.
This significantly improves the energy density and safety. Having less
volume to pack, the battery requires less support structure, hence improves
the specific energy as well. The disadvantage is that the linear pressure
versus SOC relation of the NiH2 is no longer available. Hermetically sealed
aerospace cells have been successfully tested for several thousand C/D
cycles at 50% DOD. Sanyo’s 37-Ah NiMH battery survived 15,243 cycles at
40% DOD. Their loss in capacity when operating at 0 C is much greater
than that in a NiCd battery.
The NiMH is prismatic and has the same operating voltage as the NiCd.
Therefore, the two are easily interchangeable. The NiMH, however,
produces heat during charging as opposed to the endothermic charging
of NiCd.
The key features of NiMH, as compared to NiH2 cells, are as follows:
It offers some (10 to 20%) improvement in Wh/kg

It offers significant (100%) improvement in Wh/l
Battery 209
It is less sensitive to temperature, hence it has lower thermal control

cost
It has a much lower operating pressure, hence is much safer
The cell rupture due to internal gas build-up is one possible hazard. The cell
design with vent to release the gas pressure is therefore needed. The risk of
venting hydrogen is minimized by limiting its amount during over-
charging. As a health risk, nickel and its compounds may be carcinogenic,
as in other cells.
9.3.4 Lithium-Ion
Lithium-ion (Li-ion) technology is a new development that offers signifi-
cantly higher specific energy and energy density over that of NiH2. The
prismatic shape (Figure 9.10) results in higher energy density. The large
improvement in specific energy comes from lithium’s low atomic weight of
6.9.
FIGURE 9.10 Lithium-ion cell of 35-Ah, 3.6-V ratings in prismatic casing.

(From J.A. DeGruson, Proc. 36th IECEC, ASME, p. AT-57, 2001. With permission.)
Lithium-ion batteries are widely used in laptop computers, digital

cameras, and other commercial applications. The positive electrode is
LiNiCoO2, the negative electrode is a mixture of two graphites with a non-
PVDF binder, and the electrolyte is LiPF6 salt. The cell construction uses a
stainless steel prismatic case with a safety vent between the two terminals.
The terminals are insulated from the case by polymeric seals as in the NiH2
cell. Small cells are cylindrical and large cells are rectangular prismatic in
shape. The prismatic shape gives high tolerance to shock and vibrations
environments.
Since the Li-ion is non-aqueous chemistry, it is free from electrodes
corrosion. On the negative side, lithium is highly reactive metal and must
be stabilized by encapsulating the molecules in carbon using an expensive
form of graphite. The lithium electrode reacts with the liquid electrolyte,
creating a sort of passivation film. Each time the cell is discharged and then
charged, the lithium is stripped away, a free metal surface is exposed to the
electrolyte and a new film is formed. The cell uses thick electrodes to
compensate for this. Otherwise the life would be short.
The Li-ion has the end of discharge cut-off voltage of 2.7 V, average
discharge voltages of 3.5 V and end-of-charge voltage of 4.2 V. The average
discharge voltage of 3.5 versus 1.25 for NiCd and NiH2 requires about one
third the number of cells in series for a given battery voltage, thus reducing
the assembly and test costs.
In operation, lithium-ion electrochemistry has a high charge efficiency.
The coulombic efficiency is 100% all the way to 100% charging. Other
desired characteristics of a space-qualified Li-ion cell are high-rate
capability and potentially long cycle life at high DOD. Its low temperature
sensitivity supports operation over a wide temperature range. It has low
internal impedance, less than 1 m in a 50-Ah cell, which supports high
charge and discharge rates. It can deliver high short time peak power
without an adverse effect on life.
The Li-ion cell has a potential of achieving 150 Wh/kg and 400 Wh/l,
along with abuse tolerance. However, it costs more at present, but can be
cost competitive with NiH2 in the same volume production. It is vulnerable
to damage under over-charging or other shortcomings in the battery
management. Therefore, it requires more elaborate charging circuitry with
adequate protection against over-charging. Batteries need to be trickle
charged, which requires extreme care in Li-ion. Each cell may need charge
control, which requires additional electronics.
Several roadblocks for space applications that have been recently
addressed are:
Relatively short life demonstrated so far

Sensitive to over-charge and over-discharge
High internal resistance at low temperature
Battery 211
The long calendar life for GEO and long cycle life for LEO are two issues in
Li-ion technology that still need to be proven with a sufficiently long flight
history. To date, 6000 cycles of charge/discharge at 25% DOD has been
demonstrated on 35-Ah cells with stable performance. A Li-ion battery
requires trickle charge at a rate of C/50 to C/100. The self-discharge rate is a
few percent of that in the NiH2, but increases at high temperature. The
internal impedance is low but increases at low temperature. The two must
be traded in the design.
The present development funded by NASA and the U.S. Air Force has the
following goals at the cell level:
Cell capacity of 20 to 200 Ah

Life of 30,000 cycle at 50% DOD
Life of 40,000 cycles at 40% DOD
Temperature range 40 C to þ80 C
Charge rate C/5 to C
Discharge rate C/2 to 2C
The 2001 test results on Mars Lander batteries showed specific energy of
about 150 Wh/kg.7 Nominally 5-Ah cells were tested for capacity when
discharged at a 1-A rate at various temperatures. The cells were cycled from
2.4 V to 4.1 V. The tests showed significant degradation at low temperature,
with only 60% capacity at 30 C. In internal heat generation tests on 25-Ah
cells at 10 C during charge and discharge at 5-A rate, the heat generation
spiked at the end of discharge due to increased polarization as the cell
reached full discharge. The heat generation reduction at the end of charge
was due to charge current tapered down to zero.
The tests on 10-Ah cells charged and discharged at the same temperature
have shown8 that 600 cycles at 100% DOD at 20 C resulted in 6% capacity
loss. Fellner9 reports test data on Li-ion cells cycled as follows for GEO and
LEO applications:
For LEO: 20-Ah cells, 40% DOD, charged to 4.1 V at 0.5C rate for 1 h,
discharged at 0.8C rate for 30 min
For GEO, 20-Ah cells, 100% DOD, charged at C rate to 4.1 V in 3 h,
discharged at C/5 rate to 2.5 V.
The initial test data for a discharged cell show that the internal resistance
grows exponentially with decreasing temperature for a given state-of-
charge.
The Li-ion batteries can release lithium gas and flames when over-
charged. This danger needs effective control of the over-charge. For this
reason, the maximum charge voltage and minimum discharge voltage must
be maintained for successful incorporation of Li-ion cells into the space

power system design. In this respect, Li-ion chemistry is very different to
that of the NiH2 in that the Li-ion does not tolerate over-charge or over-
discharge. As a battery of many cells in series and parallel is cycled, it often
becomes unbalanced due to small cell-to-cell differences in the rate of
capacity fade, rate of impedance increase, or rate of self-discharge that are
accumulated over many cycles. For charge balancing, various techniques
such as switched capacitor and resistance or inductor equalization have
been explored.
Li-ion has a better low temperature performance in the 20 C to þ40 C
range. It can operate even at 40 C with some capacity loss. Poor cell
performance at lower temperature is attributed to the electrolyte becoming
viscous or solid, and poor lithium diffusivity in the electrodes. An 8-cell 25-
Ah battery was space qualified in 2001 for 2000 cycles at 100% DOD for the
Mars Lander mission requiring such a low temperature range. The battery
was designed for a 3.0 to 4.1-V operation, 2.5-V cut-off voltage, and C/5 and
C/10 charge rates. Tests reported by Bruce and Marcoux10 at various
temperatures cover design and fabrication features of large capacity cells up
to 100 Ah. In cells for aircraft applications, the main concern is irreversible
capacity loss at elevated temperature. The loss in capacity at high
temperature comes primarily from the positive electrode degradation. In
cells for space use, however, the shelf life and cycle life are critical.
9.3.5 Lithium Polymer

The electrochemical basis of the Li-ion solid polymer electrolyte (Li-poly)
battery is similar to that of Li-ion, except that the electrolyte of the Li-ion is
replaced by a solid polymer that encapsulates the electrodes. The cell is
constructed with a film of metallic lithium bonded to a thin layer of solid
polymer electrolyte. The solid polymer enhances the specific energy by
acting as both the electrolyte and the separator. This construction also offers
an improvement in safety compared to Li-ion by reducing flammability.
Moreover, the metal in solid electrolyte reacts less than it does with liquid
electrolyte. There are various types of Li-polymer cell, with markedly
different characteristics. An advanced battery is under development by
NASA with an industry team for aerospace and consumer applications. The
goals of these programs are listed in Table 9.2. The 35,000 cycles life at 25%
DOD addresses the LEO satellites requirements, while the 2000 cycles life at
100% DOD would meet the GEO satellite requirements.
On the safety side, the Li-ion system is relatively less reactive and more
reliable than the present lithium metal systems, and may have performance
advantage over some solid polymer electrolyte systems. However, the
possibility of electrolyte spill is a disadvantage over a solid electrolyte.
Some materials used in the Li-ion cells may be toxic and carcinogenic, such
Battery 213
Table 9.2 NASA goals for Li-ion research programs at cell level
Performance parameter Li-poly (goals) Present Li-ion Present NiH2
Specific energy, Wh/kg 200–250 125–150 50–65

Energy density Wh/l 350–400 300 –
Life at 100% DOD, cycles 2000 1000 –
Life at 25% DOD, cycles 35,000 – –
as lithium, nickel and arsenic compounds, and possibly cobalt, copper and
polypropylene.
9.3.6 Silver Zinc

The silver zinc (AgO-Zn) secondary battery has high specific energy but
shorter cycle life. It is suitable in missions requiring a low number of C/D
cycles. It is used in military aircraft and spacecraft, launch vehicles, and
undersea applications. The specific energy ranges from 125 to 250 Wh/kg
and energy density 200 to 600 Wh/l. Prismatic cells are available in a wide
range of ampere-hours in low-rate (LR) and high-rate (HR) types. Because
resistance of the silver monoxide electrodes vary little with the state of
charge, a stable operating voltage is provided until nearly all the capacity is
drawn.
The AgO-Zn cell comes in two varieties: dry charged and dry unformed.
In the dry-charged cell, the positive and negative electrodes are charged at
factory, but the electrolyte is added at the launch site. The dry unformed
cell is charged and the electrolyte is added at the launch site. In both cases,
the cell comes to a full charge voltage of 1.86 V open circuit after the soak
time. In subsequent C/D cycles, the full cell charging is terminated at 2.05
V, and the open circuit voltage drops to 1.86 V. Full charge level is indicated
by the voltage as well as the electrolyte level on a visible scale.
9.4 Electrical Circuit Model

The battery works as a voltage source with small internal resistance. Its
electrical circuit model is shown in Figure 9.11, where Ei is the internal
electrochemical voltage and Ri is the internal resistance. The Ei decreases
and Ri increases linearly with the ampere-hour discharge. That is, the
internal voltage is lower and the internal resistance is higher in a partially
discharge battery as compared to their values — designated as Eo and Ro,
respectively — in the fully charged state. Quantitatively,
Ei ¼ Eo K1 DOD
ð9:4Þ
Ri ¼ Ro þ K2 DOD
FIGURE 9.11 Equivalent electrical circuit of battery showing internal voltage and resistance.
where K1 and K2 are constants of the electrochemistry to be found by curve

fitting the test data. Under load, the terminal voltage drops with increasing
load, and the operating point P is the intersection of the battery line and the
load line as shown in Figure 9.12.
9.5 Properties and Performance

The electrical performance of the battery depends on the electrochemistry
and on many other parameters in a highly nonlinear manner. This makes
the battery design one of the most difficult tasks for the power system
engineer. Major performance characteristics of the battery in general are
discussed in this section, mostly for NiH2, the widely used electrochemistry
at present.
9.5.1 Charge/Discharge
The cell voltage varies during a typical charge/discharge (C/D) cycle as
was shown earlier in Figure 4.3. The charge/discharge characteristic also
depends on the charge rate, discharged rate, temperature, and age of the
cell. Figure 9.13 shows such variations for a LEO satellite.
FIGURE 9.12 Battery line and load line intersecting at operating point.
Battery 215
FIGURE 9.13 NiH2 charge/discharge cycle at different rates in LEO.
Typical average NiH2 and NiCd battery cell voltages are as follows:
1.55 V when fully charged

1.45 V average during charge
1.25 V average during discharge
1.10 V at 80% DOD
1.00 V when fully discharged
After discharging certain ampere-hours, the battery requires more

ampere-hours of charge to restore the full state of charge. The charge/
discharge ratio is defined as the ampere-hours input over the ampere-hours
output for full state of charge. This ratio depends on the charge and
discharge rates and also on temperature as shown in Figure 9.14. For
example, a C/D ratio of 1.1 means the battery requires 10% more ampere-
hours for restoring to fully charged state than what was discharged.
Charging in LEO is usually at a much faster rate, which is more efficient
than the slow charging rate typical in GEO. Therefore, a low C/D ratio of
1.05 to 1.1 is needed in LEO missions, while a higher ratio of 1.1 to 1.2 is
needed in GEO missions. A high C/D ratio causes excessive corrosion of
the nickel electrodes, and large amount of oxygen leading to heating and
popping during recombination on the hydrogen electrode. On the other
hand, low C/D ratio causes reduced usable capacity, low discharge voltage,
and capacity degradation.
A new cell stores over 20% higher capacity when charged at 20 C
compared to that at þ20 C (Figure 9.15), but only about 10% higher after
about 100 cycles of charge/discharge cycles (1 year in GEO). The maximum
charge voltage also varies with temperature (Figure 9.16). The end of charge
FIGURE 9.14 NiH2 charge/discharge ratio versus temperature.
and end of discharge voltages in NiH2 remain fairly flat with the number of
C/D cycles even for several years in LEO missions, but not in NiCd.
The battery performs better under slow charge and discharge rates. A fast
discharge rate results in fast voltage degradation and low ampere-hour
capacity to the load. For these reasons, high charge and discharge rate
applications require different design considerations than the low-rate
applications.
After the battery is fully charged, the charge efficiency drops to zero. Any
additional charge will be converted into heat. If over-charged at higher rate
than the self-discharge rate for long time, the battery would overheat posing
FIGURE 9.15 Cell Ah capacity versus temperature.

Battery 217
FIGURE 9.16 Full charge voltage at C/10 for 16 h.
a potential explosion hazard. Any excessive over-charging produces

excessive gassing, which scrubs the electrode plates. Continuous scrubbing
at high rate produces excessive heat, wears out electrodes, and shortens the
life. For this reason, the battery charge circuit requires a regulator to cut
back to the trickle charge rate once the battery is fully charged. The trickle
charging produces a controlled amount of internal gassing. It causes mixing
action of the battery electrolyte, keeping it ready to deliver full charge.
9.5.2 Internal Impedance

The internal impedance of the cell is made of the resistance, inductance, and
capacitance. The resistance comes primarily from the electrodes and
electrolyte, the inductance from the leads in and out of the cell, and the
capacitance from the two electrode plates. The internal power loss during
discharge or charge can be seen as having an internal resistance Ri. The
value of Ri depends on the battery capacity, operating temperature, and the
state of charge. The higher the cell capacity, the larger the electrodes and
lower the internal resistance. The Ri varies with the state of charge as per
Equation 9.4. It also varies with temperature as shown in Figure 9.17. The
internal resistance and inductance of 50-Ah and 100-Ah NiH2 cells at 10 C
are given in Table 9.3.
9.5.3 Charge Efficiency

The charge efficiency, also known as coulombic efficiency, is defined as the
ratio of the ampere-hours being deposited internally between the plates
over that injected from the external terminals during charging. It is different
from the energy efficiency. The coulombic efficiency varies as a function of
FIGURE 9.17 Internal resistance of 50-Ah NiH2 cell at 100% SOC.
the SOC, charge rate, and the operating temperature, and is generally
higher than 97%. The charge efficiency is almost 100% when the cell is
empty of charge; the condition in which all charge received is deposited
between the plates. As the SOC approaches 1, the charge efficiency tapers
down to zero. The knee point where the charge efficiency starts tapering off
depends on the charge rate and temperature (Figure 9.18). For example, at
C/2 charge rate, the charge efficiency is almost 100% up to about 75% SOC.
At a fast charge rate of C/40, on the other hand, the charge efficiency at 70%
SOC is only about 50%.
9.5.4 Energy Efficiency

In both the charge and the discharge modes, a small fraction of energy is
converted into heat. The energy efficiency over a round trip of full charge
and discharge cycle is defined as the ratio of the energy output over the
energy input at the electrical terminals of the battery. For a typical battery of
capacity C at room temperature with an average discharge voltage of 1.25 V,
Table 9.3 Approximate impedance for IPV NiH2

cells
Parameter 50-Ah cell 100-Ah cell
Cell resistances 4.5 m at 1 kHz 2.25 m at 1 kHz

20 m at 10 kHz 10 m at 10 kHz
Cell inductance 0.70 H at 1 kHz 0.60 H at 1 kHz
0.35 H at 10 kHz 0.30 H at 10 kHz
Battery 219
FIGURE 9.18 Charge efficiency versus state of charge for NiH2 cell.
average charge voltage of 1.45 V, and the charge/discharge ratio of 1.10, the
efficiency is calculated as follows:
The energy output over the full discharge ¼ 1.25 C

The energy input required to restore full charge ¼ 1.45 1.10C
Therefore, the round trip energy efficiency is
1:25 C
energy ¼ ¼ 0:78 or 78% ð9:5Þ
1:45 1:10 C
These calculations indicate that about 22% energy is lost per charge/
discharge cycle, which is converted into heat. Most of this energy loss
occurs during discharge. The round trip energy efficiency varies with
various factors, but is typically 75 to 80% for NiH2 and 80 to 85% for NiCd.
9.5.5 V–T Characteristics

For most electrochemistries, the full charge voltage is a function of
temperature and age of the battery a shown in Figure 9.19, where the
limit numbers represent equally divided periods of the mission life. A
similar relation for the C/D ratio is shown in Figure 9.20. Such voltage
versus temperature (V–T) curves are used for on-orbit energy management
as discussed in Chapter 13. With a NiH2 battery, an alternative is to use the
pressure–temperature (P–T) curves, which exhibit similar relations with
age. The energy management software shifts the operation on-board as the
battery ages in service.
9.5.6 Self-Discharge
The battery slowly self-discharges even with no load under open circuit. It
must be continuously trickle charged to counter the self-discharge rate in
FIGURE 9.19 Temperature compensated V–T levels for NiH2 cell.
order to maintain the full state of charge. This rate is usually less than 1%
per day for most electrochemistry in normal working conditions. In NiH2
cells, the self-discharge rate shows as a proportionate loss in pressure as
seen in Figure 9.21. After the battery is fully charged, the charge efficiency
drops to zero. Any additional charge will be converted into heat. If over-
charged at higher rate then the self-discharge rate for long time, the battery
would overheat posing a potential explosion hazard.
FIGURE 9.20 Temperature compensated charge/discharge ratios for NiH2 cell.

Battery 221
FIGURE 9.21 Self-discharge rates versus time in NiH2 cell under open circuit.
9.5.7 Self-Heating
When the battery is initially put to charge, the internal heat generation rate
is negative for a while, meaning that the electrochemical reaction during the
initial charging period is endothermic (absorbing heat). The reaction
changes to exothermic and generates heat as the battery approaches full
charge. The temperature rise during this phase depends on the cooling
method used to dissipate the heat by conduction, convection, and radiation.
The internal power loss (in watts) causes self-heating in the battery. It has
a major impact on the battery temperature. During charge, it is equal to the
sum of heat generated by charging of active materials (am) and the heat
generated by over-charge reaction (or). That is,

Hc ¼ Hc;am þ Hc;or ¼ Ic c 1:45 Ec þ Ic ð1 c ÞEc ð9:6Þ
where Ic ¼ charge current, c¼ instantaneous charge efficiency, Ec ¼ charge

voltage, and 1.45 is the ratio of the enthalpy in joules equivalent to the
Faraday’s constant for NiCd and NiH2. The above equation simplifies to

Hc ¼ Ic 1:45c Ec ð9:7Þ
At the beginning of the charging process when c is near unity and Ec is low,
1.45c > Ec and Hc is negative, meaning the battery absorbs heat and self-
cools. Near the end of charging, as c approaches zero and Ec rises, the
battery generates heat and self-heats.
The heat generation rate (in watts) during discharge is given by

Hd ¼ Id 1:50 Ed ð9:8Þ
where Id ¼ discharge current and Ed ¼ discharge voltage.

As seen in the above two equations, the required information for

determining the internal heat generation rate is the current, voltage, and
the instantaneous charge efficiency. During discharge, approximately 15%
of the total energy released is heat, 5% is pressure–volume work and side
reactions, and 80% is useful electricity. In this exothermic mode, it releases
heat in addition to electrical power. Therefore, the battery temperature
rises. During charge, on the other hand, the battery temperature falls as the
chemical reaction requites heat. This is true until the full state of charge is
attained as indicated by the battery voltage of 1.55 V or so. Beyond this
point, the chemical reaction stops, and all energy supplied are turned into
heat. The battery may over-heat quickly and may explode.
During over-charge and trickle charge, the heat generated is equal to the
electrical power input. Thus, the efficiency depends upon the achievable
trickle charge rate, which in turn depends upon the charge rate,
temperature, charge regulator operation, etc. Typically, the battery is
recharged before the eclipse starts with ample margin in time. Then it is
placed into taper charge and over-charge if the V–T limit charge approach is
used. During this period, the entire over-charge will convert into heat inside
the battery. The thermal control is designed to maintain the battery within
its desired operating temperature limits.
Figure 9.22 shows the power loss in a 40-Ah 22-cell NiH2 battery
delivering a 1200-W load during eclipse in GEO with 65% DOD and C/D
ratio of 1.2. Figure 9.23 depicts the power loss in the NiH2 battery on a 3000-
W LEO platform with a maximum DOD of 35% and C/D ratio of 1.1 (EOS-
AM batteries, each 54-cells 50-Ah capacity, 1500-W average load, 80-V
battery voltage). For a lower C/D ratio, the power loss between 90 and 98
minutes would be lower. Note that the heat generation increases as the
FIGURE 9.22 Battery power loss versus time starting with eclipse for 1200-W load and 1.2 C/
D ratio in GEO orbit.
Battery 223
FIGURE 9.23 Battery power loss versus time starting with eclipse for 3000-W load and 1.1 C/
D ratio in LEO orbit.
depth of discharge increases because of the increased internal resistance. In

order to overcome possible cell voltage divergence, the C/D ratio can be
increased. However, that would increase the orbit average dissipation. For
NiH2 cells, increasing the C/D ratio from 1.05 to 1.1 can increase the orbit
average dissipation in LEO by 10% at 35% DOD and 15% at 70% DOD.
Different electrochemistry, however, generates internal heat at different
rates. The heat generation between various batteries can be meaningfully
compared in terms of the adiabatic temperature rise during discharge,
which is given by the following relation:

WHd E
T ¼ 1 v þ d ð9:9Þ
M Cp Eo
where
iT ¼ adiabatic temperature rise of the battery, in degress centigrade
WHd ¼ watt-hour energy discharged
M ¼ mass of the battery
Cp ¼ battery specific heat, Wh/kg C
v ¼ voltage efficiency factor on discharge
Ed ¼ average cell entropy energy per coulomb during discharge,
i.e., average power loss per ampere of discharge, W/A
Eo ¼ average cell open circuit voltage, volts
For full discharge, the WHd/M ratio in the above expression becomes the
specific energy. This indicates that higher specific energy cells would also
tend to have higher temperature rise during discharge, requiring enhanced

cooling design. Figure 9.24 depicts the adiabatic temperature rise iT for
various electrochemistries after a full discharge in short burst.
The thermal properties of the cell, such as the heat generation rate,
thermal conductivity, and specific heat are useful for the thermal analysis
and design. The specific heat of the NiCd cell is about 1000 J/kg C, which is
about that of pure aluminum. For the NiH2 cell, it is about 1200 J/kg C. In
terms of the ampere-hour capacity, the following equation gives a reason-
able first order approximation to the specific heat (in AhJ/ C) of the cell:
Cp ¼ 200 þ 25 ð9:10Þ
The battery temperature generally means the wall temperature. A

temperature gradient across the hydrogen gap between the wall and the
internal stack in NiH2 cell is typically 2 to 3 C.
9.6 Cycle Life

The battery fails when at least one cell fails. The cell failure is theoretically
defined as the condition in which the cell voltage drops below certain low
value before discharging the rated capacity at room temperature. The low
value is generally taken as 1.0 V in cells with nominal voltage of 1.25 V. The
remaining ampere-hours the cell can deliver from 1 V to 0 V is only a few
percent. However, this is a conservative definition of battery failure.
The battery does not fail like a light bulb, but gradually degrades in
performance. In a degraded state with one cell dropping below 1.0 V, the
rest of the cells providing voltage above 1.0 V may make up for the failed
cell, and the battery as a whole may still provide the minimum required
voltage. If the battery voltage falls below its required level at full load, it can
FIGURE 9.24 Adiabatic temperature rise under full discharge of capacity in various
electrochemistries.
Battery 225
still deliver the required voltage for a short time if the load can be shed to
lower the DOD.
Battery end of life is defined as one or more cell voltages dropping below
1.0 V at any point during the C/D cycle, usually at the end of discharge. The
cell can fail either randomly or due to wear. It can fail open, short, or
somewhere between (a soft short). A short that starts soft eventually
develops into a hard short. In a low voltage battery, any attempt to charge
with a shorted cell may result in physical damage to the battery and/or the
charge regulator. On the other hand, the shorted cell in a high voltage
battery with numerous series connected cells may work for a long time.
However, it loses the voltage and ampere-hour capacity, and hence would
work as a load on the healthy cells. An open cell, on the other hand, disables
the entire string of series connected cells.
A spacecraft battery rarely fails randomly. Its primary failure mode is
associated with electrode wear due to repeated charge/discharge cycles.
The battery life is measured as the number of times it can be discharged and
recharged before the electrodes wear out. The life depends strongly on the
electrochemistry, the depth of discharge, and the temperature. Figure 9.25
shows the cycle life characteristic of NiCd and NiH2 batteries. The life also
depends to a lesser degree on electrolyte concentration, electrode porosity,
and charge and discharge rates. The electrolyte concentration makes a
significant difference to the cycle life of the NiH2 cell. The cell with 26%
concentration gives a greater cycle life compared to one with 31%
concentration. The first set of factors is application related, whereas the
others are construction related.
The number of charge/discharge cycles in a satellite equals the number of
eclipses during the mission life. It is at least an order of magnitude greater
in LEO satellites than in GEO. Such a long cycle life requirement in LEO can
be achieved only by limiting the battery design to a low depth of discharge,
typically 30% compared to 80% in GEO satellites. Obviously, such a design
requires a proportionately much larger battery. A long cycle life at a given
DOD and operating temperature reduces the battery mass in direct
proportion.
It is noteworthy from Figure 9.25 that the life at a given temperature is an
inverse function of the depth of discharge. If the life is 100 units at 50%
DOD, then it would be about 200 units at 25% DOD. This makes the product
of the cycles to failure and the DOD roughly constant in the first
approximation. The cycle life multiplied by the DOD product decreases
with increasing temperature. Such is true for most electrochemistries. This
means that the battery at a given temperature can deliver the same number
of equivalent full charges regardless of the depth of discharge. Phrased
differently, the total watt-hour energy the battery can deliver over its life is
roughly constant in the first approximation. The battery lasts proportio-
nately longer if less energy is used per cycle. Such an observation is useful
FIGURE 9.25 Cycle life versus temperature and DOD for NiCd and NiH2 cells.
in comparing the costs of various batteries for a given application at the

conceptual design stage.
The battery does not work fully well all the way until it dies. Its capacity
remains fairly constant until it rapidly fades at the end as shown in Figure
9.26.
Since the battery cycle life is influenced primarily by the end of discharge
DOD and temperature, the cell life prediction model is generally developed
by fitting exponential curves to the test data on cycles to failure versus DOD
at temperatures the battery would be subjected in orbit. The factors such as
the charge and discharge rate and the electrolyte concentration are not
included in such a model, but are factored into the cell life data by actual
test or from previous experience. A reliability model then incorporates the
predicated cell wear-out life and the random failure rates to estimate the
mean time before battery failure by taking into account the number of cells
Battery 227
FIGURE 9.26 Capacity fade versus cycle life expended.
in series, the bypass diodes, and other components in the battery as a

whole.
The battery life testing is generally done under accelerated cycling. For
example, the GEO cycle is 24 h, but the battery testing is done in 8-h cycles.
Smith and Brill11 report such test data in NiH2 cells. Many spacecraft design
engineers believe that some electrochemical differences in actual cycling
and accelerated cycling are still not fully acceptable.
9.7 Burst Power Capability

In fast discharge applications, such as for meeting a pulse power demand,
the battery may be required to deliver the maximum possible power for a
very short time. The peak power it can deliver can be derived using the
maximum power transfer theorem in electrical circuits. It states that the
maximum power can be transferred from the source to the load when the
internal impedance of the source equals the conjugate of the load
impedance. The battery can deliver maximum power to a d.c. load when
RL ¼ Ri. This gives
E2i
Pmax ¼ ð9:11Þ
4Ri
Since Ei and Ri vary with the state of charge, the Pmax also varies
accordingly. The internal loss is I2Ri, which is also equal to Pmax delivered
to the load. The power transfer efficiency while delivering theoretical
possible peak power is therefore only 50%. Such poor efficiency may not
matter for a short-duration peak load. Both NiCd and NiH2 batteries are
capable of following pulse discharge and charge lasting a few milliseconds.

The battery voltage follows the load variations as expected even under
pulse loading. The power peaks shorter than a few milliseconds durations
primarily comes from the capacitance between the electrodes plates.
9.8 Performance Comparison

Figure 9.27 compares the specific energy and energy density of various
electrochemical cells. The voltage varies with DOD. For 60% DOD, it varies
from 1.22 to 1.26 V at moderate discharge rate. The average charge and
discharge voltages are listed in Table 9.4. Other performance characteristics
and properties of different types of cell are summarized and compared in
Table 9.5 and Table 9.6. With NiCd and NiH2 cells, the battery mass taking
into considerations all design factors and the packaging hardware is given
in Figure 9.28 for various load levels, service life, and DOD; and their
qualitative differences are listed in Table 9.7. The battery mass required to
meet typical power system performance with various cells is shown in
Figure 9.29 in percentage of the NiCd battery mass. The life and cost are
compared in Table 9.8.
9.9 Battery Design

The battery design starts with selecting the electrochemistry and the
number of batteries based on the mission requirement and the heritage
database. The following factors have strong influence the chemistry
selection:
FIGURE 9.27 Specific energy and energy density of various electrochemistries.

Battery 229
Table 9.4 Voltage characteristics of various batteries for space applications
Charge
Electro- Cut-off Discharge terminate
chemistry voltage voltage voltage Remarks
NiCd 1.0 1.25 1.55 Exhibits memory effect

NiH2 1.0 1.25 1.55 No memory effect
NiMH 1.0 1.25 1.55 Temperature sensitive
Li-ion 2.7 3.50 4.20 Safe, contains no metallic lithium
Lithium- 2.7 3.50 4.20 Contains metallic lithium
polymer
Silver-zinc 1.3 1.50 2.05 Needs good air management to
limit self-discharge rate
Table 9.5 Energy and power densities of various cells
Electro- Specific energy Energy density Specific power Power density

chemistry (Wh/kg) (Wh/l) (W/kg) (W/l)
NiCd 40–5 50–100 150–200 300–500

NiH2 45–65 35–50 150–200 200–300
NiMH 50–70 140–180 150–200 300–500
Li-ion 90–150 150–250 200–220 400–500
Lithium- 100–200 150–300 >200 >400
polymer
Table 9.6 Battery characteristics affecting thermal design
Over- Heat Mass

Electro- Operating charge capacity density Entropic heating on
chemistry range ( C) tolerance (Wh/kgK) (kg/l) discharge W/Acell
NiCd 20 to 50 Medium 0.30 1.7 0.12

NiH2 10 to 50 High 0.35 1.0 0.10
NiMH 10 to 50 Low 0.35 2.3 0.07
Li-ion 10 to 45 Very low 0.38 1.35 0
Lithium- 50 to 70 Very low 0.40 1.3 0
polymer
Specific energy and energy density

Cycle life, stability of capacity and voltage
Round trip energy efficiency
Mass and volume constrains
Temperature effects on the performance
Ampere-hour capacity ratings available
FIGURE 9.28 NiCd and NiH2 battery mass versus load and missions duration.
Table 9.7 Qualitative differences in NiCd and NiH2 batteries
Battery Advantages Disadvantages
NiCd Prismatic cell requires less Cell performance not easily scalable to high
supporting Ah ratings.
Low operating temperature, Sensitive to over-charge at high temperature
hence simpler thermal Must be maintained within narrow
management temperature range of 5 to 10 C.
Memory effect requires periodic
reconditioning, making the fully autonomous
operation difficult.
NiH2 Less sensitive to over-charge Cylindrical cells results in low energy density,
Longer life heavy supporting structure, and poor cooling.
High self-discharge rate
FIGURE 9.29 Battery mass with various electrochemistries relative to NiCd cell.
Ease and speed of recharging

Self-discharge rate
Safety issues
Battery 231
Table 9.8 Life and cost comparison of various batteries
Cycle life in full Self–discharge

Electo- discharge cycles Calendar life, (percent/month Relative cost
chemistry at 25 C in years at 25 C) ($/kWh)
NiCd 1000–2000 10–15 20–30 1500

NiH2 2000–4000 10–15 20–30 1500
NiMH 1000–2000 8–10 20–30 2000
Li-ion 500–1000a a 5–10 3000
Lithium- 500–1000a a 1–2 >3000
polymer
a
Under development.
Although NiH2 offers higher energy storage capability per kilogram, both
NiCd and NiH2 continue to find applications. One can be more advan-
tageous over the other depending on the mission-specific details. For
example, NiCd is a more likely candidate for low power missions, because
it is available in a wide range of capacity starting from a fraction of 1 Ah.
The NiH2 is generally better suited for high power satellites, and is not mass
and cost effective below 25-Ah capacity.
Regarding the number of batteries, all GEO satellites have traditionally
used two batteries, each with one half of the total Ampere-hours capacity
needed to meet the energy demand during the worst case eclipse. For small
LEO satellites, such as Iridium, one battery can be advantageous. The mass
and cost of one large battery is always less than two small batteries of one
half the capacity. On the business side, the customer generally has a strong
preference, sometimes for documented or undocumented reasons, which is
taken into account.
Once the chemistry and the number of batteries are settled, the battery
design depends on the following system parameters:
Bus voltage and load current

Charge and discharge rates and duration
Operating temperature during charge and discharge
Life in number of charge and discharge cycles
The life consideration is the dominant design driver in setting the battery
Ampere-hour ratings. Even when the load may be met with a smaller
capacity, the battery is oversized to meet the cycle life requirement. For
example, with the same watt-hour load, the battery that must deliver twice
as many C/D cycles needs approximately double the capacity.
The battery design proceeds in the following steps:
Determine the number of series cells required to meet the voltage

requirement.
Determine the ampere-hour discharge required to meet the load current
demand.
For the required number of charge/discharge cycles, determine the
maximum allowable depth of discharge.
The total ampere-hour capacity of the battery is then determined by
dividing the ampere-hour discharge required by the allowable depth of
discharge.
Determine the number of battery packs required in parallel for the total
ampere-hour capacity from packaging and other considerations.
Determine the temperature rise and the thermal control requirement.
Determine the required charge and discharge rate controls.
The actual ampere-hour capacity of each battery at operating temperature is

given by the following expression:
Pe Te
Ahb ¼ ð9:12Þ
N b dis ðNc 1Þ Vcdis Vd Vhdis DOD
where
Ahb ¼ capacity of each battery
Pe ¼ load power demand during eclipse
Te ¼ eclipse duration, in hours
Nb ¼ number of batteries in parallel
dis ¼ discharge converter efficiency
Nc ¼ number of series cells per battery
Vd ¼ voltage drop in bypass diode in case a cell failed
Vcdis ¼ voltage per cell, average during discharge
Vhdis ¼ voltage drop in harness from battery to PRU
DOD ¼ maximum allowable DOD in the worst case eclipse
Equation 9.12 allows one cell failure in each battery for single fault
tolerant design. The failed cell is bypassed using diodes as shown in Figure
9.30. There is generally one bypass diode in the charge path and three
diodes in the discharge path to limit the leakage current. Without such
bypass diodes, if one battery in a two-battery system fails, the power
available to the payload is significantly reduced, plus one arcjet would not
work. The satellite must carry more fuel to compensate for this. Some
Battery 233
FIGURE 9.30 Cell bypass diodes in charge and discharge paths for reliability.
customers may prefer relay switches to the bypass diodes to minimize the
power loss.
The battery round trip energy efficiency under the worst case with one
cell failed is determined using the following expression:
" # " #
ðNc 1Þ Vcdis Vd Vhdis AHdis
b ¼ ð9:13Þ
ðNc 1Þ Vcchg þ 3 Vd þ Vhchg AHchg
where
Nc ¼ number of series cell, Vd ¼ voltage drop in one diode
Vcdis and Vcchg ¼ cell voltage during discharge and charge, respectively
Vhdis and Vhchg ¼ voltage drop in harness during discharge and charge
AHdis and AHchg ¼ ampere-hour discharged and charged, respectively.
There is a trade in selecting the cell ampere-hour rating in fully regulated

DET bus. It is between using more small capacity cells giving closer to the
bus voltage, versus fewer large capacity cells giving lower than bus voltage.
The latter option requires greater buck and boost ratios in the battery
regulators, but saves in mass and cost. Fewer larger cells have lower mass
and volume and higher reliability. The battery voltage in any case is
regulated by the duty ratio, and the regulator mass and efficiency do not
change much with the battery voltage. A general guideline is to place cells
in series to make the nominal battery voltage during discharge equal to
about 80%, and during charge about 93% of the bus voltage. Even fewer
cells may yield more savings as long as the buck and boost converter design
is not adversely impacted.
Other factors also enter the battery design, such as the end of charge and
end of discharge voltages. They remain fairly constant in NiH2, but degrade
in NiCd over the service life.4 The ampere-hour rating thus determined
must be adjusted upwards if the battery voltage does not meet the bus
voltage requirement, cannot be fully recharged after the deepest DOD, or

the nominal eclipse starts from a partially drained battery in some orbits.
All series cells must have equal Ampere-hour ratings within a few
percent to avoid a cell voltage reversal leading to a total battery failure. For
this reason, the cell selection is a critical acceptance test conduced at the
vendor site. The cells are discharged to 1.0 V and the ampere-hour output
measured at room temperature to match the nameplate capacity. All cells
within 3 to 5% capacity variations around the average are selected for
assembly in the flight battery. Since the actual battery capacity is the
capacity of the weakest link, it is taken as the average cell capacity less 3 to
5%. Battery cells are selected also by the cell voltage variance. One may
consider the magnetic signature method to gage the internal details. Only
the stabilized cells are included in the selection. The full SOC pressure in
NiH2 cells is significantly low when new, but stabilizes after a number of
C/D cycles.
Typical specifications of the battery for 5- to 10-kW 100-V GEO
communication satellite include the following:
For 800 psi full SOC pressure, the battery should withstand 1350 psi
without permanent deformation, 5000 cycles of pressure test cycles
from fully charge to fully discharge state, and 2400 psi proof test
without rupture.
d.c. isolation resistance of 50 M at 1000 V d.c.
Reliability of 0.985 over 10 years of life based on 30 FITS per cell
The radiation dose of 5 105 rads over a 10-year mission life
Ability to withstand the required dynamic environment of the launch
9.10 Launch and Ascent Power

Generally, the battery is designed to meet the on-orbit load and energy
balance. The same battery is also used for the spacecraft power needs
during launch and ascent phases. Several possibilities that may arise during
this time are dealt in the following manner.
Case A: The battery fully charged at the time of launch is sufficient for
the launch/ascent energy requirement. This is usually the case in
most satellites, and does not require any extra step for the power
system engineer.
Case B: The battery fully charged at the time of launch is slightly short of
the launch/ascent energy need. In this case, a widely practiced
solution is to pre-cool the battery by cold air or nitrogen before the
Battery 235
launch to increase the battery capacity by several percent or more as

needed.
Case C: The launch/ascent power demand exceeds the battery capacity
beyond the case B remedy. Increasing the on-board battery capacity,
even in such a rare case, is almost never done, because it is very
expensive. Instead, a primary battery having high specific energy is
used to meet the shortage. This battery is jettisoned after the ascent.
A primary battery that is widely used in space vehicles is lithium
carbon monofluroid (LiCFx) having a specific energy of 350 to 500
Wh/kg, an order of magnitude higher than the commonly used
secondary batteries.
9.11 Thermal Design

The battery is sometimes mounted on the exterior of the spacecraft body on
a honeycomb panel on the anti-sun side. It is conductively isolated from the
spacecraft honeycomb panel by G-10 fiberglass standoffs, radiatively
isolated by multi-layer insulation blankets, and enclosed in an aluminum
honeycomb micrometeoroid shield. A part of the shield front serves as the
radiator, which is coated with silver Teflon. The remainder of the shield is
insulated with multi-layer thermal blankets. The battery radiates heat to the
radiator, which, in turn, radiates to space. Adequate thermal control for the
battery, for both cooling and heating, is extremely important. A high
temperature in a battery can lead to failure caused by electrolyte
evaporation and decreased efficiency. A high cell-to-cell thermal gradient
can cause uneven current sharing. The operating temperature significantly
influences many performance parameters in the following ways:
The capacity drops at a temperature above or below a certain range,

and drops sharply as the temperature approaches the freezing point.
The capacity significantly increases when charged at 20 C. Additional
capacity obtained at low temperature is more stable than that obtained
at 10 C when cycled.
The battery voltage and charge efficiency decrease with increasing
temperature.
The self-discharge rate increase with temperature.
The internal resistance increases as temperature decreases.
Table 9.9 shows the temperature sensitivity of NiCd cell. The process of
determining the optimum operating temperature is illustrated in Table 9.10.
It indicates that different attributes have different desirable operating
temperature ranges shown by the vertical lines. With all attributes jointly
considered, the most optimum operating temperature is the intersection of
Table 9.9 NiCd temperature sensitivity
Charge Discharge Self-discharge

Temperature efficiency (%) efficiency (%) rate, percent of C/day
40 C 0 72 0.1
10 C 94 100 0.2
60 C 75 100 8.0
all the desirable ranges. For example, if we wish to limit the self-discharge
rate below 1.0% and the charge efficiency above 90%, the optimum working
temperature range is between 10 C and 25 C as shown by thin line in the
first column. The actual temperature is then maintained between 5 C and
þ10 C shown by the heavy vertical line in order to allow some design and
operating margins.
It is necessary to perform detailed thermal analyses for every mission,
because it is mission unique. Controlling the battery temperature is a critical
requirement of the thermal control system. This often requires dedicated
radiators, conduction paths, heaters, blankets, temperature sensors, and
Table 9.10 Determination of optimum working temperature for

a nickel cadmium battery
Operating Charge Discharge Self-discharge rate,

temperature C efficiency(%) efficiency(%) percent capacity/day
40 0 72 0.1
35 0 80 0.1
30 15 85 0.1
25 40 90 0.2
20 75 95 0.2
15 85 97 0.2
10 90 100 0.2
5 92 100 0.2
0 93 100 0.2
5 94 100 0.2
10 94 100 0.2
15 94 100 0.3
20 93 100 0.4
25 92 100 0.6
30 91 100 1.0
35 90 100 1.4
40 88 100 2.0
45 85 100 2.7
50 82 100 3.6
55 79 100 5.1
60 75 100 8.0
65 70 100 12
70 60 100 20
Battery 237
control electronics. Multilayer insulating blankets are used to keep the heat
or cold away as required for maintaining the orbit-average battery
temperature. If the battery is running cold, blanketing a part of the external
radiator surface can reduce the heater power requirement. The battery
electrochemistry selection has a significant impact on the thermal system
mass and cost. The chemistry that costs more but is less sensitive to the
operating temperature can pay back at the satellite level.
Each cell in the battery pack is electrically insulated from each other and
from the ground. The electrical insulation must be good conductor of heat
to maintain low temperature gradient between the cells and also to the
ground. The maximum temperature must remain below a set value to avoid
damage to the battery or shortening the life. The minimum temperature
must not be too low to reduce the battery efficiency in the operating mode.
In the survival mode, the temperature must remain above the freezing point
of the electrolyte to avoid the battery from being inoperable.
The cylindrical shape of IPV and CPV NiH2 cells makes the thermal
cooling difficult. Moreover, it increases the volume significantly and makes
the heat conduction form the cell to the radiator plate inefficient compared
to the battery with prismatic cells. Figure 9.31 shows a battery pack with
prismatic NiCd or Li-ion cells. The cylindrical NiH2 cells are packed in one
of the following three configurations.
9.11.1 Flat Pack

The cells are laid flat on the radiator plate, offering the maximum cooling
surface. Such design has been flown in communication satellites in several
FIGURE 9.31 Battery assembly using prismatic NiCd cells.

kilowatts power range. However, it requires much greater footprint and is

vulnerable to shocks and vibration during launch.
9.11.2 Vertical Stack

This is the most common packaging method for NiH2 cells at present, which
was seen in Figure 9.6. The cells standing up and making contact with the
radiator plate only at the dome point results in poor heat conduction. The
conduction area is increased by providing conduction shims shown in
Figure 9.32. The wedge shape minimizes the mass for the required heat
transfer. The shims can be aluminum or graphite fiber composite for even
lower mass.
9.11.3 Wine Rack

The cylindrical IPVs are sometimes packaged in a wine-rack configuration.
In such design, the cell temperature can reach 15 C and maximum cell-to-
cell gradient can reach 7 C in the worst case EOL. The cell-to-cell and cell-
to-battery frame temperature can be improved by using metallic fiber,
which could reduce the cell-to-cell temperature gradient to below 2 C, and
maximum cell temperature to about 8 C.
The three NiH2 cell stacking configuration options are compared in
Table 9.11. The typical temperature requirements for large communication
satellite battery design are listed in Table 9.12.
9.12 Safety Considerations

The battery design and operation require certain safety considerations. The
most important is not to over-charge the battery. Any over-charge above the
trickle charge rate is converted into heat, which can cause the battery to
explode if allowed to build up beyond limit. This is particularly critical
when the battery is charged directly from a dedicated photovoltaic module
without a charge regulator in small science missions with short or
infrequent eclipses. In such cases, the array is sized below certain safe
FIGURE 9.32 Mass-efficient conduction shims for vertical mounting of NiH2 cell.
Battery 239
Table 9.11 NiH2 cell stacking options and their relative performance
Configuration Wine-rack Flat pack Vertical mount
Advantages Low footprint, volume, Lowest weight Best cooling

and weight. Largest foot print Moderate footprint
Better cooling and volume
Disadvantages Poor cooling Large foot print Heavy with metal
cooling sleeve
limit. As a rule of thumb, such photovoltaic array rating is kept below the
continuous trickle charge current that can be tolerated by the battery.
The battery design must consider a single failure mode of the battery
without creating any safety hazard of contamination, corrosion, explosion,
fire, injury or illness. This is required during all mission phases and ground
operations. Major failure modes are listed below. Any one or more
combinations could possibly cause venting of toxic, corrosive, and/or
flammable materials with associated fire or explosion risk:
Over-temperature from battery self-heat or ambient heat

Short circuit, internal or external to the battery
Reverse current due to inadvertent charging
Cell voltage reversal due to over discharge
Leakage of electrolyte or gases
Table 9.12 Typical thermal design requirement for NiH2 battery in large
communication satellite
Maximum gradient or any cell temperature
Gradient along cell stack as measured at cell wall 5 C

Gradient between any two points of a cell wall 7 C
Gradient between any two cell stacks within a battery during charge 7 C
Gradient between average of the cell stacks between any two batteries or 8 C
packs during charge
In transfer orbit during charge 30 C
During recharge following eclipse 15 C
During trickle charge 30 C
During EHT or arcjet firing 35 C
During reconditioning letdown 30 C
During reconditioning recharge 15 C
Minimum temperature in all scenarios 10 C
Life average temperature of cell stack in a battery 15 C
Acceptance testing (excluding EHT/arcjet firing) 10/þ30 C
Qualification testing (including EHT/arcjet firing) 15/þ45 C
Over-pressure due to over-charging or over-temperature
The design approach should preclude any of the above occurrences. If

failure does occur, venting should be prevented. If that is not possible in the
case of a major failure, it must be demonstrated that venting will occur in a
manner that is not hazardous to the mission. Safe shipping requirements
depend primarily on whether the battery is activated or not. If activated, the
applicable shipping regulations depend on the chemical reactivity of the
materials and the state of charge, both can be controlled by the battery
manufacturer to meet the shipping regulations.
Govar and Squires12 report safety tests on commercial off-the-shelf 9-V
lithium-ion battery for navy applications. Battery packs consisting of two
9-V batteries in parallel were found relatively safe and experienced no
violent behavior. On the other hand, packs with thirty 9-V batteries
experienced venting, smoke, and flames under certain circumstances.
9.13 Charge Regulation

For safety reasons, it is extremely important that excessive charging of the
battery is avoided at all times. Over-charging causes internal gassing, which
causes loss of electrolyte in vented cell and premature aging. The charge
regulator allows the maximum rate of charging until the gassing starts.
Then the charge current is tapered off to trickle charge rate so that the full
charge is approached gently.
During battery charging, the energy management software monitors the
state of charge, the overall health and the safe termination criteria. The
operating voltage, the current, and the temperature are monitored. The
charging timer is started after all initial checks are successfully completed.
Charging may be suspended (but not reset) if the software detects violation
of critical safety criteria. The timer stops charging if the defect persists
beyond a certain time limit.
The normal charging has the following three phases:
Bulk (fast) charge, which deposits 80 to 90% of the drained capacity

Taper (cut-back) charge in which the charge rate is gradually cut back
to top off the remaining capacity
Trickle (float) charge to counter the self-discharge rate after the battery
is fully charged
The bulk charge and the taper charge termination criteria are pre-loaded
in the battery management software to match with the battery electro-
chemistry and the system design parameters. For example, the NiCd and
NiMH batteries are generally charged at constant current until about 95%
SOC in 75 to 85% of the available sun time. Then onward, the charge current
Battery 241
FIGURE 9.33 Battery charge profile versus orbital sun time.
is tapered off — gradually or in a few steps — to the trickle rate as shown in

Figure 9.33. The decision is made to terminate the main charging and switch
to the trickle charge when the continuously monitored iV is detected
negative. On the other hand, Li-ion batteries, being sensitive to over-
charging, can be charged at constant voltage, and then tapering off the
charge current as needed.
Both NiCd and NiH2 batteries show voltage roll-off and sudden
temperature rise, i.e., negative dV/dt and positive dT/dt, soon after the
full charge (Figure 9.34). Both of these characteristics are used in detecting
the end of charge. In NiCd and NiH2, the voltage roll-off is 20 to 30 mV
per cell. In NiMH, on the other hand, the iV is much smaller (3 to 5 mV
per cell) and the temperature rises steadily to the final value, making the
FIGURE 9.34 Battery voltage and temperature during charge showing rapid changes at the
end of charge.
end of charge detection difficult. The rollover is attributed to the voltage

decrease subsequent to temperature increase that results from the oxygen
reaction with hydrogen. In some electrochemistries, the voltage rollover is
distinct; in some it is rather flat. When the cells is over-charged at 20 C,
the voltage rollover shows waviness with a smaller hump followed by a
larger hump.
Rapid reaction of oxygen with hydrogen during over-charge is an
undesirable feature, which causes an audible popping noise in certain
cases. Popping results in melting of the gas screen pinholes in the negative
plates, and heat stains in the core and the positive plate. In extreme cases,
popping blows away the separator, resulting in shorts. Cells with
significant positive plate swelling are prone to popping, which results in
damage to the components. Swelling also causes extrusion of active
material and reduction in the gap between the two positive plates. The
final result is a decrease in the end of discharge voltage. Positive plates that
show excessive swelling absorb more electrolytes. This is accompanied by a
reduced absorption of electrolyte by the separator.
The batteries are charged in one of the following three ways using the
voltage-limited tapered-current method or the constant-current ampere-
hour return method.
9.13.1 Multiple Charge Rates

This is the best method, in which the battery is charged gently in multiple
steps. First the battery is charged at full charge rate until 80 to 90% of the
capacity is achieved. The charge current is then cut back, in steps, until the
battery is fully charged. At this time, the charge current is further reduced
to trickle-charge rate, keeping it fully charged until the next load demand
comes on the battery. This method therefore needs at least three charge
rates in the charge regulator design.
9.13.2 Single Charge Rate

This method uses simple low cost regulator, which is either on or off. The
regulator is designed for only one charge rate. When the battery is fully
charged, as measured by its terminal voltage, the charger is turned off by a
relay. When the battery voltage drops below a preset value, the charger is
again connected in full force. Since the charging is not gentle in this method,
full charge is difficult to achieve and maintain. An alternate version of this
charging method is the multiple pulse charging. Full current charges the
battery up to the high preset voltage just below the gassing threshold. At
this time, the charger is shut off for a short time to allow the battery
chemicals to mix and voltage to fall. When the voltage falls below the low
preset threshold, the charger is reconnected, again passing full current to
the battery.
Battery 243
9.13.3 Unregulated Charging

This is the least cost method which uses no charge regulator. The battery is
charged directly from a solar array module dedicated just for charging. The
charging module is properly designed for safe operation for the given
number of cells in the battery. When the battery is fully charged, the array is
fully shunted to ground by a shorting switch (transistor). At this time, the
battery voltage must still be slightly lower than the available solar array
voltage so that the trickle charge can continue, but not much lower to avoid
over charging. The shunt transistor switch is open when the battery voltage
drops below certain value. An isolation diode would block the battery
powering the array or the shunt at night.
If one cell in one of the two batteries working in parallel fails short, the
two batteries would have different terminal characteristics. Charging or
discharging such batteries in parallel can result in highly uneven current
sharing, subsequently overheating one of the batteries. This can be avoided
in two ways. One is to charge and discharge both batteries with individual
current controls such that they both draw their rated share of the load. The
other is to replace the failed cell immediately, which is not possible in space.
In general, the individual charge/discharge control for each battery is the
best strategy. It may also allow replacement of any one battery with
different electrochemistry or different age, which would have different load
sharing characteristics. The ISS batteries are designed for replacements
several times during the life of the station.
The Li-ion battery requires additional circuitry for balancing individual
cells to obtain maximum performance from the battery. The balancing
method can vary from design to design. One method is to use d.c.–d.c.
converters with multi-tap transformer to provide energy equilibrium in
each cell. Another method is to use capacitor-switching technique called
flying capacitor that shifts small packets of energy from higher voltage cells
to lower voltage cells. The Saft Corporation has often used a resistive
dissipation method that lowers the SOC of the highest cells down to match
with the lowest common denominator cell. Sack, Croydon and Reynolds13
report yet another method using segmented charger. In this concept, the
battery charger is broken up into separate cell chargers. The concept was
tested on a 26-V 45-Ah Li-ion battery under U.S. Army funding.
9.14 Battery Management

Battery management electronics include the battery monitor and controller.
The main function is to monitor the battery status and optimize its
performance by adjusting the charging process. Electronic circuits monitor
the battery voltage, current, state of charge, and temperature. If not
managed properly, the battery performance could suffer and the life
shortened. The following are some potential performance problems:
Low charge efficiency resulting in low state of charge

Loss of capacity to hold the rated ampere-hour charge
Premature failure leading to loss of load availability
Excessive gassing and heating, leading to short life
Positive plate corrosion shortening the life
Stratification and sulfation degrading the performance
The following features incorporated in the battery management can avoid

the above problems:
Temperature compensated charging, in that the charge termination

occurs earlier if the battery temperature is higher than the reference
temperature
Individual charge control if two or more batteries are charged in
parallel
Accurate set points to start and to stop the charge and discharge modes
The batteries in modern power systems are managed by dedicated

computer software, which monitors and controls the following performance
parameters:
Voltage and current

Temperature and pressure
Ampere-hour in and out of the battery
State of charge and discharge
Rate of charge and discharge
Depth of discharge
Number of charge and discharge cycles
The above parameters are determined by a number of measurements and

limits, which include the following:
Minimum charging current threshold

Terminal voltage threshold
Absolute temperature cut-off
Rate of temperature change iT/it
Drop of the cell voltage after the peak iV
Rate of voltage change (iV/it), including the peak and the inflection
Actual charging time
Battery 245
The ampere-hour integrating meter is commercially available, which keeps

track of the ampere-hours in and out of the battery and send required
signals to the mode controller.
With age, NiCd and NiH2 batteries experience walk-downs in capacity,
voltage, and end-of-charge pressure. These are accounted by shifting the
software on board. The cell voltage divergence is another concern, more so
after reconditioning. One method of avoiding it is to over-charge, so that all
cells saturate to full voltage, provided the thermal system can handle the
associated heating.
The temperature compensation on the maximum battery voltage and the
state of charge can improve the battery management, particularly when
cold. It can allow additional charging during cold periods when the battery
can accept more charge. The low voltage alert is a good feature, as excessive
discharging below the threshold low voltage can cause cell voltage reversal,
leading to battery failure. The alert can be used to shed noncritical loads
from the battery to avoid battery damage. The battery cell voltage monitor
allows a comprehensive health check of the total battery. This is important
just before launching, when a problem can be fixed. A 22-cell voltage
monitor adds less than 200 to 300 g, but adds much more in numerous
telemetry wires.
9.15 Dynamic Model

The term impedance refers to the ratio of the cell voltage change to
corresponding change in the alternating current through the cell. It is given
by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z¼ R2 þ X2 ð9:14Þ
where R and X are the battery resistance and reactance, respectively. The
dynamic impedance characteristics discussed in this section are applicable
to all electrochemistries. They apply at a given SOC, i.e., when net d.c.
current flow or energy conversion is not involved. The internal impedance
determines the battery response to ripple, noise, and transients during
charging and discharging. It varies inversely with the capacity, as large
capacity cells have large plate area. The impedance versus frequency of a
50-Ah NiH2 cell is shown in Figure 9.35. It has a relatively constant value up
to 10 Hz, but rises rapidly at higher frequency indicating a strong inductive
component. The battery R and X are obtained by multiplying the cell values
with number of series cells in the battery. A simple dynamic model of the
battery for stability analysis with the dynamic impedance Z(s) in the
Laplace domain is shown in Figure 9.36.
Complex battery models developed by electrochemists are not expressed
in terms of the electromotive force, resistance, inductance, and capacitance.
FIGURE 9.35 50-Ah NiH2 cell impedance versus frequency.
Therefore, they are unsuitable for everyday use by the electrical engineer,
who needs a model that represents the electrochemistry in the electrical
network terms. Ceraola14 has developed a model to represent the battery’s
nonlinear behavior using two electrical branches shown in Figure 9.37. The
main branch is designated with subscript m and the parasitic branch with
subscript p. The Em and Zm are the internal electromotive force and the
internal impedance, respectively. Both are functions of the Laplace
parameters, the electrolyte temperature, , and the SOC. The parasitic
branch models the nonreversible reactions that draw some current, but does
not contribute to the main reversible reaction. The energy that is absorbed
by Ep ceases to be electrical and is converted into other forms, such as gas or
water electrolysis that occurs at the end of the charge process. The power
dissipated in the real part of Zm and Zp is converted into heat that
contributes to the self-heating of the battery. The Ro is common to both
branches and represents resistance from the battery terminals to the plates.
FIGURE 9.36 Simple dynamic electrical circuit model of the battery.

Battery 247
FIGURE 9.37 Dynamic circuit model of battery with main and parasitic reaction branches.
To make such a model useful, one need to determine the functional

dependence of E and Z on s, , and SOC, and the battery thermal model
from which the electrolyte temperature can be determined. The dependence
on s can be stated by expressing each Z in terms of the equivalent R, L, and
C values. The model can be simplified for a specific operating condition, if
known. For example, the parasitic branch can be ignored in the initial
charging phase when the charge efficiency is near unity. At the end of
charge, however, it takes almost all power entering the battery terminals.
Since the inductance is small and can be ignored, the battery model that is
valid during both the charge and discharge is shown in Figure 9.38.
The battery capacity is maximum at certain optimum temperature and
falls off on both sides of the optimum temperature. If the battery is
maintained on the cooler side of the optimum temperature, the capacity at a
constant discharge current and a constant electrolyte temperature is higher
with higher temperature and lower discharge current. At a fixed discharge
current, I, and temperature, , the capacity can be expressed as14
( )

AhðI; Þ ¼ Aho ðIÞ 1 ð9:15Þ
f
FIGURE 9.38 Dynamic circuit model of the battery during charge and discharge.
where Aho ¼ battery capacity at 0 C at discharge current I, ¼ empirical

constant, and > f, the freezing temperature of the electrolyte (such as
40 C). Equation (9.15) gives the expected value of Ah ¼ 0 at freezing
temperature when the battery becomes inoperative. It is possible that Aho(I)
can be expressed in terms of its value at a reference temperature Ir and
empirical constants K and as follows:
( )

KAho ðIr Þ 1
f
AhðI; Þ ¼ ð9:16Þ
I
1 þ ðK 1Þ
Ir
Equation (9.16) suggests that numerical values of the elements of the

circuit model have to be established for each value of electrolyte
temperature and SOC. In practice they can be established only for a few
discrete values of these two quantities, and the values in-between can be
obtained by interpolation. Determination of such parameters for the first
time would take extensive analysis and test efforts. However, having made
that investment, the model for a new battery can be established by using
previous data on similar battery suitably modified for the new battery.
9.16 Cycle Life Model

As previously stated, the battery cycle life is related primarily to the DOD
and the operating temperature. The life relationship can be expressed in
terms of C/D cycles as the dependent variable, with the DOD and
temperature as the independent variables. In developing a statistical
model for the battery life, we must first select a mathematical expression
that defines the relationship adequately, and then select a statistical
distribution that best fit the test data. The selections can be guided by
consideration of the physical behavior of the battery. For example, the
number of cycles to failure decreases with increasing DOD and tempera-
ture. Various types of relationships commonly used for analyzing life test
data can be considered to express such relationships. Among the
alternatives, such as exponential, inverse power, and Arrhenius, more
than one can give a reasonably good fit to the test data. However, the
relationship that generally gives the best fit in similar life test data is the
Arrhenius equation, which accounts for the chemical degradation limiting
the life.15,16 That is
L ¼ C1 T C2 eC3 D ð9:17Þ

Battery 249
where L ¼ mean number of cycles to the first failure, T ¼ temperature in

degress centrigrade, D ¼ DOD in each cycle, and C1, C2, and C3 are the
relationship constants to be determined by multiple regression curve fitting
techniques.
Based on the physical considerations of the wear-out mode of failure
being studied, the statistical distribution selected to represent the spread of
life around the mean value derived from Equation (9.17) at a given
temperature and DOD should have a continuously increasing hazard
function. The hazard function is a conditional failure rate, which expresses
the rate of failure for the surviving population at a given number of cycles
L. Also, the distribution should have a good shape flexibility to fit the data.
Two distributions, which can satisfy these requirements, are lognormal and
Weibull. The Weibull distribution is the most commonly used for
presentation of such life test data. It is given by
L
FðLÞ ¼ 1 eðÞ ð9:18Þ
where F(L) is the fraction of the population failing by L cycles, and positive
parameters and are scale and shape parameters, respectively. The
parameter can be assumed to be constant, while is a function of
temperature and DOD. The values of both and are derived from test
data. The curve fitting can be a simple procedure using least squares
regression techniques if the failure times of all samples are known.
However, in typical situations, the battery test data includes samples: (a)
which are known to have failed during an interval, rather than at a specific
time, and (b) many samples not yet failed at the end of the test period. Such
censored data can only be analyzed by statistical techniques developed
for dealing with incomplete life test data.17,18 They use the method of
maximum likelihood, which involves selecting as an estimate for each
unknown parameter the value that appears to be most probable on the basis
of the given test data. The computations are based on the sample log
likelihood which is a function of the unknown model coefficients (Cl, C2,
C3). The estimates of the coefficients are obtained numerically by iterative
methods on a computer, and the variances of these estimates are used to
obtain approximate confidence intervals.
When the indicated analysis is performed on a set of test data, the
expected life relationships is derived with its coefficients C1, C2, and C3
established. Estimating the life from such considerations can be relatively
less complex if the battery were used in an identical manner in each cycle
over the mission life. It becomes more complex when it is used in one way
sometimes and another way at other times, with significant variation in the
usage patterns. In such cases, the life may be estimates using the Milner’s
cumulative damage theory19,20 often used in fracture mechanics and fatigue
studies on metals and other material. The theory utilizes the cycle ratio as
its basic measure of damage. For the battery life with two independent
variables as shown in Figure 9.39, it can be modified as follows. If a battery
temperature is Ti at depth of discharge Di for Ci number of cycles, and if the
life relationship estimates Li cycles as its life at Ti and Di, then the
proportional damage, PDi, caused during that time is

Ci ðTi ; Di Þ
PDi ¼ ð9:19Þ
Li ðTi ; Di Þ
where is a positive constant to be evaluated from correlating available life

data and experience. The battery can be P expected to fail when the
cumulative damage equals unity (i.e., when PDi ¼ 1.0). The evaluation
of the exponent requires extensive data from batteries, which are much
closer in scale factor to actual battery to be modeled.
The theory keeps incremental account of the fraction of life depleted
during each segment of time depending on the life degradation factors
prevailing over that duration of use. It can be used to estimate the
remaining life, i.e.,
X
remaining life ¼ 1 PDi ð9:20Þ
FIGURE 9.39 Life model for cycles to failure versus DOD and temperature.
Battery 251
Milner’s cumulative damage theory suggests the following for use with
some judgment and experience:
The battery life ends when sum total of the accumulated proportional
damage reaches its limiting value of 1.0.
A short time at high wear depletes the same fraction of life as a long
time at low wear.
Although the life model presented in this section could not justify direct
use of the life relationships in this fashion at present, it can be regarded as a
potential future application. It is hoped that the theoretical consideration
proposed here may serve as a guide to define, plan and analyze future
research in this field.
9.17 Primary Battery

Among primary batteries, lithium/polycarbon monofluoride (LiCFx) has
the highest specific energy. It has been in used in manned space vehicles,
missiles, terrestrial and marine applications since the early 1970s, and has
accumulated a safe and reliable database. LiCFx cells are produced in many
ratings, but only a few are space qualified for NASA and Air Force
applications. Hundreds of LiCFx batteries have flown on space shuttle
flights and in many spacecraft. The reliability estimated as per the MIL-
HDBK-217 range from 1.6 to 10 106 FIT. They are insensitive to shock,
vibration, forced charging, and forced reversal. Many such cells can survive
external hard short circuit without opening the cell vent. A high-rate battery
with low heat rejection rate could potentially vent following a sudden
internal short circuit. However, sudden shorts are estimated to be highly
improbable, yielding the reliability of the cell better than 0.9999 with 90%
confidence. The space-qualified battery design using 40 Ah LiCFx cells
increases the assembly mass by 60 to 70% over that of the cells.
9.18 Advances in Battery Technologies

Incremental advances in the cell designs continue. The heavy sintered
nickel powder plaque has been the substrate of choice for many years to
support the electrochemically active material in all nickel-based batteries.
The nickel electrode is the most critical and the heaviest component in any
nickel-based battery, such as NiH2, NiCd, and NiMH. The nickel powder
plaque amounts to about one half of the electrode weight. Highly porous
light weight nickel fiber plaques are being considered to reduce the
electrode weight by 50% and the same time increase the capacity of the
active material by increasing the surface area available for deposition of the
active material. This may lead to higher specific energy for the battery.
252
Table 9.13 Replacement candidates for NiH2 cell (50 Wh/kg)
Candidate Li-ion Li-Poly Fly-wheel Super NiCd NiMH Sodium-sulfur H2/O2 RFC
Wh/kg 130 150 a 50 65 120 75

Wh/l 200 300 a 115 200 250 50
Cycle life > 10 k > 10 k High > 45 k >10 k >3k >2k
Used in space Since Not yet Not yet used Since 1990 Quald in 1992 Experimental Not yet used
since 2000 used flights
Dev. Stage Active Active Planned to fly Used in 30þ Used in 2 Inactive Was proposed
on ISS spacecraft spacecraft for ISS
a
Under development.
Spacecraft Power Systems
Battery 253
Another approach that may result in higher specific energy is a thicker

nickel electrode. Fewer thick electrodes reduce the number of other
components, such as hydrogen electrodes and separators, thus reducing
the weight of the cell.
Entirely new or derivatives of time proven electrochemistry are
continuously evaluated in search of a better battery. The following figures
of merit ultimately determine the battery mass and cost, and should be of
interest to the power system design engineer:
Specific energy and energy density

Round trip energy efficiency
Cycle life at the operating temperature and DOD
Wider operating temperature range
Better packaging
In recent years, Li-ion has emerged as the battery of choice for high
performance commercial applications — cell phones, laptop computers, etc.
It offers significant advantages over alkaline-based NiCd and NiMH
batteries. The space applications often require the battery to operate over
a wide temperature range (40 C to þ65 C), long cycle life (> 30,000 cycles
in LEO), and long calendar life (>10 years in GEO). NASA has qualified
commercial off-the-shelf Li-ion technology to power electronic peripherals
on the space shuttle and the ISS. The Spirit and Opportunity rovers that
landed on Mars in 2004 were powered during eclipse by Li-ion batteries
capable of 200 cycles at 20 C.
At present, Li-ion appears to be a candidate that may eventually replace
the NiH2 for general use in space.21 There are other candidates as well, such
as the regenerative fuel cell and the flywheel. The flywheel is under
significant research funding with a schedule to fly to the ISS in 2006. The
NiH2 replacement candidates as seen at present are compared in Table 9.13.
References
1. Parez, M. E. et al. Energy storage for space applications, in Proceedings of
the 36th Intersociety Energy Conversion Engineering Conference, ASME, 2001,
pp. 85–89.
2. Rao, G.M., Ahmad, A., and Chetty, P.K.R., Super nickel cadmium battery
operation and performance on-board the Sampex spacecraft, in Proceedings
of the 30th Intersociety Energy Conversion Engineering Conference, Vol. I,
ASME, 1995, pp. 111–116.
3. Pickett, D.F., Hayden, J.W., Lucero, D., and Johnson, Z., Comparison of
advanced NiCd space cell technology with other technologies in con-
sideration for LEO and GEO orbit and planetary missions, in Proceedings of
the 35th Intersociety Energy Conversion Engineering Conference, AIAA, 2000,

Paper No. 2989.
4. Brown, R., Nickel hydrogen life modeling, in Proceedings of the 35th
No. 2991.
5. Britton, D.L., Progress in the development of lightweight nickel electrode
for NiH2 cell, in Proceedings of the 34th Intersociety Energy Conversion
Engineering Conference, SAE, 1999, Paper No. 2537.
6. Garner, J.C., Braun, W.R., Loo, D.V., and Bowers, D., 90-Ah dependent
pressure vessel NiH2 battery, in Proceedings of the 34th Intersociety Energy
Conversion Engineering Conference, SAE, 1999, Paper No. 2590.
7. Puglia, F., Gitzendaner, R., Ehrlich, G.M., and Marsh, C., Advancing Li-ion
technology for aerospace applications, in Proceedings of the 34th Intersociety
Energy Conversion Engineering Conference, SAE, 1999, Paper No. 2637.
8. Croft, H., Staniewicz, R., Smart, M.C., and Ratnakumar, B.V., Cycling and
low temperature performance operation of Li-ion cells, in Proceedings of the
35th Intersociety Energy Conversion Engineering Conference, AIAA, 2000,
Paper No. 27-AP-B1.
9. Fellner, J.P. and Loeber, G.J., Li-ion performance testing and ac impedance
characterization, in Proceedings of the 34th Intersociety Energy Conversion
10. Bruce, G. and Marcoux, L., Large lithium-ion batteries for aerospace and
aircraft applications, IEEE AESS Systems Magazine, 16(9), 24–28, 2001.
11. Smith, R. and Brill, J, ‘‘Updated life cycle test results for flight qualified
NiH2 cell design, in Proceedings of the 36th Intersociety Energy Conversion
Engineering Conference, ASME, 2001, pp. 55–59.
12. Govar, C.J. and Squires, T.L, ‘‘Safety tests of lithium 9-volts batteries for
navy applications, IEEE AESS Systems Magazine, 16(9), 34–37, 2001.
13. Sack, T.T., Croydon, T., and Reynold, R., Segmented battery charger for
high density 28-V lithium ion battery, IEEE AESS Systems Magazine, 16(9),
15–18, 2001.
14. Ceraolo, P., Dynamic model of the electrochemical battery, in Proceedings of
Paper No. X001.
15. Hahn, G.J. and Shapiro, S.S., Statistical Models in Engineering, John Wiley &
Sons, New York, 1967.
16. Kendall, M.G. and Stuart, A., Advanced Theory of Statistics, Vol. 2, Hafner
Publishing Co, New York, 1961.
17. Hahn, G.J. and Nelson, H.B., A comparison of methods for analyzing
censored life data to estimate relationship between stress and product life,
IEEE Transactions on Reliability, R-23, April 1974, pp. 2–10.
18. Nelson, W. A survey of methods for planning and analyzing accelerated
tests, IEEE Transactions on Electrical Insulation. El-9, March 1974, pp. 12–18.
19. Miner, M.A., Cumulative damage in fatigue, Journal of Applied Mechanics,
12( Sept). 1945, A159–164, 1945.
20. Madayag, A.F., Metal Fatigue — Theory and Design, John Wiley and Sons,
New York, 1969, pp. 170–203.
21. Gross, O., Fox, C, Roller, D., and Shimanek, L., Lithium-ion polymer
batteries for space applications, Proceedings of the 36th Intersociety Energy
Conversion Engineering Conference, ASME, 2001, pp. 61–68.
Chapter 10
Power Electronics and Magnetics
10.1 Introduction
The spacecraft power electronics basically control the bus voltage and
convert voltage levels to match the operating voltages of various
components. These functions are performed by solid-state semiconductor
devices used as controlled switches which are turned on and off at high
frequency. Capacitors and inductors are used to store energy when the
switch is connected to the power source. The stored energy is then
discharged to continue powering the load when the switch is off.
Transformers are used where needed. Major power electronic components
routinely used in the spacecraft are:
Shunt regulator for bus voltage control during sunlight

Battery charge converter (buck converter)
Battery discharge converter (boost converter)
10.2 Switching Devices

A variety of solid-state devices are available for use as controlled switches.
However, the devices commonly used in space are:
Metal-oxide semiconducting field effect transistor (MOSFET)

Bipolar junction transistor (BJT)
Insulated gate bipolar transistor (IGBT)
The device selection depends on the required voltage, current and the
switching frequency. A common feature among these devices is that all are
three-terminal devices. Their generally used circuit symbols are shown in
Figure 10.1. The two power terminals 1 and 0 are connected in the main
power circuit. The control gate terminal G is connected to the auxiliary
control circuit. In normal conducting operation, terminal 1 is generally at
higher voltage than terminal 0. Since the device is primarily used for
switching power on and off as required, it is functionally represented by a
gate-controlled switch. In absence of the gate control signal, the device
255
FIGURE 10.1 Semiconductor switching devices used in space.
resistance between the power terminals is large — the functional

equivalence of an open switch. When the control signal is applied at the
gate, the device resistance approaches zero, making the device function like
a closed switch.
The switch is triggered periodically on and off by a train of gate signal of
suitable frequency. A sharp gate signal of rectangular, triangular, or saw
tooth wave shape without overshoot and ringing is generated by a separate
triggering (firing) circuit. The gate circuit for the MOSFET is simpler than
that for the bipolar transistor. Although the control circuit has a distinct
identity and much different design features, it is often incorporated in the
main power electronic component assembly. The transistor switch is turned
on and off at high frequency, typically at 50 to 200 kHz, and sometimes
higher. The duty ratio, D, of the switch is defined as
Time on Ton
D¼ ¼ ¼ Ton switching frequency ð10:1Þ
Period T
The available voltage and current ratings of the switching devices and their
gate triggering requirements vary with the device type. The presently
available ratings are listed in Table 10.1, not all of which are space qualified.
Various power electronic components built using such high-frequency
switching devices are discussed next.
Power Electronics and Magnetics 257
Table 10.1 Maximum voltage and current ratings of semiconductor switching

devices
Voltage Current
rating rating
Device Volts Amps Remark
MOSFET 1000 100 Offers higher switching speed, simpler firing

circuit.
BJT 1500 200 Requires larger current signal to turn on.
IGBT 1200 100 Combines the advantages of BJT and MOSFET
10.3 Shunt Regulator

The power output of the solar array during sunlight normally exceeds
the load plus the battery charge requirements, more so in the beginning
of life. The excess power must be diverted (shunted) from feeding into
the bus in order to control the bus voltage. The shunt load can be a
dump resistor, which would convert the solar array power into heat.
Such heat dissipation in the spacecraft body would pose a burden on the
thermal system in providing adequate cooling. An alternative commonly
used in spacecraft is to shunt some of the solar array strings to the
ground. This forces the string to operate under short circuit condition,
delivering Isc at zero voltage. In this mode, i.e., in the shunt mode, no
power is delivered to the load or to the ground. The photon energy
remains on the array, raising the array temperature and ultimately
dissipating the excess power to space. The solar array is essentially used
here as the thermal dissipator.
Another application of the shunt regulator is in small dedicated solar
array module used to directly charge the battery without a battery charge
regulator. When the battery is fully charged, the solar array module is
shunted to ground by shorting the switch. This way, the battery is protected
from overcharging.
Figure 10.2 depicts a typical shunt regulator where a transistor is used as
the switch. When the excess power is available, the bus voltage rises above
the rated value. This is taken as a signal to turn on the shunt switch across
the required number of solar array strings. Thus, the shunt is turned on or
off by a transistor controlled by the bus voltage reference. For array with
many strings in parallel, the basic configuration shown in Figure 10.2 is
used for each string separately. The same gate signal is supplied to all
modules simultaneously in small power applications. For shunting large
power, multiple shunt circuits are switched on and off in sequence to
minimize the switching transients and the resulting electromagnetic
interference to the neighboring equipment. For a fine voltage control, the
last shunt to turn on is operated in the pulse width modulation (PWM)
FIGURE 10.2 Full shunt regulator for bus voltage control.
mode, while all others are fully on or off. The constructional and
performances features of various shunt circuits are described below.
10.3.1 Full and Partial Shunts

Figure 10.2 shows the schematic of a full shunt, where the whole string of
the solar array is controlled on, off, or in-between. Figure 10.3(a) shows a
partial shunt, where the shunt switch controls a part of the solar string.
Only the lower part of the string is shunted when required. When the shunt
is on, its operation is depicted in Figure 10.3(b) on the I–V characteristic of
the string. The lower section thus works like a current source, whereas the
upper part works like a voltage source. The unshunted and shunted
sections of the string are mismatched. The mismatch thus created reduces
the power fed to the bus by operating away from the maximum power
point, leaving some of the solar power on the array where it appears as heat
and is radiated back into space. A small percentage of power is dissipated
as heat in the shunt. Since the shunting occurs only when excess power is
available, the increased solar array temperature and the resulting decrease
in photovoltaic conversion efficiency has no impact on the power system
performance. In the partial shunt scheme, the shunt transistor can be
located on the back of the solar array, which enables easy heat dissipation
into space. However, this requires more signal sliprings than power
sliprings, and the shunt transistor has to withstand wide temperature
excursion cycles of the eclipse and sunlight.
The tap voltage of the lower segment VL¼ full string voltage upper
string voltage ¼ Vsa VU. The value of VL is typically 15 V in a 22- to 35-V
sun-regulated bus. The shaded area in Figure 10.3(b) represents the shunted
power. Quantitatively, it is equal to the product of the voltage and current
in the lower shunted segments of the string, i.e.,
Pshunt ¼ VL IL ð10:2Þ
FIGURE 10.3 Partial shunt regulator with upper segment unshunted and lower segment
shunted.
Partial shunts have been used in many satellites in the past, but have fallen
out of favor in modern high power satellites, which generally use full
shunts. Both the full and the partial shunts can be any one of the following
types.
10.3.2 Linear and PWM Shunts

Figure 10.4 shows the schematic of a linear shunt. In configuration (a)
without a series resistor, the gate current drives the shunt switch in
unsaturated region when the bus voltage exceeds the reference value. In
this region of operation, the switch offers a variable resistance depending
on the gate current. The string power is partially shunted to the ground,
and the remaining power feeds the bus. If needed, an additional fixed
resistance can be added as shown in (b). A disadvantage of the linear shunt
with resistor is that a part of the solar power is converted into heat in the
resistor, which must be dissipated into space by the thermal system. For this
reason, the resistor is seldom used in large satellites. The linear shunt
without resistor keeps the unwanted power on the array for dissipation to
FIGURE 10.4 Linear shunt regulator.
the outer space, thus minimizing the sliprings, thermal management, and
EMI related costs.
Another alternative is to use the switch-mode pulse width modulated
(PWM) shunt shown in Figure 10.5. This scheme is used on most modern
high power spacecraft. Instead of operating the linear shunt in the
unsaturated region, the PWM shunt is operated in the off or all the way
on (saturated) state when the bus voltage exceeds the reference value. The
on and off duration are repeated at high switching frequency with a
controlled duty ratio. This eliminates the power dissipation from liner
FIGURE 10.5 Pulse width modulated shunt regulator.

shunts. The string power fed to the bus equals the whole string power times
the duty ratio. The duty ratio is adjusted to keep the bus voltage in a narrow
range around the reference value. A disadvantage of the PWM shunt is that
it switches full current at high frequency, causing EMI on the platform.
However, the PWM shunt is widely used for its many benefits. On the other
hand, the linear shunt is still used where EMI cannot be tolerated, or where
heat is needed to maintain temperature inside the spacecraft body.
10.3.3 Sequential Linear Shunt

When more than one string is required to be shunted, the one-string linear
shunt shown in Figure 10.4(a) is used as the building block to build the
array shunt circuit as shown in Figure 10.6. When the solar output power
exceeds requirements, the shunt drive controller first raises the gate current
of shunt 1 from zero to the value that can maintain the bus voltage at the
reference value. In doing so, if shunt-1 current reaches the saturation level,
shunt-2 is gradually turned on. If more shunting is still needed, the
controller may reach the nth shunt until the bus voltage is maintained at the
set level. Turning on the shunt switches is done in sequence.
10.3.4 Multistage PWM Shunt

Multiple solar strings are generally needed to shunt the excess power,
particularly at the beginning of life. The shunt circuit using multiple PWM
shunts is shown in Figure 10.7. The bus voltage is controlled by first placing
one string in the PWM. The duty ratio controls the power shunted. If more
power needs to be shunted, the duty ratio of the first string is raised. After
reaching the duty ratio limit of the first string, the second string is placed
under the PWM, and so on. At the end, all shunts are in PWM mode,
switching the power on and off simultaneously. This may create an EMI
problem around the shunts. The EMI is minimized by designing the PWM
Regulated bus To loads
Vbus
1 2 .... n
Solar Vref
array
Sequential
Error
drive control
amplifier
FIGURE 10.6 Sequential linear shunt regulator.

FIGURE 10.7 Multistage PWM shunt regulator.
control circuit to have the maximum duty ratio of 1. That way, all but one
shunts are operated continuously in the saturated state, and only the last
one is operated in PWM mode to shunt a fraction of power on or off at a
time.
10.3.5 Polyphase PWM Shunt

In this scheme, the multistage PWM shunt schematic shown in Figure 10.7
remains the same, but the PWM modulation control is modified to operate
simultaneously as many strings as needed for the bus voltage control. The
difference here is that the gate signals for all strings under PWM are
symmetrically out of phase with each other. For example, the gate signals
for an 8-string array are shown in Figure 10.8. This way, all strings are not
simultaneously switched on or off, but one after another in a rotation. Major
advantages of this scheme are significant reductions in ripples and EMI.
The polyphase PWM shunt regulator reduces the ripple generation,
hence lower the filter mass. Figure 10.9 depicts approximate mass saving
with number of phases used. The minimum filter mass is achieved with
four to six phases depending on the power level of the bus. The saving can
be up to 20% compared to the multistage PWM shunts.
FIGURE 10.8 Polyphase PWM gate signals with 360/n phase difference.
FIGURE 10.9 Shunt regulator mass versus number of phases.
10.3.6 Calibrated Step Shunt

The scheme of Figure 10.10 shows multiple shunts, each having power
capability in calibrated steps such that great many values of power can be
shunted by appropriate selection of the shunts and placing them in the
saturation region. It provides a good resolution on the bus voltage control,
but not as fine as other schemes.
10.4 Shunt Circuit Design

In designing the shunt circuit, the following trade options exist for the
design engineer.
10.4.1 Full Versus Partial Shunts

The full shunt has the following features:
It requires fewer slip rings and wires

More array power is lost if one circuit fails
Gives more power dissipation per circuit
The partial shunt, on the other hand, requires twice as many slip rings, one
set from the upper section and one set from the lower section.
FIGURE 10.10 Calibrated step digital shunt regulator.
10.4.2 Number of Shunt Circuits

In selecting the number of solar circuits, too few circuits leads to:
Higher power per circuit, hence more power margin is needed to allow
for one circuit failure for reliability
Higher local dissipation, hence hot spots
Fewer slip rings in the solar array
Too many circuits, on the other hand, leads to:
Smaller voltage control steps, i.e., better bus voltage resolution

Small steps require higher shunt control amplifier gain. This may be
harder to work with due to excessive amplification of the small error
signal band.
10.4.3 Typical Shunt Circuit Design

Typically, the solar array is divided into as many string circuits as required
to keep the BOL current below 5 A in order to use standard 5-A slip rings in
each circuit. For a 22 to 35 sun-regulated bus with full shunt as an example,
the maximum shunt dissipation would then be 35 5 ¼ 175 W in the shunt
just entering the linear range. This is 87.5 W per transistor with two in
parallel for redundancy. Dissipation in each shunt when saturated, on the
other hand, is (1=2 I)2 transistor on-resistance ¼ (5/2)2 0.100 ¼ 0.75 W in
each of the two switches in parallel for each solar string, or 1.50 W total in
both switches combined. Thus, the total power loss per shunt circuit in the
saturated region is less than 1% of the maximum before entering the
saturation.
The solar array shunt control should limit the bus voltage dV/dt to less
than 1% of the bus voltage per microsecond, and conduction voltage drop
below 0.1% of the bus voltage per ampere of the shunt circuit current. For
example, a 5-A circuit on a 100-V bus should keep the dV/dt less than
1 V/ms and the conduction drop below 0.50 V. The control signals to switch
shunt circuits are degraded, buffered, shielded, and double insulated for
reliability.
10.5 Bus Ripple Filter Design

The ripple produced by the PWM shunt operation is kept below the
minimum required limit by using a bus filter capacitor shown in Figure
10.11. The following design analysis establishes the required capacitor size.
The capacitor becomes charged when the shunt switch is off, and is
discharged when the switch is on. The ripple comes from two sources as
shown in the figure. As the capacitor charges and discharges, the equivalent
series resistance (ESR) of the capacitor causes the voltage drop of opposite
polarity, resulting in the capacitor voltage rise and fall. The peak-to-peak
ripple voltage Vpp on the bus can be derived in to the following expression:
Is T
Vpp ¼ Is Rc þ Dð1 DÞ ð10:3Þ
C
where
Is ¼ current switched per shunt circuit
Rc ¼ bus capacitor ESR
D ¼ duty ratio of the PWM switching
T ¼ switching period
C ¼ bus capacitor value in farads
The ripple voltage varies with the duty ratio. It has the maximum value at
D ¼ 0.5, which is equal to
Is T
Vpp ¼ Is Rc þ 0:25 ð10:4Þ
C
FIGURE 10.11 Bus voltage ripple at shunt switching frequency with bus filter capacitor.
FIGURE 10.12 Shunt regulator mass versus switching frequency.
For a known ESR and switching frequency, the minimum bus capacitor
required to meet a given ripple specification is then found from
0:25 I s T
Cmin ¼ ð10:5Þ
Vpp Is Rc
A large bus filter capacitor is desired not only to get low ripple, but also
to reduce the bus impedance at high frequency. On the other hand, fast
transient response requires a low value of the bus capacitor for a given loop
gain crossover frequency. The engineer trades these two performance
features.
Use of high frequency, i.e., low switching period, reduces the filter
capacitor size as indicated by Equation 10.5. Approximate mass saving is
shown in Figure 10.12 for the frequency range of 20 to 200 kHz.
10.6 Power Converters

The converter basically converts one voltage level into another at the
interface of two power components. The following sections describe the
topology and operation of various types of power converter.
10.6.1 Battery Charge (Buck) Converter

During sunlight, the bus voltage is maintained by the solar array output.
The battery is charged only if the battery voltage is lower than the bus
voltage. A d.c.–d.c. voltage converter is required to buck (step down) the
bus voltage to the battery voltage during charging. Figure 10.13 is the most
widely used battery charge converter circuit topology. Its operation during
one on and off cycle of the triggering signal is shown in Figure 10.14.
During the on time, the switch is closed and the circuit operates as in (a).
FIGURE 10.13 Battery charge (buck) converter.
The d.c. source charges the capacitor and supplies power to the load via the
inductor. During the off time, the switch is open and the circuit operates as
in (b). The power drawn from the d.c. source is zero. However, full load
power is supplied by the energy stored in the inductor and the capacitor,
with the diode carrying the return current. Thus, the inductor and the
capacitor provide short-time energy storage to ride through the off period
of the switch. The load current during this period is known as the free-
wheeling current, and the diode is known as a free-wheeling diode. The
voltage and current waveforms over one complete cycle are displayed in
Figure 10.15. A suitable bleeding resistor at the load terminals is sometimes
incorporated in the design to keep the converter working without the load.
The analysis of the power train of the buck converter follows. The
analytical principles that follow are applicable also to other d.c.–d.c.
converters. The power train analysis is based on the energy balance over
one switching period. Since the inductor stores energy during the on-time
and discharge during the off time:
Energy supplied to the load over total period T ¼ energy drawn from
the source during the on time, which powers the load and charges the
inductor and capacitor, and
Energy supplied to the load during off time ¼ energy discharged from
the inductor and the capacitor during off time.
FIGURE 10.14 Buck converter operation during on and off switch.

FIGURE 10.15 Current and voltage waveforms in the buck converter.
The inductor generally uses magnetic core with lumped or distributed air
gap. The flux density in the core has to be kept below the magnetic
saturation level. For that reason, the net change in flux over one cycle of
switching has to be zero. Otherwise, the core would eventually walk away to
saturation on the high side, or be depleted of energy on the low side. The
magnitude of voltage drop across the inductor is given by Faraday’s law,
VL ¼ N d’/dt, where N ¼ number of turns, ’ ¼ flux in the core and t ¼
time. The equation can be written as VLdt ¼ Nd’. As d’ must be zero over
one switching cycle, we must have
ZT
VL dt ¼ 0 ð10:6Þ
o
i.e., the change in the volt–second product over one cycle must be zero.
Thus, in the steady state condition, the inductor volt–second balance during
the on and off periods must be maintained. This in essence gives the energy
balance.
We now apply the volt–second balance in the voltage and current
waveforms shown in Figure 11.15. In terms of the inductance, the voltage
drop across the inductor is VL ¼ L dIL/dt. Therefore, during on-time,
IL L ¼ ðVin Vout ÞTon ð10:7Þ
and during off-time,
IL L ¼ Vout Toff ð10:8Þ
If the inductor is large enough, as is usually the case in practical designs, the
change in the inductor current is small, and the peak value of the inductor
current is given by
Ipeak ¼ Io þ 0:5 IL ð10:9Þ
where the load current Io ¼ Vout/Rload ¼ average value of the inductor

current.
The algebraic manipulation of the above equations leads to
Vout ¼ Vin D ð10:10Þ
Since the duty ratio D is always less than unity, the output voltage of the
buck converter is always less than the input voltage. Thus, the buck
converter can only step down the voltage, hence the name.
It is seen from Equation 10.10 that varying the duty ratio controls the
output voltage. This is done in a feedback control loop with the required
battery charge current as the reference. Modulating the pulse width of Ton
controls the duty ratio. Such converter is therefore known as the pulse
width modulated (PWM) converter.
The efficiency is calculated as follows. During the on time, the input
voltage, less the voltage drop in the transistor switch, supply power to the
circuit. The energy transfer efficiency is therefore (Vi Vs)/Vi, where Vi ¼
input voltage and Vs ¼ voltage drop in the switch. During the off period,
the inductor voltage, less the diode drop, is the output voltage. The
efficiency during this period is therefore Vo/(Vo þ Vd), where Vo ¼ output
voltage and Vd ¼ voltage drop in the diode. In addition, there is some loss
in the inductor, the efficiency of which we denote by i. The losses in the
capacitor and the wire are relatively small and can be ignored for simplicity.
The overall power train efficiency of the buck converter is, therefore, the
product of the above three efficiencies, namely

Vi Vs Vo
c ¼ ð10:11Þ
Vi Vo þ Vd i
With commonly used devices, the transistor switch and diode voltage drops
are typically 0.6 V each. If we denote both by the device voltage drop Vd,
then a sufficiently accurate simplification of the above equation is
2
Vo
c ¼ i ð10:12Þ
Vo þ Vd
The inductor efficiency can approach 0.99. The power train efficiency of
28-V output converters is therefore around 95%. The losses in the wires,
capacitors, and magnetic core may further reduce the efficiency by a couple
of percent, making it in the 92 to 94% range. The converter efficiency is a
strong function of the output voltage. It decreases with decreasing output
voltage, as the switch and diode voltage drops remain constant. For
example, a 5-V output converter would have efficiency in the 75 to 80%
range.
Since the charge converter in LEO satellites handles more power and
works longer hours, it must be designed for high efficiency and better
cooling. The LEO charge converter design, therefore, gets more complex
than its GEO counterpart.
10.6.2 Battery Discharge (Boost) Converter

In a fully regulated bus, the bus voltage is maintained within a narrow band
during the entire orbit. Since the battery voltage under discharge sags
during eclipse, a voltage converter is required to boost (step up) the battery
voltage to the bus level. The boost converter topology is shown in Figure
10.16. When the transistor switch is on, the inductor is connected to the d.c.
source. When the switch is off, the inductor current is forced to flow
through the diode and the load. The output voltage of the boost converter is
derived again from the volt–second balance in the inductor. With duty ratio
D of the switch, it can be shown, in a manner similar to that in the buck
FIGURE 10.16 Battery discharge (boost) converter.

converter analysis, that the output voltage is given by the following

expression,
Vin
Vout ¼ ð10:13Þ
1D
Since the duty ratio is always less than unity, the output voltage of the boost
converter is always greater than the input voltage. Therefore, the boost
converter can only step up the voltage, hence the name. The efficiency of the
boost converter can be shown to be

Vi Vd 2
c ¼ i ð10:14Þ
Vi
10.6.3 Buck–Boost Converter

Cascading of the buck and the boost converters as shown in Figure 10.17(a)
results in a buck–boost converter, which can step down or step up the input
voltage. A modified direct buck–boost converter that uses fewer parts is
shown Figure 10.17(b). The output voltage is obtained by cascading the
buck and boost converter voltage relations. That is,
Vin D
Vout ¼ ð10:15Þ
1D
FIGURE 10.17 Two configurations of buck-boost converter.

FIGURE 10.18 Buck-boost converter output and input voltage ratio versus duty ratio.
This equation shows that the output voltage of the buck–boost converter
can be higher or lower than the input voltage depending on the duty ratio
(Figure 10.18). In addition to its possible use to charge and discharge the
battery with one converter, the buck–boost converter is capable of the four-
quadrant operation with a d.c. machine in the variable speed applications
and with the flywheel energy storage. In the flywheel system, the converter
would step-up voltage during generating mode, and step-down during
motoring mode.
10.6.4 Flyback Converter (Buck or Boost)

Figure 10.19 shows the topology of this converter, which can buck or boost
the voltage depending on the turn ratio of the coupled inductor. The energy
is stored in the primary side of the inductor during the on time. It then
becomes inductively transformed (kicked) to the secondary side when the
switch is turned off, and is delivered to the load via diode. The polarity
marks of the two coils are of essence in this circuit. The inductor core is
generally made of potted magnetic metal power. The energy is stored in
distributed air gap of the core. The voltage ratio and efficiency are given by
FIGURE 10.19 Flyback converter (buck or boost).

!
Vo Ns Vi Vs Vo
¼D and c ¼ ð10:16Þ
Vi Np Vi Vo þ Vd i
where i ¼ efficiency of the coupled inductors, which is typically 95 to 98%.

The voltage ratio is a mere multiple of the duty ratio with the turn ratio.
This gives great flexibility in the output voltage without the limitation of the
duty ratio. It can buck or boost depending on the turn ratio less than or
greater than one.
An advantage of the flyback converter over the classical buck and boost
converter is that it electrically isolates the two sides, thus minimizing the
conducted EMI and enhancing the safety. Moreover, it stores the inductive
energy, and simultaneously operates like a classical forward buck converter
with a transformer, stepping up or down the input voltage as needed.
10.6.5 Transformer Coupled Forward Converter

The converter circuit shown in Figure 10.20 employs a transformer with
suitable turn ratio in conventional converter such that the output voltage
can be below or above the input voltage. The polarity marks are of essence.
The bleeding load resistor is required to keep the converter working
without load. The free wheeling diode is not required. Its voltage relation is
identical to that for the flyback buck–boost converter described above. That
is,
!
Vo Ns
¼D ð10:17Þ
Vi Np
and the efficiency

Vi Vs Vo
c ¼ ð10:18Þ
Vi Vo þ Vd t
where t ¼ efficiency of the transformer, which is typically 95 to 98%.
FIGURE 10.20 Transformer coupled forward converter.

10.6.6 Push–Pull Converter

This is again a buck–boost converter and is shown in Figure 10.21. The
converter topology uses a center-tapped transformer, which receives square
wave excitation. It is seldom used at low power levels, but finds
applications in systems dealing with several tens of kilowatts and higher.
It can provide desired output voltage by setting the required transformer
turn ratio. The voltage ratio and the efficiency are the same as those given
by Equations 10.17 and 10.18.
10.6.7 Inductor Coupled Buck Converter

All converters presented above need an L–C filter to control the ripple in
output voltage. The buck converter needs heavier filter than the boost
converter. Since ripples are present in both the input and output sides, it is
possible to cancel them by coupling the ripple current slopes of two sides
with matched inductors1,2 as shown in Figure 10.22. The ripple magnitudes
are matched by using required turn ratio, and the polarities are inverted by
winding in magnetically opposite directions. The net ripple on the output
side is zero at certain gap, below which it changes the polarity. A gap
adjusting screw is needed to tune the gap for a precise match. The output
capacitor Co is not really needed in a perfectly coupled design, but a small
value of Co improves the performance. The coupling capacitor C is needed
for the performance. The total value of the two capacitors is about the same
as that needed in the classical buck converter. On the output side, some
ripple may be left due to the winding resistances. The input side ripple is
the same as that without the coupled inductor. The load current changes do
not affect the ripple cancellation.
10.6.8 Resonant Converter

Another type of d.c.–d.c. converter now used in many high power
applications is the zero-voltage switching converter due to its higher
FIGURE 10.21 Push-pull converter with center-tap transformer.

FIGURE 10.22 Inductor coupled buck converter with integrated magnetics (Cuk converter.)
switching frequency, higher efficiency, smaller size, lighter weight, and

better dynamic response. It is a full bridge converter that includes inductors
and capacitors as resonant elements. The switching is done when the
voltage is passing through its natural zero, thus reducing the switching
losses.
10.6.9 Multiple Output Converter

Multiple voltage output from the same converter is obtained by using two
or more inductors or transformer coils on the secondary side as shown in
Figure 10.23(a). In such converters, the regulation in one output voltage is
influence by the load changes in the other via magnetic coupling.
Analytically predicting the load regulation of a new design can save the
prototyping cost and time, as opposed to a trial and error process. The
three-winding transformer equivalent circuit shown in (b) can facilitate the
voltage regulation analysis. It is based on the premises that between any
two coils, there is a leakage flux and the corresponding leakage reactance X.
The equivalent circuit shows the total impedance — made of the winding
resistance and the leakage reactance — denoted by Z12 between coils 1 and
2, Z23 between coils 2 and 3, and Z31 between coils 3 and 1. The value of Z12
is determined by a short circuit test between winding 1 and 2, and so on.
The analysis is significantly simplified by representing the three coil-to-coil
impedances in an equivalent Y-circuit shown in (c). Through a circuit
analysis, their values can be derived as follows:
Z1 ¼ 12 ðZ12 þ Z13 Z23 Þ

Z2 ¼ 12 ðZ12 þ Z23 Z13 Þ ð10:19Þ
Z3 ¼ 12 ðZ13 þ Z23 Z12 Þ
FIGURE 10.23 Three winding transformer equivalent circuit for multiple output converter.
The calculations become easier by expressing the impedances in percentage

of the rated impedance of the transformer. It is necessary to emphasize that
these equivalent impedances have no real meaning. Any one can be zero or
negative, as long as the sum of impedances of any two branches is equal to
the short circuit impedance between those two windings. By using the
Y-circuit, it is easy to see that the terminal voltage V2 depends not only on the
secondary load, but also on the tertiary load because of the voltage drop in
Z1. It is also easy to see that when load I2 is shed, the voltage of the other load
will rise by I2Z1. The converter with more than two outputs can be similarly
analyzed using the same general principle.
10.6.10 Load Power Converter

Many loads, particularly payloads, require input voltage within certain
regulation and quality. For example, many communications payloads are
constant power loads, which require higher current at low voltage and vice
versa. In most cases, the payload has a load power converter at its interface
with the power system bus (Figure 10.24). The LPC input voltage varies
over a wide range in sun-regulated bus, but in a narrow range in a fully
regulated bus. Loads having a stringent ripple requirement may require a
prohibitively large bus filter capacitor. Alternatively, the L–C filter shown
in Figure 10.25(a) used at the bus–LPC interface would be more mass
effective. The inductor stores much greater energy per unit mass, and
therefore significantly reduced the filter mass.
FIGURE 10.24 Load power converter with single or multiple outputs. (Source: Ratheon
Corporation E-Systems Division. With permission.)
In the converter using an L–C filter, transient analysis becomes important

on switch-on, as both currents and voltage may overshoot to damaging
levels. Large negative voltage transients could occur at the power system
and user interface when the relay contact is opened. This can result in
component damage, which can be avoided by adding a free-wheeling diode
as shown in Figure 10.25(b).
10.6.11 Voltage and Current Regulators

When a constant voltage output is required from a variable voltage input, a
voltage regulator is used at the interface. If the input voltage variation is
small, a dissipative series voltage regulator can be used, otherwise one of
the PWM converters covered in this section is used. The two types are
compared as shown in Table 10.2. The linear and buck converters are simple
and dynamically stable. All other converters are relatively complex and
may become unstable, as they do have poles in the right hand side of the
Laplace plane. The linear converter is more reliable than the switching
FIGURE 10.25 L–C filter at converter interface and free-wheeling diode protection.
converter, because of fewer parts and no high-voltage switching stress,

although it runs hotter.
When a current limiting feature is desired, the converter control circuit
regulates and limits the current. In such a design, the current is limited to
some high value during overloads, or even under a load fault during which
the bus voltage is not driven to zero. It just folds back to some low value.
Thus, the converter control loop can be designed to maintain constant
voltage, constant current or constant power. It can also be designed for a
combination of two or more performance parameters. For example, the
converter may maintain constant voltage up to certain current limit, and
then maintains constant current at the set limit, or even folds back a little in
Table 10.2 Linear versus switch mode PWM voltage

regulators
Parameter Linear–dissipative Switch mode–PWM
Series Shunt Buck Boost

Dissipation High High Low Low
Efficiency 50–90% 50–90% 85–90% 90–95%
Ripple output Negligible Negligible Moderate Moderate
Load regulation 0.1–0.2% 0.1–0.2% 0.5–1.0% 0.5–1.0%
Output impedance 1 m 1 m 20 m 20 m
current. It is all in the control system, rather than in the basic power train
design.
In all types of converter, there is a trend to increase the switching
frequency to reduce the mass of the power train inductor and the output
capacitor filter. Frequencies up to a few hundred kilohertz are common at
present. However, the design challenges increase as the frequency
approaches megahertz. At these frequencies, the lead wire inductance
and capacitance become significant due to stray and parasitic effects, which
are difficult to analyze.
10.7 Magnetics
Inductor and transformer are two magnetic components used in the
spacecraft power electronics. The inductor stores energy in the form of
magnetic flux. The transformer changes voltage level from one coil to
another via mutual flux in its magnetic core. Figure 10.26 is the simplest
magnetic circuit showing one coil on a magnetic core. The coil current
produces magnetic flux in the core. The magnetic resistance of the core,
called the reluctance, <, is given by the follow expression:
Lm
<¼ ð10:20Þ
r o
where
Lm ¼ mean length of the magnetic path, cm
Ac ¼ cross sectional area of the core, cm2
r ¼ relative permeability of the core material
o ¼ permeability of free space (air or vacuum).
The energy storage or power transfer capability of the core is approximately
equal to Ap, where ¼ 0.85 0.95 and Ap ¼ area product defined as,
Ap ¼ Ac Aw ð10:21Þ
FIGURE 10.26 Coil on a magnetic core produces flux.

where Aw ¼ window area of the core available for winding a coil through it.
The inductor generally has one coil on the core, whereas the transformer
has at least two coils, and sometimes more. The core can take various
configurations shown in Figure 10.27, all of which are used in space power
electronics. The core selection depends on the design drivers, such as
efficiency, voltage regulation, EMI, and manufacturing cost. The toroidal
core contains the flux well internally, and therefore is widely used in high
frequency designs where EMI may be of concern.
The flux density, B, in the core is proportional to the magnetizing
intensity H (ampere-turns per meter length of the flux path). The B varies
linearly with H up to certain value, and then saturates at high H. Under
alternating H, the B also alternates, but does not follow the same path while
rising and falling. This is due to magnetic resistance (hysteresis) in the
atomic structure of the core material. As a result, one complete cycle of H
Ac Ac
Aw
G Aw
D
(a) C-core (tape wound and cut) (b) Tape-wound toroidal core
Ac
Aw D G
(c) Powder or ferrite toroidal core (d) EI-laminations core
Ac
Aw G
Cross section Plan view

(e) Powder and ferrite pot core
FIGURE 10.27 Common types of magnetic core.
FIGURE 10.28 B–H hysteresis loop of the magnetic core.
traces a loop shown in Figure 10.28, known as the B–H or hysteresis loop.
The area of the B–H loop represents energy loss per magnetic cycle. The
power loss in the magnetic core is, therefore, the B–H loop area times the
frequency. It is proportional to B f, where is known as Steinmetz
exponent (or Epstein constant), which depends on the material. A great
variety of magnetic core materials having significantly different B–H loop
are available for various applications. Figure 10.29 is a collection of data on
several commonly used materials.
Under alternating flux, the core metal induces voltage as per Faraday’s
law of electromagnetic induction. The resulting eddy currents circulate
internally to the core metal and produce power loss. The eddy current
power loss is proportional to the core lamination thickness squared. For this
reason, the core is always made of insulated thin laminations or potted
powder to reduce eddy loss, as shown in Figure 10.30. The total loss (in
FIGURE 10.29 B–H loops of common core materials.

FIGURE 10.30 Eddy current loss in magnetic core.
watts) in the core is made of the hysteresis loss and the eddy loss, and is
given by the following expression:
Pc ¼ Kh Bm f þ Ke B2m f 2 t2 ð10:22Þ
where Kh and Ke ¼ proportionality constants for the hysteresis and eddy

losses, respectively, and ¼ Steinmetz constant equal to 1.4 to 1.8 for
magnetic steel laminations.
Technical data sheet available from core vendors plot the total loss per
kilogram of the core material as functions of the flux density and frequency.
Figure 10.31(a) and (b) are examples of such plots useful to the magnetic
design engineer.
Two major properties define the merit of the material for a given
application:
Maximum flux density it can carry before saturation

Total core loss per gram at the switching frequency
The magnetic saturation values of some of the common materials are listed
in Table 10.3. The higher flux density comes with a cost. It results in wider
hysteresis loop, hence higher hysteresis loss. It also results in higher eddy
current loss in the square relation.
10.7.1 Power Losses in Core and Coil

The total power loss in the core and coil under alternating flux is
proportional to the flux density, current, frequency, physical dimensions,
and the material. The loss components are broadly classified as follows
based on the source of the loss:
Losses in the core:

Hysteresis
Eddy current
Fringing flux
FIGURE 10.31 Inductor and transformer core loss curves from vendors.
Table 10.3 Composition and magnetic saturation flux

density of common magnetic core materials
Saturation flux
Core material Composition density (tesla)
Supermendur Proprietary 2.1

Magnesil 97% Fe, 3% Si 1.6
Orthonol 50% Fe, 50% Ni 1.4
Alloy 48 50% Fe, 48% Ni 1.1
Permalloy, Supermalloy 79% Ni, 17% Fe, 4% Mo 0.7
Ferrites Various 0.3
Losses in the coil conductors:

Ohmic resistance
Eddy current
Skin effect
Proximity effect
All of the above loss components must be included in the efficiency and
thermal calculations. Some losses, such as those due to fringing flux, skin
effect and proximity effects, can be significant at high frequency. Details of
their calculations can be found in books dedicated to magnetics.3
10.7.2 Inductor Design

The inductance (in henry) of an inductor is defined as the flux linkage per
ampere. In terms of the notations of Figure 10.32, it is given by
N
L¼ ð10:23Þ
I
where N ¼ number of turns in the coil, ’ ¼ magnetic flux linking (passing

through) the coil, and I ¼ current in the coil. The energy stored in the
inductor is given by U¼ 1=2 LI2. The total reluctance of the core is made of
two parts, one in the core materials itself and the other in the air gap, and is
given by
ðLm Lg Þ Lg
<¼ þ ð10:24Þ
r o Ac o Ac
where the total air gap Lg is lumped in a discrete gapped core or is

distributed in powder cores. Calculating the flux and substituting in
FIGURE 10.32 Energy storage inductor.

Equation 10.23 give the following expression simplified for most practical
designs having Lm/r << Lg:
0:0125N 2 Ac
L¼ h ð10:25Þ
Lg
In discretely gapped cores, the fringing flux around the air gap increases
the inductance by a factor
!
Lg 2G

¼ 1 þ pffiffiffiffiffiffi ln ð10:26Þ
Ac Lg
where G ¼ window height (leg height) in which the gap is located (see
Figure 10.27). A good inductor design maximizes the energy storage and
minimizes the losses. The design is driven by the following major factors:
Required inductance
Direct current it must carry
Alternating current (ripples or otherwise)
Power loss and temperature rise limitations
Mass, manufacturability, and cost
The following is an approximate expression often useful in estimating the

energy storage capability (in joules) of an inductor based on its core and coil
dimensions:
U ¼ 12 KA0:9
p Bm Jm Ku 10
4
ð10:27Þ
where
K ¼ constant of proportionality derived from similarly built designs
Ap ¼ area product of the core
Bm ¼ Bdc þ 1=2 Bac ¼ maximum flux density in core before saturation
Jm ¼ maximum current density in coil before overheating
Ku ¼ window utilization factor, i.e., fraction of the window occupied by the
conductor cross section. It is about 0.5 for pot cores, 0.7 for laminated cores
and 0.8 for tape wound cores.
It is noteworthy that the magnetic energy cannot be stored in a perfectly
magnetic core with infinitely large permeability. This is because it would
produce flux in the core with no current in the coil. The energy can be
stored only in nonmagnetic materials such as air or free space. For this
reason, the inductor core must have some air gap in the flux path, either
discrete as shown in Figure 10.32, or distributed as in potted metal powder
FIGURE 10.33 Energy storage capability versus magnetic permeability.
cores. The energy storage is mostly in the air gap, and not in the magnetic
material. As a matter of fact, the energy stored in an ideal magnetic core is
always zero.
The above point is illustrated by a practical design of an inductor with
potted core 1.35 cm OD, 0.7 cm ID 0.155 cm high with 3.1 g mass. The coil
conductor mass varies with the conductor size around 9.1 g, making the
total mass approximately 12.2 g. The energy storage capability of this
inductor depends inversely with the relative permeability of the core
material as seen in Figure 10.33. The higher the permeability, the lower the
energy storage.
For a circuit simulation, the coil inductance, ohmic resistance and
capacitance to ground can be represented by the equivalent circuit shown
in Figure 10.34. The calculated parameters often need to be calibrated by
tests due to various factors that are not well understood or are complex to
analyze.
10.7.3 Transformer Design

The transformer is a magnetic device, which steps up or down the a.c.
voltage. It has at least two coils mutually coupled via a magnetic core. Its
power transfer rating, P, is expressed as the volt–amp product and not
FIGURE 10.34 Inductor equivalent circuit.

watts. This is because an a.c. device can deliver zero watts when non-zero
voltage and current outputs are out of phase by 90 .
In steady state performance, the voltage ratio of the two coils is the same
as their turn ratio, the current ratio is equal to the inverse of the turn ratio,
and the volt–amp product on both sides remains the same. A good
transformer design maximizes the power rating while minimizing the
losses and the temperature rise. The following approximate relations are
useful in sizing a new transformer based on a similar unit built earlier. The
transformer power is the product of V and I. Since V is proportional to the
core cross section, and I is proportional to the window area available to
wind coils, the power rating is proportional to the area product of the core
Ap ¼ AcAw. That is, P ¼ KA0:9 4
p , which has the unit of cm . Alternatively, it
can be expressed (in volt–amp) as
P ¼ A0:9
p Ku Kj Bm Jm f 10
4
ð10:28Þ
where
f ¼ frequency
Ku ¼ window utilization factor
Kj ¼ current density factor
Bm ¼ maximum flux density the core can carry before saturation
Jm ¼ maximum current density the coil can carry before over heating
Assuming that the total mass is proportional to the volume, which has the
unit of cm3, the transformer mass varies as the 3=4 th power of the area
product, i.e.,
M ¼ Ku A0:75
p ð10:29Þ
where values of the window utilization factor Ku are the same as given for
the inductor design, i.e., 0.5 for pot cores, 0.7 for laminated cores and 0.8 for
tape wound cores. The total surface area Ath available for thermal
dissipation of the power losses, having the unit of cm2, varies with the 1=2
power of Ap, i.e.,
Ath ¼ Ks A0:5
p ð10:30Þ
where Ks is a surface constant, which is about 30 for pot cores, 40 for

lamination cores and 50 for tape wound cores.
We note here that higher frequency gives more power from the same
core, or it requires lower Ap for the same power. This is true for the energy
storage inductor also. That is why high frequency is preferred for reducing
the mass of the magnetics, which is relatively heavy in the power
electronics assembly. However, the skin effect at high frequency requires
FIGURE 10.35 Transformer equivalent circuit.
thin insulated wires, which results in poor space factor. The 20 kHz and
higher frequency coils are often wound with copper foil. Off course, there is
a diminishing rate of return after a certain frequency. Beyond that, the
losses, heating, and parasitic parameters may start impacting the design
negatively.
As in the inductor, the transformer can be represented in the circuit
simulation model by the winding resistance, leakage reactance, and
capacitance to the ground for each coil as shown in Figure 10.35. Looking
from the primary sides, the secondary side impedance becomes trans-
formed by a2, where a ¼ Np/Ns, the turn ratio.
10.7.4 Manufacturing and Testing of Magnetics

In the inductor and transformer manufacture, each device is impregnated in
a vacuum chamber to achieve void-free impregnation. This is done after
drying long enough at high temperature to remove all moisture. The
potting compound is degassed prior to impregnation and potting. The use
of dry air or dry nitrogen may be required in the process. The insulation
requirements that generally apply for spacecraft magnetics are listed in
Table 10.4.
10.8 Design for Maximum Efficiency

Since the power system is as efficient as its components are, designing an
efficient system means designing each component to operate at its
maximum efficiency. The electrical and electronic components while
Table 10.4 Insulation test requirement for inductor and transformers
Test on assembled unit Inductor Transformer
Dielectric withstand voltage 500 V rms 60 Hz 300 V rms 60 Hz

from 1 atm to 5 mtorr
Insulation resistance by Megger 10,000 M at 500 V d.c. 10,000 M at 500 V d.c.
transferring power from the input side to the output side lose some power
internally in the form of heat. In practical designs, the efficiency of 95 to 98%
is typical in large power equipment in tens of kilowatt rating, and 80 to 90%
in small equipment in less than a kilowatt rating. Figure 10.36 shows the
efficiency versus load power for a battery converter boosting the voltage to
28 V from three different input voltages corresponding to the battery state
of charge. Although less boosting means higher efficiency, the efficiency at
each input voltage varies with load.4 It increases with load up to a certain
point beyond which it decreases. A good design maximizes the efficiency at
the load that the equipment supplies most of the time. For example, if the
equipment is loaded at 75% of its rated capacity most of the time, it is
beneficial to have the maximum efficiency at 75% load. This is important for
battery charge converter having output current that varies from the
maximum to a trickle charge level over the sunlight duration. The battery
discharge converter, on the other hand, works near the full load current all
the time during eclipse. Such a design practice is also important in
designing redundant battery converters, which may be working in active
or dormant redundancy. The method of achieving the maximum efficiency
at a desired load level follows.
The total loss in any power equipment has in general three components:
Fixed loss representing the quiescent no-load power consumption. The

fixed loss primarily includes the eddy and hysteresis losses in the
magnetic core. It is always there as long as the unit is energized, and
remains constant at all loads ranging from zero to full load or even
overloads.
FIGURE 10.36 Typical buck–boost converter efficiency versus output load at various battery
voltage levels.
Diode and switching losses varying linearly with the load current, and
Ohmic loss in conductors varying with square of the load current.
The total loss in a component operating under a constant voltage is

therefore expressed as (Figure 10.37),
loss ¼ Lo þ K1 P þ K2 P2 ð10:31Þ
where P is the output power delivered to the load, Lo is the fixed loss and K1
and K2 are proportionality constant derived from curve-fitting the power
loss data. The component efficiency is given by
output output P
¼ ¼ ¼ ð10:32Þ
input output þ loss P þ Lo þ K1 P þ K2 P2
For efficiency to be the maximum at a given load, its derivative with respect
to the load power must be zero at that load. That is,
d Pð1 þ K1 þ 2K2 PÞ ðP þ Lo þ K1 P þ K2 P2 Þ
¼ ¼0 ð10:33Þ
dP ðP þ Lo þ K1 P þ K2 P2 Þ2
This equation reduces to Lo ¼ K2 P2 . Therefore, the component efficiency is

the maximum at the load under which the fixed loss is equal to the square
variable loss. The diode loss does not enter the efficiency optimization. This
FIGURE 10.37 Power converter loss components versus load current — fixed, diodic and
ohmic losses.
is an important design rule for all power components such as power

converters and magnetics.
10.9 Trends in Power Electronics

In striving for mass reduction in converters and the magnetics, the
switching frequency has increased steadily from early designs of 20 kHz
to hundreds of kilohertz at present. Even 500 kHz and 1 MHz converters are
under development for small power levels up to 3 kW at laboratories like
the California Institute of Technology, Virginia Tech, and General Electric.
High switching frequency basically reduces mass of the magnetic and
capacitive components, resulting in light weight and high efficiency. The
high frequency converter power density can approach 1 to 2 W/cm3
(including heat sink), efficiency up to 85% at 5-V output and up to 90% at
15 to 24 V output. They can meet EMI MIL-STD-461 with common mode
noise filter, typically 0.5 cm diameter 3 mm toroid for a 100 W converter.
Typical frequency response has a 20-kHz bandwidth.
The increased power loss at high frequency is addressed by developing
new materials, semiconducting devices, and matrix transformers. The major
cost driver is packaging. Planar magnetic provide compact packaging and
better cooling. However, it has higher leakage inductance and higher mass
per watt output. With presently available materials, the technology trend is
to increase the switching frequencies to the 500 kHz to 1 MHz range using
quasi-resonant topology with an efficiency goal of 90%. The low-cost, low-
voltage, multiple output voltage converter designs for computer applica-
tions use magnetic amplifier post regulator for high efficiency. Similar
approach may be considered for special applications in space. The tantalum
and electrolytic capacitors are being replaced with film capacitors for
further mass reduction.
The power electronics is a very wide and developing subject. The intent
of this chapter has been to present an overview of the basic circuits used in
the spacecraft power systems. Further design detail can be obtained from
many excellent books on the subject.5–7
References
1. Cuk, S., Switching dc-dc converter with zero input or output current ripple,
IEEE Industry Applications Society Conference Records, October 1978.
2. Cuk, S., Analysis of integrated magnetics to eliminate current ripples in
switching converters, Power Conversion International Conference
Proceedings, Intertec Publishers, Oxnard, CA, April 1983.
3. McLyman, W.T., ‘‘Transformer and Inductor Design Handbook,’’ Marcel
Decker, New York, 1988.
4. Fox, D., Testing of the engineering model of electrical power control unit
for the fluids and combustion facility, in Proceedings of the 34th Intersociety
Energy Conversion Engineering Conference, SAE, 1999, Paper No. 2435.
5. Wu, K.C., Pulse Width Modulated dc–dc Converters — Analysis and Design,
Kluwer Academic Publishers, Norwell, MA, 1997.
6. Wu, K. C., Transistor Circuits for Spacecraft Power System, Kluwer Academic
Publishers, Norwell, MA, 2002.
7. Mohan, N., Undeland, T., and Robbins, W.P., Power Electronics Converters,
Applications and Design, IEEE and John Wiley & Sons, 1995.
Chapter 11
Distribution Harness and Protection
11.1 Introduction
The harness includes the insulated conductors, connectors, and the shield.
Its mass is determined from the detailed layout and routing of all the
required wiring. For this reason, it is often considerably more than that
estimated without actual layout at a preliminary design stage. A typical
harness mass breakdown is given in Figure 11.1.
11.2 Ampacity of Wires

The maximum current carrying capacity (ampacity) of the wire in space is
less than that on the ground due to lack of convective cooling. This requires
derating of wire ampacity for space applications from the ground-based
rating. The wire size is measured in American Wire Gage (AWG) or British
Wire Gage (BWG). The gage number is an inverse measure of its bare
diameter, and is set on a log scale:

0:325
AWG ¼ 20 log ð11:1Þ
d
where d ¼ diameter of bare conductor in inches. Thus for every one gage
up, the diameter increases by a factor of 1.1225 and the area by 1.26. The
diameter doubles every six gages and the area doubles every three gages.
The ampacities of various gages of wires are listed in Table 11.1.
FIGURE 11.1 Harness mass breakdown in typical spacecraft.
293
Table 11.1 Maximum allowable current in amperes for wires and

connector contact pins of the same gage
Single wire in
Diameter free air on ground Wires in spacea in 70 C ambience
AWG (inches) (MIL-STD-5088) (MIL-STD-975 and GSFC-PPL-19)
Single wire Bundle or cableb

30 0.010 0 N/a 1.3 0.7
26 0.015 9 10.5 2.5 1.4
24 0.020 1 14 3.3 2.0
20 0.032 0 24 6.5 3.7
16 0.050 8 37 13.0 6.5
12 0.080 8 68 25.0 11.5
8 0.128 5 135 44.0 23.0
4 7 0.0772 260 81.0 40.0
0 19 0.0745 460 147.0 75.0
a
For TFE teflon insulated wires rated for 200 C.
For 150 C rated insulation, use 80% of values shown.
b
For cable bundles of 15 or more wires in 70 C ambience in hard vacuum. For
smaller bundles, the allowable current may be proportionately increased as the
bundle approaches a single conductor.
c
Wires sizes 10, 14, and 18 are not used in aerospace for general wiring, and
AWG 2 and 6 have no counterpart electrical connector contacts (pins).
11.3 R-L-C Parameters

The resistance (R), inductance (L), and capacitance (C) parameters of the
harness are required for detailed simulation model of the power circuit.
They are determined from the expressions given below, where the values
are per meter length of cable (lead and return wires together). When wires
are twisted, typically at a rate of one twist every few centimeters, the actual
conductor is approximately 10 to 15% longer than the straight run. The
resistance (in /m) is given by
2
R¼ ð11:2Þ
A
where ¼ resistivity of the conductor and A ¼ cross-sectional area of one-

way conductor(s). At 20 C, ¼ 1.724 108 m for copper and 2.65 108
m for aluminum, both of electrolytic grades.
The resistance R2 at temperature T2 is derived from R1 at temperature T1
using the temperature coefficient of resistance as follows:
R2 ¼ R1 ½1 þ ðT2 T1 Þ ð11:3Þ
Distribution Harness and Protection 295
where ¼ 0.00393 1/ C for electrolyte copper and 0.00390 1/ C for
aluminum.
For a cable with two parallel round conductors of equal radii (Figure
11.2a), the cable inductance (in H/m) in free space is given by

s
2o 1
L¼ þ ln
2 4 r
or

s
L ¼ 1 þ 4 ln 107 ð11:4Þ
r
where o ¼ permeability of free space ¼ 4 107 H/m, s ¼ spacing

between conductor centers, and r ¼ conductor radius.
The term 1=4 is due to the magnetic flux internal to the conductor. It is
small and approaches zero at high frequency due to skin effect. For twisted
pairs touching each other, s ¼ 2(r þ wire insulation thickness). Since the
insulation thickness is proportional to the wire radius for mechanical
reasons, the s/r ratio remains approximately constant, making L somewhat
insensitive to the wire gage. For AWG 10 to 30 wires, the inductance
remains in the range of 0.5 to 0.7 mH per meter length of twisted pair up to
hundreds of kilohertz frequency. Twisting does not reduce the inductance
per meter, but adds 10 to 15% in the length. However, it produces additive
and subtractive B fields in adjacent spans (Figure 11.2b), canceling each
other’s emission at far distance. The inductance of open untwisted wire
loops at the end of an equipment box is determined from Equation 11.4. It is
about 5 mH/m for practical loops and 7.5 mH/m for infinitely large loops. It
is noteworthy that the single point ground reduces the ground inductance
by avoiding large ground current loops. The inductance of a ground strap is
that of a single wire in free space with the ground as return. Its values are
plotted in Figure 11.3.
FIGURE 11.2 Parallel wires and resulting B field.

FIGURE 11.3 Inductance of single straight wire (ground strap).
The battery packaging influences the harness inductance and the

magnetic moment. A one-pack battery has a small harness inductance but
a two-pack battery can have a large harness inductance and the magnetic
dipole moment depending on the spacing between them. Therefore, it is
important to place all packs of the battery as close as possible to minimize
the interconnecting harness loop.
The capacitance (in F/m) between two parallel round conductors in the
cable is given by
1
ð2"o "r Þ
C ¼ 2
s
ln
r
or
"o "r
C¼ " ffi#
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
ð11:5Þ
s s 2
ln þ 1
2r 2r
where "o ¼ permittivity of free space ¼ 8.85 10(12 F/m, and "r ¼ relative
permittivity of the insulation between the conductors. The first expression
is for open wires at the harness ends, and the second expression is for
closely spaced conductors in the cable. Unlike the resistance, the L and C
parameters do not change with temperature.
Bus bars of thin rectangular cross section shown in Figure 11.4 are often
used in harness segments carrying high current. For such conductors, the
R-L-C parameters after ignoring the end fringing are
2
R¼ =m ð11:6Þ
ab
FIGURE 11.4 Rectangular bus bars for high current harness.

o 2a
L¼ dþ H=m ð11:7Þ
b 3
"o "r b
C¼ F=m ð11:8Þ
d
A typical communications satellite bus designed for 10 kW, 70 V power

with copper-clad aluminum bus bars, 5 cm wide 2 mm thick, would
typically have Rdc ¼ 2 m at 70 C, L ¼ 50 nH, and C ¼ 3 nF per meter of the
bus length. A cable using 20 AWG round copper wires would have Rdc ¼
70 m/m, L ¼ 0.375 mh/m, and C ¼ 0.077 nF/m.
The following approximate values can be used as a quick estimation
guide for wire gages generally used in spacecraft. The AWG 10 conductor
has 100 mils diameter, 40 g mass per meter and 3.3 m resistance per meter
at 70 C. Every three gages doubles the area and every six gages doubles
the diameter. With 2.3 A/mm2 typical current density, the harness mass
(conductor þ insulation þ shield, but not the connectors) is about 7 g/A one
way, and the watt loss is about 2.7 conductor mass in kg. At 100 kHz, the
skin depth is 0.22 mm and Rac/Rdc ¼ 9. At 1 MHz, the skin depth is 70 mm
and Rac/Rdc ¼ 27.
11.4 Conductor Materials

The most commonly used electrical conductor is copper for its good
performance and low cost. Annealed copper is often used for high
conductivity, but it has low tensile strength. This makes it unsuitable for
use in thin wires due to risk of breakage in handling. For this reason, wires
thinner than AWG 20 are often required to use high strength copper alloy
135, which has 40% higher tensile strength, and 10% higher electrical
resistance. Copper wire coated with tin, nickel or silver is used to resist
corrosion and oxidation. The suitability of various coatings is compared in
Table 11.2, where tin plating is included for comparison only. Tin plated
Table 11.2 Suitability of various coatings on

copper conductor
Coating on copper Silver Nickel Tin
Solderability Good Special process Good

Crimp ability Good Special process Good
Maximum operating 200 C 260 C 150 C
temperature
wires are widely used on ground, but are forbidden in space due to growth
of whiskers.
Aluminum is sometimes used in power equipment where lightweight
and/or low cost is desired. It is used in overhead transmission lines and
pole mounted power transformers on ground, and in some aircraft and
commercial spacecraft harness. The performance of aluminum is compared
with copper in Table 11.3.
For the same power loss or voltage drop, the copper can be replaced with
aluminum of relative mass given by the product of the resistivity and the
mass density ratios. An aluminum conductor, therefore, would theoreti-
cally have (2.830/1.724) (2.70/8.89) ¼ 0.50, i.e., one-half of the copper
mass. However, in practice, aluminum does not produce 50% mass saving
owing to the following reasons:
Aluminum has about 1000 times higher creep rate then copper. For
long-term reliability of the terminal connectors, thicker wire is required
to keep the crimp joint temperature below 130 C to minimize creep.
Alternatively, soldered connections are used, which are expensive and
often inconvenient.
Aluminum oxidizes instantly when exposed to oxygen. This hinders the
electrical contact between the wire and the connector. Crimp terminals
that produce cold weld are often used to keep oxygen away from bare
aluminum.
Table 11.3 Copper and aluminum conductor comparison
Characteristic Copper Aluminum
Resistivity m at 20 C 1.724 108 2.830 108

Mass density g/cm3 8.89 2.70
Temperature coefficient of resistance / C 3.93 103 3.90 103
Melting point C 1083 660
Flex life (relative) 1 0.5
Thermal coefficient of expansion (relative) 1 1.4
Creep rate at 65 C (relative) 1 1000
The poor mechanical strength of aluminum wire often requires thicker

wire for mechanical reason alone, even where it meets the ampacity
requirement.
The insulation and the EMI shield mass increases due to heavier gage
aluminum wire required to carry the same current.
As a hybrid approach, copper-clad aluminum wires have been used in

the past by some prime contractors in communications satellites with net
mass savings of 10 to 15%. However, aluminum or copper-clad aluminum
wires are generally not allowed on NASA or defense satellites.
11.5 Wire Insulation and Cable Shield

The insulation is designed to withstand the rated and abnormal transient
voltages. The transients can be higher by several times the rated value. The
insulation design must preclude corona and arcing at pressures below 10
torr, and withstand the radiation environment and atomic oxygen. In high
radiation spacecraft, such as GPS, the system specifications required that
the silicon-insulated wire be not used and the solar array wires and
interconnects be welded to withstand the high radiation.
The cable shield is wrapped around the wire bundle to prevent
electromagnetic interference from entering the cable or radiating out. The
shield can weigh 15 to 40% of the cable weight. The material options are the
braid versus tape and copper versus aluminum. The braid is used when
extreme flexibility is required. Its mass as compared to the wire conductor is
about 40% for flat cables and 20% for round cables. In tapes, 2-mil thick
copper vapor deposited on Mylar or Kapton tapes are widely used. The
tape is applied on the cable with insulation touching the cable, followed by
another layer on the top. The shield mass with such tapes is roughly 80% of
that with the braid. The EMI shielding is discussed further in Chapter 15.
11.6 Connectors
Both crimp and soldered connectors are used for connecting various wire
segments. They come in Class A and B with maximum operating
temperature of 125 and 200 C, respectively. They are further grouped in
four types as listed in Table 11.4.
Series A rectangular connectors are limited to AWG 20 pins, whereas
series B round connectors come in heavy pins, some of them in assorted
pins to handle different power circuits. The wires are crimped with pins
with a special tool. Wires thicker than pins are not permissible but can be
up to two gages thinner. In both series A and B connectors, contact pins are
further classified as Type A and B. The maximum voltage drop across type
A and B contact pins of various gages are listed in Table 11.5.
Table 11.4 Connector classification
Type Material Application
A Copper alloy General purpose

B Ferrous alloy Hermetically sealed
C Matching with thermocouples Thermocouples
D Copper alloy Shielding (including coaxial,
twin-axial and triaxial)
The general military standard isolation requirement on the connector is:
High potential withstand capability between pin-to-pin and pin-to-shell

at least 4 times the operating voltage
Insulation resistance between pin-to-pin and pin-to-shell at least 5000
M
The connector design practice as applied to the spacecraft power system

generally includes the following:
Pins in a connector and the SAD slip rings are assigned such that wires
at different electrical potentials are separated and are not on adjacent
pins or slip rings.
Twisted wire pairs are terminated on adjacent pins in the connector.
A minimum of 10% of spare pins is provided on the outer periphery of
the connector.
Each connector shell is grounded to the component case with
impedance less than 2.5 m.
Table 11.5 Permissible contact pin drop in millivolts at 200 C after conditioning
as per Military Standard MIL-C-39029
Wire Test Type A with Type A with Type B with Type B with
size current silver-plated nickel plated silver-plated nickel-plated
AWG (amp) wires wires wires wires
0 150 36 90 396 900

4 80 40 100 440 1000
8 46 45 113 495 1130
12 23 71 107 781 1177
16 13 84 126 924 1386
20 7.5 94 141 1034 1551
24 3 77 116 847 1276
28 1.5 92 138 1012 1518
32 0.5 74 111 814 1221
In defense satellites, multiple contact connectors are not used for high
voltages above 300 V d.c. or peak-to-peak a.c. voltage. Separate
connectors rated for the specified high voltage are used for each
connection.
Hermetically sealed boxes must use hermetically sealed connectors
with connector cases welded to the boxes.
11.7 Harness Mass Minimization

The harness mass is not small compared to other components. It is
primarily driven by the voltage drop specifications between various
harness segments. For example, the round trip voltage drop from the bus
voltage control point in the power regulator unit (PRU) to any load is kept
below 1 to 2% of the rated bus voltage. In a 28-V bus, the round trip drop
from PRU to any load is typically kept below 250 mV for loads below 100
W, 500 mV for heavier loads, and 1 V for heater circuits drawing power
only for a part of the orbit period. The voltage drop specifications,
particularly in high voltage spacecraft, do not often result in an optimized
design in the following sense. If a higher voltage drop in a thinner wire can
be compensated with added mass in the source that is less than the mass
saved in thinner conductor, the trade is beneficial for mass reduction at the
power system level. Therefore, the harness power loss should match with
the specific power of the source. For this reason, initially proposed voltage
drop and derating guidelines in some spacecraft should be carefully
examined before they become specifications.
The mass optimized harness design is derived from the considerations
depicted in Figure 11.5(a). The power loss in the conductor is equal to the
allowable voltage drop times the current. As the power loss increases, the
harness mass decreases, but the mass of power source increases to make up
the loss. The total of the two masses is a minimum at a certain optimum
power loss in the harness, resulting in the minimum system mass. Figure
11.5(b) shows the harness power loss and the source power generation in
watts per kilogram versus current density in the conductor. The harness
with current density Jopt would result in the minimum systems mass. At the
optimum current density, the loss per kilogram of harness equal the power
generated per kilogram of the source.
Analytically, the power loss per pound of copper conductor at 75 C is
given by 2.54 J2, where J ¼ current density in kiloamperes per square inch.
For optimum conductor size, it must be equal to the specific power of the
source Psp W/lb. This leads to
qffiffiffiffiffiffiffi
Jopt ¼ 627:5 Psp ð11:9Þ
FIGURE 11.5 Harnessing mass optimization at the power system level.
The optimum conductor diameter (in inches) is then given by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4 Irated m
d¼ ð11:10Þ
Jopt
where m ¼ margin factor depending on the protection method used,

typically 1.3 for fused circuits. The wire gage is then derived from Equation
11.1 rounded up. Such consideration leads to a simple expression for Vopt,
the optimum voltage drop (in volts) per meter run of the harness (physical
distance between the source and the load). It is given by
qffiffiffiffiffiffiffi
Vopt ¼ k Psp ð11:11Þ
where Psp ¼ specific power of the source in watts/lb. The coefficient k is 0.03
for copper wires and 0.026 for aluminum wires. Harness segments having
varying Psp over the orbit period can be designed for optimum voltage drop
using weighted average Psp.
The harness mass optimization approach leads to the following obser-
vation for the conductor sizes in view of Figure 11.6.
FIGURE 11.6 Progressive optimization of harness mass.
The harness used with a source of high specific power, such as triple
junction solar array, may operate at higher current density, allowing a
higher voltage drop as percentage of the bus voltage.
Thinner wire is justified from the solar array to PRU, because the power
loss in that segment of the wire is supplied by the solar array alone.
Thicker wire is required from the PRU to loads, because the power in
that segment has been processed by the solar array, PRU, and the
battery, all together during a part of the orbit. The whole system has
much lower specific power compared to that of the solar array alone.
For the same reason, the current density in the discharge path should be
lower than that in the charge path. This is because the charge current
comes from the solar array alone, whereas the discharge current
involves the whole system.
So, as we move closer to the loads at the end of distribution lines, the
current density should be progressively lower. Each watt becomes more
precious at the load point compared to that at the generation point. In
ground-based utilities, this is analogous to the cost of power to end users
being more than five times the cost in the central power station yard. It must
be noted that the above considerations lead to the minimum harness mass,
and not the minimum cost at the power system level. The cost of adding one
pound of source is always much greater than the cost of adding one pound
of harness. However, the harness mass optimization may be warranted
when the spacecraft mass is rigidly constrained to fit the launch vehicle or
for some other reason.
11.7.1 Example
The harness mass optimization is more important in LEO satellites. The
heavy battery in LEO makes the specific power of the entire power system
about one-half of that in GEO satellites. For example, let us consider a fully
regulated LEO power system where the battery mass is equal the solar
array plus the PRU mass. We wish to determine the optimized wire gage
for a solar array circuit carrying 5-A current from the silicon PV array to
the load through slip rings, and also for a 5-A load circuit from the PRU
to the user load. We are given that the specific power of the (array þ
PRU) is 10 W/lb, and the specific power of the (array þ PRU þ battery)
is
p 5 W/lb. The6 optimum current density for the slip ring circuit is
f10=ð2:54 10 Þ ¼ 1984 A=in2 . Allowing 30% margin, the required cross
section area of the wire is (1.3 5/1984) ¼ 0.0033 in2, which requires AWG
14 wire. Using the GSFC-PPL-10, a bundle of 14 AWG wires has a derated
current carrying capacity of 8.5 A, which is greater than (1.3 5) ¼ 6.5 A.
Thus, the mass optimized wire gage meets the preferred part list require-
ments. If the solar array were made of triple junction cells with specific
power of 5 W/lb, then the optimum current density would have been p{5/
2.54 106)} ¼ 1192 A/in2. This results in conductor cross section 40%
smaller, requiring AWG 12 wires. If such optimized wire gage results in a
local hot spot somewhere in the harness run, spreading out the heat with a
local aluminum foil should be considered first rather than increasing the
wire gage for the entire harness run.
11.8 Harness Design Process

The harness design is influenced by the following factors:
The rated current and the allowable voltage drop determine the
conductor gage
The bundle size and steady state temperature also determine the gage
The rated voltage determines the insulation thickness and the
temperature rise determines the insulation material
The derating requirement increases the gage thickness
Specific power of the source is factored in to determine the optimum
wire gage
Reliability goal may require redundant wires or double insulated wires
The harness design proceeds in the following steps:

1. Determine the maximum steady state d.c. or a.c. rms current for each
harness segment on the spacecraft. For each segment, size the wire to
carry the required current per applicable derating factors. The harness
wires always run in large bundles, so the derating factor for the bundle
of 15 or more wires generally apply.
2. Determine the voltage drop and compare with the requirement. If the
requirement is not met with the wire selected in step 1 above, use more
wires in parallel or increase the wire size until the voltage drop
requirement is met.
3. If an open wire in a current path constitutes a failure, i.e., if either the
voltage drop or the derating criteria is not met with one wire open, and
if there is no other redundant current path, then another wire should be
added to guard against this mode of single point failure.
4. Ensure that the wire(s) in a particular current path will not be damaged
to the point of propagating the damage to the surrounding wires when a
smart short occurs. A smart short is one that draws current insufficient
to blow the fuse but in excess of the wire derating. In such a case,
determine the value of the steady state current that can cause damage to
the wire insulation and could propagate to other wires. If that current is
less than the expected steady state d.c. or rms fault current, then the
harness must be increased in size to a point where the smart fault
current cannot cause damage that could propagate through the harness.
5. The harness should be secured to withstand the mechanical repulsive
forces, which can be significant during hard faults. The wires are
generally tied with each other or all in a group, often attached to
structural tubes as shown in Figure 11.7.
The mechanical repulsion force (in newtons per meter) between two
parallel conductors in configuration of Figure 11.2(a), carrying currents I1
and I2 in opposite directions, is given by
s
F ¼ 80 K I1 I2 107 ð11:12Þ
d
FIGURE 11.7 Placement and attachment of harness with structural tubes.

where K ¼ conductor shape factor ¼ 1.0 for round conductors. The force
calculation on high power rectangular bus bars is covered in Chapter 22.
11.9 Flexible Harness

The slip rings can experience cold welding in high radiation satellites, as
with a GPS. The flexible harness (coiled cable) that can accommodate the
sun-tracking rotation of the solar array can eliminate the slip rings. The
harness with 90 flex has been considered for MEO satellites requiring
20,000 cycles of flexing over 15 years of mission life. Incorporating a flex
harness in the design at present is within a development reach. Some key
features of the design are:
Flex harness is completely enclosed inside a Faraday cage (solar array

boom), hence unaffected by the deployment behavior, environment
shielding, EMI, etc.
Harness flexing is required only between the SAD and the structural
bulkhead.
Harness outside the Faraday cage remains the same, so the cable
interface with the bus is unaffected.
The cable needs to be ultra-flexible with very thin hair-like strands. This
may impose a manufacturing problem in crimping and ascertaining that no
whiskers are left projecting out. A fully developed space-qualified flex
harness could eliminate the slip rings, which weigh several kilograms per
spacecraft. The harness mass, however, would remain the same. A flexible
fluoropolymer insulated cable with 20-gage wires for power and 24-gage
wires for signals has been built and tested in the past.
The flexible harness has been used by ESA in 2002 for a Mercury mission.
In this design, the solar panels rotate 180 around one axis to keep the
operating solar cell temperature below 150 C at AU < 0.375. The sun
shielded solar array drive has no slip rings, but uses flex harness sufficiently
flexible for 180 rotation.
11.10 Fuse Protection

The fuses are provided to protect the power system from internal faults or
from faults in load circuits. The general fusing practice in the spacecraft is
as follows:
Fuses are placed near the source for protection against faults in the
wires, and also to reduce the local fuse blow voltage spikes in adjacent
loads (cross-talks). Fuses are located as close as possible to the PRU
(Figure 11.8). The fuse boards are often located inside the PRU.
FIGURE 11.8 Fuse boards located in or near PRU for various load groups.
All noncritical loads are individually fused to prevent a fault in any

load from damaging the power system.
In full-length redundant wires, fuses are placed in both lead wires
(Figure 11.9).
Critical loads are not fused to maximize the reliability, but they use
double insulated wires. When a fuse is required in a critical load circuit,
redundant fuses are used, as often required by NASA (Figure 11.10).
Suitable load path redundancy is provided such that a single fault in a
device or wiring does not permanently open or short the main bus itself
or essential bus loads.
A typical construction of the fuse used for protecting semiconducting

devices is shown in Figure 11.11. The fuse element may be silver, copper or
nickel, with silver being more common for long-term stability. It has a
reduced section for better predictability. The body is generally filled with
sand-type filler to suppress sparks when the fuse element melts and
interrupts the inductive energy of the load and the wires. The fuse life is
determined primarily by the metal aging under thermo-elastic cycles.
Because of the lack of a sufficient market, only a handful of commercially
available fuses have been space qualified in limited ratings. Among
them are PiocofuseTM series 265 and LittleTM fuses in small ratings;
and BussmannTM fuses in large ratings. They are basically commercial
FIGURE 11.9 Full-length redundant wires with fuses in both lead wires.
FIGURE 11.10 Redundant fuses in a single, nonredundant, double insulated wire.
nonsealed fuses, except that source control drawings are added for
tractability to qualify for space.
Picofuses use nickel elements in less than 1 A ratings, 50% silver and 50%
copper elements in 1 to 4 A ratings, and copper elements in greater than 5-A
ratings. All element materials have high positive temperature coefficient of
resistance. They rise in resistance with temperature, thus preventing an
absolute short.
Nonsealed fuses, however, pose design difficulties when bus voltage
exceeds 50 V d.c. This is due to arcing in partially evacuated construction.
Hermetically sealed fuses address this problem. P600LTM fuses made by
Mepcopal Corporation are such space-qualified fuses. These fuses are
made using thick film material. The fuse element is primarily gold while
the terminal pads are silver. These materials are deposited onto a 96%
alumina substrate and coated with dielectric overglaze cured at high
temperature. An arc suppressant glass of about 0.5 mm is then printed
over the fuse element. When the fuse melts under a fault, the gold
element melts and migrates readily into the arc suppressant glass coating,
thereby effectively preventing subsequent vaporization and ionization.
This suppresses arcs and restricts the circuit voltage after the fuse blow.
For these reasons, Mepcopal fuses are fast and more predictable in
clearing time. The hermetically sealed case contains the spark while
blowing. In fact, they have been developed for safe operation in explosive
mixture of chemical vapors for the National Electrical Code Class 1
environment (hazardous).
FIGURE 11.11 Typical semiconductor fuse construction.

The ground insulation requirements imposed on fuse vary with the

voltage ratings. For example, a 100-V fuse would typically have:
Dielectric withstand of 500 Vrms

Ground leakage current less than 1 mA
Resistance after the fuse clearing between 1 M and 10 M depending
on the fuse rating and the fault current it interrupts.
11.10.1 Fuse rating

The ampere rating of the fuse is the current it would carry indefinitely on
ground. The fuse clears (blows) only at higher currents as per its own I–t
relationship. The fuse is required to carry 110% of rated current for at least
4 h without blowing, blow within 1 h at 135%, and within 2 min at 200%
of the rated current. On the faster side, the fuse clearing time is in
microseconds for 10 times the rated current. The general criteria for setting
the current rating of the fuse are:
The fuse carries 110% of the rated current for a long time in hours
without exceeding 70 C above the room temperature.
The fuse clears 135% of the rated current within a specified time in
minutes, which depend of the type of fuse.
The fuse withstands full recovery voltage, which is equal to the open
circuit voltage that appears across the fuse terminals after clearing the
worst fault.
The selected fuse rating is typically 1.2 to 1.3 times the current rating of
the load circuit it protects, rounded upward to the next standard available
rating. The fuse rating must be carefully selected because both the
conservatively or liberally sized fuses are bad for the circuit protection.
The ambient temperature has an influence on the current rating of the fuse
as shown in Figure 11.12. It is desirable to up rate or down rate the fuse
accordingly.
As for the voltage rating, the fuse can be operated at any circuit voltage as
long as it can blow without suffering arc damage. When the fuse blows, the
sharp break in the circuit current cause the full circuit voltage to appear
across the blown fuse. If the load is inductive, the transient voltage across
the fuse may be substantially high due to the inductive kick. A destructive
arc may be formed across the fused element and continue to grow. The
resulting heat and pressure may cause the fuse to explode in the worst case.
The voltage rating of the fuse is conservatively given such that the fuse can
interrupt a dead short without shattering or emitting flame or expelling
molten metal.
FIGURE 11.12 Effect of ambient temperature on current carrying capacity of fuse.
The equivalent circuit during fuse current build-up and interruption is

shown in Figure 11.13. The total energy dissipated in the fuse body equals
the energy stored in the circuit inductance plus the energy supplied by the
main power source. It should not rupture the fuse body for safety. The
current build up obeys the following equation.
di 1 h i
¼ Vs Vf Ri ð11:13Þ
dt L
where Vs ¼ source voltage, Vf ¼ fuse voltage, R ¼ fuse resistance and i ¼

fuse current. The fuse clearing has two distinct phases shown in Figure
11.14.
11.10.1.1 Pre-arc period

The pre-arc period lasts until the arc voltage appears across the fuse
terminals. It is a function of the fault current, the elements neck section and
I (t)
R I (t) I
Time
+ 0 tp tt
Vs Vf
_
_ V +
f
Time
(a) Circuit (b) I and V of the fuse during operation

FIGURE 11.13 Equivalent circuit during fuse current build-up and interruption.
FIGURE 11.14 Fuse clearing process.
the material. In silver fuses, it lasts until the joule integral I2ta2 reaches
80,000 A2smm4 with adiabatic heating, where I ¼ rms current in ampere,
t ¼ pre-arc time in seconds, and a ¼ fuse element cross-sectional area.
11.10.1.2 Arc period

The arc period lasts until extinction of the arc. The arc voltage during this
period opposes the source voltage, limits the current, and eventually
extinguishes the current. The higher the arc voltage, the shorter the arc
duration. The arc energy is absorbed by the heat capacity and transforma-
tion of the surrounding material.
11.10.2 Types of Fuse

Although standard fuses are used in most load circuits, some loads may
need relatively fast or slow acting fuses. Three types of fuse are available in
the industry for different applications:
Standard fuse for general use

Time-delay (slow-blow) fuse designed to withstand inrush current
drawn by most equipment for a short time after turn-on
Current limiting (fast-blow) fuse, which limits the current by clearing
fault before current builds up from normal to full fault value through
inductive response of the circuit.
The typical blow-time and applications are listed in Table 11.6. Figure 11.15
shows current versus average blow characteristics of the three types. Even
within a given type, significant variations exist around the average value,
which must be considered in the system design to protect the circuit as
required.
Table 11.6 Fuse types and their typical applications
Typical blow time

(seconds) at 200% of rated
Fuse type current Typical applications
Fast blow <1 Transistor circuits and meter protection

Normal blow 1–10 Most standard circuits
Slow blow > 10 Circuits drawing high inrush current on
switch-on, e.g., motors, transformers,
heaters, capacitor charging, etc. It avoids
nuisance blowing.
11.10.3 Fuse Characteristics

The fuse element generates heat at the rate of I2R watts. Under a steady state
condition, this heat is dissipated in the surrounding, keeping the element
temperature and resistance stable. Under a fault, the heat generated at very
high rate raises the element to melting temperature, typically in milli-
seconds. Assuming adiabatic heating (negligible heat escaping the fuse
body in short time), the thermal energy balance requires I2Rt to be equal to
the heat capacity of fuse up to the melting temperature, which is constant
for a given fuse. If the element resistance were constant, I2t is also constant.
The clearing time t is thus inversely proportional to I2 in the first
approximation. The fuse resistance rising with temperature alters this
relationship to some extent. It is expressed as
Rf ¼ Ro et= ð11:14Þ
FIGURE 11.15 Current versus blow time for three types of fuses. (NASA/JPL Specifications
ZPP-2061-PPL.)
where Rf ¼ fuse resistance at time t after the fault, Ro ¼ initial fuse resistance
at normal operating temperature (generally at 25 C), and ¼ thermal time
constant of the fuse. Typically, the fuse clearing time tc ¼ 1.5, giving Rf ¼
4.5Ro when the fuse clears. Until then, Rf = Ro e1.5t/tc. Denoting the fault
current by If, we have
If2 tc ¼ K
or
tc ¼ K=If2 ð11:15Þ
which gives the fuse resistance at any time as

!
1:5If2 t
Rf ¼ Ro exp ð11:16Þ
K
for 0 < t < tc. The fault current decays with the rising fuse resistance. The
root mean square value of the decaying fault current should be used in the
above expression for better accuracy.
The room temperature resistance of the fuse element has significant
manufacturing variations, but it primarily depends on the current rating.
However, the voltage drop across the fuse at nominally rated current is kept
between 100 and 200 mV by design. The manufacturing variation in
clearing time is also significant. For this reason, the I–t characteristic is
generally depicted by two curves bounding the minimum and maximum
times to clear the fuse at a given current (Figure 11.16). Such curves are
derived by the vendor from numerous tests. Test data on recent
manufacturing lots are often shown by dotted lines for guidance to the
users on the latest design and manufacturing practices. Even than, the
variations in clearing time are wide and must be accounted for in the power
FIGURE 11.16 I–t characteristic band of normal-blow hermetically sealed fuses.

system design. Table 11.7 shows the variations in hermetically sealed fuses.
Variations of 1:10 or 1:15 are common. They are even wider for traditional
nonsealed fuses. The clearing time is less predictable in the normal and
slow blow fuses, because the plasma lingers on for longer duration. The
plasma behavior is difficult to predict with high certainty. The fast fuse and
sealed fuse, on the other hand, have more predictable clearing time.
Many aerospace customers require transient survival analysis to demon-
strate that each load’s input current matches the blow curves of the fuse
assigned to the load. With the actual parameters of the assigned fuse,
detailed fault simulations are run on computer to verify that:
The fuse will clear under the minimum fault current using the right
hand side of the I–t envelope in the simulation run.
The fuse will not clear under maximum inrush current during start-up
of motor and heater type equipment using the left-hand side of the I–t
envelope in the simulation run. In circuits where high start-up current
lingers for a long time, posing a risk of unwanted fuse blow, one may
use a slow-blow fuse of the same rating.
11.10.4 Fuse Derating

The arcing and heat transfer in space are much different compared to that
on Earth where the fuse rating is established by tests. The space-qualified
fuse is sufficiently derated from its normally rated current on Earth to
obtain high reliability under the actual load current it would carry in
service. The derating factor is defined as the ratio of the permissible load
current to the nominal fuse rating. The purpose of derating is essentially to
stay below the current when the resistance of the fuse element remains
stable. Otherwise, it will experience an eventual thermal runaway leading
to melting. Significant variability exists in the blow characteristic and the
atmospheric pressure surrounding the fuse element in space. For this
reason, nonsealed fuses for 1 A and smaller loads are typically derated to
Table 11.7 Typical variations in blow time at 500% rated

current in hermetically sealed space qualified fuses
Current rating Minimum blow Maximum blow

(amp) Fuse type time (second) time (second)
3 Normal blow 0.0004 0.0065

3 Slow blow 0.002 0.02
10 Normal blow 0.0004 0.0065
10 Slow blow 0.003 0.045
35%, and 70% for 5 A and larger loads. Since the blow characteristic of
hermetically sealed fuse is more predictable, the corresponding derating
factor is 80% for fuses larger than 0.5 A rating. The largest hermetically
sealed fuse available at present is 10 A. Further derating can come from
higher than normal ambient temperature. It follows the following rules:
0.5% derating per C ambient above 25 C for time-delay fuses

0.2% derating per C ambient above 25 C for regular fuses
Yet additional derating may be imposed based on the specific mission

environment. If the total fuse derating in space is say 40%, then a fuse with
nominal ground rating of 10 A must be used in 4-A load circuit in space.
This, however, creates a protection problem in ground testing, where a soft
fault may carry the current for a long time and burn the wires. For this
reason, an additional 4-A fuse (40% of the nominal 10-A rating) is inserted
in series all the way up to the launch pad. It is then removed before the
launch.
Most faults in space are hard faults that clear the fuse quickly. The fuse
closer to the main bus clears even quicker under fault. However, secondary
circuit fault at the end of a long line may clear slowly due to the voltage sag
and the resulting low fault current.
11.10.5 Fuse selection criteria

The following considerations enter the fuse selection process:
Fuse current rating after the required derating must be equal to the
nominally rated circuit current after allowing for the overload margin
and the harness gage.
Fuse voltage rating must be greater than the source voltage for a.c. or
d.c. circuits.
Maximum current interruption capability of the fuse must be greater
than the maximum possible fault current at the fuse location.
Fast fuse is not necessarily better in all applications. However, it is
required to protect semiconductor devices with I2tfuse < I2tsemiconductor.
With a fast fuse, the clearing time is less than the first peak of the fault
current, and is more closely governed by constant I2t because the rise to
melting temperature is much closer to being adiabatic.
11.10.6 Redundant and Parallel Fuses

If a load circuit uses a redundant wire pairs for reliability, then each wire
pair must have a separate fuse. Even with single wires, some customers
FIGURE 11.17 Parallel fuses increase the current carrying capacity of fuse assembly.
such as NASA require redundant fuses in most load circuits. If one fuse
fails randomly under normal operation, the redundant fuse can carry a full
load for the remainder of the mission. A redundant fuse in a 5-A circuit
cannot be achieved just by placing two 5-A fuses in parallel, which
essentially derates the fuse by a further factor of one half. Instead, an
additional series resistance equal to several times the fuse resistance is
inserted in series with the redundant fuse as was depicted in Figure 11.10.
Since the current sharing between the redundant fuse branches is inversely
proportional to the branch resistance, the primary fuse normally takes
about 90% of the load current.
Parallel fuses (Figure 11.17) are used to obtain a fuse assembly with
current rating in multiples of the individual fuse rating. Parallel fuses are
not to be confused with the redundant fuses. Parallel fuses are required to
share equal current, and therefore do not have additional series resistance
in one branch. As a matter fact, two parallel fuses must be matched for
equal resistance at operating temperature. Parallel fusing is not feasible
with nonsealed fuses due to a great variability in their resistance at the
operating temperature. One of the nonsealed fuses would always be
stressed beyond its derated value and blow prematurely. Two heretically
sealed fuses with matching resistance, however, can be connected in
parallel to share almost equal current. Even with hermetically sealed fuses,
it is difficult to balance the current with more than two fuses in parallel due
to a possible zipper effect. For example, consider a 20-A fuse assembly
consisting of four 5-A fuses in parallel (Figure 11.17). If one 5-A fuse blows
prematurely, all will blow one after another by taking progressively higher
current. If the 5-A fuses are highly derated to avoid such effect, then the
protection level of the load circuit is compromised. The zipper effect,
however, can be used to avoid a large fuse blow transient voltage by using a
few small fuses in parallel with about 10% of the fuse resistance in series in
each fuse branch. In case of an internal equipment fault, these fuses blow in
zipper sequence, reducing the fault current as they blow progressively
faster in turn. The last fuse blows fast and under smallest current, thus
resulting in less severe voltage transient.
11.10.7 Fuse Testing

Fuses are tested as follows for proper resistance after the installation, but
before powering the assembly. A four-arm resistance bridge and a suitable
voltage source are used for these tests:
A cold test with minimum current to assure minimum self-heating (as

opposed to hot test at full rated current).
The maximum test current is generally 1 mA for fuses rated 1/8 A or
less, 10 mA for fuses rated 1/8 to 1 A, and 100 mA for fuses rated
greater than 1 A. The measurements are compared to the expected
value given in the fuse specification document to verify that the fuse is
within tolerance, as opposed to merely checking the circuit continuity.
When two fuses are installed in parallel, the d.c. resistance of each fuse
must be within 50% of the value specified for the individual fuse. The
test current in this case is chosen based on the fuse assembly resistance.
The design engineer must be consulted if more than two fuses are
installed in parallel for special measuring sequence or other instruc-
tions.
11.10.8 Unfused Loads

Critical loads are not fused for reliability, and some parts of the power
system cannot be practically fused. For example, fuses cannot protect the
main bus or the harness between the bus and the fuse board, or in current
limiting designs. In such circuits, all wires are double insulated between the
lead and return lines and between the lead and the structure. The double
insulation is implemented by placing two physical layers of insulation
between two isolated conductors. An isolated conductor layer is placed
between both the insulation of the double insulated wire to verify the
integrity of each insulation after the assembly. The insulation resistance is
then measured between the intermediate conducting layer and each
insulation. Often, connections via a daisy chain or other method is provided
to test all double insulated points at the spacecraft level.
11.11 Remote Power Controllers

Numerous solid state remote power controllers (RPCs) are used on the ISS
instead of fuses for protection against faults in various items of load
equipment. The RPC has a programmable trip setting, controls the load
power on and off and provides dead-facing. When the RPC is dead-faced, it
grounds both terminals to zero voltage for personnel safety. On the other
hand, when the load is turned off or has tripped, the output terminals may
still be live with voltage. All three functions of the RPC are remotely
programmable. The basic design features of the RPC used in the ISS are as
follows:
Solid state switches in series with a dead-facing relay, which is

magnetically latched. The relay coil voltage can be 28 V or 120 V
Uni-directional or bi-directional power flow

Programmable trip setting in 256 levels (8-bit data word) with 1553
protocol for data interface
Various maximum current ratings, such as 3, 10, 25, and 50 A uni-
directional, and 25 and 130 A bi-directional. Several small RPCs are
packaged together such that each RPC box has mass of about 2 kg and
dimensions of about 15 15 7 cm. There is only one 130-A RPC
available in this size box.
Trip time of 50 to 100 ms with approximately constant I2t characteristics
Control circuit power consumption of about 5 W at 15 V and þ5 V
with quiescent power less than one watt
The on-resistance of RPC varies with rating such that the series voltage
drop at the rated current is in the 0.25 to 0.50 V range in each rating.
11.12 Early Fault Detection

Many faults start as soft high impedance arcing faults and then develop into
hard faults. It is desirable to detect a high impedance fault early, locate it,
and remove it using the redundant paths. This is particularly important in
high-power systems (>100 kW). Because of the high voltage, the fault
current is high in such systems.
The conventional fuse and RPC protect the system from hard faults only.
The high impedance soft faults, on the other hand, can be detected by
monitoring the system voltage. The bus voltage would have high frequency
hash if an arcing were present. A continuous frequency spectrum analysis
of the bus voltage can be useful in an early detection of an arcing fault. Once
detected, the fault can be located and bypassed before it develops into a
hard fault.
Chapter 12
Ancillary Components
12.1 Introduction
The major components of a PV–battery power system are covered in
Chapters 8 to 11. All other components, including the solar array drive,
deployment devices, deployment controller, thermal controller, relays,
battery monitors, current meters, and shielding equipment, are covered in
this chapter.
12.2 Solar Array Drive

The solar array drive and array drive electronics (SAD/ADE) together
provide the capability of rotating solar panels with respect to the spacecraft
body. The operation of the SAD is controlled by decoded uplink commands
from the on-board computer (OBC). It receives a 128-Hz timing clock and
0.25-Hz synchronizing signal from the OBC. In the 3-axis stabilized two-
wing geosynchronous satellite, one axis is always aligned with the local
normal to the sun and another axis along the orbit normal. Two sets of open
loop (clock controlled) solar array drive motors maintain the sun orienta-
tion. A brushless d.c. stepper motor rotates each panel separately. The slip
rings on the SAD shaft provide an interface between the rotating panel and
the fixed Earth-pointing spacecraft body. One SAD controls the north panel
and the other controls the south panel. The two are interchangeable in
design, where the ‘forward’ direction of rotation is separately selected by
external means. Each SAD has only one mechanical assembly, but has a
redundant motor winding and redundant position telemetry potentiometer.
Only one set of the winding is powered at a time. The redundant windings
are fully isolated to prevent failure propagation. A reliability of 0.99 is
typical for both sides combined over a 15-year mission. The mass of each
SAD/ADE assembly is about 5 kg in GEO communications satellites.
The SAD/ADE provides telemetry defining its status and that of the
solar panel. Each unit provides a potentiometer voltage signal which is
directly proportional to the angular position of the panel shaft, ranging
from 0 V at 0 to þ5 V at 360 . A typical SAD uses a four-phase, 16-pole
permanent magnet switched reluctance stepper motor to drive a zero
backlash harmonic drive. Each phase coil resistance is in the 50 to 100
range. A vanadium–cobalt steel stator core and neodymium–iron–boron
319
permanent magnet rotor provide high torque per unit mass. The rotor is
typically on 440C stainless steel ball bearings and titanium case. Each SAD/
ADE draws power from the essential battery bus at 22 V via a latching
relay drive. The peak input power is about 10 W and the average power
about 1 W. Switch-mode power pulses applied to the stator coils at a
suitable frequency drive the motor. The grounding scheme for ADE uses
four separate grounds, one each for the power, signal, and communication
circuits, and one for all equipment chassis.
The ADE controls all three operating modes of the solar array, namely the
normal forward, slew, and the reverse. The SAD is capable of rotation in the
forward and reverse directions at speeds from 0 (stop) to 8 Hz as
commanded by the ADE. A typical stepping motor provides the stepping
rate of 1.8 /step at 0.25 Hz in the normal mode and 1.8 /step at 8 Hz in the
slew mode. The solar panel inertia, which can be in 50,000 to 500,000 lbin2
range, determines the response of the SAD.
The motor is capable of stepping at a rate of 0 to 16 pulses per second,
with the resulting array rotation of about 0.02 per step. It can have a direct
harmonic gear drive that can drive backward and does not need a slip
clutch. The other alternative is to use two orthogonal gears, which require a
slip clutch since it cannot drive backwards. The slip clutch allows slippage
which avoids breakage should there be backlash on the solar panel. The
shaft speed is clutched down by two orders of magnitude through a helicon
gear. This protects the shaft from backlash stresses.
The solar array is initially aligned by ground. The ADE maintains the
alignment in an open loop control system. It takes the time signal from the
spacecraft clock and turns panels to the sun continuously as per the clock
time without further command. The SAD is fairly accurate, but accumulates
small errors, typically a couple of degrees per month. The sun sensor
detects the errors, which are corrected by periodic commands from the
ground or from the spacecraft computer. In some satellites, the solar array is
slewed every 6 AM until the error is zero. The spacecraft computer
commands can be overridden by ground for maneuvering or other reasons.
The control signals for the required rotation rate comes from the TT&C
system, which also selects the rotation direction and one of four speeds —
slow (0.013 mradian/s), normal (1.03 mradian/s), fast (1.047 mradian/s), or
slew (5.236 mradian/s). In some LEO satellites, such as EOS-AM, the ADE
provides control signals and power to the SAD for the commandable drive
rates of ! 2.5% at 0.1% increment and 3! 12% at 0.5% increment, where
! ¼ 1.060 mradian/s. The ADE accepts commands to select the SAD
rotation rate and the direction. The ADE also accepts commands to select
primary or redundant side. In some satellites, the solar array has two rates
of rotation, one steady at one revolution per day and another slew at one
revolution per 45 min.
The slip ring assembly, also known as the power transfer assembly,
transfers power and telemetry signals to the spacecraft body. Each power
Ancillary Components 321
and telemetry lead wire uses a separate slip ring, but some returns may
share a common ring within its derated current capability. The current
collecting slip rings and carbon brushes make the rotary joint ( joint)
between the sun-oriented solar panels and the Earth-oriented spacecraft
body as the spacecraft revolves around the Earth. The slip ring brushes may
be traditional monofilament brushes, or high conductivity fiber brushes
made with graphite or silver fibers for enhanced current collection with low
power loss. The resistance of a 5-A slip ring–brush contact at the operating
temperature at the BOL is about 5 m, which degrades to about 10 m at
the end of life. With EOL current of say 5 A, the power loss is 0.25 W per slip
ring contact, or 0.50 W per slip ring lead and return circuit.
In a typical SAD, there are 20 to 30 slip rings each rated 5 A for power
circuit, and 10 to 20 rings rated less than 1 A for signal circuits. Some units
come with heavier slip rings (Figure 12.1). Slip rings are used such that all
positive rings are on one side and the negative rings on the other, with
possibly one empty ring in the middle. The positive and negative rings are
not placed side by side to avoid the possibility of a short. Advanced slip
FIGURE 12.1 Power transfer assembly of slip rings and drive motor.
FIGURE 12.2 Biaxial solar array drive and power transfer assembly.
rings, such as those made by Litton Polyscientific Corporation, are lighter

and need no lubrication, thus extending the mission life.
Most communication satellites have only one-axis gimbals, the gimbals
driven by the SAD. As a result, the solar array is normal to the sun on
equinox days, but varies from normal by 23.45 on solstice days. The
solstice power generation is reduced, but the battery charging power
requirement is near zero (no eclipses) and the heater load is also less. Some
satellites, such as Iridium, use biaxial gimbals having two degrees of
freedom in rotation (Figure 12.2), one for the daily rotation and the other
for the seasonal rotation. Biaxials are used for both sun pointing and
antenna pointing in satellites. However, their use is limited because of the
complexities introduced in the joints, the mechanisms and the structure.
Accepting the cos reduction in power generation is often preferred over
adding the dual-axis gimbals, at least in GEO.
12.3 Deployment Devices

Three main methods of deploying various spacecraft components are
discussed below.
12.3.1 Electro-Explosive Device

This is the traditional method of deployment, which is also known as the
pyro-technique ordnance. It uses an electro-explosive device (EED), in
which electrical energy is supplied to ignite the squibb of explosive powder.
The resulting force then deploys the component, typically under a spring-
loaded force. The EED is ignited when a certain current deposits a certain
amount of energy in a specified time. For safety reasons, it must withstand a
certain minimum energy without igniting. A typical EED is rated at 1A–1W
no fire, and 4A–4W sure fire.
The EED requires heavy shielding and great care with regard to EMI
pickup. Moreover, the EED explosive is thermal and shock sensitive.
Therefore, installation of sensitive EED explosives is sometimes done at the

launch pad, which costs much more than at the factory. A pyro harness is
routed separately from the power and signal harness to minimize the EMI
concerns for safety. EED deployment technology, although used for several
decades, have some disadvantages, such as:
All spacecraft components must be designed to withstand severe pyro

shock excitation.
High safety related costs in documenting, reporting alerts, and all other
stringent regulations on EEDs must be complied with.
Some manufacturers install EED squibs at the launch pad, while most
install at the factory. In the latter case, the spacecraft is transported as a
class 1 explosive (sensitive to thermal and shock environments), and
this adds expense.
12.3.2 Laser-Initiated Deployment

Figure 12.3 shows the laser-initiated ordnance schematic, where the light
emitted by diode D1 is transmitted to the squib by an optical fiber. The
photon energy deposited in the squib fires the explosive. After the firing, a
sense light is sent by pulsing D2 and sensing the transmission, reflection or
refraction of the sense light. This is to confirm that the squib has been fired.
This technology has been space qualified and flown. It has the following
advantages over EEDs:
It is insensitive to EMI, RFI, ESD, EMP, and cross talks. Hence, it

eliminates accidental firing and related human safety issues. This
reduces the cost not only in flight, but also in the factory, in
transportation, and at the launch pad.
insensitive to harness and explosive placements
FIGURE 12.3 Laser-initiated ordnance schematic.

Requires no shielding, hence the harness mass is significantly reduced.

Lower integration and test cost, since no EMI silence clearance is
required in the factory or on the launch pad.
Insensitive explosive allows factory installation, which costs less.
Higher reliability.
12.3.3 Shape Memory Metal

Recently the shape memory metal deployment technology has been
developed for some satellites, such as EOS-AM and PM. In the working
principle, a certain metal, when heated, regains tthe shape it possessed
before forming into a new shape. In a way, the metal ‘remembers’ its
original shape and returns to it when heated. The heat is applied by a
traditional heater, which is insensitive to EMI. The most prominent benefit
of using the shape memory metal deployment is the savings in safety
related cost. More recently, a new plastic with shape memory has been
developed at the Massachusetts Institute of Technology. It can be shaped as
a string and inserted into a body through a hollow tube. When heated, it
changes into a sheet or other shapes by manipulating the temperature and
stress.
12.4 Deployment Controller

The deployment controller, also called the ordnance controller, delivers a
constant current to the deployment devices. In the traditional pyro bridge
wire, the wire resistance rises with temperature. The required voltage must
therefore rise accordingly to supply the specified minimum current,
typically a few amperes. If the bridge wire is supplied with a constant
voltage source, it must carry the specified minimum current at the peak
temperature around 800 C. The peak current is typically several times the
minimum current specified for 10 to 20 ms. The relay must be capable of
handling the peak current for sufficient number of times without over-
stressing. Figure 12.4 is a typical ordnance controller circuit powered
directly by the battery through the pyro enable plug. The safe and arm
plugs are placed between the ordnance controller and the EED squibs.
Power is supplied via a redundant path to a redundant EED, one from each
battery. When the enable and arm relays are commanded on, the EED is
ready for the fire command. The fire command turns on the EED cavity
relay, sending the pulse of power to the EED squib, which activates the
deployment mechanism that cuts a rope, or a bolt.
The ordnance controller can be powered directly from the battery or from
the bus via PRU (Figure 12.5). For reliability reasons, the deployment power
is generally drawn directly from the battery by unfused circuits. In mid-
FIGURE 12.4 Ordnance controller circuit schematic.
sized communications satellite, each battery is typically capable of

supplying a peak pyro load greater than 10 A for 10 s, 25 A for 5 s, or
50 A for 5 ms with 70 to 100% state of charge. The battery bus impedance is
typically greater than 100 m for d.c. to 10 kHz. Two independent pyro
busses are tapped from the 22nd cell of each battery without cross-
strapping. The voltage ranges from 22 to 35 V with one cell failed. Each pyro
bus — also known as the essential bus — powers the primary and back up
pyro loads, respectively.
FIGURE 12.5 Ordnance controller powering options.

12.5 Thermal Controller

Various probes are required to sense temperature at various locations of the
spacecraft for use with telemetry-thermometers and temperature control-
lers. A probe may contain any one of the following temperature measuring
elements.
12.5.1 Thermocouple
A thermocouple utilizes the Seebeck effect, which occurs at the junction of
two dissimilar metal wires. Thermally energized electrons in each metal
generate a d.c. voltage at the hot junction, which is measured at the cold
end. The magnitude of the voltage is linearly proportional to the
temperature. The measurement error is minimized by keeping the cool
end at a constant temperature and using a voltmeter that has high input
resistance.
12.5.2 Metallic Resistance Wire

In a metallic resistance wire the temperature change is derived from the
change of wire resistance compared to the reference value Ro, using the
relationship iR ¼ RoiT, where ¼ temperature coefficient of resistance
per degree centigrade. Platinum is one element that is often used in space as
it not only has high resistance, but also a high temperature coefficient of
resistance, which is linear and stable over a wide temperature range. For
platinum, ¼ 0.003 per C at 20 C.
12.5.3 Thermistor
A thermistor is a semiconductor in which a small change in temperature
causes a large change in electrical resistance. Precise measurement of the
resistance gives direct reading of the temperature at the thermistor location.
This allows a simple readout instrument and the use of long leads without
loss of accuracy. The temperature coefficient of resistance for the thermistor
is negative and two orders of magnitude greater than metallic wires. The
resistance of a typical thermistor at various temperatures is 75 k, 7 k,
3 k and 1 k at 40 C, 0 C, þ20 C, and þ40 C, respectively. The thermal
time constant ranges from 5 to 15 s. The interchangeability tolerance is
within 0.1%.
The resistance wire and thermistor probes require applying a current and
measuring the voltage across the element, from which the resistance is
derived using Ohm’s law. Since the thermal time constants are relatively
long, the temperature is measured only periodically by passive telemetry to
save energy. A constant current pulse is sent and the voltage is measured
periodically across the element. The heaters are turned on or off as needed
FIGURE 12.6 Proportional heat control circuit.
in response to the temperature telemetry. Instead of using conventional

thermostats for turning the heat completely on or completely off, the
proportional heat control is often used. It employs a transistor in series with
a fixed resistor (Figure 12.6). The control current is changed in response to
the average heat required. The heat generated equals the heat required less
the payload-generated heat. This avoids frequent on–off cycling with the
conventional thermostat where the heater is either on or off. The
proportional heat control provides a steady current with no on–off cycles,
thus eliminates the EMI. It is particularly useful near EMI sensitive
equipments, such as science instruments.
12.6 Relays
An electromechanical relay has physical metal contacts that open or close
on command as required. It offers certain advantage over transistor
switching, such as negligible power loss, and it can be used singly without
auxiliary circuit or with very simple circuit. On the negative side, it has a
limited life due to a moving contact that wears. If arcing is present under
normal or abnormal operation, it results in EMI and a much shorter life.
Relays come in a holding or latching variety. Most space-qualified relays
are rated for 28 V and up to 30 A d.c. Larger current ratings come from
Leach and smaller ratings from Genicom and Teledyne corporations. They
are typically hermetically sealed in nitrogen at 1 atm and life tested for tens
of thousands of cycles. Typical weights of 1, 5, and 10 A d.c. double pole–
double throw (DPDT) 28-V relays are 3, 15, and 45 g, respectively. All relays
for space applications are derated to 50% of the ground base current rating.
Contacts are designed such that the voltage drop at rated current is 0.1 to
0.2 V.
Relay contacts for switching high currents around 50 A are difficult to
procure, are heavy and vulnerable to vibration and shock environments
during the launch phase. In a relay having spare poles, two poles can be
connected in parallel to double the current capacity of the relay. Multiple
pole space-qualified high-voltage relays are also difficult to procure. For
this reason, many engineers find it necessary to use relays at voltages higher
than rated values after derating to a lower current. Since the arcing voltage
FIGURE 12.7 Typical 5-, 10-, and 15-A, 28-V relay derating for d.c. voltages above normal
rating without arc suppression.
on breaking the current is equal to Ldi/dt, the contact ampere rating at

higher than rated voltage decreases linearly in theory. However, manu-
facturers’ test data suggest a square function, i.e.,
2
actual current capacity rated voltage
¼ ð12:1Þ
rated current capacity actual application voltage
The derating data for 5, 10 and 15A latching relay is depicted in Figure
12.7. It is important to avoid arcing by absorbing and dissipating the
inductive energy in the relay coil and the load circuit. Various schemes
shown in Figure 12.8 for this purpose are: (a) steering and suppression
(freewheeling and series) diodes, (b) bifilar coil with one side shorted, (c)
diode and Zener, and (d) series R–C circuit. The relay is often required to
‘‘make before break,’’ or ‘‘not to make before break.’’ A relay failure in
maintaining this sequence can cause damage.
Relay coils interfacing with the command decoder discrete command is
isolated from all circuitry. Connections between the command relay coil in
the user equipment and any primary or secondary returns are prohibited.
Steering and suppression diodes are located in the user equipment (Figure
12.9).
12.7 BPM and BCVM

The battery pressure monitor (BPM) and the battery cell voltage monitor
(BCVM) are part of the PRU. Each BPM strain gage along with its dedicated
amplifier is adjusted for offset and gain to normalize the output. The output
FIGURE 12.8 Relay contact protection schemes for absorbing inductive energy.
is adjusted at the battery vendor to meet the specified offset tolerance (in
microvolts) at room temperature and zero pressure, as well as at room
temperature and 1000 psig. The strain gage voltage is nulled at zero
pressure and is typically 6 mV at high pressure. A reading is also taken at
500 psig for reference. The strain gage signal is brought to the BPM circuit,
which contains a differential amplifier and a local regulator. Each BPM
amplifier is adjusted via select-in-test resistor for offset and gain for
normalizing the 6 mV strain gage signal to the 0 to 5 V level. In operation,
the strain gage offset voltage and cell pressure follow the linear relation P ¼
mVc þ b, where b is the zero pressure offset and m is the slope of the line,
i.e., iP/iV. Having recorded the zero and full pressure parameters, they
can be provided for the look-up and offset as described earlier. Typically, a
couple of cells in each battery have the pressure and temperature telemetry.
FIGURE 12.9 Typical relay load circuit.

12.8 Battery Let-Down Unit

The battery letdown unit (BLU) is a resistor assembly used to drain the
battery completely during storage on ground, for transportation to the
launch pad, or on orbit for reconditioning. For on-orbit let down, two drain
rates are used, high in the beginning and low at the end. The drain
resistance per cell is typically 1 , or 0.5 with a thin high-resistance
harness. Three alternative design configurations of the BLU are shown in
Figure 12.10. The approach (a) is often used. It works well as long as cell
voltages are equal. However, cell voltages often diverge near zero voltage at
the end of draining. The unequal loop currents under this condition cause
unequal voltage drops in the harness, leading to errors in the cell voltage
measurements across the resistors. A measured cell voltage that is positive
could in fact be negative. This problems can be remediated by:
Software compensation for uneven voltage drops in the harness for

real-time corrections.
Dedicated resistor for each cell with its own harness shown in (b). This
adds some harness mass.
Single-resistor design shown in Figure 12.10(c). It saves some harness
mass, but also loses the ability of monitoring each cell voltage and
driving it to zero individually.
12.9 Current Meters

Three types of current meter are used in space: the shunt, the Hall-effect
transducer, and the magnetic amplifier.
12.9.1 Shunt
This sensor is essentially a known precise resistance inserted in series with
the load. The voltage drop measured across it is linearly proportional to the
FIGURE 12.10 Battery letdown resistor network options.

FIGURE 12.11 Hall-effect transducer as a current sensor.
current. It has low cost, but inserts a power loss and a reliability risk of
failing open. Moreover, it offers no electrical isolation, which may be an
issue in high-voltage busses.
12.9.2 Hall-Effect Transducer

It is an electromagnetic transducer that generates voltage when electrical
current and magnetic field are simultaneously imposed on the conducting
transducer material. With reference to the flat plate transducer of Figure
12.11, a current density J applied in the x-direction and a magnetic field B
applied in the z-direction result in an electric field EH in the y-direction with
magnitude EH ¼ R (B J), where R is known as the Hall coefficient. The
voltage measured across y-thickness is calibrated in terms of the current in
the x-direction.
12.9.3 Magnetic Amplifier

Figure 12.12 shows a schematic of the magnetic amplifier current sensor.
The current I through a magnetic core causes the core to saturate, the effect
of which is measured as the increased voltage across the coil. Magnetic
ampliers are often used in measuring the output current of the north and
FIGURE 12.12 Magnetic amplifier current sensor.

south solar array wings separately to provide a health check. In this use, the
amplifier has no mission critical failure mode. A loss of sensor is merely
that — loss of a sensor — with no impact on the bus power. It costs more,
but has no power loss in the transducer itself. However, it needs an
oscillator that requires power. Having no contact with the current path, it
provides electrical isolation and can be inserted as an afterthought.
12.10 Capacitors
Numerous capacitors are used in a typical space power system. In the bus
voltage ripple filter, they store and deliver energy over one cycle. In
snubber circuits, they absorb and dissipate transient voltage energy in solar
circuit, PRU, LPC, etc. The energy storage per unit mass of the capacitor
varies with construction and the voltage withstand capability. However,
typically low voltage (28-V) capacitors have specific energy of 20 J/kg,
medium voltage (1 kV) capacitors 100 J/kg, and high-voltage (20 kV)
capacitors 200 J/kg. The mass density of high voltage capacitors is around
2 kg/l (0.07 lb/in3). Various types of capacitor and their applications are as
follows.
12.10.1 Tantalum Capacitor

The tantalum capacitor provides relatively high-density low-cost energy
storage. The maximum voltage rating of standard production line tantalum
capacitors is 100 V with 50% military derating. Thus, a 50-V bus is the limit
for 100-V capacitor applications. The tantalum capacitor is generally heavy
and comes in two constructions:
The foil (film) type has a high ESR that gives high-power loss.
Naturally, this type finds applications in circuits where damping
resistance is a design requirement.
As a solid sintered powder bonded in one solid porous body. It has
low ESR and finds applications in ripple filters where pure energy
storage is required.
12.10.2 Metalized Polypropylene Capacitor

The metalized polypropylene capacitor is available in higher voltages. A
300-V rated capacitor of this type used on a 120-V bus provides an adequate
margin after 50% derating. However, it is bulky and offers no mass
advantage over the tantalum capacitor.
12.10.3 Ceramic Capacitor

The ceramic capacitor is made of chunks of ceramics with metal foil bonded
on two faces. It can be operated at very high temperature, with the ceramic–
metal bond as the only weak link. It has a low ESR and is compact, but
offers no mass advantage per microfarad. Ceramic capacitors are available
in high voltages. The 160 to 120 V ISS bus has used ceramic capacitors.
12.11 Filters
A filter reduces the conducted EMI entering or exiting the equipment. A
low-pass filter is most commonly used to suppress conducted high-
frequency interference using inductance, capacitance or a combination of
the two, as shown in Figure 12.13. The insertion loss IL of a filter is defined
as the noise voltage on load with the filter to that without the filter. It is
usually expressed in decibels (dB). The capacitor is an efficient low-pass
filter when connected in parallel to the high resistance load. It has IL ¼ 20
log(RC). The inductor is an efficient low-frequency filter when used in
conjunction with a low resistance load. It has IL ¼ 20 log (L/R).
When L or C alone cannot provide the desired insertion loss, the L-C filter
is constructed in L, or T configuration shown in Figure 12.14. These filters
are associated with suppression of signals in the frequency domain, such as
ripple harmonics of the switch mode power converters. They would also
suppress signals in the time-domain, such as transients spike voltages from
ESD, EMP, and lightening type sources. The response of such filter is
obtained by summing the pulse and frequency spectrum curves at each
FIGURE 12.13 Single element filters.

FIGURE 12.14 Basic L-C low pass filters for power circuits.
frequency. The filter is commonly used at the input of a switch mode

converter to reduce the reflected ripple current. The filter is usually built
inside the converter. Very high-frequency noise (hash) is effectively filtered
by using one or more filters in the following configurations:
Feed-trough capacitor
Ferrite bead around the conductor
Ferrite bead with capacitor
12.12 Telemetry and Commands

Sufficient telemetry must be provided to operate the spacecraft in a safe
manner in all phases of the mission. The following is generally provided:
The on-off status of all EPS relays and switches with positive
identification
Load currents
Bus currents and voltages
Array currents and voltages
Shunt voltages and currents with shunt switch status on or off
The open circuit voltage and short circuit current of the solar array test
coupon(s)
Charge and discharge currents and terminal voltage of each battery
Temperature and pressure co-located on two or more cells in each
battery
Individual battery cell voltage as often required by the customer.
A typical power system provides the TT&C system with about 60 analog
and 40 discrete telemetry. The telemetry accuracy is generally 3% of the
full-scale value of all analog signals, and 2 C of temperature. The types of
telemetry used on the spacecraft are listed in Table 12.1.
The bus voltage is controlled automatically, so the telemetry is only for
the information. The EPS telemetry is scanned every several seconds. The
battery current, pressure, and temperature are scanned less frequently
depending on the nature of the load and the battery SOC management
requirement. There is no criticality in the telemetry time intervals. More
frequent is better, but must be traded with the computing requirement.
Most telemetry is for information purpose, except the battery current,
pressure and temperature, in that order of criticality.
The OBC uses redundant telemetry to assure that the batteries are safe in
presence of a single EPS failure and its own failure. Its is not required to
continue automatic control after a failure without ground reconfiguration.
Commands are used to implement various functions and are typically of
three types:
Digital serial
Discrete logic
Discrete relay
Ground commands are provided to override and inhibit autonomous

functions and to manage and control the EPS. Redundant units are
individually commanded. Block-switching of redundant units in response
to a detected failure is prohibited.
The electrical power system in spacecraft like GPS may have 70 to 80
discrete commands, 30 to 35 for EPS and 40 to 45 for pyro; and 35 to 40
serial telemetry points, 20 to 25 in EPS, 10 in SAD, and 2 in pyro. A typical
5000-W communications satellite may have 50 to 100 commands and a
similar number of telemetry signals related with the power system
Table 12.1 Types of telemetry used in spacecraft
Telemetry type Nature Description
Digital serial Serial Software driven

Digital logic Bilevel þ5-V logic pulse or zero
Discrete relay Bilevel þ28-V power pulse or zero
Active analog Analog Voltage and current sensors which are active all the time
Passive analog Analog Temperature sensors, normally inactive, but periodically pulsed
with 1 A. The measured voltage is then converted in
temperature
monitoring functions. Most of them interface with PRU, which controls the
bus voltage.
The EPS command and telemetry interface with OBC via the TT&C
system. However, the EPS is capable of remaining in a safe mode in absence
of TT&C support for several minutes. It is also capable of operating
indefinitely in a mode in which all commands are issued from ground.
12.13 Electronic Packaging

The electronic packaging has multiple functions. It provides an internal
structural support, the thermal conductivity of waste heat, and the
enclosure that also provides protection against EMI and charged particle
radiation. Since about 20 to 30% of the electrical assembly mass is in
packaging, significant mass reduction potential exists by exploring alter-
native packaging materials. Aluminum (0.10 lb/in3) has been the material
of choice for its high strength-to-weight ratio, good electrical and thermal
conductivity, low cost, and easy fabrication. Other materials which have
been used are magnesium (0.06 lb/in3) and magnesium–lithium (0.04 lb/
in3). Although lightweight, magnesium–lithium is expensive, flammable,
and a poor thermal conductor. In recent years, fiber–epoxy composites have
been developed with properties tailored to meet a wide variety of
requirements. Aluminum filled with graphite fibers or carbide particles
may prove to offer reduced mass with acceptable thermal conductivity and
stiffness. A composite fiber enclosure with a thin layer of aluminum foil for
EMI shielding is about 50% lighter than an aluminum-based enclosure.
If a hermetically sealed component is specified, then the following is used
in the construction:
Hermetically sealed connectors with connector cases welded to the

hermetically sealed enclosure.
All circuit boards are conformally coated with an opaque polyurethane-
type compound by dipping, and then properly cured with no cracks
in the finished card. Wide cracks due to mishandling damage or poor
manufacturing is a risk. In a good circuit board, any spark due to a
loose contact or arcing under a fault current would remain inside the
conformal layer.
The circuit boards add significant mass to the electronic assembly. Some of
the mass reduction potential may lie in using lower density Kapton (0.06
lb/in3) versus glass (0.08 lb/in3), and thinner and narrower copper traces
(0.005 versus 0.010 inch). Circuit board fabrication is done on fiberglass
board, copper plated on one side that works as the common return plane.
All return leads are locally soldered to the copper back. A transformer
weighing less than a pound is almost always toroidal, not potted. It is
screwed to the board with a washer of the same outer diameter as the
transformer and 1/8 inch diameter machine screws with no silicon or other
adhesive. A transformer weighing less than 1=4 pound, however, is glued
directly on the board with silicon adhesive tab at the bottom and also at the
top. Aluminum heat sinks are bonded with the base plate to dissipate heat.
Nonmagnetic components may be potted using urethane resin filled with
alumina for thermal conduction. Large magnetic components are potted
using magnesia, which penetrates well. Small magnetics are merely bonded
to the board with alumina filled urethane resin.
Most electrical parts are required to be electrically isolated while
maintaining good thermal contact for heat transfer. The insulators must
be mechanically strong enough to withstand the short circuit forces. Most
commonly used washers as spacers are beryllia (0.10 lb/in3) and alumina
(0.14 lb/in3), 20 to 40 mil thick.
Advances are continuously being made in heat sinks. A detailed look at
what is currently available in the market can prove beneficial.
12.14 Radiation Shield

Energetic particles passing through material lose energy by interaction with
the atomic structure of the subject material. Electrons become highly
scattered by the material, while the protons, being heavier, penetrate deeper
into the material. A particle loses all its energy after penetrating through a
certain thickness of the material. This thickness is known as the range,
which approximately varies linearly with the energy level of the particle
and inversely with the mass density of the material. For this reason, the
range is commonly expressed as the product of the depth and density
(g/cm2) or in units of depth in aluminum as the basis of comparison as
shown in Figure 12.15. A practical range, Rp, is defined as the most probable
range of particle of a given incident energy.
Normal space-qualified devices are generally used in commercial space
applications. These devices generally have sufficient shielding against the
natural radiation of charged particles. This is because the device is in its
own case, then often behind a card, and also in an overall metal enclosure.
A more sensitive electronic device is protected by placing it in a heavy
metal enclosure, such as cast aluminum of several millimeters thickness.
Under a man-made radiation threat, such as a nuclear detonation around a
defense spacecraft, the device may need additional radiation hardening.
Such radiation hardened semiconductor devices are expensive.
The metal enclosure further reduces damaging effects of the charged
particles to a level that is tolerable by the built-in radiation hardness of the
electronic part inside. Selection of the enclosure metal and its thickness is a
function of estimated fluence and the tolerance dose. For example, the
radiation fluence entering the enclosure decreases with increasing product
FIGURE 12.15 Range of radiation damage in aluminum.
of the wall thickness and mass density. Since aluminum is the most widely
used enclosure material, the fluence penetration versus thickness curve is
generally plotted taking aluminum as the reference (Figure 12.16). The dose
absorbed by an electronic part inside the enclosure therefore depends on
the wall thickness. For example, the total ionization dose comprising mainly
the electrons and solar flare protons versus equivalent aluminum thickness
in three different orbits is shown in Figure 12.17. Such curves are used to
determine the required enclosure wall thickness as described below.
FIGURE 12.16 Radiation fluence penetration over 10-year mission versus shield thickness.
FIGURE 12.17 Total radiation dose over mission life versus shield thickness for three orbits.
12.14.1 Example of Radiation Shield Design

Referring to Figure 12.18 as the charged particle radiation dose versus
enclosure wall thickness curve, say we wish to protect a piece part having a
tolerance for a dose of 2 104 rads of ionizing radiation. The part has its
own case that is equivalent to 1.5 mm of aluminum, which will result in a
radiation dose 1 105 rads. In order to reduce the dose by a factor of 10, i.e.,
bring it down to 1 104, we must have a total shield thickness of 4 mm.
FIGURE 12.18 Example of shield thickness determination for protection against charged
particle radiation.
This requires a 2.5-mm-thick aluminum enclosure on the top of its own

1.5 mm case.
The following are general guidelines in shielding the electronic devices
against radiation.
Shield the device individually if only a few devices require shielding.

Place a shield only on one side if major dose of the radiation comes
from one direction. Flat and deep packages are shielded with strips on
top and bottom.
Provide a shielding enclosure if a large number of devices require
shielding.
May use thin tantalum shield using the density equivalency of 1 mil
tantalum ¼ 5 mils aluminum.
Investigate the common mode noise under solar flare radiation.
The enclosure of electrical component works as a dual-function shield

against the charged particles radiation as well as the EMI radiation. Its
design as the radiation shield is discussed above, whereas the EMI shield is
discussed below.
12.15 EMI Shield

The conducting metal enclosure shields the equipment inside from radiated
EMI. The metal walls provide eddy current shielding and control the near
and far field coupling of the radiated electromagnetic fields. It attenuates
the electrical and magnetic fields from leaving or entering the enclosure.
Performance of the shield can be viewed in terms of the transmission line
analogy as shown in Figure 12.19.
When an electromagnetic wave traveling along a transmission path in
free space with the characteristic wave impedance Zo impinges on a metal
surface (load impedance ZL), the surface absorbs some wave power, some
reflect back and some continues beyond the shield wall thickness and
reaches the victim equipment. The reflection coefficient, <, is given by
ZL Zo
<¼ ð12:1Þ
ZL þ Zo
The higher the value of <, the more wave energy is reflected from the shield
surface. For total reflection, i.e., for < ¼ 1, ZL must be small compared
with Zo. On the other hand, ZL ¼ Zo gives < ¼ 0, there is no reflection and all
the wave energy is absorbed by the shield wall thickness. This condition is
known as the impedance matching. The far field transverse electromagnetic
wave in free space has Zo ¼ 120 ¼ 377 . The electric dipoles have
FIGURE 12.19 Reflected, absorbed, and transmitted EMI energy.
dominant e-field and higher Zo. The current loops have dominant B-field
and lower Zo. The load impedance of a metallic surface is given by ZL ¼
1=
(2f ) 2 where f, and are frequency, relative permeability, and
resistivity of the surface material, respectively. Their values for commonly
used enclosure metals are listed in Table 12.2. With these values, all metals
except stainless steel have values of ZL much lower than 377 . Therefore,
they all provide good shielding from high impedance sources (dipole
sources), but provide poor shielding from low impedance sources (loop
currents).
The reflection loss is defined as the ratio in dB of the wave field intensity
passing through the shield to the field impinging upon it. For aluminum
and steel, the two most widely used metals for enclosures, the reflection loss
at various frequencies is listed in Table 12.3.
Once the outside interference enters the shield box, a part is absorbed in
the metal wall as a resistive loss, and the remainder propagates further. The
I2R loss absorbed in the wall thickness depends on frequency and the skin
depth of the wall, and is given by
¼ ½ =ðfÞ0:5 ð12:2Þ
Table 12.2 The relative permeability and

resistivity of commonly used shield metals
Relative Relative
Material resistivity permeability
Copper 1 1
Aluminum 1.6 1
Mild steel 10 100–1000
Stainless steel (no-mag) 50 1
Table 12.3 Reflection loss in aluminum and steel at various

frequencies
Frequency E-field loss (dB) B-field loss (dB) Plane wave loss (dB)
Alum Steel Alum Steel Alum Steel

60 Hz 280 240 20 0 150 115
1 kHz 240 200 30 10 140 100
150 kHz 180 130 50 20 115 80
15 MHz 115 80 75 40 95 70
If the wall thickness is five times the skin depth, all the energy entering the
metal wall is absorbed. For other thickness, the absorption loss, A (in dB), is
given by
A ¼ 0:134tðfr = r Þ0:5 ð12:3Þ
where t ¼ wall thickness in mm, r ¼ resistivity relative to copper, and r ¼

permeability of the box interior relative to free space. The relation shows
what is intuitively expected, namely the greater the wall thickness, the
greater the power absorption. The absorption loss values for aluminum and
steel enclosures at various frequencies are given in Table 12.4.
Table 12.3 and Table 12.4 show that the reflection and absorption of the
energy depends inversely on the resistivity. The permeability, on the other
hand, has a beneficial effect on absorption, but a detrimental effect on
reflection. This property can be used in low-frequency magnetic shielding.
The reflection and absorption losses in dB are added to obtain the total loss
in dB. The table values are theoretical for a perfect shield. In reality, the total
dB loss is lower by a few fold due to seams and ports for connector and
displays. However, most practical boxes provide adequate shielding except
for very low frequency magnetic fields, as long as they are metallic and
thick enough to support the parts inside.
Table 12.4 Absorption loss in aluminum and steel

at various frequencies
Frequency Absorption loss in Absorption loss in

(Hz) aluminum (dB/mm) mild steel (dB/mm)
60 0.8 13
1000 3.2 56
150 k 40 680
15 M 400 4240
FIGURE 12.20 B-field of current loop entering metal shield.
The overall shielding effectiveness (SE) is often measured by testing the

noise signal amplitude A with and without the shield, and using the
expression
Awith shield
SE ¼ 20 log ð12:4Þ
Awithout shield
Enclosures made of solid aluminum or aluminum backed composite

laminates provide 50 to 100 dB shielding effectiveness for high frequency
noise, in megahertz. The metallic backing can be less than 0.1 mm in most
applications. Other conductive materials may be used, such as metal mesh
and conductive fabrics. The shield does not have to be grounded for the
FIGURE 12.21 EMI shielding effectiveness of various materials.

shield effectiveness. It is grounded for safety reasons in case of a fault to the

chassis and for protection against electrostatic buildup.
If a solid plate is used to shield a component by eddy currents (Figure
12.20), not only is it required to have t > , but also Rt >> 2 for good low-
frequency shielding against radiation from small current loops. Here t ¼
shield thickness, ¼ resistivity of the shield material, and ¼ skin depth for
the signal frequency to be filtered. Only magnetic metal can provide low
frequency shielding, although it is rarely required. However, when needed,
high permeability magnetic metal tape having the required saturation flux
density is considered. The tape loses permeability after bending and
straightening. Therefore, it often needs laser annealing after the wrapping
process. Shielding effectiveness of various metal meshes and honeycomb
are shown in Figure 12.21.
Part C
Power System Performance
Chapter 13
Energy Balance and Power Management
13.1 Introduction
When we speak of designing the spacecraft electrical power system, we
really mean designing the electrical energy system. The satellite has a
limited time in orbit to generate power, but the loads need to be powered all
the time. The battery stores the required energy during sunlight and
delivers to the loads during eclipse. The energy balance between the battery
charge and discharge over one orbit period must be, on average, positive
with some margin. In a simple case of constant load shown in Figure 13.1,
the lightly dotted area must exceed the dark area with an acceptable
margin. Otherwise, the battery would become totally depleted after some
time. The power to and from various components must therefore be
managed in order to maintain the energy balance in both the transfer orbit
and the operational orbit.
In LEO satellites, the orbit period is short, about 1.5 h. The eclipse
duration, although short in clock time, is much longer as a percentage of the
orbit period. Consequently, the sunlight duration as a percentage of the
orbit period is much shorter, imposing a stringent demand on the power
system. It must not only collect the required solar energy in a relatively
short time, but also manage it efficiently over the entire orbit period. The
energy balance over the orbit period is maintained by a combination of
hardware and software control as described in this chapter. It amounts to
maintaining the following integrals equation:
Z
ðsolar flux PV conversion efficiencyÞ dt ¼
sunlight
Z
ðloads þ charge power þ shunt power þ lossesÞ dt ð13:1Þ
sunlight
Z
þ ðloads þ lossesÞ dt
eclipse
In most satellites, the energy balance is maintained autonomously on

board. However, the ground control intervenes when it is necessary to (a)
347
FIGURE 13.1 Energy balance over one orbit period with constant load power.
override the autonomous controls for any reason, (b) trim the system to
account for predicted or unpredicted degradation in the performance, or (c)
recondition the battery, if required.
The battery telemetry available — all with redundancy — for this
purpose are:
Two cell temperatures per battery, one at the hottest spot and one at the
coldest spot
Two cell pressure readings per battery, temperature compensated
All cell voltages for comparative performance analysis
Each battery voltage and charge/discharge current
The charge rate is selected between the full charge and trickle charge
rates as controlled by the PRU mode controller to maintain the bus voltage.
Typical charge rates in GEO satellites are:
C/120 for trickle charge during solstice periods (no-eclipse seasons)

C/60 for trickle charge following eclipses around equinox to maintain
high state of charge, also for recharge during solstice after N–S station
keeping
C/20 normal charge rate
C/10 maximum charge rate to charge the batteries sequentially for high
efficiency if the power budget requires so
During the transfer orbit, continuos ground control is not possible and
the on-board computer (OBC) automatically controls by the battery state of
charge (SOC). The OBC can adjust all charge rates from default values and
can inhibit automatic functions. At launch, the battery charger’s minimum
rate is set at C/120 for charge maintenance, and the maximum rate is set at
C/10 for sequential charging for high efficiency up to 85% SOC.
Energy Balance and Power Management 349
In operational orbit, normally the batteries are charged concurrently at

C/20 when they are under the autonomous control of the OBC. The ground
control has the option of charging them sequentially at C/10 by program-
ming the OBC. Charging at C/10 is more efficient up to a certain SOC,
beyond which it is switched to C/20.
Using the telemetry data, the OBC estimates the battery SOC using the
battery cell pressure telemetry compensated for the telemetered tempera-
ture value. Using the hysteric dead-band control, the OBC initiate charge if
the cell pressure falls below the preset value selected for charge initiation
less the hysteresis level of about 2 to 3%.
During charging, the OBC compares the cell pressures against the
predetermined state, 85% being a typical default value. The OBC terminates
the charge when the SOC rises above a preset value, and commands the
chargers to one of the two trickle charge rates pre-selectable by ground
command. The ground command also has an option to preset an additional
amount of charge to reach full charge without excessive temperature and
pressure rise, followed by automatic charge termination. All preset values
in the battery charge and discharge management are alterable by the
ground control.
13.2 Energy Balance Analysis

The energy balance analysis is performed at the design stage by simulating
the power flow and energy account on the computer. The program is
generally structured to allow analyses on the baseline design and its
derivatives, and to answer many what if questions in normal and abnormal
operation.
The program basically determines whether the EPS is in energy balance
over a desired period, e.g., over every orbit or over several orbits, for a
given load profile, solar array and the battery. Such analysis is performed
during all phases of the mission for a given launch date (year, month, and
day). Various fault conditions, including battery cell failures (short or open)
and loss of a solar array circuit, are simulated to determine energy balance
under the worst-case condition(s).
Other equally important uses of the energy balance computer program
are:
Determine and/or optimize the load capability of a given EPS

Derive the system component ratings based on the maximum power
flow through each component
Determine the power dissipation for thermal design of each component,
particularly the battery, since its performance is highly temperature
sensitive
13.3 Computer Program Structure

The computer program is generally developed around variable parameters
with no hard-coded numbers. Such a structure allows greater flexibility in
using this tool for a wide variety of applications. Figure 13.2 depicts EPS
components contributing to the energy balance program. The basic
equation of currents in the energy balance analysis is:
during sunlight,
Isa ¼ Iload þ Icharge þ Ishunt Idischarge ð13:2Þ
during eclipse,
Iload ¼ Idischarge ð13:3Þ
The battery DOD at any time is computed typically every second or so as

follows:
DODðtÞ ¼ DODinitial þ ðAHdelivered =AHactual Þ ð13:4Þ
where DODinitial is the initial DOD, AHdelivered is the sum of AH delivered,

and AHactual is the actual AH capacity. The entire program is divided into
several software modules, each representing various components. Due to
the nonlinear nature of the battery cell and the solar array performance
parameters, the program uses static lookup tables to determine the cell
performance characteristics as a function of the battery current, tempera-
ture, state of charge and the solar array operating voltage. Programming
with a computer language, instead of modeling on a spreadsheet,
significantly improves the capability of the program.
Major software modules of the programs can be structured as follows.
13.3.1 Battery Module

The battery_module determines a number of battery parameters, including
the state of charge (battery_soc), battery voltage (battery_volt) and battery
FIGURE 13.2 Power system components participating in the energy balance.

dissipation (battery_disp). The ampere-hour (Ah) integration is performed to

predict the battery SOC in terms of actual capacity. Ah returned to the
battery and the Ah removed from the battery are calculated at each
program integration step which is typically in seconds. The battery C/D ratio
is determined by the ratio of the ampere-hours put into the battery to the
ampere-hours taken out of the battery. The battery voltage is determined
using a lookup table, which provides the battery cell voltage as a function of
the charge or discharge rate (þ for charge and for discharge), battery
temperature, and the battery SOC.
The battery dissipation is determined as a function of the difference
between the cell voltage and the thermal neutral potential (therm_neut_pot)
of the battery cell, the battery current, and the battery charge/discharge
efficiency. The thermal neutral potential for the NiH2 battery cell is 1.51 V.
The battery cell efficiency during discharge can be assumed 100%. The
battery efficiency during the charge portion of the orbit varies during
charge as a function of the battery charge rate, state of charge and the
temperature. It is determined using a static lookup table and performing
linear interpolation between points.
The following features are desirable in the battery module:
Variable charge management, V–T and/or V–P limiting, and full, cut
back, and trickle charge rates
Variable maximum allowable charge/discharge currents
Variable battery configuration (number of cells, number of open/short
cells)
Variable battery temperature
Tracking the performance of each battery individually
Dissipation in individual batteries and cell bypass diodes
Dissipation in battery charge and discharge converters
Variable bus voltage regulation profile
Constant or time varying efficiency from test data
13.3.1.1 V–T Limit for overcharge control

Controlling the battery overcharge is extremely important for maintaining
the battery temperature, which has a strong influence on the battery life.
The battery when fully charged reaches the maximum voltage beyond
which all input power is converted into heat. For every electrochemistry,
there is a definite relationship between the voltage and temperature (V–T)
at the end of charging. One can use that relation to determine that the
battery has been fully charged, and then to cut back to the trickle charge rate
equal to the self-discharge rate. The V–T relation varies with age of the
battery. The fully charged voltage is higher when the battery is new, and
then slowly degrades with age. For each battery, several V–T curves, each
representing an age period, is stored in the computer. For example, Figure
13.3 shows six curves for a mission divided into six periods. The top six
curves are for healthy battery with all cells working, and the bottom six are
with one cell failed and bypassed by the diode.
13.3.1.2 Charge cut-back and trickle charge

When the full charge is approached, the battery charge current is often
tapered down to the trickle rate linearly or in a few steps. The battery
current and voltage profile over one orbit period shown in Figure 13.4 is
typical for LEO satellites. A constant charge rate is maintained up to 95%
SOC, typically in 50 to 60% of the sunlight duration. The charge rate is then
tapered off to trickle charge in the C/100 to C/50 range.
Figure 13.4 is an example of the battery current analysis results in a LEO
satellite. It starts at the beginning of eclipse. During eclipse, the discharge
current rises (becomes more negative) as the battery voltage drops because
of the constant power nature of the load. As the battery is placed in charge
at the beginning of sunlight, the battery voltage rises under constant charge
current. About half-way in the sunlight, the charge rate starts tapering and
is reduced to the trickle rate at the end to maintain full voltage until
inception of the next eclipse.
FIGURE 13.3 V–T limits for overcharge control in a 22-cell battery.

FIGURE 13.4 Battery current and voltage versus time for energy balance over one orbit
period.
13.3.1.3 AH counter and sensor

The battery voltage is not a reliable indicator of the SOC. In the past, the
battery SOC had been monitored and managed by analog controls using
various sensors on-board. Many present-day satellites employ software
instead, which is a mass-efficient approach. The software keeps the account
of the ampere-hours in and out of the battery to determine the SOC and
then proceed accordingly. The accumulated error can be detected by
temperature and/or pressure sensors, and the ultimate safety of the battery
is maintained by a suitable hardware circuit. One such self-explanatory
computational flow chart often used in communication satellites is depicted
in Figure 13.5 as an example.
In recent developments, monitoring the battery SOC using fuzzy logic
analysis has been tested with coulomb counting by digital software and
with the battery impedance measurement by analog hardware. The
coulomb counting alone does not allow variations in the internal battery
characteristics to be monitored; hence the advance warning of premature
failure of the battery is not possible with this technique. The SOC
monitoring with impedance measurement alone is not useful in practice
either, because of the complexity of instrumentation and analytical
techniques required to derive the equivalent circuit parameters of the
cell. However, the combination of the fuzzy logic data analysis and the
internal impedance technique for determining the battery SOC has a
potential of providing a powerful approach for determining the battery
SOC.
FIGURE 13.5 Logic flow chart for energy balance software.
13.3.2 Solar Array Module

The sa_module is programmed to incorporate the following:
Solar cell I–V characteristics

Array configuration (number of series cell number of parallel
circuits)
Cell and array performance degradation factors
Array temperature
13.3.3 Bus Module

The bus_module sums all currents available at the bus. They are the
discharge current, charge current, solar array current, load current, and the
shunt current, as shown in Figure 13.2. The bus voltage is determined by
using the net current as a derivative offset to the previous bus_volt. The finite
difference equation used to calculate the bus voltage at the kth time step in
the numerical iteration process is:
VbusðkÞ ¼ Vbusðk1Þ þ ½ðIsaa þ Idischarge Icharge Ishunt Iload Þk t=Cbus capacitors

ð13:5Þ
where it is the time interval.

The bus voltage regulator typically maintains the bus voltage as the
source and the load currents vary to maintain the dc equilibrium of the two
currents. A bus capacitor bank is generally used to absorb the leading and
trailing edge transients of fast changing loads or sources.
13.3.4 Input Variables

It is desirable to have the energy balance program flexible in incorporating
the following system inputs in various simulation runs:
Time variable load, plus a constant load, over one or multiple orbits
BOL or EOL conditions
Off-pointing of the solar array
Launch and descent phase of the mission
13.3.5 Output Format

Several output files are produced on running the program, such as the
battery parameters (battery_out), EPS performance parameters (eps_out),
BPC parameters (bpc_out) and solar array parameters (sa_out). The output
files contain variable names and values and may be loaded into a
spreadsheet program for subsequent plotting or a word-processing
program for inclusion in reports.
The output files print out numerous performance parameters at selected
print interval over the study period. For example, the battery_out file prints
out all the following parameters:
Orbit Sequential number of the orbit being printed

Minutes Time from the beginning of the study period
Vbat1 and Vbat2 Voltage across Battery 1 and Battery 2
Ibat1 and lbat2 Currents through Battery 1 and Battery 2

DOD1 and DOD2 Depth of Discharge in Battery 1 and Battery 2
Qbat1 and Qbat2 Dissipation in Battery 1 and Battery 2
Qcelll and Qcell2 Dissipation in cells in Battery 1 and Battery 2
Qopen1 and Qopen2 Dissipation in cell-open bypass diode in Bat1 and 2
Vopen1 and Vopen2 Voltage across cell-open bypass diode in Bat1 and 2
bat1 and bat2 Charge efficiency of Battery 1 and Battery 2, (from
lookup tables representing measured data)
AHout1 and AHout2 Ampere-hour going out of Battery 1 and Battery 2
AHin1 and AHin2 Ampere-hour going in Battery 1 and Battery 2
Table 13.1 is an example of the battery_out file. It prints key battery

performance parameters such as the battery voltage, current, depth of
discharge, power dissipations in the whole battery, in individual cells and
in cell bypass diodes of Battery 1 and Battery 2 in every orbit every minute
or two.
13.3.6 Program Verification

Before using any such energy balance program, it must be calibrated and
verified using a simple load profile for which the performance can also be
calculated by hand. Figure 13.6 is an example of the computed battery
DODs compared with the hand-calculated values. They must compare well
within the limitation of hand calculations. As seen in the figure, the
assumption of constant battery charge efficiency in hand calculation results
in slightly different DOD as the battery approaches zero DOD.
13.4 Energy Balance Simulation Runs

On run command, the program initializes all input parameters, including
the battery lookup tables and the solar cell lookup tables by calling the
init_module. In this module, the simulation start and end times and the
number of orbits (single or multiple) are specified. The BDR/BCR
efficiencies and the harness resistances are entered in this module. The
print interval is also specified, allowing data to be printed to an ASCII file in
the designated increment (say, 1 min).
The load_module determines the load current requirement at the bus
voltage according to the load power requirement specified in the input file.
The load power can be either constant or can vary as a function of time with
a time increment of 0.1 to 1 min.
Static lookup data files (battery cell data and solar cell I–V data for
example) are located in separate files.
Table 13.1 Typical output format of the energy balance computer simulation
(partial list)
Orbit Minutes Vbat1 Ibat1 DOD1 Qbat1 Qcell1 Qopen1 Vopen1 nbat1 AHout1 AHin1
1 2.0 23.89 13.81 0.0082 24.64 1.449 0.000 0.000 1.00 0.466 0.000
1 4.0 23.73 13.90 0.0163 26.97 1.587 0.000 0.000 1.00 9.28 0.000
1 6.0 23.58 13.98 0.0244 29.27 1.722 0.000 0.000 1.00 1.392 0.000
1 8.0 23.43 14.07 0.0326 31.59 1.858 0.000 0.000 1.00 1.860 0.000
1 10.0 23.27 14.16 0.0409 33.94 1.996 0.000 0.000 1.00 2.331 0.000
1 12.0 23.12 14.25 0.0492 36.33 2.137 0.000 0.000 1.00 2.804 0.000
1 14.0 22.98 14.34 0.0576 38.55 2.267 0.000 0.000 1.00 3.281 0.000
1 16.0 22.87 14.41 0.0660 40.41 2.377 0.000 0.000 1.00 3.760 0.000
1 18.0 22.76 14.48 0.0744 42.18 2.481 0.000 0.000 1.00 4.242 0.000
1 20.0 22.67 14.53 0.0829 43.53 2.561 0.000 0.000 1.00 4.725 0.000
1 22.0 22.59 14.58 0.0914 44.89 2.641 0.000 0.000 1.00 5.210 0.000
1 24.0 22.51 14.63 0.0999 46.18 2.716 0.000 0.000 1.00 5.697 0.000
1 26.0 22.44 14.68 0.1085 47.47 2.792 0.000 0.000 1.00 6.186 0.000
1 28.0 22.40 14.70 0.1171 48.08 2.828 0.000 0.000 1.00 6.675 0.000
1 30.0 22.37 14.72 0.1257 48.64 2.861 0.000 0.000 1.00 7.166 0.000
1 32.0 22.34 14.74 0.1343 49.14 2.891 0.000 0.000 1.00 7.657 0.000
1 34.0 24.73 12.00 0.1391 13.36 0.786 0.000 0.000 0.92 8.012 0.088
1 36.0 24.78 12.00 0.1327 16.46 0.968 0.000 0.000 0.91 8.012 0.488
1 38.0 24.83 12.00 0.1264 19.53 1.149 0.000 0.000 0.90 8.012 0.888
1 40.0 24.85 12.00 0.1200 22.33 1.314 0.000 0.000 0.90 8.012 1.288
1 42.0 24.88 12.00 0.1138 25.25 1.485 0.000 0.000 0.89 8.012 1.688
1 44.0 24.90 12.00 0.1076 28.26 1.662 0.000 0.000 0.88 8.012 2.088
1 46.0 24.93 12.00 0.1015 31.26 1.839 0.000 0.000 0.87 8.012 2.488
1 48.0 24.95 12.00 0.0954 34.37 2.022 0.000 0.000 0.86 8.012 2.888
1 50.0 24.98 12.00 0.0894 37.84 2.226 0.000 0.000 0.85 8.012 3.288
1 52.0 25.01 12.00 0.0835 41.34 2.432 0.000 0.000 0.84 8.012 3.688
1 54.0 25.05 12.00 0.0776 44.80 2.635 0.000 0.000 0.83 8.012 4.088
1 56.0 25.08 12.00 0.0718 49.18 2.893 0.000 0.000 0.82 8.012 4.488
1 58.0 25.11 12.00 0.0661 54.2 3.194 0.000 0.000 0.80 8.012 4.888
1 60.0 25.15 12.00 0.0605 59.31 3.489 0.000 0.000 0.79 8.012 5.288
1 62.0 25.18 12.00 0.0551 66.35 3.903 0.000 0.000 0.77 8.012 5.688
1 64.0 25.20 12.00 0.0498 74.89 4.405 0.000 0.000 0.74 8.012 6.088
1 66.0 25.21 12.00 0.0447 83.86 4.933 0.000 0.000 0.71 8.012 6.488
1 68.0 25.23 12.00 0.0398 92.37 5.433 0.000 0.000 0.68 8.012 6.888
1 70.0 25.25 12.00 0.0351 100.49 5.911 0.000 0.000 0.66 8.012 7.288
1 72.0 25.29 12.00 0.0306 107.20 6.306 0.000 0.000 0.64 8.012 7.688
1 74.0 25.30 10.98 0.0264 115.41 6.789 0.000 0.000 0.58 8.012 8.075
1 76.0 25.30 0.85 0.0233 17.12 1.007 0.000 0.000 0.20 8.012 8.413
1 78.0 25.30 0.85 0.0232 17.15 1.009 0.000 0.000 0.20 8.012 8.442
1 80.0 25.30 0.85 0.0231 17.17 1.010 0.000 0.000 0.20 8.012 8.470
After editing input file(s) to reflect the case to be simulated, the energy
balance program is recompiled using the new input file(s). The new
executable code is then run. The simulation is run for the following three
distinct phases of the mission.
13.4.1 On-Orbit Simulation

On-orbit power flow analyses and energy balance is carried out for the
worst case at the beginning of life and also at the end of life. Since the BOL
FIGURE 13.6 Calibration example for battery DOD 1 and DOD 2. (–––– Computed, Hand
calculated.)
power generation is much greater than that at the EOL, the BOL power
analysis establishes the current rating of the sliprings and the shunts, both
of which always carry the greatest currents in the beginning of life. The end
of life energy balance, on the other hand, must be verified using the worst-
case orbit conditions, namely the yearly minimum solar flux, minimum
operation altitude and the maximum eclipse time. If applicable, the solar
angle of 0 usually results in the longest eclipse time.
Besides being a design tool, the energy balance analysis is also carried out
on board the spacecraft to detect and correct potential problems. This is
done by performing the following forward computations.
For a given load power profile the spacecraft is about to enter, the battery
DOD is projected at the end of the eclipse. If the projected DOD exceeds the
maximum allowed, an alert is sent to the spacecraft computer for an
approaching battery limited conditions and for a possible load shedding.
For the allowable DOD and the required charge power during the next
sunlight duration, the maximum allowable load is computed. If it is less
than the planned load, an alert is sent to the spacecraft computer for an
approaching solar array limited condition and for a possible load shedding.
13.4.2 Transfer Orbit Simulation

This phase of the mission in communication satellites begins with
separation of the spacecraft from the launch vehicle (approximately 30
min after lift-off) and ends with the spacecraft in full 24-h orbit around the
Earth. The first orbit after separation is nominally a 10.5-h elliptical orbit
having perigee at the spacecraft–launch vehicle separation altitude and
apogee at the geosynchronous altitude. During a transfer orbit period
spanning 1 to 2 weeks, a series of timed apogee engine burn raises perigee

to the geosynchronous height. During this period, the solar panels are
stowed against the north and south equipment panels, with only the outer
panel of each wing exposed. The GEO spacecraft is slowly spun around roll
or yaw axis typically at 0.1 rpm to maintain acceptable component
temperature. Because the solar array panels are folded against two opposite
sides of the spacecraft body which is spinning, the power output
approximates a full-wave rectified sinewave. Consequently, the batteries
are alternately charged and discharged at twice the spin rate (Figure 13.7).
The selection of a high charge rate permits use of all available solar array
power more efficiently.
The following are three basic spacecraft orientations in transfer orbit
during which the solar panels of communication satellite are typically
subjected. They experience just about every conceivable level of sun
illumination as follows.
13.4.2.1 Roll spin

At times during the transfer orbit the spacecraft spins around the roll axis
(east–west axis). The north and south outboard solar panels are alternately
illuminated. The sun angle (the angle between the spin axis and the sun-
pointing vector) is anywhere between 50 and 130 . Shadowing of solar
panels during this phase does not occur. Each outboard panel sees a half-
wave rectified sun input profile with the peak ranging from full sun to cos
40 .
13.4.2.2 Yaw spin

At times the spacecraft spins around the yaw axis (nadir–-antinadir axis).
As with the roll spin mode, the north and south outboard solar panels are
alternately illuminated. For most of the orbit, the sun angle is held within 85
FIGURE 13.7 Transfer orbit energy balance with full-wave rectified sine wave variation in
power generation.
to 95 . Each outboard solar panel sees full sunlight once per spin. During
apogee engine firings, however, the sun angle range increases to 60 to 120 .
Peak sunlight during rotation can be between full sun and cos 30 . In this
orientation, the communications antenna reflectors can create shadows on
the solar panels at spacecraft spin angles of 40 from the sun-normal.
13.4.2.3 Initial attitude fix

The spacecraft initiates an Earth search and lock maneuver approximately
1 h after separation from the launch vehicle. The spacecraft is spun slowly
until the Earth sensors find the Earth. Once the Earth lock is achieved, the
spacecraft remains in an inertially fixed attitude for up to 10 min. During
this phase, each solar panel can see full sun, no sun, or anything in between.
If one of the solar panels is pointed directly at the sun, the solar cell
temperature may exceed 115 C.
13.4.2.4 Maneuvers to improve the energy balance

In case the transfer orbit energy balance with any one of the above three sun
orientations is in deficit, the following changes in the operating maneuvers
can improve the energy balance:
Install more efficient soar cells on the outer panels. Such cells can be
selected from the given lot by cell selection without added cost, or by
procuring a class of more efficient cells at added cost.
Partially deploy the solar wing in transfer orbit. The risk with this
approach is a plume impingement by the control thrusters and some
propellant penalty.
Make outboard panel larger than the inboard panels. This may make
the packaging difficult and add some cost.
Improve the sun angle. For example, GPS II spacecraft spins at 55 rpm
in the transfer orbit, and the worst-case sun angle is 16 . This can be
improved by 10 by expending half a kilogram of fuel (one time cost).
This cost may be recovered at the end of the transfer orbit when the
spacecraft needs to move to apogee kick motor angle, which may be,
say, 90 . Thus, not all fuel is wasted in orienting the spacecraft in
transfer orbit for a better sun angle.
Power-down the transfer orbit operations as much as possible. Ask
whether we need all that heater power? Can we turn some loads off at
some times during the transfer orbit? Is the power usage by all other
equipment reasonable? Can they be reduced by better design or smart
utilization? Is it possible to design them with a special low-dissipation
pre-operational mode without adding much cost? Such questions may
lead to a solution or a partial relief.
Use smart energy management and operational work-around. For

example, pausing the spin for a while when the sun angle is favorable
to allow the outer panel collect more energy as shown in Figure 13.8 can
help the transfer orbit energy balance. The dwell time in such a
maneuver is determined by the allowable temperature limit on the
panel or inside the body, which can become roasted under a long dwell
time. Such stop-and-go operation, however, consumes additional fuel.
The mass of this fuel is charged to the power system.
13.4.3 Transfer Orbit Load Budget

Since only one panel is facing the sun at a time as the spacecraft spins in the
transfer orbit, the power output of the array is a full wave rectified
sinewave as was shown in Figure 13.7. The power generation varies with
sun angle as
P ¼ Pmax sin ð13:6Þ
where
Pbol
Pmax ¼ f ð13:7Þ
Np cs
and ¼ sun angle measured from the horizon (90 gives maximum power),
Np ¼ number of panels on both wings, and fcs is the improvement in
outboard panel performance due to better cell selection (about 1.02). The
sun angle is tightly constrained between 80 and 100 , except during
maneuvers, when it can be between 30 and 140 .
FIGURE 13.8 Dwelling at favorable sun angle to improve energy balance during transfer
orbit.
The average power during transfer orbit, ignoring the thermal effects, is
given by
2
Pt:o:avg ¼ Pmax ð13:8Þ

Also denoting PL ¼ load power, including the orbit average losses
Edis ¼ total energy discharged during half a turn (dark shaded area)
Echg ¼ total energy charged during the same half turn (light shaded area)
1 ¼ crossover point between the charge and discharge mode
we have

PL
1 ¼ sin1 ð13:9Þ
Pmax
Z1
PL Pmax sin
Edis ¼ 2 d ð13:10Þ
d
0
Z2
Echg ¼ ðPmax sin PL Þ c b d ð13:11Þ
1
where 1 is in radians, and Edis and Echg are in watt-hours. For the satellite to
be in energy balance during the transfer orbit, Echg must equal or exceed Edis
with a suitable margin compatible with the uncertainty at a given design
stage. Such analysis determines the required peak power Pmax, the
spacecraft inclination or the maximum permissible load during the transfer
orbit. It does not account for eclipse during the transfer orbit, which must be
considered separately, if applicable.
13.4.4 Launch and Ascent Simulation

During the launch and ascent and until separation, the battery powers the
computers, EEDs and other essential equipment until sunlight is acquired.
In an LEO mission, the battery designed for a shallow (30 to 35%) DOD
during normal operation orbit can be discharged to a higher (80 to 90%)
DOD for one time during the launch/ascent phase. However, once the sun
is acquired, the battery is first fully charged before undertaking other
maneuvers. Figure 13.9 depicts a typical battery SOC for an LEO satellite.
If the battery is short of energy in this phase, two options are:
FIGURE 13.9 Energy flow and battery SOC during launch and ascent phases.
Cool the battery to a lower temperature by using a jet of cool air or

nitrogen on the launch pad just before launch. This allows charging
the battery to a higher Ah capacity. This is a widely used option.
If that is not enough, a small primary battery may be used to augment the
main battery. The primary battery having an order of magnitude
greater specific energy gives a mass efficient solution. Carrying a
larger main battery just for this phase is almost never done, as it adds
significant mass, and also costs much more.
13.5 Battery State of Health Monitoring

Since the battery is a critical component in the spacecraft power system,
monitoring its state of health and the remaining life is important. The data
collected for the on-board energy balance management can also be used for
this purpose. The battery overall state of health on board is generally
monitored by the following two key indicators.
The pressure strain gage readings monitor the battery’s overall condition.
Any unexpected reading is taken as an alert flag. Initial matching of the
cells ensures that two cell pressures represent all other cells.
Cell divergences are detected using the individual cell voltage monitored
by the BCVM. Trend analysis of the cell voltages is performed to detect
individual cell voltage divergence. Any such divergence flags a battery
weakness.
The battery state of health indirectly measures external behavior such as a
capacity loss. It also measures internal behavior such as severe corrosion,
the remaining useful life, and the remaining useful energy. The changes in
FIGURE 13.10 Circuit model for battery with wiring resistance and inductance.
the electrode surface, diffusion layer, and electrolyte are not measurable.
However, the terminal voltage and body temperature is used to indirectly
determine the performance. Accurate modeling of the transport mechanism
inside the cell and the electrochemical and thermal models are needed for
this purpose. The cell internal impedance frequency spectrum technique
developed by Kozlowski et al2 can retrieve useful information about the
internal electrochemical conditions. The internal impedance of the cell is
measured by applying a small signal excitation to the cell and measuring
the response. A range of frequencies is used to account for the frequency
dependence of the of many electrochemical impedance parameters. The
impedance model of the cell between positive and netgative electrodes
separated by an electrolyte is shown in Figure 13.10,3 where R and L are the
wiring resistance and inductance, respectively. It represents the physical
electrochemical process via electrolyte resistance, Re, charge transfer
resistance, , double layer capacitance, Cdl, and the Warburg impedance,
Zw. These parameters represent the physical electrochemical process, such
as the charge and mass transfer during cycling. For example, the model
identifies the electrolyte resistance, Re, versus the state of charge for a NiCd
battery as shown in Figure 13.11. Such information is then used in a neural
network and fuzzy logic state of charge prediction model. This technique is
FIGURE 13.11 Electrolytic resistance of NiCd battery from the model.

under development at the U.S. Naval Research Center and holds promise
for monitoring the state of charge, state of health, and remaining life of the
spacecraft batteries.3
13.6 Battery Latch-up in Sun-Regulated Bus

In the sun-regulated power system, the difference between the stable
operation of the solar array and that of the battery often leads to an
operating condition known as the battery latch-up. It occurs when the
satellite is coming out of an eclipse. At this time, although the solar array is
capable of delivering full power, the battery continues to deliver the load
power, instead of going into the charge mode. The bus remains latched to
the battery voltage and is unable to recover to the shunt controlled voltage
even during sunlight. Subsequently, it is not possible to maintain the energy
balance, either by hardware or software. This is obviously an anomaly that
should be immediately corrected to prevent the battery from further
discharge. The condition is typical only in the sun-regulated system. The
fully regulated system is inherently free of such latch-up.
The phenomenon is explained here with reference to Figure 13.12. The
dotted parabola is the constant power load line for the spacecraft. The
heavy solid line is the V–I characteristic of drained battery at the end of
eclipse. The system operates at the stable point of their interaction, A1. The
other point of their intersection, U, is unstable. The faint dotted curve B is
the I–V characteristic of the solar array soon after entering the sunlight. This
curve keeps rising as the spacecraft moves from eclipse to penumbra to full
FIGURE 13.12 Battery latch-up in sunlight–regulated bus.

sunlight. Until it rises to curve D, the battery continues discharging to the

load at point A1, as curve D and load line have no point of intersection.
When the illumination rises to full sunlight curve E, there are intersection
points A2 and A3 with the load line. But, since point A1 is stable, it does not
move to a higher voltage point A2. The system thus remains latched at the
battery voltage A1. At this point, the current drawn from the solar array is
V1E1 and the current drawn from the battery is E1A1. With curve F, the
battery and the solar array match in their current capabilities. The system
operates at point A1, where it is stable because of being latched with the
battery (the solar array is unstable at A1).
When the illumination rises to curve G (full sun), the operating point A1
moves to point A5 on curve G, but since it is unstable, it starts moving
towards the stable point A6. On the extreme right hand side, however, the
sunlight shunt regulation control set at 35 V limits it at point A4, where Ishunt
¼ A4A7. The operating point with load becomes A4 with the shunt in
control of the system now.
Thus, the battery latch-up can occur under the following two conditions
at the end of eclipse:
The battery voltage drops below certain low point, drawing heavy
current to power constant power loads, and
The solar array current output upon entry to the sunlight is less than the
current drawn from the battery at the end of eclipse.
The remedy for avoiding the battery latch-up is to size the solar array
more than that required for meeting the end of life load power with
sufficient margin. In Figure 13.12, the I–V curve of fully illuminated solar
array should be at least above curve F (point A1) to unlatch the system from
the low battery voltage during eclipse. This generally required 5 to 10%
extra power margin compared to that in the fully regulated bus. The added
mass and cost in the solar array constitutes a major disadvantage of the sun-
regulated system. For this reason, large power systems, where this penalty
can be high, find fully regulated system advantageous. If the battery voltage
is too low during eclipse, even more solar array power is required to
unlatch the system. On the other hand, if the battery voltage is too high
during sunlight, the charge converter works inefficiently at low duty ratio
(close to zero). The design engineer trades these two features.
Another remedy for the battery latch-up is to add one (rarely two) cell.
With a 23-cell battery instead of the traditional 22-cell battery, the higher
battery voltage prevents the latch-up. That extra voltage of one cell is often
enough to avoid the latch-up. However, the 23-cell battery becomes fully
charged at 23 1.6 ¼ 36.35V, which is outside the 22-35 V limits on the bus.
This may be allowed in a system where the battery is charged directly from
a dedicated solar array without a charge regulator.
If the battery charge power is derived from a dedicated section of the

solar array with a charge regulator, yet another remedy could be connecting
the charging array on the bus until the latch is cleared up.
After exhausting all possible remedies, if the bus is still latched on the
battery, there is nothing one can do except to shed some non-essential or
noncritical loads temporarily. Some heaters can always be momentarily
shut off. The reduced load current then would allow the sunlight voltage
regulating shunts to regain control of the bus voltage. The loads can then be
brought back into service.
References
1. Singh, P., Gaddam, V., Arey, S., and Yang, Z., Battery state-of-charge
meters for high performance batteries based on Fuzzy logic methodology,
in Proceedings of the 34th Intersociety Energy Conversion Engineering
2. Kozlowski, J.D., Crawley, T., and Byington, C.S., Model Based Predictive
Diagnostics for Primary and Secondary Batteries, Pennsylvania State
University, Applied Research Laboratory, June 1999, Technical Report
No. 99-076.
3. Kozlowski, J.D., Watson, M.J., Byington, C.S., Garga, A.K., and Hay, T.A.,
Electrochemical cell diagnostics using on-line impedance measurements,
state estimation and data fusion techniques, Proceedings of the 36th
Intersociety Energy Conversion Engineering Conference, ASME, 2001, pp.
981–986.
Chapter 14
Dynamic Performance and Stability
14.1 Introduction
The dynamic performance of power system under an internal or external
transient perturbation is influenced by the dynamic bus impedance and the
control loop gain. Key performance attributes emerging from the dynamic
study are the bus voltage ripples, transient deviations, fault and fuse blow
transients, and the control loop stability under harmonic ripple excitation.
On the other hand, the static performance under a slow change or after the
dynamic response has settled in time is largely influenced by the static
impedance of the bus. The voltage regulation long after a load change is an
example of static performance. Since the dynamic and static bus impe-
dances are similarly defined, they are jointly covered in the following
section.
14.2 Bus Impedance and System Stiffness

A complex electrical network having a number of sources and loads
between any two terminals can always be reduced to a simple equivalent
source, known as the Thévenin equivalent source, consisting of one source
voltage Vs with an internal series impedance Zs (Figure 14.1). The two
source parameters are determined as follows.
With the system operating at no load between points 1 and 2 in Figure
14.1, but with all other parameters at rated values, the voltage between
FIGURE 14.1 Complex electrical network reduced to Thévenin equivalent source.
368
Dynamic Performance and Stability 369
terminals 1 and 2 equals the source voltage Vs (since the internal voltage
drop is zero). Therefore,
Vs ¼ open circuit voltage of the system at the load terminals ð14:1Þ
With terminals 1 and 2 shorted, the internal voltage is now totally

consumed in driving the current through the source impedance only.
Therefore,
Zs ¼ open circuit voltage=short circuit current at the load terminals ð14:2Þ
The short circuit current is determined by calculations in the actual

circuit. If it is desired to determine by tests, a reduced voltage is applied to
limit the current to rated value. The full short circuit current is then
calculated by scaling to the full rated voltage. Any nonlinearity, if present,
must be accounted for.
The source impedance of most spacecraft bus architectures is highly
nonlinear due to the use of multiple bus regulators along with dead-band
regions. The dynamic performance of such systems is largely driven by the
transient nature of switching from one mode to another, such as solar array
shunt control to battery discharge control. Linear modeling may be
acceptable for steady-state regulation and small signal load changes within
the control range of each controller, but not for modeling mode change
transition, which must use transient models for the individual controls.
Two alternate ways of determining Zs are as follows:
Set all independent voltage and current sources in the system to zero
and determine the impedance between open terminals 1 and 2.
With the system energized and the load connected between points 1
and 2, measure the voltage drop iV for a slow load change iI. Then,
Zs is the ratio iV/iI.
The Thévenin equivalent source model derived under the steady state
static condition gives the static bus impedance Zs. The source impedance
derived under the dynamic condition — that is, for an alternating or
incremental load — is the dynamic bus impedance Zd. The Zd varies with
frequency and can be either calculated or measured with the test set up
shown in Figure 14.2. With the bus in operational mode delivering the rated
load, a small high frequency ac current Ih is injected into the bus using an
independent current source. The value Vh, the high frequency voltage
perturbation in the bus voltage, is measured. The dynamic bus impedance
at that frequency is then
FIGURE 14.2 Dynamic bus impedance measurement by injecting ripple current Ih.
Vh
Zd ¼ ð14:3Þ
Ih
Since Zd has a strong influence on the system’s dynamic performance, it

is kept below specified limits. Figure 14.3 and Figure 14.4 are examples of
the main bus and the battery pyro bus source impedances for mid-size
communication satellites.
One way of evaluating the relative fuse blow speed in a power system is
to evaluate the system stiffness, which is inversely related with the static
bus impedance. After the initial inductive rise of dc fault current takes place
in five time constants (L/R), the steady fault current magnitude is inversely
proportional to the static impedance of the source looking back from the
fault location. We now define the following terms leading to the system
stiffness:
FIGURE 14.3 Main bus source impedance requirement for typical GEO communications
satellite.
FIGURE 14.4 Pyro (battery) bus source impedance requirement for typical GEO commu-
nications satellite.
Rated system impedance Zrated ¼ Vrated/Irated

Normalized static impedance Zn ¼ Zs/Zrated
System stiffness ¼ fault current in multiples of the rated current
¼ 1=Zn ð14:4Þ
The stiffer the system, the quicker the fuse blow. This concept is useful in
top-level comparison of two power systems of different designs.
14.2 Voltage Regulation and Transients

The steady-state voltage rise on removal of the full rated load current is
then iV ¼ IratedZs, and the static voltage regulation (as a percentage) is
defined as
V
voltage regulation ¼ 100 ð14:5Þ
Vrated
The dynamic voltage regulation is defined as the change in the steady

state voltage after the feedback control dynamics have taken effect. It
primarily depends on the control loop design, rather than on the dynamic
bus impedance.
Under a load step change — partial or full — the voltage oscillates until
the transient settles to a new steady-state value as depicted in Figure 14.5. A
good design limits and damps the oscillation in a short time to avoid the
FIGURE 14.5 Transient voltage undershoot and overshoot following load step change.
system becoming unstable. If the load current rises in step, the voltage
oscillates before settling down to a lower steady state value. The steady-
state change in the bus voltage is then given by V ¼ I Zs if there were
no feedback control. The feedback voltage control loop, however, responds
to bring the deviated bus voltage back to the rated value as shown in Figure
14.6. The bus voltage is maintained within specified limits at a specified bus
FIGURE 14.6 Main bus transient voltage limit for typical 100-V bus.
sense point, generally the power distribution point inside the power
regulator unit. In order not to flutter the system more than necessary, the
control system is designed with suitable deadbands. For example, Figure
4.9 in Chapter 4 illustrated a 120-V PV–battery system with two deadbands
in the control loop.
The bus voltage can deviate from its nominally rated value due to many
reasons. Computers and microelectronic circuits are more susceptible to
voltage transients than rugged power equipment such as motors, transfor-
mers and heaters. The deviation that can be tolerated depends on its
magnitude and the time duration. Small deviations can be tolerated for
longer time than large deviations. The tolerance band is generally defined
by voltage versus time (V–t) limits. The system voltage must be maintained
within the specified V–t envelope shown in Figure 14.7 as an example. The
right-hand side of the band comes primarily from the steady-state
performance limitations of the load equipments. The middle portion
comes from the voltage flicker considerations. And the left-hand side of
the band comes from the electronic load susceptibility considerations,
which is in microseconds range based on the volt-second capability of the
power supply magnetics and the dielectric withstand capability of the
insulation.
FIGURE 14.7 Bus voltage deviation limits versus duration for typical 3-kW, 120-V LEO
satellite. (CE07 limits for EOS-AM.)
14.4 High-Frequency Ripples

The ripple is the term used to describe peak-to-peak voltage deviation from
the normal steady state operating voltage under repetitive current pulses.
The ripple is sometimes expressed in terms of the rms value also, defined as
VrðppÞ
VrðrmsÞ ¼ pffiffiffi ð14:6Þ
2 2
The ripple is caused by switching a current on or off, such as in the PWM

operation of solar array shunts, battery power converters, and communica-
tion load modulation. Figure 14.8 illustrates the ripple voltage generation
under repetitive on-off pulses of current. The PWM switching of shunt
current is a major source of the bus voltage ripples. The ripple frequencies
are integer multiples of the switching frequency. The ripples are periodic
but not sine wave, and are superimposed on the fundamental voltage, a.c.
or d.c. The ripple voltage induced on the bus due to ripple current of certain
frequency is given by
Vripple ¼ Iripple Zd ð14:7Þ
where Zd is the dynamic bus impedance at the frequency of interest. The

impedance is generally in hundreds of milliohms and the ripple voltage
limit varies over a wide range in hundreds of millivolts. Figure 14.9 is
a typical voltage ripple limit imposed on a main bus of a mid-size
communications satellite.
The ripple is minimized by bus filter capacitors connected to the bus, or
preferably at the component terminals causing the ripples. The ripple
FIGURE 14.8 Bus voltage ripple under repetitive load pulses.

FIGURE 14.9 Bus voltage ripple limits for typical 5- to 10-kW, 100-V GEO communications
satellite.
current is then supplied or absorbed by the capacitor, rather than by the

bus, thus improving the quality of power. For a 28-V sun-regulated bus for
communications satellite with 4000-W payload, a typical bus filter may be
about a 4000-mF capacitor bank with an ESR less than 60 m at 25 C. The
filter design is made two-fault tolerant by using a bank of many capacitors
connected in parallel with internal fuse protection for each.
Equal output source impedance of the PRU in the sun and the eclipse
operations can benefit the bus design for meeting the ripple limit.
Otherwise, the higher of the two must meet the limit, while not receiving
any benefit from the lower ripple in the other mode of operation.
Controlling the input ripple to the PRU is also important to avoid excessive
common mode noise at the battery monitoring circuit.
Multiple ripple frequencies are often present in the system. The system
performance may suffer if ripples of numerous frequencies having about
the same p–p value are present, although each individually meets the p–p
ripple specification. For this reason, it is a general practice to specify not
only the p–p limit at any one ripple frequency, but also the total ripple
distortion factor (TRDF) due to all ripples combined. The TRDF is defined
in terms of a Fourier series, where a periodic voltage of any shape is
decomposed in a series of sinusoidal components, i.e.,
X1
VðtÞ ¼ V1 þ n¼2
Vn sin ðn!t þ n Þ ð14:8Þ
The first component on the right-hand side of the above equation is called
the fundamental component, whereas all other higher frequency terms
(n ¼ 2,3 . . . 1) are called the harmonics. The TRDF is then defined as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V22 þ V32 þ ::: Vn2
TRDF ¼ ð14:9Þ
V1
The TRDF is useful in comparing the quality of power with respect to the
ripples between two or more power systems, or at various locations of the
same power system. In a pure ripple-free voltage source, TRDF ¼ 0. The
greater the value of TRDF, the more distorted the bus voltage. The major
operational concern arising from the ripples is the EMI. Additionally, the
interaction of ripple voltage and current do not contribute to the average
power. The ripple currents merely result in more I2R loss for the same
useful power delivered. Poor quality of power therefore adversely affects
the power system efficiency also.
The harmonic distortion on bus voltage caused by the nth harmonic
current In drawn by any nonlinear load is given by Vn ¼ InZn. This
distortion in the bus voltage that causes the harmonic current to flow in
other loads is called the victim load. A harmonic-rich bus voltage supplies
distorted current to all loads, linear or nonlinear. Therefore, harmonics are
generally filtered out before feeding power to the bus. A bus voltage having
the TRDF less than 3 to 5% is generally considered acceptable.
14.5 Ripple Measurement

Two simple test methods to measure ripples are shown in Figure 14.10.
They consist of connecting the oscilloscope probe to the output terminal of
the equipment under test either (a) directly, or (b) with a capacitor. The set-
FIGURE 14.10 Ripple measurement methods.

FIGURE 14.11 Ripple measurements using terminated coaxial cable.
up (b) is widely specified method for power converters. It typically uses

AWG 16 twisted pair of output wires terminated into a 50 mF capacitor and
an oscilloscope of bandwidth greater than 50 MHz.
Figure 14.11 shows another way to measure output ripple voltage using a
terminated 50 coaxial cable going directly to the oscilloscope. A ground
plane is used between the power converter and the scope. Extraneous noise
pick-up is minimized in this low impedance shielded measurements. In
addition, an input balun is used with the d.c./d.c. converter to minimize
any common mode noise pickup. Any noise pickup in this set up may be
checked by connecting both input and ground leads to the same converter
terminal and looking for noise on the scope. With the termination resistance
matching with the coaxial cable characteristic impedance, the measured
ripple voltage is equal to one half the actual value of the p–p ripple.
It is sometimes required to measure the reflected ripple current from a
d.c./d.c. converter back into the source. Such measurements are done with
a circuit shown in Figure 14.12. It uses a wide-band current probe and
oscilloscope, both of bandwidth at least 20 MHz to give accurate wide band
measurements. A non-inductive current sampling resistor to make a voltage
measurement is another method, but it takes considerably more care to get
accurate results.
FIGURE 14.12 Reflected ripple current measurements.

14.6 Quality of Power

There is no generally acceptable definition of the quality of power as yet.
However, the power quality is regarded to have three major performance
criteria:
The total ripple distortion factor generated by the power electronics

The transient voltage sags caused by system disturbances and faults
Periodic voltage flickers
The power quality specifications are generally written in two parts. Part I
address the source and the distribution system. It specifies the following
attributes at the output of PRU that various loads can expect:
Steady state voltage limits

Transient voltage limits
Source impedance
Ripple voltage at the bus
Part II, on the other hand, specifies the requirements the loads must
comply with in order to be connected to the bus. Specifically, it limits the:
Steady state voltage deviation withstand capability

Transient voltage deviation withstand capability
Load impedance
Isolation voltage and resistance
Turn-on in-rush current
Ripple emission
The systems engineer ascertains that the source power quality specified
in Part I is compatible with the load power quality specified in Part II. For
example, the voltage transient created by the source should be less than the
voltage transient the load is required to withstand with some margin.
14.7 Minor Fuse Blow Cross-Talk

A voltage spike occurs in the system when a fuse blows. If many load
equipment are connected to a common fuse board, fuse blowing in one
equipment may generate a voltage spike in another. This is often known
as the cross talk between equipments. Various equipments should be
individually fused to minimize cross talk between them. Full-scale circuit
simulation can predict the cross talk voltage spike. When the loads are not
well defined in an early stage of the program, the energy approach

presented below may give approximate, but quick and directly usable,
upper limit on the cross-talk spike. It is based on a simple circuit model of
Figure 14.13 using only the cable parameters shown in (b).
Various wires and other resistances in series determine the steady state
fault current just before the fuse blow. With that current, the wire
inductance would store energy equal to 1=2 LI2. When the current is
interrupted on the instant of the fuse blow, this energy is dumped on C,
the capacitance of the wire or the EMI filter, or both. Thus, the cable
inductive energy gets transferred to the capacitance, and the capacitor
voltage rises. Ignoring the energy loss in the transfer and the initial energy
in the capacitor at rated voltage, the inductive and capacitive energy must
be equal, i.e.,
1 2
2LI ¼ 12CV 2 ð14:10Þ
Thus, the bus voltage on the instant of current interruption would rise to the
value given by, in the first approximation,
rffiffiffiffi
L
V¼I ð14:11Þ
C
If the cable capacitance is insufficient to absorb the inductive energy, an

external capacitance can be added so that the total capacitance keeps the
voltage spike under the specified limit.
14.7.1 Example
Consider a 28-V bus cable 5 m long, made of four parallel wires, each with
inductance of the 0.375 H/m and capacitance of 0.077 nF/m. If a fault
current in each wire before the fuse blow is 70 A, the energy equation gives
1=2 (0.375 5 m) 106 702 4 ¼ 0.018 J ¼ 1=2 C V2. To keep the voltage
FIGURE 14.13 Cross-talk between equipment following a small fuse blow.

rising above say 80 V, we need a capacitance C ¼ 2 0.018/(802) ¼ 5.6 mF.

Since the estimated wire capacitance (0.77 5 4 nF) is much less than
what we need (5.6 mF), we must ascertain that the load converters’ input
EMI filters have at least 5.6 mF at the front end. Alternatively, one can place
5.6 mF in the fuse board to limit the fuse blow voltage spike from rising
above 80 V.
The hermetically sealed fuse blows suddenly because no plasma is
generated in the sealed vacuum. Therefore, it can cause severe voltage spike
following a fuse blow, and may need the above estimated capacitance at the
fuse location. The regular nonsealed fuse may require less capacitance
because it blows slowly due to the lingering plasma. An alternate method of
spike suppression often used is a freewheeling diode in parallel with the
fuse. This technique, however, cannot be used to suppress a transient
generated by a source inductance common to the affected power distribu-
tion. The most effective means of such transient control is to minimize the
source inductance and use capacitive damping.
Some load faults can result in large but short duration transient currents
in hundreds of amperes. Such fault currents would result in bus voltage
transient that are much greater than the example above. They may
momentarily collapse the bus as discussed in the next section.
14.8 Major Fuse Blow Transient

The bus voltage rapidly drops under fault in large equipment connected to
the bus. For example, a fault in one of the arcjets with a 50 A fuse is
simulated as Load1 in Figure 14.14. The solar array, battery and the flitter
capacitors, all feed current to the fault as shown by arrows in the figure.
After the 50 A fuse blows, the available solar array power starts recharging
the battery and the capacitors. The voltage takes a sharp v-turn and starts
rising until it reaches the nominal bus voltage. The power system control
loop then stabilizes the bus voltage within the tolerance band. The fall and
FIGURE 14.14 A circuit model for a major fuse blow transient analysis.
rise of the bus voltage under such major fuse blow is depicted in Figure
14.15. Thus, the entire power system is involved in clearing a large fuse. The
larger the rating of the affected fuse, the deeper the voltage sag. The power
system specifications generally limit such voltage sag to some minimum
value set by a critical equipment requirement. For example, the central
computer on a 70-V system cannot perhaps tolerate the voltage sag below
35 V for more than 5 ms. The power system design engineer must therefore
run the fuse blow simulation analysis and show that such a requirement is
met.
Most faults are between the line and chassis involving the ground, rather
than between the line and return. The fuse-blow transient is sensitive to the
ground return resistance, which may vary from fault to fault. For this
reason, the fuse blow circuit simulation is carried out with various possible
fault resistances. The results are parametrically plotted as shown in Figure
14.16. This assures that under the worst-case fault resistance the bus voltage
meets the requirement. In addition to the random fault resistance, there is
always some minimum ground impedance depending on how the system is
grounded to the structure. For example, the structure with graphite
composite face-sheet has higher ground contact resistance. The resulting
spreads of the current give higher inductance also. For this reason, the
ground contact impedance of the graphite face-sheet used in some modern
communication satellites is about three orders of magnitude higher than
that with aluminum face-sheet. This significantly changes the transient
characteristic for faults involving ground. High ground impedance reduces
the final fault current only slightly, but greatly increases the L di/dt when
the fuse blows, thus raising the EMI/EMC concerns.
FIGURE 14.15 Bus voltage decay and recovery transient during a major fuse blow.
FIGURE 14.16 Bus under-voltage and over-voltage versus fault resistance during major fuse
blow transient.
A typical main power bus fault response requirement is shown in Figure

14.17. In the past, fuse transient has not been a major problem, except some
telemetry may give an erroneous reading on that instance, but recovers
soon. Such a glitch is generally tolerable in commercial satellites, but not in
military and NASA satellites, because, in an extreme case, it may upset the
CMOS computer gate.
The EPS is generally designed to accommodate payload shedding in case
the bus voltage drops to 90–95%, and is capable of powering essential
spacecraft functions up to 75–80% of the nominal bus voltage.
14.9 Stability and Bus Impedance

The steady state operating point is stable if it gets restored after an
arbitrarily small momentary disturbance in the source or the load. For
example, at operating point A1 under constant power P1 in Figure 14.18,
suppose a disturbance causes a drop of iV in voltage. At the new voltage,
the source current is less than the load current. Consequently, the voltage
decreases again and the operating point moves further away from A1, and
again further away, and again in a runaway situation. Similarly, a small
increase in voltage due to a disturbance will make the system move away
from point A1 in the other direction. Thus point A1 is an unstable operating
point, where the system works like a positive feedback system.
On the other hand, at point A2, a small change in voltage in any
direction, due to any reason, produces counterchanges to bring the system
back to its original operating point A2. Thus, point A2 is a stable operating
point, where the system works like a negative feedback system.
Analytically, we can derive the following condition for static stability. An
FIGURE 14.17 Main power bus fault response requirements for a typical 5- to 10-kW, 100-V
GEO communications satellite.
FIGURE 14.18 The stability of a solar array operating point under a constant power load.
operating point is stable if and only if the following conditions are met at
that point

dV dV
> ð14:12Þ
di source di load
If we think in terms of the absolute values of the dynamic impedances and

define Zs as source output impedance and ZL as the load input impedance,
then the system is stable only if
Zs < ZL ð14:13Þ
at all frequencies. When Zs > ZL, further analysis is needed to determine the
system stability.
The operation of solar array under a resistive load is shown in Figure
14.19. The load line is a straight line from zero with the slope equal to the
resistance value. For such a load, the above condition is always met, hence
the solar array powering a purely resistive load is always stable.
A nonlinear solar array source coupled with a nonlinear load may result
in multiple equilibrium points under a given operating condition. The
actual equilibrium point that the system resides at is determined by the
stability nature of the equilibrium points and the past history of the
operating point.
During eclipse when the battery directly supplies a constant power load,
the intersection of the source line and the load line always gives the stable
operating point as shown in Figure 14.20. The unstable point in this case is
outside the practical range of the constant power load, which its tripped off
before the voltage drops to the point where the battery operation may
become unstable.
FIGURE 14.19 The stability of a solar array operating point under a resistive load.
FIGURE 14.20 The stability of a battery operating point under a constant power load.
Figure 14.21 shows an example of the load and source impedances

overlapping and the unstable operating region. The source and load
impedance specifications are primarily set to ascertain the system stability
under all conditions. This task can be complex even for a small spacecraft. It
can be extremely complex for a large space platform with a variety of loads
that can be connected, not all of them known in advance, such as on the
ISS.1 In those cases, the load impedance specification for every load can be
established from the fundamental systematic considerations presented
below.
Any complex system with a number of sources and loads can be reduced
to the equivalent network with Thévenin parameters Vs and Zs. Any
additional load with impedance ZL must be greater than this value of Zs for
stability. A critical factor in the design of a large platform such as the ISS is
the stability of the system under all expected conditions of loading and
transient perturbations. The ISS power systems engineers have developed a
set of impedance criteria for both the source converters as well as the load
converters to assure stability under significant variations in the load
connectivity over the life of the station. When a new load is added on the
large platform, the old source and load impedances combine in parallel to
form the new source impedance (Figure 14.21). Thus, the new Zs will
always be less than the old Zs. The new system would therefore be stable as
long as the new load impedance is greater than the greatest load impedance
in the old system. This logic may be used to set the dynamic input
impedance of any load that can be connected by any user on the ISS type
platform.
If the system is stable, then any additional load in parallel will keep the
system stable.2
FIGURE 14.21 Thévenin equivalent source and dynamic load impedance for a stability study
in a large power system.
14.10 The dynamic Stability of a Control System

A dynamic stability study determines whether the system will be stable
under sustained small harmonic perturbation coming internally from the
power system or from an external source via conducted EMI. Verifying the
stability involves modeling the total system with all control loops. A
detailed theory of the control system is outside the scope of this book, but a
short review follows.
The negative feedback control system is widely used for controlling many
system, subsystems and components, often with multiple nested loops.
Figure 14.22 represents a feedback control system in terms of the transfer
functions (TFs), where G is the main equipment TF and H is the TF of the
feedback control loop. It can be shown that the TF of the entire system is
Laplace transform of the output yðtÞ GðsÞ

TF ¼ ¼ ð14:14Þ
Laplace transform of the input x ðtÞ 1 þ GðsÞHðsÞ
FIGURE 14.22 A classical feedback control system model.
This equation indicates that 1 þ G(s)H(s) ¼ 0 results in an infinitely large

output for an arbitrarily small input. The solutions (roots) of Equation 14.14
are called the system poles, which are in general complex, say s ¼ þ j.
We say that the output blows up, i.e., the system is unstable, if
GðsÞHðsÞ ¼ 1 ð14:15Þ
In other words, the open loop gain of 0 dB and the phase angle of 180 give
an unstable system. The stability is often judged two ways. In one method,
the gain in dB and phase angle in degrees are plotted with respect to
frequency in what are known as Bode plots shown in Figure 14.23. The
system is generally considered stable if the gain is below zero with about a
30-dB margin, and the phase angle is away from 180 with about a 30
margin. These margins vary with the spacecraft requirements and heritage
data on similar bus. In one 3000-W, 120-V LEO satellite, the PRU and mode
controller amplifier were required to provide stability phase margin of at
least 45 and gain margin of at least 10 dB.
FIGURE 14.23 Dynamic stability criteria from open loop Bode plots.
FIGURE 14.24 The dynamic stability criteria from poles plot in complex plane.
In another method for determining stability, the poles are plotted in the
complex plane. The system is stable only if all poles have negative real part
which results in damped oscillations. Or, in other words, all poles must be
in the left-hand side of the complex plane (Figure 14.24). In case the system
is found unstable, a compensating lead–lag network is introduced in series
with the main equipment as shown in Figure 14.25. The combined forward
TF is then C G, which is used in evaluating the stability of the
compensated feedback system. Since the circuit parameters of the compen-
sating network may vary with temperature and age, it is customary to check
the stability under the worst compensation and load conditions as
illustrated in Table 14.1. The table in this example shows that the system
is stable with comfortable margins under the worst conditions.
14.11 Dynamic Simulation Model

Considering the size and complexity of modern spacecraft power systems,
the digital computer is the only viable tool for system modeling and
simulation. Two general classes of simulation programs appear in the
literature. They are: (1) generalized circuit and system simulation models,
FIGURE 14.25 Series compensation (lead–lag network) added for stability.

Table 14.1 An example of a stability study under the worst compensation and
load combination for a 100-V bus
Payload
configuration With maximum compensation gain With minimum compensation gain
System Phase Gain System Phase Gain

band-width margin margin band-width margin margin
(Hz) (dB) (Hz) (dB)
All loads off 1700 105 40 1100 70 46
1 TWTA on 1400 104 40 1000 70 44
2 TWTAs on 1400 103 40 900 69 48
3 TWTAs on 1200 102 40 900 70 46
or (2) dedicated models of specific system. The generalized computational

models as a class are built using commercial programs with tradenames
such as PSPICE, SABER, SYSCAP, ICAP, and a number of others. All have
similarities in that they offer a large library of component templates capable
of analog, analog/digital mixed signal and mixed technology simulation
(electrical, mechanical, thermal, optical, and even chemical). The templates
are assigned parametric values to simulate the actual component used in
the system. The simulation can be at any level: component, behavioral,
functional, or control level. The scope of total simulation model generally
consists of developing the following models:
Component level models of
Solar array
Battery
Behavioral or component level models of
Solar array shunt control

Battery charge and discharge converters
Load types: constant power, constant current or resistive
Power distribution and protection: cables, relays, fuses, RPCs
Control loops: usually nested in multiple layers
14.11.1 Solar Array Model

Two approaches can be taken for modeling the solar array: micro and
macro. In the micro, each cell is modeled by its I–V characteristic equation
presented in Chapter 8. The advantage of this approach is that cell-to-cell
variations internal to the array can be modeled. For example, a local cell
failure or shadowing can be inserted into the model. However, the number
of cells needed for a full simulation in a large satellite can be prohibitive and
inefficient. The macro model, on the other hand, can represent the entire
array as one component with the total I–V characteristic of all cells
combined in series–parallel connections.
The shunt control loop may show some sensitivity to the capacitance of
the solar string circuits. However, in practical shunt control loop designs,
the gain and phase margins in the stability analysis are found to be
insensitive to the cell capacitance varying over a wide range around the
nominal value. The solar cell capacitance can be calculated from the solar
array impedance measurements under sweeping frequency with d.c.
voltage and current bias conditions. The capacitance of 2 4 cm silicon
solar cell is typically 0.25 mF when in dark. Under illumination, it is about
0.50 mF near short circuit current and 0.75 mF near the peak power point. A
typical dynamic a.c. model for one 5-A, 70-V solar circuit is shown in Figure
14.26.
Lindmyer3 gives the a.c. impedance (in ohms) of the solar circuit as
Rd
Zac ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ð14:16Þ
ð1 þ !
where Rd ¼ dV/dI, the slope of the solar cell I–V curve, i.e., the dynamic
resistance, and ¼ diffusion time constant of electrons in the p-region of the
cell. The Zac includes the a.c. resistance of the equivalent diode as well as
the Rs and Rsh components shown in Figure 8.5. It varies with the d.c.
operating point, and also the cell temperature via the term. The
capacitance, Cs (in farads), can be analytically obtained from NASA/JPL
data4 as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
q " Na
Cs ¼ A ð14:17Þ
2ðVb Va Þ
FIGURE 14.26 The solar array circuit dynamic model for system simulation.
where
A ¼ cell area,
q ¼ electron charge,
" ¼ permittivity of silicon,
Na ¼ acceptor density in the p-region,
Vb ¼ barrier voltage,
Va ¼ applied voltage (positive in forward bias).
A typical 2 m silicon cell with " ¼ 1 pF/cm, Na ¼ 11016 cm3, and Vb ¼
0.8 V at room temperature gives a value for Cs, in mF/cm2, of
0:0288
Cs ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð14:18Þ
0:8 Va
The maximum and minimum values of Cs occur at open circuit and short
circuit, respectively. Not much has been published on the variation in Cs
with temperature, probably because the changes in d.c. operating voltage
far outweigh the temperature effects. However, Cs increases with increasing
temperature at about 0.1% per C.
14.11.2 Other Component Models

The battery model can be simply a constant voltage source with an internal
resistance shown in Figure 9.36, or a complex nonlinear voltage source
shown in Figure 9.37. The choice depends on the degree of precision desired
from the model. For a 100 Ah NiH2 cell, the battery internal inductance is
about 0.35 mH/cell.
The solar shunt regulators, and the battery charge and discharge
regulators can be modeled at their own component levels in the micro
sense, or at their behavioral levels in the macro sense. Both approaches are
valid, although the latter is perhaps simple and yet sufficiently accurate for
a wide variety of analyses.
The communications loads (TWTA and SSPA) are constant power load.
Many bus loads such as arcjets and reaction wheels are also constant power
loads. They are cut off when the voltage drops below certain values. The
time for which they stay on before tripping off is a function of the
undervoltage as shown in Table 14.2. During on-time, the constant power
load is modeled as a variable resistance of value R ¼ V2/P, where V is the
actual bus voltage at any given time. If the voltage remains below the left-
hand column value for time greater than that in the right-hand column, the
load is cut off. On the other hand, constant resistance loads like heaters
remain connected all the time. Before trip-off, the constant power loads
Table 14.2 The trip-off voltages and the duration of

under-voltage for typical constant power load
Bus voltage transient minimum level Time

(% of nominal bus voltage) duration before trip-off (ms)
85 10.0
71 2.6
58 1.6
45 1.1
31 0.9
18 0.7
5 0.6
0 0.5
reflect as negative resistance on the converter, while all other loads reflect as
positive resistance.
14.12 Simulation Runs

The labor involved in building the dynamic model is rather extensive. Once
the model has been built, independent calibration with known solutions is
the next major task. It is used for the analysis only after verification has been
carried out satisfactorily and reviewed by a team of independent engineers.
The operating point of the system is determined by the source (solar array
or battery), the power conditioning equipment and the load characteristics.
Since the solar array output power and the load demand vary over a wide
range, several modes of operations exist, and all must be analyzed. Some
uses of the dynamic simulation are:
Stability studies
Impedance plots versus frequency
EMI–EMC compliance verifications
Trade studies and parametric studies
Failure effects and worst-case analyses
Fuse blow under-voltage and over-voltage
Troubleshooting
However, the dynamic simulations are performed primarily to determine

that all four primary control loops are stable and have adequate gain and
phase margins. In shunt mode, the regulated bus voltage control loop
(shunt loop) and the battery control loop (BCR loop in charge mode) are
two loops that are independently active. Also in shunt mode, the battery
charge current is a constant current load, which is a part of the overall solar
array load. In discharge mode, the bus voltage regulation (BVR) is

performed by the BVR loop. For these loops, linear circuits can be modeled
using PSPICE or SABER.
There are two primary areas of interest. The small signal analysis is to
obtain phase/gain characteristics of each control loop under various
conditions and to obtain the output impedance characteristic. The large
signal analysis is to assure that each loop meets its specified performance
characteristic for the specified disturbances, i.e., the loop intra-mode large
signal response. Often the inter-mode computer simulations may be
necessary to combine two or more control models to simulate the inter-
mode response, i.e., the transfer of control from one loop to another.
Spacecraft level tests can detect inter-modal transient problems, which
may require re-compensation of the BCR and BVR control loops. Examples
of changes that can be implemented to remedy such problems are:
Addition of output filter for each BVR

Mode controller error amplifier to reduce inter-modal transients
BVR switching circuit that affects the average switch model
BVR error amplifier
References
1. Wilde, R.K., Aintablian, H.O., and Gholdston, E.W., International Space
Station U.S. major elements laboratory power quality test, in Proceedings of
the 34th Intersociety Energy Conversion Engineering Conference, SAE, 1999,
Paper No. 01-2434.
2. Fassburgh, H.D., Gholdston, E., and Mong, A, Development and
implementation of stability requirement for the International Space
Station electrical power system, in Proceedings of the 36th Intersociety
Energy Conversion Engineering Conference, ASME, 2001, Paper No. AT-48,
pp. 281–288.
3. Lindmayer and Wrigley, Fundamentals of Semiconductor Devices, D. Van
Nostrand, 1965, pp. 47–48.
4. NASA/JPL, Solar Cell Radiation Handbook, 3rd edition, 1982, Report 82-69,
pp. 1–19.
Chapter 15
Electromagnetic Interference and
Compatibility
15.1 Introduction
All spacecraft systems are required to be compatible with the electro-
magnetic interference expected from internal and external sources. For
decades, the electromagnetic interference (EMI) and electromagnetic
compability (EMC) requirements have come from MIL-STD-461. It specified
the maximum emission limit of the potential culprit, and the minimum
susceptibility level of the potential victim equipment. The companion MIL-
STD-462 defined the test methods for verifying that MIL-STD-461 require-
ments are met, and MIL-STD-463 defined the applicable terms and units.
The first two standards are now merged into one, MIL-STD-461. For
complying with the EMI/EMC requirements, the contractor of commercial
and defense spacecraft is required to develop the following three
documents and submit to the customer as deliverables:
1. EMC control plan containing the analysis of expected EMI in the
spacecraft versus the imposed requirement, and the steps to be taken for
making the two compatible.
2. EMI test plan with detailed test setups, cables to be used, the grounding
details, frequency range, point of monitoring, and the pass–fail criteria.
3. EMI test report, summarizing the findings of the tests, and steps to be
taken to rectify the failures, if any.
The EMI requirements broadly fall in two general groups, the conducted
EMI and the radiated EMI. In addition, some spacecraft-specific require-
ments may also be imposed. The EMI can enter the equipment either by
conduction via wires, or by radiation in space. For verifying that the
spacecraft would meet such requirements, the first task is to determine the
conducted EMI and the radiated EMI from potential sources, and the
degree of coupling to the victim equipments.
15.2 Sources of EMI

In the commercial world, radiated EMI comes from radio and TV towers,
mobile radio vans, lightening, and nearby power lines. The conducted EMI
394
Electromagnetic Interference and Compatibility 395
comes mainly from power cables. In space systems, the main sources of EMI
are:
Switching large currents or voltages at high frequency, causing large

dI/dt and dV/dt
Electrostatic discharge
Nuclear detonation
15.2.1 Conducted EMI

Active components, such as switches and diodes, are the main sources of
conducted EMI in the switch mode power converter. As was seen in Section
14.4, the switching frequency voltage ripple is primarily influenced by the
dynamic impedance of the equipment. The high-frequency EMI, however,
arises due to the parasitic ringing in voltage following a current switching.
Therefore, refined component models are needed to predict accurate
ringing frequency, amplitude overshoot, and rising or falling transients.
Such models must be verified by tests. The commercially available circuit
modeling tools, such as PSPICE and SABER, may not be accurate enough
for the EMI prediction. Specifically, accurate circuit models for the
following high frequency components are needed in the applicable
frequency range.
Active components (switches, diodes, etc.)

Passive components (transformers, capacitor, inductor, resistor, etc.)
Printed circuits board patterns used in the spacecraft
Accurate models of these components cannot only predict the high-

frequency EMI level, but also can identify the cause. An optimized EMI
solution can then be achieved before fabricating the hardware.
15.2.2 Radiated EMI

The power system harness is a major source of low frequency electro-
magnetic radiation and coupling. Around a short wire (< 10% of the signal
wavelength), generally one type of field dominates, either the magnetic
field (B-field) around high-current lines or the electric field (E-field) around
high-voltage lines. The B-field dominates around a typical power cable. It is
derived from Ampere’s law. In the commonly encountered geometries
shown in Figure 15.1, it is given by the following expressions:
FIGURE 15.1 Magnetic field due to current carrying conductors in various configurations.
For a single wire,
o I
B ¼ ð15:1Þ
2 r
For a parallel pair,
o I d Id
Bz max at z¼o ¼ ffi o ð15:2Þ
2 rðr þ dÞ 2 r2
at far distance.
For a current loop,
o I loop area
By max at y¼o ffi ð15:3Þ
r3
at far distance.
For a common mode current,
o I 2h
Bz max at z¼o ¼ ð15:4Þ
2 rðr þ 2hÞ
The B-field is negligible around a coaxial cable with small eccentricity,

and also at some distance from a twisted wire pair carrying a differential
mode current. For the wires depicted in Figure 15.1, the B-field falls off as
1/r for a single wire, 1/r2 for two parallel wires carrying equal and opposite
currents, and 1/r3 for a current loop.
The high-frequency electromagnetic field travels at the speed of light in
the medium of propagation. The field is characterized into two types: the far
field and near field in terms of the wavelength , which is the ratio of speed
of light over frequency.
15.2.2.1 Far field at r > k/2p

This is also known as the radiation field. It travels long distances in wave
motion with a characteristic wave impedance Zo. In vacuum or air, Zo is
377 . In coaxial cables it is about 50 . The field strengths, both electric
and magnetic, fall off as 1/r.
15.2.2.2 Near field at r < k/2p

This is also called the induction field. Its intensity at a point is determined
by the characteristic of the source. The E-field falls off as 1/r2 if the source
impedance Zs << Zo, and as 1/r3 if Zs >> Zo. On the other hand, the B-field
falls off as 1/r2 if Zs >> Zo, and as 1/r3 if Zs << Zo.
15.2.2.3 Transition field at r ¼ k/2p

At this distance, the field characteristic changes from the near field to the far
field.
15.3 Modes of Coupling

The electromagnetic field coupling between the two components can be via
one or more of the following parameters:
Stray capacitance
Mutual inductance
Ohmic conduction
The degree of coupling can be quantified in terms of the following

parameter:
voltage across victim ðsucceptorÞ

transfer impedance ¼ ð15:5Þ
current through culprit ðemitterÞ
Figure 15.2 depicts capacitive coupling between various conductor config-

urations on a relative basis. The E-field coupling for an ESD type of
transient is governed by the mutual capacitance between the source and the
victim cable. For two straight conductors shown in Figure 15.3, the circuit
analysis leads to the capacitive coupling between conductors 1 to 2 to be

C12
j!
V2 ðC12 þ C2G Þ
¼ ð15:6Þ
V1 1
j! þ
RðC12 þ C2G Þ
The B-field coupling between a culprit and a victim cable can be strong
around the high-current bus, transformer or motor. The voltage induced is
given by V ¼ A dB/dt, where A is area of the loop formed by the victim
cable. If the radiation is sinusoidal at frequency f, then V ¼ 2fAB volts, or V
FIGURE 15.2 Capacitive coupling between various conductor configurations.

FIGURE 15.3 Capacitive coupling between two straight conductors.
(in dBmV) ¼ K þ B (in dBmT), where K ¼ constant. In terms of the mutual

inductance M, the term V (in volts) is then
V ¼ MdI=dt ¼ 2fMI ð15:7Þ
The total inductive and capacitive couplings between two parallel

unshielded wires over a ground plane are determined by the mutual
inductance and capacitance, which, in turn depend on the geometry of the
cables. The two mutual couplings are shown in Figure 15.4, where C and M
are determined from the classical transmission line theory. The capacitive
coupling circuit between an unshielded culprit and shielded victim wires is
shown in Figure 15.5. At very low frequency, the capacitive coupling is
FIGURE 15.4 Capacitive and inductive coupling between two parallel wires over ground
plane.
FIGURE 15.5 Capacitive coupling between unshielded culprit and shielded victim wires.
dominant when the termination resistances are high, so that the current
flow is low and the magnetic coupling (inductance) is ignored.
At power frequencies up to 100 kHz, the coupling is primarily magnetic.
In this frequency range, the following are some practical design guidelines
for controlling the cable coupling:
The coupling increases with the common run length and frequency and
decreases with an increase in spacing.
The noise voltage of the victim’s termination load is a function of the
load impedance. A capacitive filter connected from the victim line to
ground can be used to decrease the load impedance at noise frequency,
provided the victim’s interface can accommodate such capacitance.
The cable shield characteristic is defined by the transfer impedance.
Lower impedance results in more effective cable shielding. Good
impedance values are in the milliohm per meter range.
The twisted shielded wire pair is effective up to 2 MHz, above which
the coaxial cable is required.
High current power lines and low impedance loads are prime sources
of low frequency magnetic coupling. Therefore, the power line leads are
always twisted with their returns.
The twisting effectiveness increases with the number of twist per meter
and decreases with the frequency. It is effective only on balanced or
single point ground system and up to 100 kHz. Above this frequency,
capacitance from the wires to the ground causes ground current to
circulate, which generates common mode coupling. Practical twist pitch
is 30–50 per meter.
The capacitive coupling of the electric field is best avoided by shielding
the culprit, victim or both cables, and grounding the shield (Figure
15.6).
Single point ground is used at low frequency where the wavelength of
the frequency to be suppressed is longer than 10 times the shield length.
FIGURE 15.6 Shield protection from capacitive coupling.
If the culprit wire is shielded, the ground point is at the exit from the
culprit source. If the shield is on the victim wire, the ground point is at
the entry to the victim’s electronics (Figure 15.7).
The shield grounding at high frequency should be at both ends of the
shield. This directs the return current away from the ground and into
the shield, thus decreasing the ground loop currents and common
mode radiation pickup (Figure 15.8 and Figure 15.9).
Single point grounded shield does not reduce the common mode EMI
(Figure 15.10).
A 360 peripheral low-resistance bond should be used via a connector
backshell assembly to assure low-impedance shield grounding in high-
quality shield cable. This bond resistance limits the shield performance.
FIGURE 15.7 Victim cable with a shield grounded at one end.

FIGURE 15.8 A shielded cable with both ends grounded.
FIGURE 15.9 A cable with shield grounded at one end for low frequency and at both ends for
high-frequency signals.
FIGURE 15.10 A shielded cable with only one end grounded.

15.4 EMI/EMC Specifications

The EMI standard establishes the maximum interference an equipment or a
system, such as the power system as a whole, can conduct or radiate into
the load equipment. The EMC standard, on the other hand, establishes the
minimum conducted and radiated interference that an item of equipment
connected to the power system must withstand without being susceptible to
a functional impairment. Obviously, the susceptible level must be higher
than the emission level with comfortable margin, usually tens of dB
depending on the criticality of the equipment in the system.
The EMI/EMC levels are specified in dB, which equals 10 log10 (signal
power/reference power), or 20 log10 (signal voltage or current/reference
voltage orcurrent). Thus, 3 dB doubles the power, or 6 dB doubles the
voltage or current.
The EMI/EMC requirement is specified in two parts, the conducted and
radiated emission levels of the culprit equipment, and the conducted and
radiated susceptivity level of the victim equipment. The general require-
ments for all spacecraft and launch vehicles come from MIL-STD-461. Those
applicable to power equipments are listed in Table 15.1. The spacecraft
specific requirements are also specified. Some such representative require-
ments are shown in Figure 15.11 through Figure 15.17.
The safety margin specified between the emissions levels and the
susceptibility levels are generally:
20 dB for critical equipment

6 dB for all other equipment
In addition, certain self-derived EMI/EMC requirements that have been

established in the past on some spacecraft are listed below as examples of
what may be required of the power equipment:
The EMC specification applies at the load points, not at the PRU
location.
Table 15.1 MIL-STD-461 requirements

applicable to spacecraft and launch vehicle
power equipment
Requirement Conducted (C) Radiated (R)
Emission (E) CE01 RE01

CE03 RE02
CE07 –
Susceptibility (S) CS01 RS02
CS02 –
CS06 –
FIGURE 15.11 Repetitive transient current limits for equipment on a typical 10-kW
communications satellite.
The bus output impedance is estimated excluding the bus loads.

However, for estimating the bus ripples the effect of major loads on the
bus impedance is taken into account.
The main bus power return, the signal return, and the structure are
connected to a single reference ground point. All electrically conductive
FIGURE 15.12 Conducted emission limits for equipment on a typical 10-kW communications
satellite.
FIGURE 15.13 The CE07 limit for main bus voltage for a 3-kW, 120-V, LEO science satellite.
FIGURE 15.14 CS01/CS02 susceptibility requirements.
FIGURE 15.15 CS06 transient susceptibility requirements.

FIGURE 15.16 RS01 B-field susceptibility requirements for a 3-kW, 120-V, LEO science
satellite.
surfaces are connected to the bus structure to form a common satellite

ground cage.
The case ground lead for each component is connected to a pin on the
respective component multi-pin connector. The connector shell is
always conductive and bonded to the case.
Typically, large commercial communications satellites designed for a 5- to

10-kW load and 100-V bus generally have the following self-derived
requirements:
The magnetic dipole moment of each solar array wing is limited to 3 to

5 Am2.
FIGURE 15.17 RS03 E-field susceptibility requirements (14 kHz to 16 GHz) for a 3-kW, 120-V,
LEO science satellite.
The magnetic moment of each battery is limited to 20 Am2, which is

aligned with the pitch axis of the spacecraft. In two-battery systems, the
magnetic dipoles of the two batteries oppose each other to cancel out at
a distance.
The magnetic moment of the whole spacecraft is limited to 5 to 10 Am2.
Any load step is limited to less than 5% of the total load current
capability of the bus, and the dI/dt is limited to a few tens of milliamps
per microsecond. The undershoot and overshoot voltages are kept
below 1% of the bus voltage.
The load is required to withstand the peak-to-peak ripple voltage of 2 to
4% of the bus voltage, depending on the load equipment sensitivity.
The dynamic impedance is limited to 250 m in the 50 Hz to 20 kHz
range and 500 m for frequencies below 50 Hz.
The bus voltage ripple at any load point resulting from the load current
ripple is kept below 2 Vp–p from d.c. up to 50 MHz.
In the class of small satellites, as an example, one 28-V fully regulated

defense satellite bus with load less than 1 kW, the ripples were limited to
500 mVp–p at 36 kHz due to the battery discharge regulator. The user turn-
on or turn-off dI/dt were kept below 10 A/ms for load current changes
exceeding 2 A. Any transients that did not meet this rate had the I–t product
limited to 100 mAs. The bus broadband ripples were required to be 6 dB
below that specified in the EMI/EMC specifications for the spacecraft. User
equipment dynamic impedance was kept below 8 from 3 to 10 kHz.
For EOS-AM type platform delivering a 3000-W load at 120-V bus, the
bus source impedance at the capacitor bank was kept is 0.1 d.c. to
100 kHz, and the ripples were limited to 0.5 Vp–p.
15.5 EMI Suppression Methods

Various methods of controlling and/or suppressing the EMI, in the order of
their importance, are listed below and some discussed further in the next
section:
Minimize the EMI generation in the first place by

Minimizing the current loop area in switching circuits
Minimizing the switching transients’ dI/dt rate in large current
loops
Using snubber capacitors to minimize the voltage transients’ dV/dt
rate
Minimize the E-field and B-field couplings between the culprit and the
victim equipments by
Minimizing the inductive coupling by twisting wires or using

coaxial cables
Minimizing the capacitive coupling by using shields and by
reducing area of exposed metal and keeping it far from the ground
(since C ¼ A"/d)
Divert the energy impinging on the victim equipment to ground by
using
Proper grounding scheme
Faraday shield, single or double
Protect the equipment from the coupled energy by using
L-C filters for conducted EMI
Enclosure shield for radiated EMI
15.5.1 Twisting Wires

The twisting not only reduces the spacing between conductors, it also
produces self-contained B-fields of opposite polarities around the adjacent
spans. Thus, it results in a much lower B-field at far distance due to the
cancellation effect. It does not, however, reduce the local B-field close to the
cable.
Note that twisting two wires carrying a common mode current cannot
null the B-field. The only effective way to eliminate the common mode EMI
is to block the common mode current by breaking the ground loop. An
alternative is to use a common mode choke (balun) discussed in Section
15.5.
15.5.2 Grounding
Two configurations of the single point ground (SPG) used for d.c. or a.c.
power systems up to 20 kHz are shown in Figure 15.18. Configuration (a) is
preferred. Configuration (b) is simple but less effective for high frequency
EMI, since the ground wire inductance increases the Zground between two
pieces of equipment. The grounding at a single point in the system
effectively minimizes the radiated EMI by avoiding ground current loops.
FIGURE 15.18 Single point grounding schemes.

Power return is grounded only at one point. All ground leads are directly
connected to one point, which is then solidly connected to the structure.
Single-point ground reduces the ground inductance. It is used when the
circuit dimensions are less then 3% of the wavelength to be suppressed.
As shown in Figure 15.19(a), a single point ground is used to suppress
low frequency EMI. On the other hand, multiple point ground (MPG)
shown in Figure 15.19(b) is used for radio and higher frequencies EMI
suppression. MPG is used when the circuit dimensions are greater than 15%
of the wavelength at the operating frequency. Each ground lead is
connected to the closest ground conductor. The ground plane minimizes
Zground between two pieces of equipment. In a box having circuit boards
sensitive to different frequencies, a hybrid of the SPG and MPG approaches
may be needed.
15.5.3 Cable Shielding

Military and science satellites having stringent EMI requirements cannot
tolerate any unshielded segment of the harness, or even an open fuse board.
All harness groups — power, signal, and pyro — are kept in separate
bundles with separate shielding. Commercial satellites, on the other hand,
may have fuse boards or some segments of the harness unshielded. All
shields (power and signal) are grounded about every meter to a SPG.
FIGURE 15.19 Single versus multiple point grounds.

The cable shielding alternatives are the braids versus tape and copper
versus aluminum. The braids are used when extreme flexibility is required.
The shield mass as compared to the wire conductor is about 40% for flat
cables and 20% for round cables. In tapes, 2-mil-thick copper vapor
deposited on Mylar or Kapton tapes is widely used. The tape is applied
on the cable with insulation touching the cable, followed by another half-lap
layer on the top (Figure 15.20). The shield mass with such tapes is roughly
80% of that with braids. The bend diameter must be less than about 10 cm to
avoid unwrapping of the tape. Aluminum backed tape can be used, but
being brittle, it needs greater bend radius to keep it from cracking at cold
space temperatures. It does not create oxidation-related problems, as long it
does not come in contact with copper or other dissimilar metals.
15.5.4 Bonding
In fabricating a shielding enclosure or in making ground connections, the
conductors are bonded together by various processes. The bond is
schematically shown on the circuit diagrams by a heavy dot before the
conducting part meets the ground symbol. The solid bond generally
requires having impedance less than a few milliohms. Movable parts may
be bonded using a strap having resistance less than a few milliohms and
inductance less than a few tens of nanohenries. The mechanical and
electrochemical stability is required in the bond. Welding is recommended
for magnetic enclosures, while brazing and soldering are recommended for
conductive enclosures. The corrosion in bonding two dissimilar metals,
such as a copper strap to steel box using steel fasteners, is controlled by
FIGURE 15.20 The cable shielding method using metal backed insulating tape.
selecting two materials in contact with about equal galvanic potentials. This
requires the electric potential differences between them to be near zero.
Otherwise the condensed humidity between the two acts as an electrolyte,
creating a voltaic cell. Copper, silver, platinum, and gold are cathodic,
having positive galvanic potential, while many other commonly used
metals are anodic, having negative galvanic potential.
15.6 Common Mode EMI

Even a small common mode current can cause trouble. We recall that the
induced voltage is given by e ¼ L dI/dt. A typical value of the wire
inductance L is about 80 nH/m. If the rise time of say 100 mA is 1 ns (ESD
type discharge), it induces voltage equal to 80 100/(1 109) ¼ 80 V/m
length of the cable. This can upset many sensitive electronic and control
circuits.
Common mode noise can be detected by a circuit shown in Figure 15.21.
The noise can be minimized by using a balun filter as shown in Figure 15.22.
A transformer or a coupled inductor inserted between components 1 and 2
reduces the common mode propagation. Its magnetic core provides two
large inductances to the common mode current, but provides zero net
inductance to the differential mode current. Therefore, the inserted
impedance in this mode equals two times the self (mutual) impedance of
the transformer plus the self-impedance of the wire with respect to the
ground. Bifilar winding or twisting of wires does not reduce the common
mode noise. This is in contrast to that in the differential mode noise, where
the impedance to the noise signal is equal to the leakage impedance of the
transformer plus the wire, which is very small and can be further reduced
by using bifilar windings and twisted wires.
FIGURE 15.21 A common mode noise detection method.

FIGURE 15.22 A common mode noise filter (balun.)
15.7 Broadband EMI

EMI is classified as narrow-band or broadband, depending on the signal’s
spectral distribution relative to the resolution bandwidth of the receiver
used for the measurement. It is essential to know which type of signal is
being measured to avoid erroneous test results and the wrong interpreta-
tion of data. In this context, the knowledge of the instrument’s impulse
bandwidth is necessary to correctly determine the absolute amplitude of an
emission. Determination of the receiver’s impulse bandwidth can be done
with a series of measurements, but not simply by multiplication of its 3 dB
or 6 dB bandwidth with a constant factor. Only the accurate knowledge of
this parameter allows a meaningful comparison of the measured emission
levels to a limit line and verification of a product’s compliance with an EMI
standard.
When measuring impulsive signals, it is necessary to know the impulse
bandwidth of the interface filter of the EMI receiver or the spectrum
analyzer. Due to the broadband nature of impulsive signals, some of the
spectral components are not contained in the interface filter pass-band at
any one time. As the filter bandwidth becomes wider, more spectral
components are included and the amplitude of the signal at the filter output
increases until the filter is wide enough to enclose the entire signal
spectrum. Filters implemented in EMI receivers or spectrum analyzers are
usually defined by their 3 dB or 6 dB bandwidths, which is suitable for
measuring narrow-band signals. However, this specification is insufficient
for measuring broadband signal because the filter’s amplitude and phase
response affect the results.
A comparison of test data taken with different EMI receivers is very

difficult (if not impossible) to make. A study conducted in Japan in 1996
showed that the difference in measurement results of up to 6.4 dB is
possible when measuring broadband impulsive signals, based on a 3-dB or
6-dB bandwidth specification. This could lead to situations where equip-
ment under test passes or fails a compliance test, depending on the receiver
being used in the tests.
15.8 Commercial Off-the-Shelf Equipment

In recent years, NASA and defense agencies have considered using
commercial-off-the-shelf (COTS) equipment in space for reducing procure-
ment cost. In Europe, interest in using commercial equipment in space has
resulted in the EMC Directive issued within the European Union (EU) and
enforced since 1996. The EU undertook a revision of the Directive in 1998,
and recently has published its final revised draft for review and discussion
among its 25 member countries. The fully and finally revised new Directive
was published in mid-2004. It will become effective 30 months after the date
of publication, possibly with a 2-year transition period. During transition,
products can comply with the old or new Directive as advantageous to the
manufacturers.
The EMC Directive requires commercial manufacturers to use EMI test
methods similar to those required for aircraft, military, and space
equipment. However, this requires acceptable methods to interpret and
compare space system EMI test data with commercial test data, and that is
difficult.
For example, NASA Specification SSP30237A establishes the electromag-
netic emission and susceptibility requirements as well as design require-
ments for electronic, electrical, and electromagnetic equipment and
subsystems designed or procured for use on the International Space
Station. The applicability of the emission and susceptibility requirements
is wholly dependent on the intended location or installation of the
equipment or subsystem within the space station. SSP30237A designates
the equipment or subsystem as an internal or external equipment
depending on the intended location site. Internal equipment applies to
equipment located inside a module or node. On the other hand, external
equipment applies to all equipment located outside the modules and nodes.
The SSP30237A relies on MIL-STD-461 as its framework document, which
has been tailored based upon previous work on known electromagnetic
environment for the space station. The MIL-STD-461 limits were adjusted
based on power requirements and receiver sensitivity as well as margin for
safety and to limit system-level compatibility concerns. The detailed test
procedures for SSP30237A are contained in SSP30238A — Space Station
Electromagnetic Techniques.
NASA has performed tests as per SSP30237A and has compiled EMI test
data on various components. The system designers use this database for
locating and co-locating various components within the space vehicle. If
SSP30237A data cannot be correlated with data obtained from the
corresponding commercial test requirements, then the equipment may
either be placed in an environment where it can be susceptible or can cause
susceptibility without violating the system level performance requirements.
In addition, if correlation is not possible, the system designer may decide to
remove the component from a questionable environment to a benign
environment. This may cause an extensive redesign and impact on cost and
schedule of the entire system.
15.8.1 EMI Requirements in Commercial Equipment

The following are three major EMI test requirements used extensively in the
commercial market.
15.8.1.1 Code of Federal Regulations — Part 15 (FCC Part 15)

This Code addresses conducted and radiated emissions in the frequency
range of 450 kHz to 1 GHz. The FCC divides the EMC environment into
Class A and Class B requirements for office and household environment,
respectively.
15.8.1.2 EMC Directive 89/336/EEC

This Directive is for use in the European Community (EC), which specifies
both emission and susceptibility (immunity) requirements. The EC docu-
ments divide the EMC environment into household and light industrial
environment in one class and industrial environment into another class. The
European EMC requirements rely on a series of test methods developed by
the International Electrotechnical Commission (IEC) and designated as IEC-
1000-X-X, the European Union’s emission and immunity requirements
designated as EN-550XX, and the immunity testing documents designated
as EN-610XX.
15.8.1.3 RTCA/DO-160C
This Standard is used in avionics industry for use in commercial airlines.
The original version of the document was tailored from MIL-STD-461, but
has been updated over the years and contains requirements beyond the
MIL-STD-461 test methods, such as lightening and environment tests. Since
the space station requires equipment similar to that used in commercial
airlines, this standard is important when purchasing navigational and
communications equipment.
15.8.1.4 Other requirements

There are areas where the commercial requirements do not fully cover the
International Space Station requirements specified in SSP30237A. For
example, when comparing CE01 to IEC-1000-3-2, the commercial standard
does not fully cover the frequency range. The CE01 requires the measure-
ment of conducted emissions from 30 Hz through 15 kHz, whereas IEC-
1000-3-2 requires the measurement of the harmonic current emission only
up to the 40th harmonic. Therefore, comparison between IEC-1000-3-2
certified equipment and SSP30237A qualified equipment can only be
applicable if the inter-harmonic emissions are not of concern and if the
equipment does not have local oscillators less than 10 kHz such as in d.c.-to-
a.c. switching mode power supplies. The SSP30237A requires CE01 testing
for d.c. power leads only, whereas IEC-1000-3-2 is not applicable to d.c.
systems. Since NASA may procure COTS equipment designed for 115
or 220 V, 50 or 60 Hz operation, analyzing a.c. equipment against the CE
requirements of SSP30237A to provide a reference point for NASA
engineers is necessary. The scope of that work must include identifying
differences between various measuring instruments as well as components
used in test setups, such as the line impedance stabilizing network (LISN)
and feed through capacitors. Measuring instruments are generally similar
in related specifications. However, there are some differences that require
careful examination. For example, the LISNs called out in the NASA
specifications are different from the LISNs in the DO-160-C specifications
and the LISNs specified in the majority of IEC-1000-4-X requirements are
still of different design.
To narrow the difference, one must include the cable parameters, such as
the length of the cable, height above the ground plane, whether the cable is
run behind, in front or alongside the equipment under test, and other
configuration issues. These all can be addressed by including the cable
capacitance, inductance, resistance and ground conductance (C, L, R, and G,
respectively) for the type and length of the cable used in the actual
laboratory setup. These parameters come from standard high-frequency
transmission line theory that covers the entire frequency range from low to
high frequencies. The parameters can be lumped for short cables, or can be
distributed if necessary for better accuracy in long cables.
15.8.2 NASA Versus Commercial Test Methods

The basic approach must be to analyze corresponding test methods and
limits from SSP30237A and the applicable requirements in the FCC
regulations, the EC EMC requirements, and/or RTCA/DO-160C. Then it
is necessary to develop fitting transfer functions based on the comparison of
the two sets of data for translating one of the commercial data into the
NASA SSP30237A electromagnetic environment envelope.
We first consider the NASA CE01 and CE03 requirements, and the
corresponding commercial requirements, which are listed in Table 15.2. The
table shows how each of the SSP30237A requirements are analyzed against
the various commercial standards. The analysis must be performed by
evaluating various instrumentation, measurement techniques, and applic-
able limits for each of the test methods. This includes evaluations of
different line impedance stabilization networks (LISNs), feed-through
capacitors, spectrum analyzers, oscilloscopes, and other test instrumenta-
tions that are used in different test methods, which could lead to differences
between measured quantities. For example, CE01 and CE03 require the use
of a 10-mF capacitor on all power lines and measuring the RF current
directly on the power line using an RF current probe and EMI receiver. FCC
Part 15 and EN55022 each require an LISN on the power line under test and
the measurement of voltage across a resistor in the LISN. Therefore, the two
limits are in different units and cannot be simply compared. This requires
an analysis to determine the relationship between the measured quantities
and thereafter for developing the transfer function between the two.
15.8.3 Testing COTS for Space Compliance

One of the major concerns in comparing different EMI test requirements is
the test setup. For instance, the majority of the C-1000-4-X requirements
specify that the equipment under test be placed on an insulated surface
approximately 80 cm from the ground plane, whereas SSP30237A requires
the unit to be placed approximately 5 cm above the ground plane. This
would cause a major difference in test results that must be carefully
investigated. For the purpose of conducted emission testing, the cable can
be modeled as a transmission line having C, L, R, and G parameters in an
equivalent T network.
Table 15.2 NASA versus commercial EMI

requirement
SSP 30237A FCC ECLIPSE DO-160C

requirement requirement requirement requirement
CE01 – IEC-1000-3-2 –
CE03 Part 15 EN55022 Section 21
CS01 – IEC-1000-4-13 Section 18
CS02 – IEC-1000-4-6 Sec. 18 & 20
CS06 – IEC-1000-4-5 Section 17
RE01 – – Section 15
RE02 Part 15 EN55022 Section 21
RS02 – IEC-1000-4-8 Section 19
RS03 – IEC-1000-4-3 Section 20
The standards being considered use very similar measurement tech-

niques. They use measurement receivers to measure voltages, and current
probes to measure currents. They also use current probes to measure
current across a known resistor to determine voltage with respect to a given
frequency range. Conversion factors are then applied based on calibration
of the equipment used to perform the measurement and then a comparison
is made to a given emission limit. However, each method has various
differences in the approach for maintaining consistency within testing. The
IEC-1000-4-X series of documents recommend the use of coupling/
decoupling networks to match the typical installation impedance of various
cables to the test cable configuration. The idea here is to minimize the
distortion between the laboratory measurement and a typical field
measurement. The SSP30238A relies on proper calibration of the equipment
being used to perform the measurements. The values obtained from the
various test specifications require adjustments in order to compare limits
and test results. These adjustments need to be handled on a test method
basis.
As for the frequency range, the test requirements must be reviewed to
determine if they overlap the entire specifications. If not, the differences are
taken into account as needed.
The simulations and the laboratory results on the four test setups have
shown that the type of cable and its configuration and layout used in the
actual setup make significant difference in the results. A parametric study
made with PSPICE simulation showed sensitivity of the results to one more
parameter, the cable resistance, in determining the peak of the response.
This is expected, as low resistance results in a high-Q circuit having high
and sharp peak at the resonance frequency.
A seemingly minor variation in the laboratory test setup, in particular the
cable length, routing and types such as coaxial, twisted or randomly laid
out, and the cable resistance makes significant difference in the correlation
between any two test setups. This indicates that for properly comparing
similar specifications, a careful and robust analysis of the similarities and
differences in the setups is required.
Since the cable length and layout are equipment specific, it is extremely
difficulty, and may perhaps be misleading, to correlate one test method to
another in space systems, unless the cable length, layout and types are
specified in the test standards.
15.9 Electromagnetic Pulse/Nuclear Threat

The electromagnetic pulse (EMP) is a man-made threat that arises from a
nuclear detonation around the spacecraft. It can destroy satellites in low
and mid Earth orbits. (Figure 15.23). A nuclear burst can upset electronic
systems hundreds of miles away. Therefore, it is usually included in the
FIGURE 15.23 A nuclear burst in LEO altitude can destroy many satellites.
defense spacecraft power system requirements. The nuclear threat require-

ment is generally included on defense satellites in any orbit. For example,
it was included in the original GPS satellites, which were procured
exclusively for defense use of the U.S. Air Force until mid-1990s. It was
removed from GPS-IIF and later versions, which were made available for
commercial use also.
The EMP is basically the burst of electron flow resulting from the gamma
radiation. It is an intense time-varying electromagnetic field of very short
time pulse in the nanosecond to microsecond range. The radiation and
propagation of the electromagnetic waves in space obey Maxwell’s
equations. The wave carries little energy to the equipment at far distance,
but is deposited in a fraction of a microsecond, making the rate of energy
deposition (power) extremely high. As a result, it can cause the following
effects in the power system:
Radiation dose degradation, particularly on solar array

Common mode noise in power system equipments
Temporary upset of sensitive electronic parts
Semi-permanent upset (latch-up), requiring power shut-off and restart
Permanent damage to sensitive semiconducting devices
The most dominant damage comes from the charged particle radiation
dose on solar array and sensitive electronic parts. A high-altitude burst test
named STARFISH conducted in 1962 produced a large number of beta

electrons, which became trapped in the Earth’s magnetic field. Its yield was
1.4 MT at 400 km altitude above Johnson Island in the South Pacific. The
intense man-made radiation belt created by this burst ended the functional
life of seven satellites within 7 months, primarily by solar cell degradation
and command decoder failures. This radiation lingered for 10 years until
1972. Today, since LEO space is crowded with satellites, a severe nuclear
detonation can cause the loss of hundreds of satellites resulting in
enormous economic loss and national security concerns. It is estimated
that a 10-kT burst over Japan at 150 km altitude can shorten the life of many
LEO satellites such as Iridium and NOAA from 5 years to 5 months. The size,
magnitude and duration of the nuclear radiation is calculated by the Air
Force computer code that models the nuclear burst in space and calculates
its effects on satellites. Using this unique code, the Air Force and the
Institute to Defense Analysis predict that a 20-kT burst over Iraq would
accumulate a radiation dose of 100 krad on Iridium-type LEO communica-
tions satellites in about 10 months, and the enhanced radiation would
persists for several years. In some cases, the electron level would remain
over 100 times higher than the natural level even 2 years after the burst.
This makes the rapid replacement of destroyed satellites infeasible because
they will also fail permanently due to the lingering radiation. For this
reason, the replacement must wait and still use a much thicker radiation
shield.
The EMP effects can be studied by quantifying the energy deposited on
the victim equipment, which is the time integral of power. The power in an
electromagnetic wave is given by the Poynting vector, which is the cross-
product of E and H fields. The energy is therefore,
Z Z
energy ¼ power dt ¼ ðExHÞ dt ð15:8Þ
We then think in terms of energy-coupling to various systems compo-

nent. Wires exposed to EMP can couple damaging amount of energy into
the system. The degree of coupling is a complex function, and is outside the
scope of this book. However, it depends on the cable configuration,
geometry of the cable run, and its orientation with respect to the
electromagnetic field vectors. Once the energy is coupled to the wires, the
high frequency transmission line theory applies for propagation along the
wires. A part of the coupled energy is absorbed by the load (victim
equipment). The remaining energy is reflected, the amount of which
depends on how the load resistance matches with the characteristic
impedance of the incoming wire. A good conductive paint on the surface
can reduce the coupling when the EMP energy impinges directly on the
equipment’s enclosure surface, rather than entering via cables.
FIGURE 15.24 Main bus voltage transient requirement following nuclear burst on a 28-V
fully regulated MEO communications satellite.
As an example of the power system requirement arising from a nuclear

threat, the main bus voltage transient following a nuclear burst on a fully
regulated 28-V MEO communications satellite was required to be kept
within the envelope of Figure 15.24. In another example, the bus voltage of a
28-V fully regulated GEO communications satellite was required to be
within the envelope shown in Figure 15.25. All loads must survive and
return to the operating state after such abnormal transient is over.
FIGURE 15.25 A main bus voltage transient requirement following a nuclear burst on a 28-V
fully regulated GEO defense communications satellite.
Chapter 16
Electrostatic Discharge
16.1 Introduction
Electrostatic charge accumulates on any probe in space regardless of being
inside or ouside the Van Allen belts. The accumulated charge raises the
electrical potential of the probe, causing a current flow from the probe to the
surrounding plasma (Figure 16.1). If the current cannot maintain a balance
of charge, the probe potential will keep rising until arcing takes place. Such
a problem can occur particularly when the spacecraft leaves or enters an
eclipse when the interaction with space suddenly changes. Arcing can also
arise due to differential charging of insulated surfaces electrically not
connected together. Each isolated surface acts as an independent probe in
space, which floats to a potential that results in no net current to or from the
space plasma. That potential is of the order of the plasma kinetic energy.
Insulating surfaces do not distribute surface charge, hence can charge up to
a much higher differential potential until discharge takes place by way of
arcing and/or flashover. The charging, the subsequent electrostatic
discharge (ESD), and their remediation in GEO and in LEO are significantly
different as discussed below.
16.2 ESD in GEO

In GEO’s high-energy plasma environment, the major concern is the
differential charging of different parts of the spacecraft, leading to high
electric fields and subsequent arc discharges between components. The
FIGURE 16.1 A spacecraft floating in space plasma.
421
ambient thermal plasma current densities are insufficient to discharge

spacecraft surfaces rapidly. However, the electron charge accumulated on
insulating surfaces increases the electric field to adjacent conductors above
the breakdown level, leading to arcing. The resultant arcing currents
traveling through conductors can upset electronic components and induce
spurious signals. A common design solution for GEO is to coat all outside
surfaces of the spacecraft with conducting materials. This prevents
differential charging by distributing the charge over all surfaces and
equalizing their potentials. Coating materials having surface resistivity less
than 5 k/square is considered adequate to eliminate differential charging.
On a solar array, an indium oxide type conductive and transparent coating
is applied on the coverglass when required.
16.3 ESD in LEO

In LEO, because of high thermal plasma current density, surfaces do not
ordinarily collect much differential charge. The major concern in LEO is the
absolute charging of spacecraft surfaces with respect to the surrounding
plasma. Normally, the collected plasma current bleeds off the absolute
potential rapidly. However, some spacecraft impose differential charging
on themselves by using distributed high voltages. If conductors at different
voltages are not exposed to space plasma, all spacecraft surfaces float
within a few volts of the surrounding plasma potential. However, if
conductors of similar area at different voltages are exposed to space plasma,
a rule of thumb is that the most negative surfaces float negative of the
plasma about 90% of the total voltage difference between the surfaces. This
is true, for instance of solar arrays, where interconnects between solar cells
or cell edges are exposed to the plasma, yet operation of the array depends
on the voltage distribution. Spacecraft designs not using solar array also
place different surfaces at different potentials. High voltage spacecraft in
LEO environment have experienced significant interactions with their
environment. The Upper Atmospheric Research Satellite (UARS), with the
array voltage reaching 100 V when coming out of eclipse, was seen to
charge to 90 V negative of its surrounding plasma because of plasma
interactions with the solar array. This made data interpretation from some
of its instruments difficult. The International Space Station floats at 140 V
negative of its surrounding plasma. Because this make it liable for arcing
and sputtering damage, a plasma contactor to control the floating potential
is incorporated in the ISS design. The payload of SP-100 space nuclear
power system as designed would float at about 100 V negative of LEO
plasma because its structure ground is discontinuous between the power
system and the payload. Locally, however, insulating surfaces would
charge to only a few volts negative of the surrounding plasma. Therefore,
arcs in LEO can occur from conductor to insulator junctions (including
Electrostatic Discharge 423
Table 16.1 Electrostatic charging in LEO and GEO compared
Parameter LEO GEO
Plasma environment Low energy High energy

E 0.1 eV E 10,000 eV
Potential difference due to V B high V B negligible
spacecraft motion in the Earth’s ( 0.3 V/m) ( 0)
magnetic field
Potential difference on insulating 1V > 1000 V
surface
Potential difference on conductors 5-5000 V 1V
Remedy Cover exposed Drain surface charge off
conductors with insulators using conductive
insulation coating coating
holes in the cable insulation) when the conductor is highly negative of the
surrounding plasma or from anodized or other dielectric surfaces when the
underlying conductor is at a negative potential higher than the dielectric
breakdown strength of the coatings. Micrometeoroid and/or debris impact,
dielectric breakdown, etc may expose conductors. Arc currents may flow
out into the surrounding plasma, with the return currents distributed over
wide areas of other surfaces of the spacecraft. Arcs also can occur through
the plasma between closely spaced conductors at differing voltages.
Conducting surfaces highly negative of the plasma will attract high-energy
ions, and are liable to sputtering. Nearby surfaces may acquire a sputtered
conducting coating, changing their electrical, optical, and thermal proper-
ties. At high positive potentials, electrons are collected, leading to localize
heating and significant power drain.
The LEO has low energy plasma environment (E 0.1 eV), but high
electric field generated in the conductor moving in the Earth’s magnetic
field (V B 0.3 V/m). This can add up over a long conductor and may
cause arcing. The remedy is not to expose bare parts of large conductor, and
apply insulating coating on all large conductors in LEO. In GEO, on the
other hand, conductive coating and grounding the structural parts together
at one point equalizes the potential of the parts and eliminates the ESD.
Thus the causes and remedies of EDS in LEO and GEO are quite different as
compared in Table 16.1.
16.4 Dielectric Breakdown

The electrostatic discharge from one surface to another involves dielectric
breakdown. It is not like lightening, as no discharge can ionize the vacuum.
However, the rise time is much faster than lightening, in the order of a
nanosecond as shown in Figure 16.2.
FIGURE 16.2 Typical ESD rise and fall times.
Ground tests for the International Space Station found that dielectric
coatings often break down at 100 V in LEO plasma. Rated dielectric
strengths and the strengths measured were not consistently reached before
breakdowns occurred in simulated space plasma. It is suspected that this is
due to porosity of the coatings, allowing plasma ions to reach much closer
to the underlying conductor than the nominal dielectric coating thickness.
Even coatings with rated dielectric strengths of much more than 100 V
seemed porous enough to the plasma ions to have real dielectric strengths
of less than 100 V. Thus, for surfaces that must be exposed at high negative
potentials, it is important that dielectric coatings be used. Or, its outer
surfaces are strong enough to stand off the full system voltage in plasma.
Care must be used in selecting surface coatings of high dielectric strength.
In particular, the chromic acid anodization commonly used on aluminum
exposed to space often have insufficient strength to stand off 100 V in
plasma. Thickening the coating usually influences its thermal properties, so
a thicker coating may not be a good answer. Sulfuric acid anodization
coatings have greater dielectric strengths, but very different thermal
properties. For LEO missions, the sulfuric acid anodization coating is a
preferred remedy where thermal control is not an overriding concern. On
all other exposed high voltage surfaces, AO-protected aluminized Kapton
of 1000-V dielectric strength is preferred.
16.5 ESD Entry Points

The following are the ways the ESD energy can enter the sensitive electronic
equipment and the ways of avoiding the potential damage.
16.5.1 Strike Through Gaps

The ESD striking to the enclosure’s outside wall causes voltage to rise
locally until the charge distribution equalizes. Small gaps from the
enclosure to the inside circuit boards may breakdown and arc under the
resulting high voltage. A remedy is placing a grounded metal barrier
between the striking point and the vulnerable board. The air breaks down at
about 3000 V/mm, and ESD specifications require testing up to 15,000 V.
Therefore, all air gaps less than 5 mm are critical spots during ESD testing
on ground.
16.5.2 Puncture of Solid Insulation

Most power electronic equipment use solid insulation of puncture strength
7 to 8 kV/mm. For 15-kV withstand, all insulation less than 2 mm thick
must be carefully evaluated for a possible strike during ESD tests.
16.5.3 Indirect Interaction

ESD current flowing through the enclosure couples energy into wires,
which is then conducted to the circuit boards inside. Large dE/dt and dB/
dt associated with the ESD-induced transients cause electromagnetic
radiation, which in turn propagates to the inside circuits. This must be
analyzed along with the withstand level for every component with ample
margin in order to avoid circuit board damage in case of an ESD failure
during tests or in operation. A safety factor of 10 is considered adequate for
ESD-protection.
16.6 Effect of ESD

The ESD failure can affect the spacecraft equipment in the following ways:
Human safety
Operational disturbance
Component damage
The operational disturbances and component failures can be as follows:
Noise in radio communication and flicker in video display

Switching of logic gates
Component failure by over-voltage on semiconductor junction insula-
tion
Ignite spacecraft EEDs (critical failure)

Ignite fire in hazardous areas (Class 1, explosive fuel vapors)
16.7 ESD Mitigation

The primary options for arc discharge mitigation in the spacecraft are:
Controlling the spacecraft floating potential in order to eliminate the

initiating arcs
Limiting the bus voltage and/or string current that could feed an arc to
level below the threshold for sustained arcs
Use of proper grounding scheme
The above options can be implemented as follows:
Increase conducting area on the spacecraft, thus lowering the floating

potential.
Use plasma contactor, thus lowering the spacecraft potential (as done in
ISS).
Lay out the solar array strings to reduce the potential difference
between the adjacent cells below 60 V.
Use grout between cells to insulate the gaps between cells.
Limit the string current below 1 A, and use diode isolation.
In circuits using transformers, one of the windings and the chassis are
always grounded to the spacecraft structure ground. The core, on the other
hand, is kept floating with respect to the chassis by maintaining some gap
between the chassis and the core. It is possible that the charge from the
spacecraft can enter the core by flashover from the chassis, but cannot enter
the winding unless the major winding insulation punctures. The floating
core design is required by military standards, but the issue is debatable as
to its superiority over the grounded core.
The electrostatic discharge effects are analyzed and controlled in the
following steps:
Examine every surface of the spacecraft. If the electrical resistivity of

any surface is greater than 10 k/square, apply conductive coating
(ITO) if practical on that surface.
Apply grounded shield on all cables outside a Faraday cage.
Determine the expected ESD-induced transient voltage and the with-
stand level for every component. If the analysis shows a safety factor of
less than 10, ESD testing is performed to establish the survivability. If
the component does not survive with acceptable margin, additional

shielding and/or filters are added in the design.
An extensive study by Ferguson1 on environmental effects on spacecraft

gives the following recommendations for the power system design:
Wherever there are discontinuous structure potentials of more than

100 V on the spacecraft, they must have an insulating material of
sufficient thickness between them.
Attention must be paid to geometry and materials where there are
exposed conductor–insulator junctions in LEO. In particular, conduc-
tors known to have plasma arcing threshold higher than the conductor
potential relative to the surrounding plasma must be used. Plasma
testing to identify such materials is desirable.
There is some evidence that arc rates are higher for materials that
adsorb water and/or other volatile materials. Such materials should be
avoided near conductor–insulator junctions, such as the solar cell
edges.
Geometry can influence plasma arcing as well. The geometry that helps
prevent ions from entering the conductor–insulator region is preferred.
For LEO missions, the structure and payload surfaces at high negative
potentials should avoid exposed conductors, or at least exposed
conductor–insulator junctions in order to avoid the same sort of arcing
considered above.
For GEO missions, fully conductive surfaces should be used every-
where to avoid differential charging.
Implementing a proper grounding scheme is important in controlling the

surface potentials in the spacecraft structure relative to the surrounding
plasma. A negatively grounded power system collects sluggish ions, and
even the large surface area may only shift the floating potential away from
the most negative system voltage by a few volts. On the other hand,
positively grounded high voltage solar array in LEO may control its floating
potential. However, space-qualified high voltage electronics for this polarity
is scarce, slow, and may be prohibitively expensive.
16.8 ESD Control in Solar Array

The entire array (the cells, coverglass and substrate) is designed to perform
at full specification under the effects of electrostatic charging. In GEO, this
often requires indium–tin–oxide (ITO) coating, which reduces the solar
array power output by a couple of percent or more. For example, 700
Angstroms (0.07 mm) coating reduces the cell power output by approxi-
mately 4%.
Regarding the grounding and conductivity of the array, the following are
noteworthy:
The back of the array is grounded to the aluminum honeycomb core of

the panels in at least three places on each panel. The honeycomb from
each panel is then connected to a ground lead from the boom.
The panel end and the SAD end of the boom are both electrically tied
together and connected to the spacecraft structure via a set of two 20-k
bleed resistors, as shown in Figure 16.3.
If a conducting tape is used on the boom, it is grounded by over-
wrapping the boom with the bus wire, which is grounded to the
array structure in the same way as the boom.
In an early design stage, the best option is the array layout to
eliminate closely spaced cells with large voltage gradients.
As a protection against ESD, the following design and/or construction
features are generally implemented on the solar array:
The back surfaces of the panels are coated to have electrical resistivity
of less then 100 M/square.
The booms have conducting outer surfaces. The use of non-conducting
tapes or paints over the conducting boom surfaces is minimized. The
conducting tapes used on the boom are grounded.
Blankets and loose surfaces are grounded to the array structure by
means of the appropriate number of bonding straps.
Snubber capacitors are used in the SAD before the slip rings and/or in
the PRU.
FIGURE 16.3 A solar array grounding scheme for ESD protection.

16.9 Part Sensitivity

The ESD results in high potential (up to 20 kV) of short duration (100 to
1000 ns). Energy levels depend on the surface areas involved. The current
injected into the spacecraft depends on local structure configuration and
impedance of the propagation paths. The surge impedance of the biaxial
and twin-axial cables is 50 to 60 . For twisted shielded wire pairs, it is
about 60 for AWG 20 wires and about 120 for AWG 30 wires.
Although the energy level in the ESD is small, typically in several
millijoules, the energy required to damage sensitive components such as
CMOS, FET and pin diodes is even smaller (microjoules to millijoules).
Robust component such as magnetics and vacuum tubes can withstand
much higher energy levels. On-line components are more vulnerable than
off-line components.
The ESD sensitivity is generally measured in terms of the voltage the part
rises to under the charge. MIL-HDBK-263 lists ESD sensitive parts in terms
of the ESD voltage ranges. The list is in three classes, class I parts sensitive
to less than 1000 V, Class II sensitive to 1000 to 4000 V, and class III sensitive
to 4000 to 15,000 V.
Table 16.2 lists the ESD damage threshold of some electronic piece parts
commonly used in the spacecraft power system. The upset usually takes
about 10% of the damage threshold voltage.
ESD specifications require testing up to 15,000 V. An ESD simulator, such
as shown in Figure 16.4, is used for this purpose. It basically consists of
charging a capacitor from a d.c. source via high resistance (slow charge) and
then discharging to the equipment under test via low resistance (fast
discharge).
Table 16.2 ESD damage

threshold levels of power
electronic devices
ESD damage
Device threshold (Volts)
MOSFETs 10–200
GaAsFET 50–2000
JFETs 150–7000
Bipolar transistors 300–7000
Laser diodes 100–2000
PIN diodes 200–1000
Schottky diodes 300–2500
Schottky TTL 500–2500
CMOS 150–3000
Operational amps 200–2500
Film resistors 300–3000
FIGURE 16.4 ESD simulator for typical circuit board testing.
16.10 Safety During Tests

While testing sensitive electronic parts, particularly the circuit boards on a
test bench, care must be exercised for protecting the parts from ESD caused
by human body charging from the static charge in air. The body potential
can rise significantly as listed in Table 16.3. Sensitive parts can get damaged
if picked up without draining the charge to the ground via conducting
bracelet worn by the test personnel. ESD training and certification is,
therefore, a part of the test floor standard practice.
Table 16.3 Typical electrostatic voltages on human body
Means of static generation Electrostatic voltage
10–20% 65–-90%
Humidity Humidity
Walking across carpet 35,000 1500
Walking over vinyl floor 12,000 250
Working at bench 6000 100
Opening vinyl envelop for work instruction 7000 600
Picking up common polyvinyl bag from bench 20,000 1200
Working on chair padded with polyurethane foam 18,000 1500
Source: MIL-HDBK-263.
Reference
1. Ferguson, DC., Interactions between spacecraft and their environment, in
Proceedings of the Intersociety Energy Conversion Engineering Conference, 1992,
Paper No. 0092-X.
Chapter 17
Reliability and Derating
17.1 Introduction
The economic success of any system depends on its availability for the
intended use. The overall availability has two components: the reliability
and the maintainability. For satellites once launched, the availability
depends only on the reliability, as the maintenance is generally beyond
reach.
The system reliability is based on probability of numerous components
working together successfully over the entire mission duration. The
reliability of a given competent is derived from tests on a large number
of identical components under actual operating conditions. In the spacecraft
industry, an actual flight history is available on only a small number of
satellites. This makes the reliability estimate difficult, sometimes debatable,
or even questionable. However, the estimate is extremely useful, at least on
a relative basis, in identifying and correcting weak links in the chain of
reliability. The perspective of the system reliability developed during the
process is perhaps more important than the final reliability number.
The probability of failure of a component comes from four sources:
Random failure, which is just a matter of chance.

Wear-out failure, which is well understood and accounted for in the
design.
Design failure, which occurs if operating stress exceeds the design
limit under an anticipated or unanticipated disturbance. It is avoided
by allowing adequate design margin.
Manufacturing failure, which depends on the consistency and
sensitivity of manufacturing processes and the quality assurance
procedure in place. Two manufacturing problems that can impact the
power system reliability are:
Nicked wire that may occur during removing the enamel insulation
from wire strands. This degrades the wire both electrically and
mechanically, and may lead to a short or open circuit.
Protruding screws in a box, puncturing the surrounding wire
insulation and shorting to another piece part. It may also cause
dielectric stress concentration, leading to arcing under ESD.
431
The component reliability estimate is based on its failure rate in time

(FIT), also known as the hazard rate, which is quoted in number of failures
in million hours. The failure rate of any component varies with age and
follows the well-known bathtub shape shown in Figure 17.1(a). The initial
high failure rate — the infant mortality rate — is caused by manufacturing
defects. The rate decreases rapidly with time in use, particularly in
electronic components. Therefore, newly manufactured components are
placed in operation for some time in a process called ‘burn-in’ to weed out
infant failures. For power electronic components, most early failures occur
in the first 12 h of burn-in operation.
A low constant rate of failures for a long time in service makes the flat
stable part of the bathtub curve. In this region, the failures are random in
nature, which are treated probabilistically in engineering designs. The
rising failure rate near the end of life is due to wear out mechanisms that are
generally well understood and accounted for by the design engineer.
The flat part of the curve is considered to be the useful life of the
component. Only the random failure rate is used in developing the
reliability estimate over the useful life. Electronic components with no
moving parts have a relatively longer flat part of the tub. On the other hand,
mechanical components with moving and physically wearing parts have a
relatively shorter flat part as shown in Figure 17.1(b).
FIGURE 17.1 Component hazard rate versus time in service.

Reliability and Derating 433
17.2 Random Failures

If a large population of the component is put to test simultaneously, the
number of random failures in a given time duration is constant regardless
of the beginning or end of life. In other words, under random (chance)
failure, if the component has not failed at a given time, it is as good as new.
However, under the accumulative nature of chance failures, the probability
of component still working decreases exponentially with laps of time. The
reliability is, therefore, an exponential function of time, known as the
Poisson distribution function, expressed as
RðtÞ ¼ e t ð17:1Þ
where ¼ failure rate, i.e., the number of failures in unit time, also known
as the hazard rate. The probability of failure in time t, i.e., unreliability, U,
is given by
dU
UðtÞ ¼ 1 RðtÞ ¼ 1 e t and ¼ et ¼ fðtÞ ð17:2Þ
dt
The probability of component failing during a time interval it at any time t

is f(t)it. For this reason, f(t) is known as the failure probability density
function. For a constant hazard rate shown in Figure 17.2(a), f(t) is an
exponentially decaying function as shown in Figure 17.2(b). The component
unreliability is the area under f(t) curve from time 0 to t, and the reliability is
the shaded area from t to 1.
FIGURE 17.2 Constant hazard rate and failure probability density function versus time in
service.
The expected mean time between failures (MTBF) in a large population of

a given component is
MTBF ¼ 1= ð17:3Þ
That is, if numerous identical parts were put to test under identical
conditions, then the mean time between failures would be 1/, always the
same because is constant. The probability of component working at any
time t in terms of the MTBF is therefore,
RðtÞ ¼ exp ðtÞ ¼ exp ðt=MTBFÞ ð17:4Þ
and the probability of component not working at any time t is
UðtÞ ¼ 1 RðtÞ ¼ 1 exp ðtÞ ¼ 1 exp ðt=MTBFÞ ð17:5Þ
17.3 Wear-Out Failure

Components such as bearings, solenoid valves, frequently operating relays,
and battery electrodes wear out. The life under wear-out failure mode is
expressed in terms of the mean life and the standard deviation . If f(t)
represents the probability of failure in time t, it follows the normal bell-
shape probability density expressed by the Gaussian distribution function
( )
1 1 t 2
fðtÞ ¼ pffiffiffiffiffiffi exp ð17:6Þ
2 2
The probability of a wear-out failure at time t is the integral of f(t) from 0 to

t. The reliability at time t, or the probability of not failing in time t, is 1
minus that integral, i.e.,
Zt ( )
1 1 t 2
RðtÞ ¼ 1 pffiffiffiffiffiffi exp dt ð17:7Þ
2 2
0
17.4 Fundamental Theorems of Reliability

The reliability of any complex system having components in series, parallel
or any combination thereof, is determined by the following two funda-
mental theorems of reliability:
1. If A and B are two independent events, with their individual

probability of occurrence P(A) and P(B), respectively, then the
probability of both A and B occurring is
PðAÞ PðBÞ ð17:8Þ
2. When there are n identical components each with reliability R, then the
probability that there will be exactly k components working at any time
t is given by the binomial theorem,
n!
Pk ¼ Rnk ð1 RÞk ð17:9Þ
ðn kÞ ! k!
Using the first theorem, the total reliability of a component under all failure
modes is the product of the reliability under each mode separately. For
example, the overall reliability under the four failure modes identified
above is
Ro ¼ Rr Rw Rd Rm ð17:10Þ
where the four reliabilities on the right-hand side account for the random,
wear-out, design, and manufacturing failures, respectively. The component
is designed with an adequate design margin. The reliability under wear-out
mode is made almost equal to 1.0 by making the mean wear-out failure time
much longer than the design lifetime. Eliminating potential failures using
consistent manufacturing and quality assurance procedures leaves only the
random failures to account for in the total reliability estimate, and so is done
in practice.
17.5 Series–Parallel Reliability

Two or more components are said to be working in series in the reliability
sense if all components have to work for the whole assembly to work
successfully. On the other hand, two or more components are said to be
working in parallel in the reliability sense if only one component has to
work for the assembly to work successfully. The series–parallel working of
various components for reliability of an assembly is not to be confused with
the series–parallel connection in the electrical circuits. Electrically parallel
components can be in series in the reliability sense, and vice versa.
If two components having individual reliability R1 and R2 are working in
series as shown in Figure 17.3, then using the theorem (17.8), reliability of
the assembly is given by
R ¼ R1 R2 ¼ exp ð1 tÞ exp ð2 tÞ ¼ exp ½ð1 þ 2 Þ=t ð17:11Þ

FIGURE 17.3 The assembly of two series components.
Or, the unit failure rate ¼ 1 þ 2, and
1
MBTF ¼ ð17:12Þ
1 þ 2
If two components having individual reliability R1 and R2 were working

in parallel as shown in Figure 17.4, then the assembly would not work only
if both were not working. That is, U ¼ U1 U2, and R ¼ 1 U, or
R ¼ 1 U1 U2 ¼ 1 ½ð1 e1 t Þð1 e2 t Þ ð17:13Þ
By using some algebra, it can be shown that for such unit,
1 1 1
MTBF ¼ þ ð17:14Þ
1 2 1 þ 2
17.5.1 Examples
Applying the series–parallel reliability principles presented above we
examine a few examples below:
The reliability of an assembly made of two series component, each

having a reliability of 0.99, is 0.992 ¼ 0.9801.
Another assembly made of the same two component in parallel has the
reliability of 1 0.012 ¼ 0.9999.
A system of 15 series component, each having 0.95 reliability, has the
overall reliability of 0.9515 ¼ 0.4633.
If a system made of 150 components in series, each having equal
reliability R1, has the overall system reliability requirement of 0.99, then
the minimum requirement on the component reliability is derived from
equation 0.99 ¼ R1150, which gives R1 ¼ 0.999933.
FIGURE 17.4 The assembly of two parallel components.

Practical systems are made with some components in series and some in
parallel. For example, we calculate the reliability of the overall system
shown in Figure 17.5 as follows. Dividing the system in sub-assemblies A,
B, C and D, we determine their unreliabilities and reliabilities as
UA ¼ (1 0.90) (1 0.90) (1 0.90) ¼ 0.001, and RA ¼ 1 0.001 ¼ 0.999
UB ¼ (1 0.80) (1 0.80) (1 0.80) ¼ 0.008, and RB ¼ 1 0.008 ¼ 0.992
UC ¼ (1 0.992 0.99) (1 0.97 0.98) ¼ 0.000885, and RC ¼ 1

0.000885 ¼ 0.999115
and the sub-assembly D, having only one part, has RD ¼ 0.998.

The overall system reliability, Rs is therefore,
Rs ¼ RA Rc RD ¼ 0:999 0:999115 0:0998 ¼ 0:99612
For an assembly such as a battery and solar array made of numerous

identical series cells having equal reliability, Figure 17.6 depicts the
assembly reliability versus number of components. The rule of 72 is
noteworthy here:
The reliability of the assembly reduces to one-half when the product of

the number of series component and the percent unreliability of each
component equals 72.
The assembly reliability increases with the number of parallel components,

but with rapidly diminishing rate of return. This is seen in the last column
of Table 17.1, which gives the percent improvement in reliability by adding
parallel components starting with a single unit. It is given by (Rn R1)/R1,
FIGURE 17.5 The assembly and sub-assemblies of series and parallel components.
FIGURE 17.6 System reliability versus the number of series component and the reliability of
each component (rule of 72.)
where Rn is the reliability with n component in parallel, each having

the reliability of R1. If each component has 80% reliability, then two
components in parallel would improve the assembly reliability by 20%,
three by 24%, and it would take infinite many components in parallel to
improve by 25%. This clearly shows that using two components signifi-
cantly improves the reliability, and more than three component in parallel is
almost fruitless.
The cross-strapping of solar array sections shown in Figure 17.7 is often
used to improve the array reliability. It is also used for other assemblies if
required. The improvement is quantified here for a system shown in Figure
17.8. It is made of two identical parallel paths, each having four series
components with reliability indicated in the component boxes. In config-
uration (a) without cross-strapping, the reliability of the assembly is
Ra ¼ ½1 ð1 0:99 0:98 0:97 0:96Þ2 ¼ 0:991
Table 17.1 Diminishing rate of return in system reliability using

parallel units
Improvement in
Number of Overall system Incremental system reliability over single
components reliability reliability component (%)
1 0.800000 Reference Reference

2 0.960000 0.160000 20.00
3 0.992000 0.032000 24.00
4 0.998400 0.006400 24.80
5 0.999680 0.001280 24.96
6 0.999936 0.000256 24.99
FIGURE 17.7 Cross-strapping of solar array sections.
With cross-strapping, the assembly reliability increases to
Rb ¼ ð1 0:012 Þ ð1 0:022 Þ ð1 0:032 Þ ð1 0:042 Þ ¼ 0:997
The cross-strapping thus reduces the assembly failure rate to one third,
from 9 to 3 in 1000.
17.6 Redundancies
Redundancy is obtained by placing more units in parallel than necessary for
the specified operation. Redundancy has a value only if the failure is
instantaneously detected and acted upon to switch over to the backup unit.
It is therefore important to continuously monitor the unit performance.
Deviation in one or more performance parameters can be interpreted as
FIGURE 17.8 System without and with cross-strapping.

failure. It is needless to say that the failure detector must be of high

reliability to match the targeted reliability of the unit it monitors.
The redundant units can be operated in two ways as shown in Figure
17.9: (a) active (hot) all the time, or (b) dormant (cold) until the primary unit
fails and then switched to the active state. Various types of redundancy are
described below.
17.6.1 Active n for (n 1) Units

For an assembly made of n identical units active all the time in parallel,
where (n 1) units must work — or any one may fail — for the assembly to
perform the specified function, the probability of assembly failing, i.e., all
units failing, is
U ¼ U1 U1 . . . n times ¼ U1n
where U1 is the probability of failure for each unit, assumed to be the same
for all units. The reliability of the assembly is then
R ¼ 1 U ¼ 1 U1n ð17:15Þ
For a given series reliability, using two large units (n ¼ 2) in parallel, one
active and one spare, would result in a heavier assembly than three small
units, two needed and one spare. On the other hand, using 10 small units
would also be mass heavy, as there is always some fixed mass for each
small unit. Obviously, there is an optimum number of units in between for a
FIGURE 17.9 Active and standby redundancies.

given reliability with minimum assembly mass. We examine this for a

battery charge or discharge converter made of multiple small units working
in parallel.
The mass of power converter in the 100 to 200 W range is approximately
given by 500 þ 3Po g, where Po is in watts. It has little sensitivity to the
voltage and bus voltage regulation. The unit mass in general is expressed as
mass ¼ þ Po, where is the fixed overhead and is the power
proportional mass constant. If n units are to be used for (n 1) needed to
meet the converter power requirement with one unit failed, the optimum
value of n for minimum mass can be shown to be
rffiffiffiffiffiffiffiffiffiffiffi
Pout
noptimum ¼1þ ð17:16Þ

Thus, if the fixed mass is small, many small units will result in the
minimum assembly mass, although it would cost more. Practical experience
suggests that n ¼ 3 is the optimum for all values of Pout/ between 2 and 7.
17.6.2 Active n for m Units

If a total of n units are active in parallel, and m units must work to meet the
full rated operation, the probability of exactly k units working is given by
the binomial distribution
n!
Rnk ð1 R1 Þk ð17:17Þ
ðn kÞ!k! 1
The probability of at least m units working is the sum of all terms from
k ¼ m to n, i.e.,
X
k¼n
n!
R¼ R1nk ð1 R1 Þk ð17:18Þ
k¼m
ðn kÞ!k!
where R1 is the reliability of each unit.
17.6.3 Dormant n and Active m Units

For an assembly with (m þ d) units in parallel, m units active which must
work and d units dormant in spare, any one of which can replace a failed
active unit, the mean time between failures is given by

dþ1
MTBF ¼ MTBF1 ð17:19Þ
m
where MTBF1 ¼ 1/1 for each unit.

For example, if there were three for two battery chargers (two active and
one spare), the MTBF of the assembly would be [(1 þ 1)/2] MTBF1, which is
the same as that for each unit. If there were only two for two units with no
spare, the MTBF of the assembly would be 1=2 MTBF1. Here all units are
assumed to have identical failure rates, and the reliability of the switching
element is assumed to be one. In reality, the dormant unit under no
operating stress would have much lower failure rate until switched on.
With two parallel units, active A and backup B, with a switch at one end
as seen in Figure 17.9(b), the reliability is estimated with the switch in series
with the backup unit. Because, for the assembly to continue working after
one unit fails, the unit B and the switch both have to work at the time of
switching.
17.6.4 Shared Versus Separate Battery Chargers

The battery charge/discharge converters can be configured two ways as
shown in Figure 17.10(a) and (b). In (a), each battery separately has two
chargers, each rated 100% of each battery load, or 50% of the total system
load, allowing one converter failure. In (b), both batteries jointly share three
chargers; each rated 50% of the total system load so that any two converters
will meet the full system load. Shared chargers (three for two for both
batteries) require fewer chargers, and hence result in lower mass. However,
if one cell fails (open or short) in one battery, say B1, one cell must be
removed from battery B2 also and placed on trickle charge to keep the
removed cell alive. Otherwise, B1 will share a much-reduced load due to its
loss of 1.25 V since the internal battery resistance is small. This is true for
any two voltage sources working in parallel. Their voltages must be equal
for equal load sharing. For this reason, the actual implementation of shared
chargers becomes complicated; hence it is seldom used in practice. Separate
FIGURE 17.10 Battery charge/discharge converter configurations.

chargers for each battery require more parts resulting in greater mass.
However, being simple in implementation, it is the most widely used
redundancy configuration for battery charge and discharge converters.
17.6.5 Diode O-Ring for Active Redundancy

The primary and backup units often use a diode o-ring circuit as shown in
Figure 17.11. Such a circuit makes the primary unit take almost all the load
and the backup unit remains active in parallel taking only 5 to 10% of the
load. The efficiency of the assembly is optimized by making the primary
unit work more efficiently near full load. The uneven load sharing is
achieved by R1, R2 and R3 settings in the control circuit of the voltage
dividers of the primary and backup units. The value of R2 > R3 makes the
control voltage Vc1 > Vc2, so that the primary unit takes almost all the load.
Figure 17.12 displays the method of determining the load sharing
between two units. The terminal voltages drop lines of both units, Vp and
Vb, are plotted versus load in individual units between two vertical lines
separated by the total load to be shared. Since the two units are electrically
connected in parallel, their common terminal voltage must have the same
value Vt at the point of their intersection. The load is then Lp on the primary
unit and Lb on the backup unit, and the total load is Lp þ Lb.
The diodes are necessary in this circuit to isolate the two units against the
reverse current flow under two possible conditions, namely:
An internal fault in any one unit with the other unit feeding the fault, or
The drift in Vc1 and Vc2 can cause the primary unit to take, say, 110%
load and the backup unit to take 10%, i.e., the backup unit works as a
load.
FIGURE 17.11 Diode o-ring circuit.

FIGURE 17.12 Load sharing between two parallel components.
17.7 Failure Rate Statistics

The failure rate is derived from tests on a very large population of identical
components. It is not practical to carry out such tests to end until all
components in the population fail, for it would take very long time. Instead,
testing is done for a preset duration, failed components are replaced as they
fail, and the testing is terminated after long enough time duration. This
gives the results as intended because the failure rate is constant and the
unfailed component at any time is as new as a brand new component. For a
given population, the failure rate in unit time under the test environment is
then simply given by
total number of failures

¼ ð17:20Þ
total test duration
Statistics on failure rates under particular operating conditions are

generally available from the manufacturer. Various organizations are also
active in compiling such information and in publishing various documents.
For example, MIL-HDBK-217 compiles voluminous data on various
components under various operating environments. A comprehensive
review of these data shows that passive components, such as capacitors,
resistors, diodes, and connectors, are the most reliable. The least reliable
parts are the traveling wave tube amplifiers (TWTAs) and components with
moving parts such as sliprings, bearings, potentiometers, and relays. This is
in line with what is expected.
17.8 MIL-HDBK-217
This military handbook establishes the uniform method of predicting
reliability of military electronic parts, equipment, and systems. It lists the
base failure rates of numerous parts under base thermal, electrical, and
mechanical stresses. Any deviations from the specified operating conditions
would alter the failure rate as given by the following expression:
a ¼ b p ð17:21Þ
where
a ¼ failure rate under actual operating conditions in the design

b ¼ failure rate under base operating conditions specified in the
handbook
k ¼ factor to modify b for environment, operating stress, and
construction differences, if any, for k ¼ 1,2, . . . n
p ¼ 12 n ¼ product of all modifying factors.
Major factors that modify the failure rates of electrical and electronic
parts are:
E ¼ Environmental factor to account for factors others than tempera-

ture.
Q ¼ Quality factor to account for difference in quality levels (class A, B,
S, etc.) For example, this factor may be 1.0 for a class B part
(commercial) and 0.01 for a class S part (space qualified.)
T ¼ KiT ¼ Operating temperature deviation factor, where iT is
temperature deviation from the nominal rated value in C, and K is a
constant varying from 1.05 to 1.15 depending on the part construction.
With an average value of K around 1.10, it is noteworthy that the failure
rate factor doubles with every 10 C rise in the operating temperature,
and cuts in half for 10 C drop.
D ¼ (Vop /Vrated) ¼ dielectric stress factor, where ¼ 2 to 4 near rated
voltage, and 5 to 8 above corona inception voltage. Obviously, the
corona takes away life at a much faster rate. Therefore, corona in high-
voltage equipment should be avoided.
V ¼ vibration factor, which depends on the fatigues life, which in turn
depend on many factors such as the vibration amplitude, frequency,
stress concentration, fracture toughness of the material used in the part
fabrication, etc.
Table 17.2 Base failure rate for some selected electronic parts
Failure rate, k, Failure rate, k,

Component per 106 h Component per 106 h
CAPACITORS TRANSISTORS
Ceramic 66 22 Signal, NPN 71 14
Film 78 26 Signal, PNP 100 20
Tantalum 200 63 Power, NPN 650 130
Electrolytic 1600 480 Thyristor, power 1800 360
ICs, digital 1050 18
RESISTORS
20 gates
Fixed, composition 45 45 ICs, linear
32 Qs 900 33
Fixed, wire wound 100 100
Fixed, film 3 3 TRANSFORMERS
Variable, cermet 2760 1380 Signal 10 3
Variable, wire wound 300 100 Power 82 32
Switches, 225 3
DIODES thermal
Signal 28 5.6 Connectors, 32 4
Power 2160 432 contact pairs
Fuses 300 100
Source: MIL-HDBK-217.
There are other modifying factors listed in the handbook, which must also
be accounted for in the reliability estimates. The base failure rate b for some
selected commercial and military power converter parts before applying the
modifying factors are reproduced from MIL-HDBK-217 in Table 17.2.
17.9 Parts-Count Method of Reliability Estimate

For a bid proposal and early design stage, the unit reliability can be
estimated by using a simple part count method outlined in MIL-HDBK-217
as follows. The information needed to apply this method are: (1) generic
part types and quantities, (2) part quality levels, and (3) operating
environmental of the unit. The general expression for equipment failure
rate with this method is given by
X
i¼m
eqp ¼ Ni bi pi ð17:22Þ
i¼l
where
eqp ¼ total equipment failure rate (failures/106 h)

bi ¼ base failure rate for the ith generic part (failures/106 h)
pi ¼ product of all modifying factors for the ith generic part to account
for the environment, operating stresses and quality class
Ni ¼ Number of ith generic parts used in the equipment, i ¼ 1, 2, 3, . . . m
m ¼ number of different generic part categories used in the equipment.
The above method applies only if the equipment is used in one

homogenous environment. Otherwise, it is applied to portions of equip-
ment in each environment. Then the failure rates of those portions are
added to determine the total equipment failure rate.
17.10 Derating for Reliability

The failure rate of a component depends on operating stresses, such as the
voltage, temperature, etc. The electrical insulation at high temperature
oxidizes, becomes brittle and may crack, leading to failure (short circuit).
The oxidation is a chemical degradation, which follows the Arrhenius
exponential growth. Data on numerous items of equipment show that
failure rate doubles, or life shortened to one half, for every 7 to 10 C rise in
the operating temperature. In the reverse, the life doubles for every 7 to
10 C reduction in the operating temperature. Similar degradation (wear)
takes place above certain voltage, although not as well understood as that
for the temperature. The rise in failure rate with the operating stress level is
not to be confused with the wear-out failure rate described earlier in Section
17.3. It is to be seen as raising the flat part of the bathtub. The failure rate is
still constant per unit time, although another constant at another operating
stress level.
Derating is the reduction of electrical, thermal, and mechanical stress
levels applied to a part in order to decrease the degradation rate and
prolong the expected life. Lowering an operating stress on the component
reduces the failure rate. Operating the component at lower than the
nominally rated stress level is called ‘derating’. It is routinely practiced to
decrease the wear-out failure rates in military and space worthy designs.
The derating in current is often done to lower the temperature. On the other
hand, the current derating in some active devices, such as transistors, may
be to control the di/dt rates, which can upset the semiconductor operation.
Derating increases the margin of safety between the operating stress level
and the actual failure level of the part. It provides added protection from
system anomalies unforeseen by the design engineer.
The derating can come from many sources, and all should be considered.
The primary derating comes from the environment and operating condi-
tions of the subject part. The derating factor when multiplied by the normal
rating gives a maximum recommended design stress level and does not
preclude further derating from other factors. Parts having no derating
factors listed on an approved part list may mean that inadequate data is
available to make any recommendation. In that case, a judgment based on

data from similar parts may be considered to derate the part in question
with generous margin.
Most programs maintain their own preferred part list based on the failure
rates. Such a preferred part list also gives derating factors, which must be
applied to all designs on those programs. Both the selection of parts and the
derating factors in the preferred part list are based on experience on the
heritage designs successfully flown. Space agencies, such as NASA and the
European Space Agency, maintain their own preferred part list. If a desired
component is not in the preferred part list, it must be qualified by rigorous
testing under the same environment as those in the list.
17.11 Quick Estimate of Failure Rate

The full scale reliability testing is time consuming and expensive. A quick
estimate of the reliability for screening a new proposed part for a potential
application can be obtained by testing several units in parallel until the first
one fails in normal or accelerated testing. If T is the operating time to the
first failure, obviously T is not the mean time between failures, since it is not
derived from a large population. But it can be used to roughly estimate the
MTBF as follows:
1
MTBF ¼ 1=½1 ð1 CÞT ð17:23Þ
where C is desired confidence level, which is generally 60 to 90% in

commercial world, and higher than 99.9% in defense and nuclear power
equipment.
17.11.1 Example
To understand the above method of quick estimate, suppose we put to test
20 units of a new part, and the first unit failed at 3000 h. The total operating
time of all units is then 20 3000 ¼ 60,000 h before the first failure occurred.
The MTBF of the population is then estimated using Equation 17.23 for
different confidence levels as follows:
Confidence level C 50% 63.2% 90% 99%

MTBF for the population 72,130 60,000 26,058 13,030
Note that for exponential distribution, 1 – (1/e) ¼ 0.632, and that is why the
MTBF to the first failure has a confidence level of only 63.2%.
17.12 FMECA
The reliability analysis provides an estimate of the inherent reliability of the
system. The failure mode and effect analysis (FMECA) on the other hand,
goes beyond that. Its primary purposes is to identify not only the failure
modes, but also their effects and criticality on the mission success. For
example, criticality 1 means that if the component fails, it can cause injury to
human, or cause the mission to fail completely. Obviously, such a
component needs sufficiently high reliability. In other components, the
probability of failure must be at the minimum possible level that is
acceptable for the mission under design.
17.13 Non-Redundant Components

The top-level reliability specifications generally read as follows:
The power system reliability of meeting all the specified performance

requirements shall exceed 0.95 over 15 years. No single point failure
should cause loss of redundant paths for any critical function, nor
should propagate to other systems. It should not connect the battery
directly and permanently to the main bus.
To obtain such high EPS reliability of 0.95 over 15 years, many high-
reliability components are used with redundancy. However, redundancy
cannot be provided in the following power system functions, because it is
either not practical even for a critical function, or is not really needed for a
noncritical function:
SAD shaft, bearing, stepper motor rotor, and gears cannot be practically
made redundant from the spacecraft fabrication point of view
Telemetry not critical to the spacecraft operation is nonredundant, such
as
Battery cell voltage monitoring, and
Battery reconditioning circuit, if used.
It may be useful to cite some general design guidelines for achieving high
reliability in the power system.
For each unit, the wear-out failure rate is kept well below 10% of the
random failure rate, and the MTBF is designed to be 40 to 60% longer
than the mission life.
The slip rings can fail in seizure because of the cold welding, which
must be avoided.
Laser welded connections in the bus and other wiring improves

reliability.
Wire break is a failure mode, often caused by a nicked wire that can
occur during removal of the enamel insulation prior to soldering. Wire
nicks should be avoided by a well-controlled manufacturing process.
Single-point failure (short) in nonredundant wires is avoided by using
double insulation.
The American practice in the power system design has been to use hard
low-resistance single-point ground. However, such systems would trip
or clear the fuse under any fault involving the ground. The system
reliability may be improved by soft grounding through a high
resistance so that one fault may not shut down the power. This
makes the system single ground fault tolerant. The concept of
grounding to the structure through a several thousand ohm resistor
was thoroughly investigated by NASA/Jet Propulsion Laboratory
several years ago, but has not been implemented as yet in any
spacecraft due to a some concerns, the major one being electrostatic
buildup.
17.14 Example of Reliability Calculation

Consider a solar array that is required to have 99.2% reliability over a
15-year mission. Say it needs 30 shunt circuits to generate the require power.
Each shunt circuit consists of three strings, 126 4 6-cm cells in each string,
and has an estimated 15-year reliability of 0.95, including the string wiring,
slip rings, shunts and the random cell failure at 1.0 FIT (failures per million
hours). An array of 30 such shunt circuits would have the reliability of
(0.95)30 ¼ 0.21, way below the required value of 0.992. As seen in Table 17.3,
for achieving 0.992 solar array reliability, we must use 32 circuits, two more
than the needed 30 for the required power.
Table 17.3 Example of determining number of shunt circuits

for achieving required reliability
No. of shunt circuits Probability of failure Solar array reliability
1 for 1 0.050 0.950

30 for 30 0.786 0.214
31 for 30 0.114 0.886
32 for 30 0.008 0.992
33 for 30 0.001 0.999
Chapter 18
Integration and Testing
18.1 Introduction
If an assembled spacecraft malfunctions, either on the test floor or in orbit,
or anywhere between, a great deal of time and money are spent in detecting
the problem, investigating why it happened, and in fixing it. A thorough
testing during various fabrication and integration phases is required to
weed out potential problems at an early stage, and to ascertain a timely
delivery within the contracted budget. The spacecraft customer requires the
prime contractor to prepare the Spacecraft Level Test and Verification Plan,
which becomes the guiding document for all systems and components
testing. The plan generally draws heavily from MIL-STD-1540 for environ-
ment testing and MIL-STD-1541 for EMC testing.
Once a component or an assembly is manufactured to the final design,
acceptance testing is done on the floor to verify that no manufacturing
errors or defects exist in the final product. Components are tested before
they flow into their respective system assemblies, and the systems are tested
separately before integrating into the spacecraft. The operating environ-
ment in space is simulated during the tests as much as possible and
practical.
The tests typically performed on key power system component are listed
in Table 18.1. If the thermal shock torture test on a component is required, it
is often carried out by quenching the part from boiling water (100 C) to
liquid nitrogen (193 C). The conformal coating and ceramic parts may
crack under such shocks.
Analyses may be used in lieu of testing to verify compliance to some
requirements which may pose undue test difficulties. The selected
analytical techniques typically include engineering analysis, comparative
analysis, statistics, analog and digital modeling and computer simulation.
The analyses must be preformed based on the data available at the critical
design review stage.
18.2 Test Limits

All system components are subjected to extensive testing to prove their
ability to perform reliably in the launch and orbit environment. Newly
developed components undergo most stringent qualification tests at the
451
Table 18.1 Tests performed on key power system components
Battery power
Solar cells Battery packsa converters Pre- assembly
Cell performance d.c. isolationb Efficiency Tri-channel controld

Deformation Thermistors Input voltage range SAS1, SAS2 voltage drops
Solar panels Strain gauges Output voltage Close loop phase shift and
regulation gain
Circuit check Ampere-hour Input/output ripple Bus ripple
capacity
High-voltage Charge Transfer function Command and telemetry
isolation capability
Illumination and Retention of Min/max loads BPM accuracy
continuity charge
Thermographic Reconditioning Local over-voltage BCVM accuracy
map protection
Battery blankets Electrolyte Command and SAD functions
leakage telemetry checks
Grounding Vacuumc Him on/IUm off (?) SAD currents
i/face u?
Effective Heater Battery current Auxiliary power redundancy
conductance operation regulation and regulation
By-pass Failure detection
protection
a
Battery cells are tested in acceptance tests at the vendor site.
b
Resistance of 100 M minimum at in both pre-vibration and post-vibration Megger2 tests.
c
Exposure to 5 105 torr for 8 h.
d
Self-diagnostics test.
highest stress levels. The components previously flight qualified undergo

routine acceptance tests before being placed in the spacecraft assembly. The
test levels under these tests are higher than the expected stress levels under
normal operating range. Typical temperature bands for various classes of
tests are shown in Figure 18.1. Each spacecraft follows its own bands as
specified in the requirements. But in all cases, the qualification temperature
band is the widest, then the acceptance band, the design band, and then the
normal operating band, in that order. Meeting these temperature require-
ments at various phases of testing assures adequate margins for the mission
success.
The test temperature limits for components are derived from the
following considerations:
The flight heritage and temperature capability of the component

Knowledge of expected temperature range within the spacecraft
enclosures
Integration and Testing 453
FIGURE 18.1 Temperature bands during normal operation and during acceptance and
qualifications tests.
Evaluation of thermal environment external to the spacecraft as a

function of the thermal design variables
Appropriate margins
The Test Plan is the contractual document that establishes the basis for
derivation of component temperature limits. The plan generally requires
that components be qualified at temperatures of 10 to 15 C more severe
than the minimum and maximum levels expected during the entire flight
mission. It requires a 5 to 10 C margin with respect to the same expected
mission extremes for acceptance testing. The minimum and maximum
temperature levels likely to be encountered during the flight mission are
based on thermal analysis of the spacecraft design, environmental
exposures, and internal dissipation profiles associated with all modes of
the spacecraft operation. Because such analysis is subject to physical and
mathematical modeling uncertainties, it is a standard practice to add 5 C
uncertainty to the temperatures predicted by the analysis to derive
qualification and acceptance test temperatures for the component. The
customer specification defines expected environmental extremes and
associated qualification and acceptance test levels for various components.
It also defines several other parameters for component thermal testing

including the number of temperature cycles required, cumulative soak
times, and the temperature transition rates. In certain cases, adequately
sized thermostatically controlled heaters built in the design of a specific
component permit reduction of the standard temperature margins.
Table 18.2 lists the component temperature limits for a typical GEO
communications satellite. The basic considerations for defining the survival,
qualification, acceptance, and design analysis temperature limits in the
table are as follows.
Table 18.2 Component temperature limits for typical GEO communications

satellite
Survival (non- Qualification Acceptance Design analyses

Component operational) (operational) (operational) (operational)
Solar array 180/75 180/75 175/70 165/60a,b

Solar array drive 35/70 35/70 30/65 20/55c
NiH2 battery 15/45 15/45 10/30 5/25d–g
Battery let-down 35/70 35/70 30/65 20/55
unit
Battery cell 35/70 35/70 30/65 20/55
voltage monitor
Bat pressure 35/70 35/70 30/65 20/55
monitor
Power regulator 35/70 35/70 30/65 20/55
unit
Ordnance 35/70 35/70 30/65 20/55
controllers
Other power 35/70 35/70 30/65 20/55
electronic boxes
a
60 C is for a 15-year mission. The PV conversion efficiency degrades more in a long
mission, which increases the array temperature. For a 5-year mission, it could be 50 C.
b
In the pre-operational mission phase, the array temperature can reach 110 C, so the array
needs to qualify for 130 C.
c
The maximum allowable instantaneous temperature difference between inner and outer
races of associated bearings is 10 C in either direction.
d
During trickle charge.
e
Temperature margin for this item in cold conditions is provided by thermal control heaters
with a minimum of 30% excess heating capacity.
f
Maximum allowable battery cell temperatures during transient events are: 30 C during
charge in transfer orbit, 30 C during reconditioning letdown, and 15 C during
reconditioning recharge. The maximum allowable mission mean temperature of battery cells
is 15 C.
g
Maximum allowable battery cell temperature gradients are: 5 C within a cell surface, 7 C
from the pack average to the dome of any cell, 7 C between pack averages within a battery
assembly during charge, and 8 C between battery assemblies during discharge.
18.2.1 Survival (Non-operating Mode Only)

The component design should be capable of surviving exposure to the
extreme survival temperatures such that:
The component does not necessarily operate at the design performance

level, but
The structural integrity is maintained to prevent damage, contamina-
tion, or degradation of other spacecraft hardware, and
Full functional performance is achieved when the temperature is
subsequently brought within the qualification temperature range.
18.2.2 Qualification (Operating Mode Only)

The component design should be capable of providing functional perfor-
mance as defined by the component specification at the qualification
temperature limits.
18.2.3 Acceptance (Operating Mode Only)

Each component delivered to the spacecraft assembly should demonstrate
the functional performance as defined by the component specification at the
acceptance temperature limits.
18.2.4 Flight Design Analysis (All Modes)

The spacecraft thermal design should maintain the applicable interface or
component element within this temperature range as demonstrated by the
worst-case thermal analysis. Where no temperature range is shown in the
non-operating column,
Either the component is operational during the entire flight mission, or

A specific temperature control is not required to preserve the integrity
of the component and the spacecraft when the component is not
operating.
18.2.5 Temperature Location

Location of the temperature must also be specified. It can be the area-
average temperature at the interface of the mounting structure that is in
direct conductive contact with the component, or the area-average
temperature of a major structural element (usually a baseplate or mounting
chassis) within the component. The location of this element is selected

mutually by the electrical and thermal engineers.
18.3 Test Sequence

Each protoflight or flight unit, built in accordance with the documentation,
procedure, and controls as reviewed and accepted during the unit’s final
design review is subjected to protoflight tests for qualifications or prior to
acceptance for integration into the spacecraft. The input levels and range for
the required acceptance and qualification tests are as specified in the
component environmental test specification. The tests are carried out in
certain sequences shown in Figure 18.2 and Figure 18.3, respectively. The
description of each test follows.
18.3.1 Ambient Functional Performance Test

These tests include measurement of all performances in all modes of
operation at ambient pressure and temperature. They are repeated a few
times, starting with the initial performance tests, and ending with the final
performance tests.
18.3.2 Leak
All pressurized or hermetically sealed units are subjected to leak tests. The
leak test is conducted before and after the thermal vacuum, vibration, and
pressure tests. The leak test is performed with the component pressurized
at the maximum operating pressure and then at the minimum operating
pressure if the seals are dependent upon pressure for proper sealing. The
duration of the leak test is 6 h except for units which are part of the
propulsion system. The duration of the propulsion component leak test
should not exceed the time that the pressurized condition is normally
experienced during propellant loading with a minimum of 6 h.
FIGURE 18.2 Component acceptance test sequence.

FIGURE 18.3 Component qualification test sequence.
18.3.3 Pressure
All pressurized units are subject to pressure tests. The pressure test is
performed on the unit prior to and after successfully completed thermal
vacuum and vibration tests. The pressure level for qualification is at least
1.5 times the maximum operating pressure. The hold time is at least 5 min.
The test is then repeated for a number of cycles equal to at least 4 times the
mission life at the maximum operating pressure.
18.3.4 Random Vibration

Each unit is subject to a random vibration test except for those tested for
acoustic. The unit is mounted to a fixture through the normal mounting
points and operated in the mode in which it is launched. The unit is
vibrated once per axis in each of the three mutually perpendicular axes.
18.3.5 Sine Vibration

Each unit is subjected to a sinusoidal vibration test. The unit is mounted to a
fixture through the normal mounting points and operates in the mode in
which it is launched. The unit is vibrated once per axis in each of the three
mutually perpendicular axes.
18.3.6 Post-Vibration Performance

These tests are to verify that there is no performance degradation caused by
the above dynamic testing.
18.3.7 Acoustic
For units with large surface areas and relatively low density, it may be more
practical to expose the qualification unit to an acoustic test instead of a
random vibration test. If random vibration test is considered too severe, the
unit is subjected to an acoustic test. The unit is installed in an acoustic

chamber capable of generating the desired sound pressure level. During the
test the unit is operated in the same manner as during the launch phase. It is
monitored for malfunctions in accordance with the requirement of the
applicable unit test procedure. In electrical power system, the solar array
panels and the batteries are generally subjected to the acoustic test.
18.3.8 Thermal Vacuum Cycles

Each qualification unit is subjected to a series of ten thermal cycles in a
vacuum less than 105 torr. During the thermal vacuum cycle, the unit’s
turn-on and turn-off capability is demonstrated, as well as the thermal
design is verified. The performance test is conducted during the first and
last cycles.
18.3.9 EMI/EMC
These tests are carried out using the test setup and procedures as per the
EMI/EMC Test Plan.
18.3.10 ESD
Each unit is subject to an ESD susceptibility test. The unit is mounted to a
graphite epoxy panel in a representative flight configuration. A current is
injected into the panel in order to demonstrate that the unit is not
susceptible to ESD induced transients. The test levels and duration is in
accordance with the ESD Control Plan.
18.3.11 Life Test

All units which have not been previously life tested and are susceptible to
noticeable wear, degradation, fatigue, thermally induced creep, or preload-
induced creep during its projected lifetime is subject to life testing. During
the life test, the unit is subjected to the conditions representative of those
experienced on-orbit and during ground testing. When accelerated life tests
are performed, each unit must demonstrate 50% life margin above its
projected mission lifetime. In the power system, the battery cells, solar cells
and SAD are subjected to life tests.
18.4 Solar Cell Tests

The incoming quality of solar cells is subjected to I–V performance,
radiation, thermal shock, humidity, and peel tests as explained below.
The cell’s I–V performance is tested under equivalent of one-sun

illumination of a xenon arc lamp with infrared filters to match the sunlight
spectrum. The voltage and current measurements are made over the
complete range by varying the load resistor from zero to infinity. Since the
cell performance is temperature dependent, the tests are made under
controlled temperature. The short-circuit current, the open-circuit voltage,
and the maximum power point are read from the I–V graph and compared
with the specifications to which the cells were procured.
The performance degradation under space radiation is conducted on a
test coupon using lab-generated radiation from a Van de Graaff electrostatic
accelerator. The accelerator produces charged particle radiation at rates
many orders of magnitude greater than those found in space, so that several
minutes of test simulate the total equivalent 1 MeV fluence over the mission
life. Such tests are conducted with and without the coverglass.
Thermal shocks are normal on solar array when going into eclipse (hot to
cold) and coming out of eclipse (cold to hot). Test coupons are heated to
high temperature (120 C) and then dipped in liquid nitrogen (193 C) to
simulate the first condition, and the reverse to simulate the second
condition.
Humidity of around 95% is applied for many hours to verify that the
moisture penetrating under the cell electrodes does not weaken the bond.
Sticky tape is applied to the cell and then peeled off to test the bond
strength.
18.5 Tests on Wires

Isolation testing between return wires of two circuits is conducted to verify
that the following typical requirements:
d.c. isolation test to verify that dc resistance of the insulation between

wires is greater than 1 M
a.c. isolation test to verify that capacitance between the wires is less
than 0.1 mF.
18.6 Power System Tests

The power system as a whole is subjected to the following major tests
during fabrication and acceptance:
Solar array illumination test under flash to verify that the bus power
generation meets the BOL requirement
Performance test to verify that all connections are made correctly, and
the currents and voltages are measured as expected within the specified
tolerance limits
Electromagnetic interference and compatibility tests, with a particular
emphasis on the conducted emission, which originates primarily form
the power system
Magnetic moment measurement on the assembled spacecraft. The solar
array and electrical harness are primary contributors to the total
moment
Corona test under out gassing to demonstrate the minimum time to
turn on and off high voltage and high frequency components when
exposed to thermal vacuum
Mechanical tests such as shock, vibration, and acoustic
Thermal vacuum bake-out test to clean contamination, if any
The power system is again tested as follows when the spacecraft is fully
integrated:
Flash light go–no go tests on solar wings to check continuity of all

connections in the power system
Acoustic tests at wing level to check all joints, hinges, gimbals, etc.
Continuity checks and insulation tests on batteries
Continuity and insulation tests on all power electronics boxes
Major power system tests are described below.
18.6.1 Solar Array

The electrical performance of the solar array is measured at room
temperature using a large-area pulsed sun-simulator. Demonstration of
workmanship is accomplished by subjecting the solar array panels to 10
thermal cycles, followed by an electrical performance test. High and low
temperatures correspond to the predicted highest and lowest temperatures
in orbit. The solar array is subjected to the following tests after mounting
the solar array drive:
Electrical performance test

Structural integrity by sine and acoustic tests
Release deployment at zero gravity conditions
The electrical performance tests on solar array include the following series
of tests.
18.6.1.1 Wiring loss

All wiring losses within the solar array assembly are predicted by analysis
and verified by measurements.
18.6.1.2 Current–voltage
Using a solar simulator, an I–V curve is developed for each circuit on a
panel, both before and after the vibration test. These data are used in
predicting the solar array performance.
18.6.1.3 Power generation

The solar array panels on each wing are simultaneously illuminated by
lamps at approximately one third the maximum solar flux. The array output
and array shunts are monitored for qualitative operation. The telemetry
circuits of the array are checked, and the spacecraft power system interfaces
are verified.
18.6.1.4 Array life

The degradation of the solar array is computed by analysis to verify that the
mission life requirements are met. Power distribution and other losses are
included in the analysis.
18.6.1.5 Stability test

Worst-case conditions applicable to the solar array current, capacitance,
constant power load, etc. are simulated to demonstrate stable regulation
operation. This test is an alternate way to demonstrate stability margin since
no direct measurement can be made with only the solar array assembly.
18.6.1.6 Shunt control

Measurements are made for each shunt control cut-in operation, voltage,
and regulation for each unit and all units together. Bus stability simulation
upon eclipse entrance and exit is demonstrated. The shunt drive capabilities
at the unit level, system level, and the spacecraft level are demonstrated at
various stages except at the launch site.
18.6.1.7 Shunt regulator and shunt

The solar array impedance measurements are made with the shunt
regulator in the linear mode for each shunt unit. The test for the shunt
demonstrates the ability to detect the detector failure and/or the controller
failure and to switch over to the redundant detector and/or controller.
18.6.1.8 Thermal cycling test

The solar panel is installed and instrumented in a large temperature
chamber or in some other facility with comparable performance. Electrical
tests are performed to verify the thermal test instrumentation prior to

closing the chamber door. After the chamber has been dehumidified, the
panel temperature is increased until the highest reading thermocouple
reaches þ60 C. The panel temperature is held at this level for 16 h. It is now
decreased until the highest reading thermocouple reaches the specified
lower temperature level, and then increased until the highest reading
thermocouple reaches the specified upper temperature level. The solar
panel temperature is subsequently decreased to ambient conditions. This
constitutes one complete thermal cycle.
18.6.1.9 Functional tests

The solar panels are subjected to pre-thermal functional test to determine
whether manufacturing defects have been introduced into the fabrication,
and whether the panels meet the specified performance requirements.
Functional tests are performed before and after thermal cycling. As a
minimum, functional tests include the following tests, unless specified
otherwise.
Illumination tests: All sections of each circuit and full panel are
illuminated at AM0 equivalent intensity at þ35 C with a standard solar
cell selected in accordance with the power system specifications. The
current–voltage (I–V) curves of the solar cell circuits are generated and
electrical parameters such as Isc, Voc, and illumination intensity are
recorded.
Diode tests: The isolation diodes connected to the output of each circuit
is tested for forward bias voltage drop at the operating current and for
reverse leakage current at the operating voltage.
Shunt dissipator tests: Each shunt dissipator section is tested to
demonstrate its capability to meet all the applicable electrical perfor-
mance requirements.
Temperature telemetry test: Calibration data is acquired by recording
temperature sensor data at ambient, hot, and cold temperatures.
Substrate isolation test: The resistance between each solar cell circuit
and the substrate to which it is bonded is measured and recorded.
18.6.2 Solar Array Drive

Testing each solar array drive individually includes the following:
The rotational speed of each drive is tested.

The drive torque for each redundant drive motor is demonstrated to
satisfy the worst-case static load and temperature. This test demon-
strates both the step size and the drive rate to satisfy the panel
orientation and the dynamic criteria.
Maximum predicted current at maximum predicted operating tem-
perature is passed through the sliprings for one complete revolution.
The slip ring voltages are monitored to verify the slip ring operation
and the contact resistance.
Noise, vibration, acoustic, deployment shock, and steady state accel-
eration (spacecraft spin) are the mechanical tests performed on the solar
array drive.
18.6.3 Battery Assembly

Each battery assembly undergoes acceptance level test sequence shown in
Figure 18.4. In spacecraft with two battery assemblies, one is typically
mounted on the north bus equipment panel and the other on the south bus
equipment panel. Each equipment panel then undergoes the following
acceptance tests with a nonflight battery that is used for in-house testing:
Temperature and pressure sensors check-out

Thermal cycle tests with charge and discharge cycles of the battery sub-
assemblies as units
Launch site capacity tests at spacecraft operating temperature
Launch environment testing with the spacecraft
A functional checkout of the complete battery system with the
spacecraft during the system level launch environmental exposure,
and the launch site–operations–integrated system testing. These tests
include:
Thermal tests
Low- and high-temperature capacity tests
FIGURE 18.4 A battery acceptance test sequence.

High discharge rate capacity test

Overcharge and charge retention test
Available capacity at minimum charge rate test
Launch environment tests
18.6.4 Power Electronics

Power electronics, such as the control electronics and power converter
assemblies, undergo acceptance testing in typical sequences shown in
Figure 18.5.
18.6.5 Bus Impedance and Stability

The bus dynamic impedance is tested using a set-up shown in Figure 18.6.
A small a.c. current of desired frequency is injected in the bus and the a.c.
bus voltage at that frequency is measured. The dynamic impedance of the
bus is then Vac/Iac. Such a test is repeated over the entire frequency range of
interest using the frequency sweep of the instrument.
Figure 18.7 is a set-up for testing the small signal dynamic stability of the
entire power system and the control loop. The power system and the load
circuit are broken and a small a.c. voltage, Vstab, is inserted in the closed
loop. The Vac1 and Vac2 are measured and plotted. The measurements are
repeated over the frequency range of interest by the frequency sweep, and
the gain and phase shift are plotted on Bode plots from which the gain and
phase margins are determined and compared with the required values. This
assures the power system stability under small signal perturbations in the
test frequency range.
FIGURE 18.5 Power electronics assembly test sequence.

FIGURE 18.6 Dynamic bus impedance test set-up.
18.7 Spacecraft Level Tests

The spacecraft undergoes the following acceptance level tests after the
integration of all major systems, including the power system.
18.7.1 Initial Power Turn-On Test

The initial power turn-on test is used to provide an orderly sequence for
verifying that the application of power to the spacecraft does not damage
previously integrated components or systems, and that the items being
tested are not adversely affected. This test is used to initialize each
component and system that makes up the spacecraft. A sequential turn
on procedure is developed that allows initialization of each of the
spacecraft’s systems. Data sheets are developed on which the voltage and
current values for each system are recorded. This allows a comparison
between the current values and those obtained during the system testing
and the component/system specifications. Where possible, breakout boxes
and test tees are used for measurements of the electrical values.
18.7.2 Electrical Power System Test

This is the first full electrical test performed on the electrical power system
after completion of the satellite assembly. This test is comprised of a
performance baseline test conducted on each system while the remaining
systems are active in their maximum EMI states. Both the primary and
FIGURE 18.7 Small signal dynamic stability test set-up.

secondary path redundancies are measured for their compliance with the
specification. Alternate modes of operation for the TT&C and payload
communications systems are functionally verified and cross-strap switches
are exercised during the test. The spacecraft RF signature is obtained in
parallel with the electrical systems tests.
The objectives of the EPS tests at this stage are to:
Verify that the spacecraft assembly operations are completed properly

Demonstrate that the interfaces are operating and are interconnected
properly between the bus and all other systems, and
Establish the baseline payload performance data for the transponder/
antenna interfaces
Subsequent electrical system tests demonstrate that the spacecraft

systems continue to meet all electrical specifications and provide a basis
for trend analysis.
Each spacecraft system is initialized and checked to verify that it can be
commanded into all its operational modes, and the electrical and/or
mechanical performance are measured. These data, when compared with
the data obtained during component or system testing, become the baseline
for all following spacecraft tests and trend analysis.
Following the initial turn-on of the spacecraft, a part of the electrical
system tests is performed during which the TT&C system is ON to ensure
that a reliable communications link can be established between the
spacecraft and the ground electrical equipment. This communications link
is maintained with the spacecraft for the remainder of all electrical system
testing. It provides a baseline test for the TT&C system.
18.8 Test Points

Sufficient test points are provided to permit testing of all on-line redundant
and all failure detection circuits at the spacecraft level. Test points are also
provided to permit open-loop stability testing (Bode plot test), large signal
load transient testing on low-voltage and high-voltage busses, and bus
impedance testing at the spacecraft level. All test points are buffered to
protect from the external shorts, and are properly isolated so that a direct
short to ground of a test point does not degrade spacecraft performance, nor
overstress spacecraft hardware.
The following test points are generally provided as a minimum. The
voltage test points are provided with sense lead carrying no current:
Bus voltage test points typically at two places

Battery voltage test points
Battery charger out-of-limit current simulation test points

Battery discharge converter failure simulation points
Load current access port from the bus capable of carrying up to 10 A in
satellites within a few kilowatts load
Spacecraft star ground test points
Spacecraft star ground access points for external load current return
capable of carrying up to 10 A
Load current access port from the low-voltage-regulated bus capable of
carrying up to 10 A
Shunt control system failure simulation test points, one for each control
system
Bus stability test points
18.9 Ground Test Equipment

External power sources, such as a solar array simulator and a battery
simulator (or a workhorse battery), are needed for ground testing of the
integrated spacecraft. Adequate protection must be provided to protect
from erroneous input voltage to the spacecraft during these tests. Blocking
diodes are placed in series with component input power leads for
protection against application of the reverse voltage. The ground test set-
up should also allow the spacecraft start-up and shunt testing for voltage
regulation with the solar panel installed. Figure 18.8 is a typical ground test
set-up. It shows only one simulator and one group of solar array circuits
representing one wing of the spacecraft. Another group will be in parallel to
represent the other wing.
The solar array is simulated with one or more current sources, each with
several ampere current limit, and the voltage limit a few percent above the
nominal bus voltage. Each source output is connected via umbilical
connector to one of many solar array shunt circuit. A T-connection in the
PRU connects the simulator to the shunt and the array circuits. Batteries are
charged using the solar array simulator and the spacecraft battery charge
converters. The initial power turn-on is accomplished with the solar array
simulators providing power via the battery–PRU harness connectors
without the battery connected. Once proper operation of the power system
is confirmed, the workhorse battery is connected in its place. The battery
relay turns the battery on and off.
18.10 Launch Site Testing

At the launch site, external power via an umbilical cable from the ground
power source supports the spacecraft testing and the housekeeping loads
FIGURE 18.8 Ground test equipment set up.
FIGURE 18.9 Hubble Space Telescope satellite in integration and testing before going to launch
site. (Source: NASA.)
prior to launch. The isolation diodes in the ground power supply prevent
the current flowing backward when the ground source is de-powered or
shorted. Typically, several amperes at nominal bus voltage are needed from
the ground supply interface after the isolation diode. A battery trickle
charge power supply may be connected to keep the battery fully charged
when the spacecraft is turned off, such as for transportation between
facilities. Such a feature minimizes the number of battery charge operation
at the launch site.
Figure 18.9 shows the new Hubble Telescope spacecraft in the final
integration and testing before going to the launch site for a planned launch
in late 2004.
Part D
Special Power Systems
Chapter 19
Interplanetary and Deep Space Missions
19.1 Introduction
The power systems for interplanetary and deep space missions differ from
those used in the Earth orbits because of the significantly different
environments they encounter. The mission environment depends heavily
on the spacecraft distance from the sun (Figure 19.1). Nine planets of the
sun are divided into two groups: the inner planets (Mercury, Venus, Earth,
and Mars) and the outer planets (Jupiter, Saturn, Uranus, Neptune, and
Pluto). The inner planets are small and are composed primarily of rock and
iron. The outer planets are much larger and consist of hydrogen, helium,
and ice. The inner planetary missions are characterized by higher solar flux
FIGURE 19.1 The planets in our solar system with the distance from the sun.
473
and elevated spacecraft temperature. Mercury orbit temperature can reach

300 C. Venus has a CO2 atmosphere 90 times denser than on Earth, causing
the greenhouse effect and the resulting temperature up to 477 C on the
surface. Protecting the spacecraft from heat is a major design concern. On
the other hand, missions to the outer planets, comets, asteroids, and deep
space are characterized by lower solar flux and cooler spacecraft
temperature. The temperature in Saturn orbit can be as low as 250 C.
Pluto is even farther from the sun and so cold that methane freezes on its
surface. Generating electrical power and protecting the spacecraft from
extreme cold are major concerns in such missions.
If one could look at our solar system from far above the Earth’s North
Pole, the planets revolves around the sun in the counterclockwise direction.
All planets, except Venus and Uranus rotate on their axes also in this same
direction. The solar system is remarkably flat — only Mercury and Pluto
have inclined orbits. The moons of various planets mimic the behaviors of
their planets and move in the counterclockwise direction with a few
exceptions. Some remarkable patterns also exist. For example, Mercury
rotates on its axis three times for every two revolutions around the sun; and
Uranus and Neptune are deficient in hydrogen.
The asteroids are small rocky bodies that move in orbits primarily
between Mars and Jupiter. Numbering in thousands, they range in size
from 1000 km diameter to microscopic grains. Comets circle the sun in two
main groups. The Kuiper belt is a ring of debris that orbits the sun beyond
Neptune. Many comets with periods less than 500 years come from the
Kuiper belt.
For missions away from the Earth, either closer to the sun or farther away
from the sun, the solar array, battery, and the power electronics designs
differ significantly because the extreme temperature on either side, high or
low, has a large impact on their performance.
The solar flux at any distance in deep space is given by
IEarth
I¼ ð19:1Þ
R2
where IEarth ¼ solar flux in the Earth orbit (1358 W/m2), and R ¼ distance
from the sun in astronomical units (AU). This equation assumes the sun to be
a point source, and may give some error at distances less than a few sun
radii. The solar array power output varies linearly with the incident solar
flux. Therefore, as the spacecraft moves away from the sun, the power
decreases inversely with the distance squared. In the PV power system, the
array temperature also decreases in the same ratio, which results in a higher
PV conversion efficiency. The combined effect of the flux and the
temperature changes is such that the photovoltaic power generation varies
not inversely with the distance squared, but to a lesser exponent in R, where
Interplanetary and Deep Space Missions 475
Table 19.1 The solar flux and PV power generation in the orbits of the various
planets in our solar system (relative to those in the Earth’s orbits)
Planet Distance from Solar flux relative to PV power generation accounting for
the sun (AU) Earth orbits temperature difference
Mercury 0.31–0.47 10.40–4.52 Severe loss of voltage

Venus 0.72 1.93 1.63
Earth 1.0 1.0 1.0
Moon 1.0 1.0 1.0
Mars 1.66 0.36 0.59
Jupiter 5.20 0.037 0.084
Saturn 10.08 0.0098 0.031
is approximately 1.5. Thus, the power generation is less than two times
with two times the solar flux, and more than one half with one half solar flux.
Table 19.1 lists planets in our solar system with their distances from the
sun, in AU, and approximate power generation rates in their orbits
considering both the solar flux and the temperature variations. The data
transfer rate to Earth from outer space depends on both the distance and the
power available as depicted in Figure 19.2.
19.2 Temperature in Deep Space

The Stefan–Boltzmann law determines the temperature of an object which
receives and radiates heat. With the spacecraft surface radiating to absolute
zero in deep space, the surface temperature is determined by
1
I 4
T¼ ð19:2Þ
"
where T ¼ absolute temperature in K, I ¼ solar flux in W/m2, ¼ Stefan–

Boltzmann constant (5.67 108 W/m2K4), ¼ absorptivity, and " ¼
emissivity of the surface. For a solar panel with one side absorbing and two
sides radiating, the above equation becomes
1
I 4
T¼ ð19:3Þ
2"
The /" ratio of various surfaces varies from 0 to 1 (Table 19.2). It is near
zero for shiny mirror type surfaces, near 1 for flat black surfaces, and 2 to 8
for highly absorbing surfaces. Typical solar cells have this ratio near 0.8.
As an example of using the above equation, we determine below the solar
array temperature in the Earth’s orbit during sunlight after ignoring the
FIGURE 19.2 The data transfer rate from outer space at two power levels.
Earth albedo and infrared radiation on the spacecraft. The array with /"
ratio of 0.8 would give rise to
1
1357 0:8 4
T¼ ¼ 313K or 408C ð19:4Þ
5:67 108 2
And, the array with an /" ratio of 0.6 would give rise to
1
1357 0:6 4
T¼ ¼ 291K or 188C ð19:5Þ
5:67 108 2
On the other hand, a spacecraft in deep space at 2 AU distance from the sun
receives solar flux 1/4th compared to that on Earth. Its solar array with an
/" ratio of 0.8 would give rise to a temperature of
1
1357=4 0:8 4
T¼ ¼ 221K or 528C ð19:6Þ
5:67 108 2
As for the power generation, missions beyond Mars cannot effectively

generate PV power because of decreasing solar intensity. Moreover, the
Table 19.2 Absorptivity and emissivity of various surfaces
Type of Surface Absorptivity, Emissivity, " Ratio /"a
Solar reflector (mirror) Low High < 0.1

Solar reflector Low High 0.1-0.2
Flat black reflector Low Low 1
Flat black absorber High High 1
Solar absorber High Low 2-8
a
A typical PV array has this ratio around 0.8.
greatly varying sun angle, illumination intensity, and temperature in such

mission cause wide variations in I–V response. The NEAR Shoemaker
spacecraft’s PV power system shown in Figure 19.31 is designed for use at
AU distances varying from 1 to 2.2, farthest from the sun to date and much
beyond Mars. Although the peak power tracking architecture is well suited
under such conditions, the direct energy transfer PV–SuperNiCd battery
system were used in the NEAR Shoemaker spacecraft to overcome certain
unique design challenges.1
Missions to the inner planets and near the sun would also be difficult to
design with a photovoltaic power system because of the strong solar
intensity and the resulting high temperatures. However, some examples
where PV cells have been successfully used in the past follow. Mariner-2
and Mariner-5 went to Venus, and Mariner-10 went to Venus and Mercury
in 1973. A long duration Magellan mission went to Venus in 1989. For
missions either too far or too close to the sun, the radioisotope thermo-
electric generator (RTG) becomes a strong candidate, which is covered in
the next chapter.
The power system related environmental aspects in some interplanetary
and deep space missions undertaken by NASA in the past are discussed
below. The interplanetary space is relatively free of charged particles.
However, the charged particles trapped in the planet’s orbit due to the
planet’s magnetic field do impact the power system design. If the planet has
no magnetic field, its orbit is free of Van Allen type trapped particle belts.
However, untrapped particles traveling to the surface of the planet
unimpeded can cause some damage there. Since Mars has negligible
magnetic field, it has no radiation belts. On the other hand, Jupiter has a
high magnetic field, hence more trapped particles in its orbit.
FIGURE 19.3 Power system for NEAR-Shoemaker spacecraft to asteroid.

(From: J.E. Jenkins and G. Dakermanji, Proc. 36th IECEC, ASME, p. AP-15, 2001. With
permission.)
19.3 Lunar Mission

The moon, although not a planet, is the closest neighbor of significant size
that has drawn our attention for thousands of years. More recently, it has
been the subject of human exploration by actual landing on its surface.
Apollo-11 was the first manned mission that landed on the moon in 1969. In
2004, China announced its own unmanned lunar program with a goal of
landing in 2010. Also in 2004, in the US, President Bush announced a long-
term space program for human landing on the moon by 2015, and then on
Mars. The following are the environmental features that impact the power
system design for lunar missions.
The moon, being the Earth’s natural satellite, is approximately at the

same distance from the sun as the Earth. The solar flux in the moon’s
orbit is therefore essentially the same as that in the Earth’s orbit.
There is no atmosphere around the moon. The pressure ranges from
108 to 102 torr. This is essentially a vacuum, which can support a
high-voltage power system. However, out-gassing from the surround-
ing equipment can alter the local pressure. This should be accounted for
in the high-voltage breakdown considerations.
There is no space plasma. Plasma means ionized gas, and if there is no
gas in the atmosphere, there cannot be any plasma.
The temperature on the lunar surface ranges between 175 C during
night and þ75 C during day, about the same as that in the Earth’s orbit.
The gravity is about 1/6th of that on the Earth’s surface.
The moon has negligible magnetic field. This means that the sun’s
charged particles can reach the moon’s surface unimpeded, without
being deflected and trapped in belts. The solar array degradation rate
due to charged particles would therefore be higher than on Earth.
The meteoroid impact rate is higher on the moon, giving a higher rate of
mechanical damage on the solar array.
19.4 Mercury Mission

Mercury is the closest planet to the sun, and has a diameter of 4880 km,
gravity 1/18th compared to that on Earth, and no air or water. It is very
dense with a large iron core. Its surface has countless craters, the largest
being about 1300 km in diameter. It has a weak magnetic field. It orbits the
sun in 88 days at a distance varying between 0.31 AU and 0.47 AU. The
solar intensity varies from 10.6 suns to 4.6 suns, with the resulting solar
array temperature rising to 430 C at the equator. Since the planet has no
atmosphere and rotates slowly on its axis once in 60 Earth days, most of the
heat is lost at night with the temperature plunging to 160 C.
Due to its proximity to the sun, the electrical power system for a Mercury
mission must meet the harsh thermal and radiation environment. Haines2, of
the European Space Agency, has recently reported the power system design
for a mercury sample return mission. It consists of three independent power
systems for each phase of the mission. For example, a 20-kW, 100-V high-
power system shown in Figure 19.4 is used for electric propulsion, which is
jettisoned just before the orbit insertion. After that, a 500-W, 28-V system is
used for the orbiter, and a smaller power system for the surface landing,
sample collection, and return phase of the mission. The electric propulsion
power system array has the following design features:
Peak power tracking power system in which the array is rotated as it

approaches the sun in order to limit the temperature
The solar array has a thermal shield along the length of the panel to
assure that the panel edges are not overheated when tilted
The solar cells are GaAs with a 100-mm-thick ceria microsheet cover-
glass coated with infrared and Earth albedo reflectors
Adhesives used in the solar array fabrication are compatible with the
high temperature operation
The cells are interleaved with about 100-mm ceria microsheet surface
mirrors to reflect a part of the solar flux for controlling the temperature
Solar panels are rotated 180 around one axis to keep the operating
solar cell temperature below 120 C from 1 AU to 0.375 AU and below
150 C at AU < 0.375. The sun shielded solar array drive is used with no
FIGURE 19.4 Power system for electric propulsion to Mercury.

(Source: European Space Agency/J. Haines.)
slip rings, but with flex harness (coiled cable) sufficiently flexible for
180 rotation
Lithium ion battery is used to meet the orbit parameters
19.5 Near-Sun Mission

For missions near the sun, the solar intensity increases to 100 suns at 0.1
AU (21 sun radii, 1 sun radius equals 0.00476 AU), and to 2500 suns at
0.02 AU (about 4 sun radii). The PV power system design for near-sun
missions between Mercury and the sun needs special considerations due
to high temperature. The PV cell loses power generation capability at
such temperature due to loss in the open circuit voltage. Various options
to limit the temperature below 1000 C include array tilting, adding
mirrors on the surface to decrease absorptivity and increase emissivity,
partially silvered cover glass, and various louvers and shades to control
the solar flux. Moreover, the PV cell having high band gap is needed.
Figure 19.5 from a paper by Brandhorst and Chen3 shows effective
power output as a function of AU distance and band gap of various PV
cells operating below 1000 C. At distances greater than 0.5 AU, the
bandage has no significant effect on power generation. At a distance of
less than 0.5 AU, the higher band gap PV cell generates more power up
to 0.1 AU. Closer than 0.1 AU, the PV cell becomes useless. The curves
assume that the cell temperature is limited to 1000 C in all cases.
It is apparent from Figure 19.5 that the PV power system has limitations
in approaching the sun at a close distance. An alternative approach may be
FIGURE 19.5 The distance from the sun versus power output of PV cells of various band gaps
operating under 1000 C.
to use the thermo-photovoltaic (TPV) direct energy conversion. There are

several advantages, including easy coupling to a thermal source operating
above 2000 K. The feasibility of such an approach has been demonstrated
under DoE funding, but has not been fully developed.
Another alternative is to use the thermoelectric converter (TEC) with the
sun as the heat source. Such system is feasible for solar probes requiring
instrument power under a few hundred watts. For example, NASA/JPL
designed for launch in late 2003 the sun-TEC power system for a fly-by
probe to Jupiter and then towards the sun to study the coronal heating and
origin and acceleration of solar winds. It will become the first spacecraft to
fly in the sun’s atmosphere. In the power system design reported by Choi4,
the probe’s distance from the sun varies greatly, from 5.2 AU near Jupiter
(gravity assist orbit) to less then 0.1 AU (4 solar radii) near the sun. The
corresponding solar flux varies over 5 orders of magnitude from 50 W/m2
to 4 106 W/m2. The spacecraft bus is shaded by the primary sunshield
blocking the sun (Figure 19.6). The shield outside temperature is estimated
to be 2100 C at 4 solar radii. A high-temperature multi-layer thermal
blanket keeps the spacecraft components cool. The shield and the blanket
are made of carbon–carbon composite.
The solar probe has three different power systems: (a) low temperature
solar array for AU > 0.5, (b) high-temperature solar array operating at or
below 400 C between 0.5 and 0.1 AU, and (c) TEC for AU < 0.1 when the
solar array is retracted inside the umbra of the sunshield.
Near the Earth, the low temperature solar array generates 2700 W of
power. Since this is far more than needed, most of it is shunted in the
power system. Between 0.5 and 0.1 AU, the power is provided by the
high temperature solar array, which is retracted inside the umbra of
the sunshield at 0.1 AU when the array temperature exceeds 400 C. Inside
0.1 AU up to 4 sun radii, 200 W of power is generated by using infrared heat
from the primary sunshield on the TE converters. In a concept described by
Choi (Figure 19.7), silicon germanium/gallium phosphorus (SiGe/GaP) TE
converters use infrared radiation from the spacecraft primary heat shield as
the primary energy source. A liquid sodium high-temperature cooling
system to maintain the TE hot side to 1200 K. The (SiGe/GaP) converter
efficiency is about 7% at 1200 K. The remaining 93% heat is radiated back into
space by using carbon–carbon composite and liquid sodium heat pipe
radiators.
19.6 Mars Mission

A spacecraft takes more than 7 months to reach Mars — also known as ‘the
red planet’. By contrast, going to the moon takes just 3 days. Although
Mars, as a planet, is close to the Earth, it has proved to be very inhospitable
to us. Of about 35 missions since 1960, more than 22 have crashed en route
FIGURE 19.6 A solar probe in the intense heat of the sun at an AU distance of 0.1.
(From M.K. Choi, Proc. 36th IECEC, ASME, p. TM-17, 2001. With permission.)
or otherwise failed before they could complete the mission. And yet, in June
2003, we launched four spacecraft, two by NASA and two by ESA, to study
the history, geology, and biology of Mars. NASA’s rovers, Spirit and
Opportunity, reached Mars in January 2004.
Mars probably once had liquid water on its surface, but now its CO2
atmosphere is so thin that the planet is dry and cold. The Mars environment
has the following features:
The distance from the sun is 1.66 AU, with the average solar flux 36% of
that in the Earth’s orbit.
The typical operating temperature for large area solar array range
between 100 C and 0 C. The low-temperature affects material proper-
ties and PV cell performance.
The surface gravity is about 40% of that on the Earth.
FIGURE 19.7 The TEC layout for NASA’s solar probe.
The atmosphere consists of 95% CO2, 3% nitrogen, and 2% all other

gases.
The atmospheric pressure is low — in the order of 10 torr. This happens
to be the worst possible value on the Paschen voltage breakdown curve
for CO2. At 10 torr, the breakdown voltage between bare conductors in
CO2 is 50 V. Therefore, the power systems design with exposed
conductors must be limited to 35 V to leave some design margin.
Higher voltages must be encapsulated or shielded in an enclosure with
a suitable pressurized gas, which can conveniently be CO2.
The dry low-pressure atmosphere is conducive to electrostatic charging.
The spacecraft surfaces are likely to collect a large amount of
triboelectric charge via moving surfaces and wind-born dust particles.
The temperature ranges between 100 C and 30 C, somewhat cooler
than in the Earth’s orbit.
The great dust storms on the Mars surface cover an area greater than
1 million km2 with durations over 100 days and a peak opacity
exceeding 6. The dust is about 60% SiO2. Such storms occur with a
yearly probability of 30 to 80%. The atmospheric dust scatters and
absorbs solar wavelength radiation to affect the magnitude, angular
distribution, and the solar radiation spectrum.
Local dust storms covering less than 1 million km2 area occur with a
probability of 5% in the equatorial regions. They have only minor
impact due to their limited size, short duration of a few days, and
opacity around 1. Frequent dust storms occur in gusts of 30 to 75 miles
per hour.
The dust haze, ground fog, CO2, and water ice clouds obscure the
sunlight. The dust particles precipitated and collected on surfaces
following the dust storm activity reduce the transmission of light and
decrease the surface emissivity. This raises the array temperature. The
PV cell current loss of 0.28% per Martian day due to dust precipitation
was measured over a short period of time in 1997 during the Pathfinder
mission. The dust collection problem makes the solar array effective
only for 3 to 6 months on Mars.
The wind speed in the upper atmosphere can exceed 100 m/s.
However, near the surface boundary layer, the dust wind speed can
range from 3 m/s at the surface to about 50 m/s at 5-m elevation.
With an atmospheric pressure of only 6 torr, the PV array aerodynamic
loading is modest in high wind, about twice the body force.
Mars has a negligible global magnetic field — about 3 105 times that of
Earth. However, NASA’s Mars global survey data suggest that the planet
has pockets of magnetic areas that can be 10 to 20 times stronger than the
Earth’s magnetic field.
The electrical power system design issues in low Mars orbit are
somewhat similar to that in LEO, including the AO degradation and arcing
to plasma. On the surface of Mars and the moon, Paschen breakdown of
atmospheric gases in a low-pressure neutral environment, and chemical
and electrical interactions with dust may also be of concern.5,6 The Mars
environment is thick. It provides effective shielding from meteors and
proton/electron radiation. Therefore, the PV array design can ignore the
radiation degradation and meteoroid damage. The design must, however,
consider the radiation protection of human crews and sensitive electronics.
NASA is studying a novel wind power system using the sand storms on
Mars. Wind turbines designed to general electricity at Earth’s South Pole
and in remote regions of Alaska, where there are about 6 months of
darkness each year, may some day lead to similar wind machines for Mars
bases. One reason to propose wind power on Mars is that wind turbines
could generate electricity during the month-long Martian global dust
storms that can make days on the red planet as dark as night. Wind power
and solar power may complement each other on Mars. When a large dust
storm blocks the sunlight, a wind turbine can generate electricity.
19.7 Missions to Jupiter and Saturn

Jupiter is the largest planet, with an immense magnetosphere. Its hydrogen
and helium atmosphere contains pastel-colored clouds. Saturn is similar to
Jupiter in many ways. The sun is not bright enough near Jupiter and Saturn,
making the PV power system ineffective. The electrical power is therefore
generated using the RTG, which is covered in the next chapter. A few
missions to Jupiter and Saturn are described below.
NASA’s Galileo spacecraft to Jupiter at 5.2 AU distance from the sun was
a mission in two parts, the orbiter and the probe. The probe was designed to
enter the Jupiter environment. After delivering the probe, the orbiter was
routed to the Jupiter’s moon Ganymede. The probe recorded a temperature
of 171 C above the clouds, 305 C near the surface, and the wind speed of
400 miles per hour. The probe was destroyed due to high atmospheric
pressure. In 2003, the orbiter concluded its 14-year exploration of Jupiter
and its moons. This 3000-pound spacecraft traveling at 108,000 miles per
hour ran out of the on-board fuel used to steer it during its 35 orbits of
Jupiter. Galileo was then purposely plunged into its own atmosphere to
avoid crashing and contaminating Europa, one of Jupiter’s watery moons.
The radiation hardening of the power system components is more
important in Jupiter orbits than in orbits of other planets. This is because
Jupiter has a strong magnetic field and strong Van Allen type radiation
belts. Spacecraft encountering the powerful radiation belts of Jupiter have
undergone single event upsets. This becomes a significant design issue.
Pioneer-10 spacecraft was built by TRW Corporation (Figure 19.8) and
launched in 1972 to probe the environment of Jupiter. It was designed for a
3-year life, but kept ticking for more than 30 years. It passed Pluto’s orbit in
1983. In 1997, it was some six billion miles away from Earth, much beyond
Pluto, cruising at 27,700 miles per hour to become the farthest man-made
object in deep space. It used RTGs with plutonium-238, and produced more
than two thirds of its 1972 power in 1997. Citing diminishing return, NASA
FIGURE 19.8 The Pioneer-10 spacecraft with two RTG power sources at the end of two booms.
(Source: NASA Ames Research Center.)
halted all regular communications with Pioneer-10 in 1997. The last regular
signal received from it was 1 1021 W, and took 9 h to reach Earth. In 2002,
scientists at NASA’s Jet Propulsion Laboratory sent a signal more than 12
billion kilometers away in an attempt to contact the spacecraft. After 22 h,
they detected Pioneer’s response on a 70-m dish antenna in Madrid, Spain.
Even without communication to and from Earth, it will continue its journey
in deep space. It carries pictures of terrestrial man and woman, and our
address, i.e., the location of the Earth in our solar system. It cannot reach
Proxima Centauri, the nearest star next to our sun, as it would take 100,000
years at the current speed.
Voyager-1 is another spacecraft, launch in 1977, which passed by Jupiter
and Saturn. In 2004, it left our solar system and entered the interstellar
space, when it was more than three times as far as Pluto from the Earth.
However, it is still working after more than a quarter century since its
launch. If it continues at its present speed of over one million miles a day, it
could take 40,000 years before reaching the nearest star.
The Cassini spacecraft was launched in 1997 on a 7-year mission to
Saturn. It is designed to send back data after entering the orbit of Saturn in
late 2004 and making 60 orbits in various orientations around the planet
over a 4-year period. The mission is a joint undertaking of NASA, the
European Space Agency, and the Italian Space Agency. It carries RTG with
30 kg of plutonium on board. Because of such a large amount of the
potentially deadly radioactive element, antinuclear groups opposed its
launch. Public concerns have negatively affected the future of nuclear
power in space, although Russians have built and flown many small
nuclear reactor–TEC power systems similar to the U.S. built SNAP-10.
19.8 Deep Space Missions

The power system for a deep space mission typically includes RTGs, power
electronics, and a small battery located inside the spacecraft body. The RTG
heat may be sufficient to protect the power system from low temperature. If
not, additional isotope heat is needed to keep the electronics at the required
temperature. For example, an unheated interplanetary spacecraft launched
to explore the rings of Saturn would experience an average temperature
of about 190 C, the liquid nitrogen (LN) temperature. For this reason,
low temperature power electronic circuits have the potential of finding
applications in deep space missions, and also in cryogenic instrumentation
in superconducting energy storage, which is covered in Chapter 26. Such
circuits designed and operated at low temperature may result in a more
efficient system layout than the room temperature circuits. The advantages
include reducing or eliminating the thermal shutters and the need for an
isotope heat, which can cause overheating during launch. Understanding
the performance of power electronics at extreme low temperature is needed
for this purpose. The following is known about the operation of power
electronic components near liquid nitrogen temperature.
The performance of certain semiconductor devices improves with
decreasing temperature down to liquid nitrogen temperature. At low
temperature, the majority of carrier devices demonstrate reduced leakage
current and reduced latch-up susceptibility. In addition, these devices show
higher operating speed resulting from increased carrier mobility and
saturation velocity. An example is the power MOSFET, which has lower
conduction loss at a low temperature due to the reduction in drain-to-
source resistance Rds(on) resulting from increased carrier mobility. NASA
has tested other components such as resistors, capacitors, and magnetics
that are needed for various power converters at liquid nitrogen operating
temperature. Many of them have been found suitable for operating at
unheated interplanetary spacecraft.7
The battery can be a roadblock at very low temperature. The Li-ion
battery offers a somewhat favorable combination of the energy and power
density. However, its temperature performance below 40 C is poor. Tests8
have shown the following about the Li-ion cells:
Between room temperature and 20 C, variations in electrolyte

resistance and the anode to electrolyte resistance are negligible, but
the cathode electrolyte interface resistance increases substantially.
The cell voltage and Ah capacity fall to approximately one half at
40 C (Figure 19.9). As a result, practically no energy was delivered at
40 C. This is due to substantial increase in the total internal resistance.
Poor cell performance at low temperature can be attributed to the

electrolyte becoming viscous or solid as the temperature is lowered.9 It is
FIGURE 19.9 The capacity of a 5-Ah Li-ion cell at low temperature in interplanetary missions.
also attributed to the poor lithium diffusivity in the electrolyte. Work is

under way to improve the low temperature performance of the Li-ion cell.
References
1. Jenkins, J.E. and Dakermanji, G., Near Earth asteroid rendezvous—
Shoemaker spacecraft power system flight performance, Proceedings of the
36th Intersociety Energy Conversion Engineering Conference, ASME, 2001,
Volume I, pp. 251–256.
2. Haines, J.E., Inner planets sample return missions, the challenge for power
systems, Proceedings of the 34th Intersociety Energy Conversion Engineering
Conference, SAE, 1999, Paper No. 2483.
3. Brandhorst, Jr, H.W. and Chen, Z., PV approaches for near-sun missions,
SAE, 1999, Paper No. 2631.
4. Choi, M.K., Power and thermal systems with thermoelectric generators at
930 C for solar probe inside 0.1 AU, Proceedings of the 36th Intersociety
Energy Conversion Engineering Conference, ASME, 2001, Vol. II, pp. 1161–
1163.
5. Kerslake, T.W. and Kohout, L.L., Solar electric power system analysis for
Mars surface missions, Proceedings of the 34th Intersociety Energy Conversion
Engineering Conference, SAE, 1999, Paper No. 01-2482.
6. Kilecki, J.C. and Hilalrd, G.B., Proceedings of the Electrical and Chemical
Interactions at Mars Workshop, NASA Report CP-10093, 1992.
7. Elbuluk, M.E. et al. Low temperature performance evaluation of battery
management technologies, Proceedings of the 34th Intersociety Energy
Conversion Engineering Conference, SAE, 1999, Paper No. 01-2543.
8. Nagasubramanian, G, Low temperature electrical performance character-
istic of Li-Ion cells, Proceedings of the 34th Intersociety Energy Conversion
Engineering Conference, SAE, 1999, Paper No. 01-2462.
9. Smart, M.C, Huang, C.K., Ratnakumar, B.V., Surampudi, S., and
Sakamoto, J.S., Factors affecting Li-Ion cell performance, Proceedings of
the 37th Power Sources Conference, Paper No. 239, 1996.
Chapter 20
Radioisotope Thermoelectric Generator
20.1 Introduction
Deep space and outer planetary missions cannot effectively use photo-
voltaic power generation due to insufficient solar flux. For those missions,
an on-board nuclear energy source or a radioactive isotope is often used to
generate electrical power. The radioisotope heat is directed at a thermo-
electric junction, which generates electrical potential just as in a thermo-
couple. This concept was briefly presented in Section 3.6, and is covered in
detail in this chapter.
The radioisotope thermoelectric generator (RTG) has been fully devel-
oped and used for decades for power levels in several hundred watts. Such
a power source has an advantage of supplying power all the time, thus
eliminating the need for a battery in a base load system having no peak
power requirement. An obvious disadvantage is the heavy radiation
shielding required around electronic components. Also, the nuclear fuels
that are safe and easy to handle with little shielding, such as curium-244
and plutonium, are expensive. Inexpensive and easily available fuel, such
as strontium-90, is unsafe.
High-energy particles emitted from the radioactive isotope material are
the primary sources of energy, which heats the absorbing material. The
mass of the isotope decay exponentially at a rate characterized by half-life,
T1=2 . The thermal radiation decreases proportionally with the remaining
mass. Therefore, the thermal power P(t) radiated at any time decays
exponentially from its initial value Po, as given by
PðtÞ ¼ Po exp ð0:7t=T12 Þ ð20:1Þ
Table 20.1 compares the isotope fuels presently used in RTGs with their
half-life and specific power achievable.
The advantages of an RTG are:
It provides power for a long period of time, independent of the

spacecraft orientation and distance from the sun.
It is suitable for missions far away from the sun, too close to the sun, or
lunar missions with long eclipse periods.
489
Table 20.1 Characteristics of three alternative isotopes for RTGs
Half-life Thermal power radiated Radiation level

Isotope (years) per gram of isotope (W/g) (Curies/W)
Curium-242 0.45 100 27

Strontium-90 28 0.25 153
Plutonium-238 86 0.55 30
The power output is not affected by radiation damage in the Van Allen
belts or due to man-made nuclear threats.
The RTG consists of numerous thermoelectric converter (TEC) cells

connected in series–parallel combination to obtain the required voltage and
current. Each TEC converts the isotope thermal energy into electrical
energy. The power conversion efficiency, , of the TEC is defined as
electrical power output

¼ ð20:2Þ
thermal power depletion
The efficiency depends on the material properties and the hot and cold
junction temperatures Th and Tc, respectively. The theoretical limit on this
efficiency is the Carnot efficiency, which is
Th Tc
max ¼ ð20:3Þ
Th
Practical designs yield about one half the theoretical maximum efficiency.
The most widely used material, plutonium-238 with Si–Ge TE cells, gives
about 7% conversion efficiency. Removing the remaining 93% of the system
energy as waste heat poses a significant design challenge. The specific
electrical power output of RTGs is typically low. Based on the total power
system assembly mass, it was 2 W/kg in SNAP-19 spacecraft, and 5 W/kg
in Galileo spacecraft. Plutonium-238 costs more than a couple of million
dollars per kilogram. The DoE inventory of this isotope in 2002 was merely
10 kg. The U.S. currently does not produce it, but purchases it from Russia.
20.2 Thermoelectric Fundamentals

The working principle of the TEC is based on the Seebeck effect, which
generates electrical potential when any two dissimilar materials are
maintained at different temperatures. It involves electron or hole transfer
between the two dissimilar materials under thermal energy. The two
materials can be conductors or semiconductors. The TEC for space power
Radioisotope Thermoelectric Generator 491
applications uses semiconducting materials, one p-type and the other

n-type, as shown in Figure 20.1. If two such dissimilar materials are held at
temperature difference iT ¼ Th Tc, an electric potential difference V12 is
produced at their junction. It is given by
V12 ¼ 12 T ð20:4Þ
where 12 is known as the Seebeck coefficient of the couple, generally

expressed in mV/ C. The 12 is often called the TE power, although it is not
really a power. It is a characteristic constant, which depends on the material
properties. The coefficient 12 is considered positive if the Seebeck voltage
polarity produces current in the p-type material from high temperature to
low temperature. The current under this voltage then delivers electrical
power to the external load connected in the loop.
Each arm of the two materials has electrical resistivity and thermal
conductivity k. The parameters , k, and all vary with the operating
temperature. Since and k cause loss of power and reduces the conversion
efficiency, the figure of merit, Z, for each material in the couple is defined as
212
Z¼ ð20:5Þ
k
Another figure of merit for each material is the power output per unit mass.
It is defined as
212
Z0 ¼ ð20:6Þ

The figure of merit of both materials together in the couple is defined as
2
Z12 ¼ pffiffiffiffiffiffiffiffiffi 12pffiffiffiffiffiffiffiffiffi2 ð20:7Þ
1 k1 þ 2 k2
FIGURE 20.1 Single-stage unicouple thermoelectric converter.

where and k are the electrical resistivity and thermal conductivity of the
two materials 1 and 2, respectively.
Equation 20.7 suggests that a good TE junction has the following
properties:
High Seebeck coefficient , which generates a high voltage.

Low electrical resistivity , which results in low ohmic loss.
Low thermal conductivity k, which gives high iT.
For a junction of two metallic arms, such as in thermocouples used to

measure temperature, the Seebeck coefficient is low around 50 mV/ C. Both
and k are large, resulting in poor efficiency around 0.1%, which is
unsuitable for power generation. Some semiconductors have high Seebeck
coefficient in the 300 to 500 mV/ C range. Their conductivity is relatively
low, resulting in relatively high conversion efficiency in the range of 5 to
7%.
The fraction of heat removed from the junction and delivered to the load
in the form of electrical power is
8 9
>
>
<
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi >
>
Tc 1 þ Z T 1 =
Pe ¼ 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi T > ð20:8Þ
Th >>
: 1 þ Z T þ c > ;
Th
where
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
T ¼ Th þ Tc ð20:9Þ
Low-grade commercial TECs have potential terrestrial applications such

as recovering waste heat from various industrial processes and vehicle
exhaust to decrease fuel consumption. Typical space-qualified RTGs uses
high-grade SiGe TECs with plutonium-238 as the heat source. The SiGe
material is doped with phosphorus for the n-leg and with boron for the
p-leg. For widely used SiGe couples, Z12 is around 0.001 with 30%
variation depending on the material grade and quality of the manufacture.
Various TEC configurations used in the RTG designs for space applications
are described below.
20.2.1 Single-Stage Unicouple

This is the most basic elementary construction of the TEC converter that is
shown in Figure 20.1. One end of the p and the n arms are kept in a common
reservoir at hot temperature Th. The other end of the arms is kept at cold
temperature Tc. The d.c. voltage is generated between the p and n terminals
at the cold end, which is proportional to the temperature difference iT ¼

Th Tc. A load resistance R would draw electrical current I and power I2R.
Any form of heat can be converted into electricity. The conversion efficiency
obtained with the state of the art materials is about 7%.
If the cold temperature reservoir is a liquid metal bath, as is typical, the
p and n terminals must be kept at two different reservoirs electrically
isolated in order to withstand the generated voltage. Otherwise, the
resulting electrical short would prevent the power transfer to the load.
This configuration derives its name because there is only one TE couple,
which sees iT in one stage only. The heat energy at Tc is discharged from
the TEC system as waste heat. The efficiency is therefore relatively lower
than the other configurations described hereafter.
20.2.2 Single-Stage Multicouple

Two or more p–n couples are stacked together in the single-stage multi-
couple TEC assembly. Figure 20.2 show one such assembly with two
couples, which are thermally in parallel and see the same temperature
difference iT ¼ (Th – Tc) in one stage only. Electrically, however, they are
in series. The total voltage output of the assembly is therefore twice that of
one element. Material B in the figure is an electrical insulation with high
thermal conductivity, so that the copper straps C are at the same
temperature as the adjacent heat reservoir.
This configuration gives twice the voltage using twice the TE material.
Therefore, the specific power per kilogram of the active material remains
the same. However, the specific power of the system is higher due to the
overhead mass less then double.
FIGURE 20.2 A single-stage multicouple thermoelectric converter.

20.2.3 Multistage Multicouple

In this configuration, the heat is extracted in two stages as depicted in
Figure 20.3. The temperature drops from Th to Tm in the first stage having
two couples, and from Tm to Tc in the second stage also having two couples.
The material B (unshaded) is an electrical insulation of high thermal
conductivity to minimize the temperature drop across it. The copper strap
C (checkered) is a material with high thermal and electrical conductivities.
It conducts the electrical current and distributes the heat uniformly across
the device so that all couples in the same stage see the same temperature
difference.
20.2.4 RTG Assembly

The Seebeck voltage per junction is around 300 to 500 mV/ C. Numerous
junctions are connected in series to obtain the desired voltage, and
numerous strings of series junctions are connected in parallel to obtain
the desired current. The series–parallel combination is a matter of design
that results in the maximum power transfer to the load. The RTG is
FIGURE 20.3 A multistage multicouple thermoelectric converter.

basically an assembly of all TECs, which is typically a cylindrical canister as

shown in Figure 20.4. Once ignited, the isotope heat generation cannot be
turned off. The refractory metal insulation shield and the graphite isotope
capsule could be damaged if they were subjected to atmospheric oxygen
while operating at high temperature, and could experience sublimation
damage even before the mission starts. For this reason, the RTG is filled
with pressurized xenon gas until after launch. The very low thermal
conductivity of xenon makes it possible to store and launch the RTG hot
and generate the needed pre-launch and launch power. Once in space, the
xenon is vented.
20.3 Electrical Model of RTG

The electrical equivalent circuit of an RTG is shown in Figure 20.5(a). The
total voltage generated due to the Seebeck effect works as an internal
voltage source. With an open circuit (zero load current), the external
terminal voltage V is the same as that generated internally. We designate
this voltage as the open circuit voltage Voc. When electrical current is drawn
by load resistances RL, there is an internal voltage drop. This is represented
by an internal resistance Ri, which is approximately constant at a give
temperature. The external terminal voltage V therefore decreases linearly
with increasing load current, i.e.,
V ¼ Voc IRi ð20:10Þ
With the external terminals shorted, the maximum current flows to the
load. We designate this current as Isc. The above equation gives
FIGURE 20.4 A radioisotope thermoelectric generator assembly. (From IEEE Spectrum, p. 63,
March 1998. With permission.)
FIGURE 20.5 Electrical equivalent circuit model and operating characteristics of TEC.
Isc ¼ Voc =Ri ð20:11Þ
Equation 20.10 can be rearranged using Equation 20.11 as
I ¼ Isc V ð20:12Þ
where ¼ Isc/Voc, a characteristic admittance of the RTG power source.

Equation 20.12 sets the I–V characteristic of the RTG. It is a falling straight
line from Isc at zero voltage to zero current at Voc, as shown in Figure 20.5(b).
On the other hand, the voltage, V, required to circulate current I through the
load increases with the current in a straight line. This line, known as the
load line, has a slope of RL. An RTG having the source line shown by solid
line and the load line shown by dotted line operates at the point of their
intersection. At that operating point, the electrical power transferred to the
load is
P ¼ Vop Iop ð20:13Þ
20.4 Maximum Power Transfer

The power, P, transferred from TEC to the load at any operating point P in
Figure 20.5(b) is
P ¼ VI ¼ VðIsc VÞ ¼ VIsc V 2 ð20:14Þ
The power system design for extracting the maximum power from RTG to
the load must operate at a voltage such that, at the operating voltage,
dP=dV ¼ Isc 2V ¼ 0 ð20:15Þ
Equation 20.15 gives the corresponding operating voltage Vm ¼ Isc/(2),

which is also 1=2 Voc, and the current at that voltage is 1=2 Isc.
FIGURE 20.6 A TEC operation for maximum power transfer to load.
The maximum possible power transfer therefore is
Pmax ¼ 1=2Voc 1=2Ioc ¼ 1=4Voc Isc ð20:16Þ
It occurs when the RTG is operated at the voltage that is one half of the open
circuit voltage. This point is shown as Pmax in the power versus voltage plot
shown in Figure 20.6(b). The power at zero voltage is obviously zero. It is
also zero at Voc since the current is zero there. In between, the power rises
with the operating voltage, reaches the maximum value Pmax, and then falls
to 0 at Voc. The RTG conversion efficiency is maximum at the maximum
power transfer point. The I–V, P–V and –V curves with the operating
voltage as an independent variable are shown in Figure 20.7.
20.5 The Effect of Temperature and Aging

The characteristics displayed in Figure 20.7 are for a given iT across the TE
junctions. The I–V line of the RTG shifts upward for a higher iT, and
FIGURE 20.7 TEC current, power, and efficiency versus operating voltage.
FIGURE 20.8 TEC current versus voltage at different temperature gradients.
downward for a lower iT, as shown in Figure 20.8. The amount of shift is a
characteristic of the couple material.
Aging has little effect on the RTG output, because the basic heat source
has a half-life in decades. For this reason, the power generation degrades
little, as shown in Table 20.2. Most power degradation is due to slow
precipitation of the phosphorus doping in the n-type leg of the thermo-
couple. The I–V and P–V curves shifts uniformly with time and the
temperature such that the maximum power point remains at the same
voltage (Figure 20.9). This is a happy coincidence for the design engineer.
The conversion efficiency is a function of the contact resistance and the hot
and cold side temperatures as shown in Figure 20.10 and Figure 20.11,
respectively.
Table 20.2 Measured power levels of the RTG in Voyager II

versus time
Voyager II power Percent of initial

Date output, (W) power generation
August 1977 (launch date) 478 100

December 1980 436 91
December 1998 327 68%
(21 years) (after 21 years)
(Source: Data from Mondt, J.F., Proc. of the 36th IECEC, ASME, 2001,
Vol. I, pp. 133–139.)
FIGURE 20.9 I–V–P characteristics of RTGs with age.
FIGURE 20.10 TEC conversion efficiency versus contact resistance.
FIGURE 20.11 TEC conversion efficiency versus temperature.
499
20.6 Flight History of RTGs

RTG technology has been developed and flown in many interplanetary and
deep space missions for a couple of decades. RTGs have been used in
Pioneer, Explorer, Voyager, Galileo, Ulysses, and Cassini missions. These
missions have used essentially a standard RTG assembly designed to meet
mission requirements listed in Table 20.3. A number of such assemblies are
used as needed for a particular mission. For example, Galileo spacecraft for
exploring Jupiter and its moons used two RTGs (Figure 20.12), whereas
Cassini spacecraft exploring Saturn and its rings carried three RTG (Figure
20.13). These and other outer planet missions that have used RTGs in the
past are briefly described in Section 19.7. A new application of RTG in a
solar probe close to the sun as reported by Choi1 is discussed in Section 19.5.
20.7 Segmented TECs

In a unicouple TEC, the same heat and the same current is seen by each
segment in each leg. High conversion efficiency can be obtained by
operating the TEC material over a wide temperature range, and using
different materials in each temperature range where they posses optimum
performance. This can be accomplished in two ways:
Multistage TEC, where each stage is electrically isolated but thermally

in contact with each other, operates over a fixed temperature difference.
Segmented TEC, where each of the p and n legs is made of multiple
segments of different materials placed in series. Such a TEC is estimated
to have a conversion efficiency around 15% versus 7% using the state-
of-the art BiTe, FeSi, PbTe, and SiGe alloys. In the segmented TEC
concept, the length and cross section of each segment is adjusted in
accordance with its thermal and electrical conductivity to achieve the
desired temperature difference across each material and the maximum
Table 20.3
Performance characteristic of the standard RTG in use

Mission life 5 years
BOL power 285 W
EOL power 255 W (notice the negligible decay over mission life)
Suitable for space shuttle launch
RTG module 42 cm diameter 114 cm long
Total mass 55 kg
TE 572 SiGe unicouple in 16 circumferential rows
Operating voltage 28 to 30 V
Operating temperature gradient iT ¼ 707 K
Operating temperature: 1273 K hot junction, 566 K cold junction
FIGURE 20.12 Galileo spacecraft using two RTGs attached to two long booms.
(Source: NASA.)
efficiency of the couple. The overall operating temperature drops from

1275 C to 300 C in typical segmented TEC designs.
20.8 Advanced RTGs

The DoE recently issued a directive to develop and maintain the nuclear
space power program to support defense and deep space missions. The
program has funded research and development for the Advanced Radio-
isotope Power System (ARPS) with a goal to triple the conversion
efficiency of the present TECs. A key thrust is on using advance SiGe
cells with improved thermal conductivity and contact resistance of less
than 10 m/cm2. The program includes building an engineering model of
FIGURE 20.13 Cassini spacecraft in operational configuration with three RTGs.

(From IEEE Spectrum, p. 62, March 1998. With permission.)
100 W output and flight qualified unit around 2005. Table 20.4 summarized
the ARPS’s goals.2
Three technologies under evaluations to achieve these goals are: (a)
segmented alkali metal thermal to electric converter (AMTEC) with 15%
efficiency and 7 to 9 W/kg specific power, (b) Stirling engine with 25%
efficiency and 6 to 7.5 W/kg, and (c) thermo-photovoltaics (TPVs). In each
case, the heat comes from the natural radioactive decay of nonweapons
grade plutonium-238 pellets, clad in protective layers of graphite and
iridium. The TPV was briefly described in Chapter 3, the Stirling engine is
described in Chapter 21, and the AMTEC is described below.
Table 20.4 Goals of advance RTG research programs
Performance Preset state of the art Goals of the Advanced

parameter RTG capabilities Radioisotope Power System
System efficiency 6.5% 13–25%

Specific power 5 W/kg 9–10 W/kg
Life > 20 years TBD
TEC junction SiGe Improved SiGe
Operating temperature 1275 C to 575 C 1000 C to 700 C
Isotope Pu-238 Pu-238
In the AMTEC, the isotope heat vaporizes sodium that flows through a
-alumina solid electrolyte tube to produce electricity. -Alumina is an alkali
metal conducting ceramic: a solid electrolyte that conducts sodium ions. An
electrochemical potential is generated when sodium is present at two
differential pressures separated by an electrolyte. The free electrons are
captured by the electrodes, which are used to power the load and then get
recombined with the sodium ions. The temperature in the vapor circulation
drops. The conversion efficiency is a function of the differential temperature.
The electrical characteristic of AMTEC module is similar to that of the
conventional TEC. The terminal voltage drops linearly with increasing
current. The power output peaks at certain voltage, approximately one half
of the open circuit voltage. As in the traditional TEC, the open circuit
voltage and the maximum power in AMTEC are higher at higher-
temperature gradient.
20.9 Thermoelectric Cooler

The TEC junction can also be used to cool a small device locally by
replacing the load by a battery and sending the current backward into the
TEC. This is actually the TEC working in the reverse mode. The junction
works in the reverse by taking the electrical energy and converting into a
temperature drop. Such reverse operation is called the Peltier effect. The
Peltier heat absorption is given by
Qp ¼ 12 J12 ð20:17Þ
where
Qp ¼ rate of Peltier heat absorption per unit area of the junction,

12 ¼ Peltier coefficient of the couple, J12 ¼ current density taken as
positive for current flowing from material 1 to material 2 at the
junction.
Assuming that the cold junction is perfectly insulated thermally, it produces
a temperature drop
Tmax ¼ 1=2 ZTc2 ð20:18Þ
The cooling effect can also be created by using the TEC as a heat sink in
the forward mode for cooling a delicate small device. In this mode, the heat
near a device is converted into electrical energy, conducted by wires, and
dissipated into a load away from the device to be cooled.
References
1. Choi, M.K., Power and thermal systems with thermoelectric generators at
930 C for solar probe inside 0.1 AU, Proceedings of the 36th Intersociety
Energy Conversion Engineering Conference, ASME, 2001, Vol. II, pp. 1161–
1163.
2. Mondt, J.F., Advanced Radioisotope Power System Technology develop-
ment for NASA mission 2011 and beyond, Proceedings of the 36th
pp. 133–139.
Chapter 21
Dynamic System with Alternator
21.1 Introduction
Solar energy can be used in systems other than the photovoltaic. A dynamic
energy conversion system is an example, where the sun’s energy is collected
in the form of heat using a concentrator. The heat, in turn, is used to
produce steam and drive a rotating turbo-generator or a reciprocating
alternator to generate electricity. Such a system was a primary candidate for
the space station design in the 1980s with an estimated power requirement
of 300 kW. The system configuration is shown in Figure 21.1. A parabolic
concentrator focuses the sun’s heat on to a receiver, which boils a fluid. The
fluid can be a suitable liquid metal, such as potassium chloride. A high-
pressure stream of liquid metal produced in the receiver would drive a
turbine based on the Rankine cycle. The fluid can also be a gas, such as a
mixture of helium or xenon having the molecular weight around 40. The
heated compressed gas would drive a turbine working on the Brayton cycle.
A gas-based system minimizes erosion and sloshing problems in transport-
ing the liquid metal. A functional schematic of dynamic power system for
space is shown in Figure 21.2. In either a liquid metal or a gas-based system,
the high-pressure high-temperature fluid drives the turbine, which in turn
drives an electrical generator. Waste heat transferred to the liquid coolant is
dissipated via radiator panels to space. The energy conversion efficiency is
FIGURE 21.1 A solar dynamic power system schematic.
505
FIGURE 21.2 The functional blocks of a dynamic power system.
much higher than the photovoltaic system. This minimizes the deployed
collector area and the aerodynamic drag in low Earth orbit.
The usable energy extracted during a thermodynamic cycle depends on
the working temperatures. The maximum thermodynamic conversion
efficiency that can be theoretically achieved with the hot side temperature,
Thot, and the cold side temperature, Tcold, is given by the Carnot cycle
efficiency, which is
Thot Tcold
carnot ¼ ð21:1Þ
Thot
where the temperatures are on the absolute scale. The higher the hot side
working temperature and lower the cold side exhaust temperature, the
higher the efficiency of converting the captured solar energy into electricity.
The hot side temperature, however, is limited by properties of the working
medium. The cold side temperature is largely determined by the cooling
method and the environment available to dissipate the exhaust heat.
An indirect but major advantage of this system is that the energy storage
is interwoven in the system at no extra cost. It resides in the latent heat of
phase change at high temperature — around 1000 K. The systems can store
thermal energy for hours with no degradation of electrical performance, or
longer with some degradation. This feature makes the technology capable
of meeting peak power demands with no added mass or cost of separate
energy storage. It eliminates the battery requirement altogether.
Today’s space power systems using well-proven PV technology provide
up to a few tens of kilowatts with system specific power of 10 W/kg and life
up to 15 years. Although the solar dynamic technology is not yet proven in
space flights, it offers potential advantages in efficiency, weight, scalability,
and the overall cost in high power spacecraft. The cost advantage comes
from elimination of costly semiconductor PV cells. The systems can be
Dynamic System with Alternator 507
extremely cost effective in a few kilowatts to hundreds of kilowatts power

range. The concept is sufficiently developed for use in the near future,
particularly in high-power LEO missions. It may also find applications in
high-power defense spacecraft where a large solar array can make the
mission nonmaneuverable and vulnerable to detection and attack by
enemy. It has been considered for a 300-kW space station and for a
dynamic isotopes power system (DIPS) for space defense.
The efficiency advantage in the dynamic system comes from the higher
efficiency of the engine (20 to 40%) as compared to silicon solar cells (15 to
20%), and higher efficiency of thermal energy storage of the receiver (85
to 90%) as compared to the battery efficiency (70 to 75%). The greatly
improved overall system efficiency as compared to the PV system translates
into less solar collection area. This results in reduced drag and less concern
regarding station dynamics, approach corridors, and experimental viewing
angles. The reduced drag is particularly important because it allows lower
flight altitudes within given constraints of drag–makeup fuel and orbit
decay time. At power levels near 100 kW, such as for space-based radar, the
PV solar array collector area becomes prohibitive. It is in this power range
that solar dynamic power system is expected to find advantageous
applications. It offers the following additional advantages:
Low solar collection area

High voltage a.c. power generation
Highly scalable and mass producible
Potentially long life components
Inherent radiation tolerance
Recent prototype testing of a 2 kW non-optimized solar dynamic systems

reported by Mason1 demonstrated a conversion efficiency of almost 30%
using 1990s component technologies. Significant improvements in effi-
ciency can be realized for large systems in ratings above 100 kW using
newer technology components and optimized design parameters.
The dynamic system uses an electromechanical energy converter in
which an armature coil moves in a rotary or linear motion under an
alternating magnetic field of north and south poles. The generated voltage
in the armature is therefore alternating at a certain frequency. The terminal
voltage output equals that generated in the armature less the internal
voltage drop. The mechanical power source driving the armature is
typically a steam or gas turbine or a reciprocating engine. The electro-
mechanical energy conversion efficiency is in the 85 to 98% range
depending on the system rating and configuration. The machine size and
mass primarily depend on the power rating, and secondarily on the
adopted thermal design to remove the internal power losses. A smaller
machine generally results in lower efficiency, hence relatively higher power
loss and greater thermal system mass. A rough rule of thumb in trading
weight and loss over a narrow range using the same materials and design
configuration is to assume that the product of the machine mass and the
power loss remains constant.
21.2 Thermodynamics
On the thermodynamics side, the following two major configurations are
used.
21.2.1 Stirling Engine

The Stirling concept consists of two isotope heat source modules
surrounded by multifold thermal insulation, two Stirling engines driving
electrical alternators, and housing that acts as the primary structure (Figure
21.3).2 Isotope heat vaporizes a salt under pressure to drive a steam turbine
and electrical generator. The working temperature is around 600 to 700 C
on the hot side and 100 to 120 C on the cold side.
The Stirling technology demonstration prototype has been built and
tested successfully on the ground. It used a linear alternator for rated
electrical power output of 55 W at 90 V and 80 Hz frequency. The alternator
FIGURE 21.3 Stirling radioisotope power system — baseline concept.

(From R. Cockfield, Proc. 35th IECEC, AIAA, p. 2843, 2000. With permission.)
output was electronically converted into 28 V d.c. suitable for use on the
spacecraft.
The reliability can be an issue in such concepts with only two engines. If
one fails, the other becomes overloaded unless both are designed for
delivering full power, which would result in a heavy mass penalty. For this
reason, the number of Stirling engines in practical systems designs is likely
to be four. The total specific power of a few hundred watts power source
with a matching converter is estimated to be around 8 to 10 W/kg in two-
engine systems and 4 to 5 W/kg in four-engine systems.
As reported by Thieme, Qiu and White,3 the baseline electrical power
rating of the present Stirling converter is 55 W. Multiple units will be
connected in parallel to achieve the desired power levels. Key design and
operating issues are parallel operation of two thermodynamically indepen-
dent free-piston Stirling converters and vibration reduction to about 2% of
that of single unbalanced converters over the entire mission life even with
one engine failed. An EMI compatible design for science missions is also a
challenge, which is addressed by Sargent.4
NASA GRC and DoE with two industry partners are actively developing
Stirling converters for an advanced radioisotope power system to replace
the currently used RTG systems for deep space missions. A standard
module of 100 W using a thin long alternator is being developed, which can
be connected in parallel for higher power. Boeing is developing a multi-
mission modernized 100 W RTG that will be flexible and adaptable to both
the orbiter and the lander missions to a planet. On the other hand, Lockheed
Martin is developing a Stirling radioisotope generator that will use
plutonium-238 heat and a reciprocating alternator driven by a Stirling
engine.
The multi-mission RTG is still 7% efficient, whereas the Stirling generator
would have about 25% efficiency requiring less plutonium. The disadvan-
tage of the Stirling is the moving piston that could reduce the reliability in
space. The multi-mission RTG uses a lead telluride TEC, which operates at a
lower temperature than the SiGe. This does not require a heat resistant
refractory metal structure, which oxidizes or erodes relatively easily in any
CO2 and CO in a planet’s atmosphere. One of these two new power systems
is targeted for use for a Mars mission in 2010.
The Stirling system is the most developed converter option for the
advanced isotope power system concepts presently being funded. It has the
following advantages:
High efficiency in excess of 20%, resulting in less heat rejection

Reduced isotope inventory by a factor of at least 3 compared to RTG
Specific power about 1.5 times that of RTG
It has proven to be reliable compared to the Brayton and Rankine
cycles.
21.2.2 Brayton Cycle

In this system, solar energy concentrated on to a receiver is transferred to a
Brayton heat engine for conversion to electrical power. It utilizes a turbine,
compressor, and rotary alternator. The working fluid is an inert gas. The gas
from the receiver is expanded through a turbine, cooled through a heat
exchanger where the waste heat is transferred to a liquid coolant and is
pressurized in a compressor before being re-heated by the solar receiver
(Figure 21.4). A recuperative heat exchanger between the turbine discharge
and receiver inlet is often used to improve the thermodynamic cycle
efficiency. The rotary alternator typically provides 3-phase a.c. at 1 kHz
frequency for further conditioning and distribution. The closed cycle
Brayton energy conversion system has significant advantage over other
energy conversion system due to high efficiency (up to 40%), long life up to
30 years and low recurring cost. These advantages are important where it is
necessary to supply large amount of electrical power and compressed gas.
NASA Glenn has demonstrated the operation of a 2-kW closed Brayton
cycle power conversion unit with a xenon ion thruster. The testing was
conducted as part of a nuclear electric propulsion test bed at Glenn. The test
successfully demonstrated that such power converter could supply and
regulate high-voltage electrical power to an ion thruster.
21.3 Electromechanical Energy Conversion

Various types of electrical alternators that can be considered for the
electromechanical energy conversion in the dynamic power system are:
Permanent magnet (PM) synchronous or d.c. machine

Synchronous machine with solid rotor (Rice or Lundell)
Synchronous wound rotor machine, brushless design
FIGURE 21.4 Brayton cycle schematic.

Homopolar induction machine with no rotor excitation

d.c. machine, brushless design with solid state commutator
Homopolar d.c. machine with solid or liquid metal current collector
Superconducting synchronous machine working at cryogenic tempera-
ture, with efficiency exceeding 99.5% and mass about 10% of the above
machines
The exciting current that produces the magnetic field in these machines is
always d.c., which must be controlled for regulating the voltage output.
Therefore, the permanent magnet machine, which inherently gives an
unregulated voltage output, is unsuitable here. The turbo-generator in the
energy conversion assembly often has two field excitation coils to produce
the required voltage. The series field supplies the bulk of the excitation
during normal operation and provides all excitation during start up from
rest. The shunt field supplies the remaining excitation for voltage control
during normal operation and for power factor control during start-up.
Since power is the product of torque and speed, the power and speed
ratings determine the shaft torque, and subsequently the overall machine
size and mass. If the machine must operate over a range of speed, its mass
will depend on the lowest speed at which the required power must be
delivered. For a.c. machines, the significant parameter becomes the volt–
ampere product per rpm, since the machine size depends on both the real
and reactive power output. The speed in turn depends on the frequency
and number of poles. Overall, the specific power improves significantly at
higher power and speed ratings.
Each of the machines, except the PM machine, is capable of producing
approximately twice the rated output for a short time. In addition, these
machines are usually capable of delivering several times the rated current
into the system fault to clear the fuse or trip the circuit breaker. This
additional output capacity is extremely useful when the machine must
operate over a variety of operating conditions. In a system where such
added capabilities are not needed, a mass saving of 15 to 20% is possible.
On the negative side, drawing twice the load for short time imposes a
transient torque of 3 to 4 times the rated value on the prime-mover unless
specific steps are taken to prevent it. The machine mass is also impacted by
the required quality of power expressed as departures from the rated
voltage and frequency, and the transient voltage deviation during load
switching. Normal power quality requirements add about 10% in the
machine mass.
As for the environment, the shock and vibration during launch and in
orbit may not pose a severe limitation on the rotating machines design.
However, the radiation damage on the insulation, bearings, and semi-
conductors used in construction may be of concern.
The reliability and life are limited by fatigue in the metal parts, and by
thermal degradation of the insulation at operating temperature. Typical
electrical machine used in the aircraft industry has an MTBF ranging from
2000 to 10,000 h. Attaining an MTBF over 100,000 h is necessary for a
10-year mission in space. Since this is an order of magnitude higher than the
aircraft machines, it is apparent that the electromechanical machines for
spacecraft must be design using significantly better materials and design
configuration.
21.4 Electrical Generator Technologies

For a 300-kW solar dynamic power system, NASA’s early space station
designs screened 22 generator (alternator) concepts from which four
candidates were selected for further evaluation and development in the
mid-1980s. The configurations of these machines are shown in Figure 21.5.
21.4.1 Brief Description of Each Machine

21.4.1.1 Synchronous generator with PM rotor
This machine has a very simple construction. The stator has one set of
output windings, and the rotor has permanent magnets to produce the
magnetic field. Compared to other machines, it has a higher efficiency and
lower specific volume. The output voltage is regulated by external
electronics. Until recently, such machines were limited to small ratings
due to limitations in the permanent magnet strength. However, new
anisotropic neodymium–iron–boron permanent magnets not only have
high energy exceeding 20 MGOe, but also have high heat resistance with
capability of operating above 100 C. Magnequench Corporation produces
the anisotropic powders, which are used to produce bonded magnets by a
rapid solidification process.
21.4.1.2 Synchronous generator with solid rotor (Rice or Lundell)

This machine is a brushless synchronous generator. The field excitation coil
is stationary, and the flux is carried to the rotor through two auxiliary air
gaps at each end of the rotor. This improves the speed characteristics. It has
a smooth solid rotor with alternating magnetic portions of the rotor
separated by nonmagnetic portions, which are primarily structural
elements that provide strength and pole separation in the rotor. The flux
alternates under rotation of the alternating magnetic and nonmagnetic
portions.
21.4.1.3 Synchronous generator with wound rotor and brushes exciter

The rotor of this machine has four sets of active elements on a common
shaft, namely, the main field winding, the armature of the main exciter, the
FIGURE 21.5 Constructional features of candidate electrical generators.
pilot exciter’s permanent magnets, and the rectifier diodes. The d.c. power
is fed to the exciter via rectifier that eliminates brushes. The voltage
regulation is controlled by varying the excitation.
21.4.1.4 Induction homopolar generator

The induction homopolar machine uses a fixed toroidal field coil. When
power is applied to the field, a unidirectional magnetic flux is created and it
follows the path as shown in the figure. The stator output voltage of the
homopolar machine is varied by varying the current of the toroidal field coil
in the stator. It has a solid rotor with notches. The d.c. field winding on the
stator applies variable flux to the output winding through the notched
rotor, which results in a.c. power generation. This generator is mechanically
simpler but has a higher specific weight and larger specific volume.
21.4.1.5 Advantages of each machine

A second level screening ranked these four candidates on a relative basis as
shown in Table 21.1. The synchronous PM rotor machine ranked the first,
Lundell generator ranked the second, the third was the synchronous wound
rotor brushless generator, and the last one was the induction homopolar
generator. Different criteria and/or different weighting factors may lead to
different results. However, the synchronous generator with PM rotor and
the Rice generator appear to be the most suitable generator candidates. The
final selection of a generator and the power conditioning is based on the
most effective power system within the technical constraints, such as d.c.
or a.c., frequency, voltage level, power level, engine speed, etc. The Rice
generator has several advantages such as a relatively higher speed
capability and simple voltage regulation. However, it has a relatively
higher specific weight, higher reactance, and more complex excitation coils
when compared to the synchronous PM generator. Therefore, for a three-
phase 400-Hz a.c. power distribution system, the synchronous generator
with PM rotor using the cycloconverter approach similar to that used in the
aircraft appears to be the preferred candidate for the power generation and
conditioning. This type of generator is matched to the optimal engine speed
(45,000 rpm) to generate power in the frequency range of 1000 to 3500 Hz,
and to provide a naturally commutated cycloconverter to produce precise
Table 21.1 Key performance parameters of four candidate generators for

dynamic power system
Synchronous Synchronous Synchronous Induction

Generator type PM rotor Rice or Lundell wound rotor/brushless homopolar
Maximum rotor 230 250 180 300

tip speed (m/s)
Specific mass 0.011 0.031 0.011 0.213
kg/(kVAkrpm)
Specific volume 0.004 0.006 0.011 0.041
liter/(kVAkrpm)
Efficiency% 88–95 87–94 82–92 80–91
Voltage regulation Complex simple simple simple
Excitation Rotating PM Stationary Rotating diodes Stationary
d.c. windings from PM exciter d.c. windings
Reactance Low Medium Lowest Highest
Can operate as Yes Yes Yes No
motor
400 Hz output power. In addition, the cycloconverter approach has a higher

efficiency and reliability than the conventional d.c. link inverter technology.
21.4.1.6 Other criteria in machine selection

The following criteria are also important in the final generator technology
selection:
Parallel operation
Overload capabilities
Voltage regulation
Speed and frequency limitations
Start-up operations
21.4.2 Parallel Operations

Two or more identical generators are connected in parallel either to increase
the power capability or to improve the reliability of the group. For this
reason, only those generators which can work in parallel are suitable for
reliable space applications. Moreover, they should be modular and scalable
to higher ratings without designing and qualifying new designs for
different missions. All of the four candidate machines in Table 21.1 can
be operated in parallel.
21.4.3 Overload Capability

The ability to withstand temporary overload and to provide a high fault
current for fuse clearing are desired performance features of any machine.
Typical machine overload capabilities of the four machines are summarized
in Table 21.2. The synchronous PM and induction homopolar machines
have the greatest overload capability, and the synchronous wound rotor has
the least.
21.4.4 Voltage Regulation and Field Excitation

Voltage regulation of the four machines is compared as follows.
21.4.4.1 Synchronous generator with PM rotor

The output voltage of the PM generator cannot be regulated directly since
the flux is provided by the permanent magnets. The voltage regulation of
this generator must be provided by external electronics such as the
cycloconverter.
Table 21.2 Electrical generator overload capabilities
Overload Short-time Short-circuit

Generator type for 5 min overload for 5 s current for 5 sa
Synchronous 150% 200% 300% minimum

PM rotor
Synchronous 115% 150% 300%
Rice or Lundell
Synchronous wound 115% 150% 200%
rotor brushless
Induction 150% 200% 300%
homopolar
a
Internal losses are different during short circuit, since the voltage
output is zero.
21.4.4.2 Synchronous generator with solid rotor (Rice or Lundell)

There are two field coils, one at each end of the alternator located adjacent
to the stator frame. Each coil has two sections, a series section and a control
section. The series coils provide excitation from current transformers in the
leads to the load. Therefore, this excitation is proportional to the load
current. The control coils provide excitation from the voltage regulator
(feedback control system) as necessary to maintain constant voltage
regardless of the load, speed, or temperature variation in the system.
21.4.4.3 Synchronous generator with wound rotor and brushless exciter

The voltage regulation of the main generator is controlled by varying the
current in the exciter field. This system has three rotors (main generator,
wound exciter, and PM exciter) and rectifiers on a common shaft.
21.4.4.4 Induction homopolar generator

Voltage regulation is controlled by using a voltage regulator exciter system
with magnetic amplifier feedback control. Since this type of machine has the
highest internal reactance of the four machines considered, the field current
from no load to full load would have the greatest variations.
21.4.5 Speed and Frequency Limitations

The output frequency (in hertz) of a generator is given by the following
equation:
rpm number of poles

frequency ¼ ð21:2Þ
120
Thus, doubling the speed of a generator doubles the frequency and the
electrical power output capability without requiring a change in the
magnetics. However, the mechanical stress increases with increasing
diameter and the speed squared, requiring increased structural material
for high-speed machines. The combined effect of the diameter and the
speed result in the stress proportional to the linear tip speed of the
outermost fiber of the rotor. Therefore, it is advantageous to define the
maximum speed as a function of peripheral tip speed in meter per second
rather than in rpm, since the diameter of the rotor depends on the machine
rating. The maximum rotor tip speed in meters per second for the four
generator types is shown in the first row of Table 21.1, which sets the
maximum frequency these machines can be designed for.
The maximum rpm is also a function of the diameter of the rotor and the
kilovolt-ampere rating of the alternator. For a lower frequency application
(such as 400 Hz) based on the near optimal turbine speed of the solar
dynamic engine (such as 45,000 rpm), the minimum frequency of a two-pole
generator would be 750 Hz. Therefore, for 400 Hz standard frequency, an
effective frequency conversion using a cycloconverter or d.c. link inverter is
needed. Rectifiers are needed for d.c. distribution. All of the generator types
considered are capable of producing 3-phase power up to 2400 Hz. On the
high frequency side, none are capable of producing 20 kHz, although it may
be possible to design a 20 kHz, 1-phase, homopolar induction alternator.
References
1. Mason, L.S., A solar dynamic power option for space solar power,
2. Cockfield, R., Radioisotope Stirling generator concepts for planetary
mission, Proceedings of the 35th Intersociety Energy Conversion Engineering
Conference, AIAA, 2000, Paper No. 2843.
3. Thieme, L.G., Qiu, S., and White, M.A., Technology development for a
Stirling radioisotope power system for deep space missions, Proceedings of
the 34th Intersociety Energy Conversion Engineering Conference, SAE, 1999,
Paper No. 2454.
4. Sargent, N.B., The electromagnetic compatibility design challenge for
scientific spacecraft powered by a Stirling power converter, Proceedings of
Paper No. CT-37, pp. 447-451.
Chapter 22
High-Power High-Voltage Systems
22.1 Introduction
The electrical power requirement in spacecraft remained under several
hundred watts for decades. It was traditionally met with the sun-regulated
28-V d.c. power system. However, the power levels for commercial, science,
and military spacecraft have been rising steadily (Figure 22.1).1 It has
doubled approximately every 7 years. Because high-voltage systems are
lighter and more efficient, the rising power levels have been accompanied
by rising voltage levels. For example, the mass saving in just the
distribution cables at various power and voltage levels are shown in
Table 22.1. Today’s GEO communications satellites use power in the 7 to
15 kW range at 70 to 100 V. The ISS uses 105 kW power at 120 V. Some
strategic Defense Initiative weapons platforms may require steady power in
several megawatts and burst power in hundreds of megawatts at voltages
up to 100 kV or more. This chapter first presents high-power, high-voltage
power system architectures and then discusses concerns and issues in
designing such systems.
FIGURE 22.1 Spacecraft electrical power requirement growth trend.
518
High-Power High-Voltage Systems 519
Table 22.1 Power distribution cable mass with copper

conductors versus power and voltage levels
20 kW over 100 kW over

Voltage level 10 m distance 30 m distance Present use
28 V 155 kg 7000 kg Heritage voltage level

70 V 20 kg 900 kg Modern GEO satellites
120 V 7 kg 290 kg ISS and EOS
200 V 3 kg 140 kg SP-100
Power systems up to 100 kW have been designed at voltages up to 200 V.

Distribution systems above 200 V are considered high-voltage (HV) in space
due to various environmental conditions the spacecraft must undergo
during the launch and operation. The space plasma, the wide variations in
operating pressure and temperature, out-gassing of the materials, corona,
and atomic oxygen are the key design issues.
22.2 High-Voltage PV Array

For satellites with power requirements over 25 kW, high-voltage solar
arrays are needed for the system mass optimization. In LEO, the system
voltage is limited to less than 200 V due to the leakage current and arcing
concerns in high-density space plasma. In GEO, having low plasma density,
higher operating voltages are possible after reaching the orbit. An array
voltage of 1000 V has been considered for high-power solar arrays and long
cables. If the system were to operate at a nominal voltage of 1000 V, a
100 kW array would require current of 100 A, thus limiting I2R loss in the
conductor. In addition, the resulting low magnetic field interacting with the
Earth’s would keep the drag and the re-boost cost under control. However,
micrometeoroid impacts damaging the array leading to arcing and voltage
breakdown between high-voltage components is a concern. Recent tests2 on
a GaAs solar array with 6-mil coverglass indicated no arcing in designs
with sufficient coverglass overhang to avoid voltage creep and the wiring
contacts insulated by encapsulation. The test were conducted for 7 years
worth of micrometeoroids simulated by 100 mm soda lime glass spheres at
velocities up to 11.6 km/s in a plasma environment typical of the GEO orbit.
No arcing was detected under a bias voltage up to 438 V even where the
coverglass was penetrated and cell contacts were damaged. These tests
suggest that designing a solar array for voltages around 400 V may be
possible in GEO orbits.
A large solar array designed for generating high power beyond a few tens
of kilowatts becomes prohibitive due to the following reasons:
FIGURE 22.2 Solar dynamic power system schematic.
The solar array becomes difficult to fabricate

The large moment of inertia poses difficulty in maneuvering the
spacecraft
The spacecraft becomes highly vulnerable to offensive enemy attack, as
the large exposed array would be an easy target even from a far
distance
The solar dynamic power system covered in Chapter 21 is an alternative

candidate for high power. Its key components are shown in Figure 22.2.3 In
such a system, the energy can be stored as thermal energy in the form of
latent heat of a phase change material, thus eliminating the battery, which is
generally heavier than the solar array in LEO satellites.
22.3 High-Power Nuclear TEC

The nuclear isotope with thermoelectric converter (TEC) covered in Chapter
20 is suitable for generating a few hundred watts of power. For large power
in a few hundred kilowatts, the U.S. Air Force and DoE have jointly funded
significant research and development since the 1980s to design, build, and
qualify a nuclear reactor–TEC power system under the SP-100 program.
The program was dormant in the 1990s and has been revived recently
under NASA’s Project Prometheus for further development and deploy-
ment for high-power interplanetary propulsion (Figure 22.3) and space
defense missions with nuclear threats or with irregular orbits. The first
nuclear powered space probe is planned on the Jupiter Icy Moon Orbiter
(JIMO) for a possible launch in 2011. The static TEC offers a better
alternative to the rotating turbo-generator. The effect of the high gyroscopic
effect of large rotating machines in space is a major technological risk. No
rotating machines of such a size have been used in space before.
FIGURE 22.3 SP-100 power system for electrical propulsion to Mars.

(From Bell and Esch, IEEE Spectrum, p. 48, August 1991. With permission.)
In the SP-100 power system, the TEC modules are made of multicouple
SiGe/GaP thermoelectric (TE) cells having a TE conversion efficiency of
about 7%. However, the working heat losses reduce the overall system
efficiency to 5%. A 100-kWe electric power system, therefore, requires the
reactor thermal power of 100 kWe/0.05 ¼ 2000 kWt. The remaining 95% or
1900 kWt of the reactor heat must be radiated back to space in order to
control the spacecraft temperature within the material limits. The radiator
design, therefore, is a principal design driver. Figure 22.4 shows the
FIGURE 22.4 SP-100 power system configuration with nuclear reactor and TEC.
(Source: U.S. Department of Energy.)
radiator with panels fanning out in a conical shape for heat rejection to
space. The reactor core and the radiation shield are at the tip of the cone.
The user payload and sensitive equipment are located away from the heat
by placing them at the other end of a long 50-m boom.
The SP-100 uses a fast spectrum liquid lithium cooled reactor and SiGe/
GaP TE with Z ¼ 0.0007 for the state-of-the-art design and 0.0014 for the
future advanced TEC design. Each TE cell has a Voc of 0.50 V, and Isc of
12.8 A. Thus, the maximum power from each TE cell is 0.50 12.8/4 ¼
1.6 W. The specific power of the TE cell is 33 W/kg.
The nuclear reactor research and developments for space defense have
been funded in the past for electrical power in the 30 kW to 300 MW range.
The reactor is similar to a ground-based nuclear power plant, only much
smaller in scale.
In the baseline concept, the nuclear reactor boils the liquid metal. The hot
liquid is circulated on the platform by a liquid lithium pump loop,
transferred through an intermediate heat exchanger and delivered to an
array of heat pipes mounted to the 12-sided conical main body as shown in
Figure 22.4. The TE panels receive heat radiantly from the heat pipe array.
Thus, the heat is transferred from the reactor core to the TE modules, each
of which generates 4.6 kWe at 100 V. The heat not converted into electricity
is wasted. It is radiated into space from the outer surface of the TE panels.
The panels are deployed for effective radiation to the outer space. The
lithium coolant is pumped by an electromagnetic pump driven by a
dedicated TE module. The TE for the pump is powered by the temperature
drop between the working fluid and the pump radiators to assure pumping
as long as the reactor is at a temperature.
Twelve 4.6-kWe TE modules are connected in parallel to produce 50 kWe
at 100 V. Two such groups are connected in series with the mid-point
electrically grounded to the spacecraft structure. The voltage between two
outer conductors is therefore 200 V (Figure 22.5). The usable voltage to
avoid electrical breakdown in the SP-100 mission orbits is in the 150 to 200 V
range. However, the optimum voltage from the electrical efficiency point of
view was determined as exceeding 200 V. The finally selected bus voltage of
200 V is derived by using a center ground conductor between 100 V
conductors. This, in effect, gives only 100 V operating voltage from the
structure ground, while giving the load voltage of 200 V, keeping the
current level and ohmic loss low.
The SP-100 power system uses a regulated voltage bus with a shunt
dissipater. The TE modules continuously generate 100 kWe whether the
payload needs that much power or not. When payload power consumption
is less than 100 kWe, the excess power is absorbed by the shunt dissipaters
and dissipated into space. The shunts are made of traditional high-
temperature tungsten filament resistance wires. The shunts are turned on
automatically in response to the bus voltage error signal. The power output
of the TE panels are diode-isolated and are combined in a single 200 V d.c.
FIGURE 22.5 SP-100 thermoelectric converter assembly for 200-V system between 100-V
lines.
bus at the power module, which is located on the opposite end of the
extension boom. Such physical isolation helps reduce the amount of
shielding required.
A 300-W, 28-V auxiliary power supply from a separate RTG is provided
for startup, maintenance, and the shunt-down period. Under certain
mission constraints, solar or battery-only options may also be considered
for this purpose.
The following were the SP-100 system requirements:
7-year mission life

Survival to a classified nuclear threat
100 kWe EOL power, scalable to 30 kWe to 1000 kWe power levels
200-V fully regulated bus
Reliability goal of 0.990
Heat source upper temperature 1310 K
Although significant development work has been done in the USA, no SP-
100 unit has flown. The Russians, on the other hand, have flown dozens of
such units. The U.S. research funding was reduced in the early 1990s after
the U.S. procurement of similar technology from Russia following the
collapse of the former Soviet Union. However, from 2003 to 2008, the SP-100
budget is expected to be over one billion dollars, primarily targeted at space
defense and electric propulsion. A key thrust of this research is the

development of high-temperature materials. High power requires high
temperature to effectively reject waste heat with small compact radiators.
Multi-megawatt burst power systems for strategic defense may consider
nuclear reactors with rotating machines. For minimizing the gyroscopic
disturbance, such systems would necessarily use counter-rotating machines
in pair(s), as discussed in Section 22.11.
22.4 High-Voltage Design Issues

The design issues that are common to all power systems using high voltage
are discussed in this section.
22.4.1 Paschen Breakdown Voltage

The breakdown voltage between two bare conductors in a given gas
depends on the p–d product, where p is the gas pressure and d is the
separation distance. It is shown in Figure 22.6, known as the Paschen curve.
The spacecraft passes through a wide pressure variation, ranging from
atmospheric pressure on Earth before launch to 1 1010 torr in high orbits.
During this transition, it always passes through the minimum breakdown
voltage point on the Paschen curve. The power distribution system
normally has numerous bare spots of conductors and terminals. They
would spark during the spacecraft’s journey from the Earth to a high orbit
regardless of the separating distance between them. Out-gassing can make
FIGURE 22.6 Pressure-spacing product versus breakdown voltage in various gases (Paschen
curves).
the pressure prediction difficult. For this reason, a high-voltage system can
be turned on only after reaching the orbit (0.1 Pa vacuum). Alternatively,
the equipment must be filled with insulating gas, such as nitrogen, or
completely encapsulated with a solid insulation.
Preliminary data have shown that material out-gassing causes the local
pressure around the high-voltage component to increase to the point that it
is no longer a good vacuum. This, in turn, results in lower corona inception
voltage. The approach to increasing the corona voltage, therefore, is to select
the material with a low out-gassing rate in the first place. Figure 22.74
depicts out-gassing test results on a few materials. It indicates that Teflon
has a low out-gassing rate among electrical insulation materials, making it a
good insulation candidate for high-voltage space applications.
22.4.2 Dielectric Stress Concentration

If a HV rectangular conductor is placed in an enclosure filled with gas, the
dielectric stress at the ends is not uniform, but is concentrated near the
conductor edges. If the conductor radius is r and end distance e (Figure
22.8), the maximum dielectric stress is almost5
V e
Emax ¼ 1 þ 0:25 ð22:1Þ
e r
For Emax to be equal to V/d, i.e., the same as the uniform stress in the flat
part with ends fully radiused, i.e., r ¼ a/2, the end distance is derived by
equating

V 2e V
Emax ¼ 1 þ 0:25 ¼ ;
e a d
which gives
ad
e¼ ð22:2Þ
a 0:5 d
FIGURE 22.7 Breakdown voltage versus out-gassing time.

FIGURE 22.8 Voltage stress concentration at flat conductor edge with radius.
For example, with a ¼ 3 mm and d ¼ 1 mm, e must be 1.2 mm, i.e., 20% more
than d, the insulation thickness. For a smaller radius, say r ¼ a/4, e ¼ ad/
(a d) ¼ 1.5, i.e., 50% more insulation is required at the ends.
The stress concentration at sharp corners and their effects on corona
voltage and dielectric breakdown in atmospheric conditions on the ground
are well studied and reported in the literature. Such is not the case in the
space environment. A thorough systematic study on this aspect of the
insulation design is required. Until then, the available data must be used
with interpolations and extrapolations using wide margins to minimize the
associated design risks.
22.4.3 Corona Degradation

The insulation design should be such that partial discharges (corona)
should not start. If started at some higher transient voltage, it must
extinguish during the down-slope of the transient at some voltage higher
than the steady state operating voltage. Corona has two degrading effects
on the insulation performance. It chemically erodes the material, thus
shortens the life; and it produces radio frequency EMI that may interfere
with the surrounding equipment. Both of these effects are undesirable, and
should be avoided.
The corona voltage is a function of the type of material and its thickness
used in the design. For a 1000-V electrical component, corona tests on Earth
on polypropylene insulation gave the test results reported in Table 22.2.
These results must be modified to account for the Paschen minimum and
plasma effects depending on the orbit in which the spacecraft is designed to
operate or pass through.
A critical quantity needed for evaluating solid insulation subject to
corona is Ei, which is the partial discharge inception voltage stress. This
can be obtained directly by measurements, from extrapolation of the
Table 22.2 Corona test results on polypropylene

insulation
Thickness (mils) Inception(Vrms) Extinction (Vrms)
5 880 835
10 1570 1420
(Data source: M.R. Patel, NASA Report No. CR-175071,

1986.)
breakdown voltage–time curve, and from careful analysis of the voltage

stress at the conductor edge or in a void if necessary geometric parameters
are known. Large voids must be avoided in solid insulations. The dielectric
stress is "E in a laminar void and 3"E/(l þ 2") in a spherical void much
smaller than the insulation thickness. Here, " is the relative permittivity and
E is the uniform stress in the solid insulation. Partial discharge occurs in a
void when the product of the pressure in the gas and the void thickness
(electrode separation) exceeds the breakdown voltage according to the
characteristic Paschen curve. It can be aggravated by dielectric stress
concentration around sharp corners. Good HV conductor design avoids
sharp corners to keep the dielectric stress concentration low and corona
inception voltage high.
The following analysis establishes the largest permissible void size in
space-worthy high-voltage insulation. The laminar void is assumed, since it
produces more stress concentration than the spherical void. If a laminar
void of size do exists in a solid insulation between two conductors due to
any reason, the breakdown voltage across such a void will depend on the
gas pressure in the void. Since this pressure may vary over a wide range in
and around the spacecraft at different times of the mission, the minimum
Paschen breakdown voltage must be used in the calculations, which is
about 297 Vpeak or 210 Vrms between copper electrodes. There are now two
dielectrics in series, a gap of laminar thickness do and the solid insulation of
thickness d. If the dielectric stresses are Eo and E in the gap and the solid,
respectively, the voltage across the conductors would be V ¼ Eodo þ Ed.
Polypropylene insulation with a relative dielectric constant of 2.25 would
take 1/2.25 times the gap stress, i.e., E ¼ Eo/2.25. With d ¼ 20 mils as an
example, the above voltage then becomes

Eo 8:89
V ¼ Eo d o þ 20 ¼ Eo do þ 8:89 Eo ¼ Eo do 1 þ ð22:3Þ
2:25 do
where do is in mils. The product Eodo is the voltage across the gap, which has
the minimum Paschen value of 210 Vrms. The corona would start at this
voltage across the gap. The corona inception voltage of a cable with
polypropylene insulation is then

8:89
Vi ¼ 210 1 þ ð22:4Þ
do
Equation 22.4 is plotted in Figure 22.9, which shows that it takes about a 2-
mil delamination gap for corona to start at 1200 Vrms. This condition would
not normally lead to an immediate failure, but eventually would do so
under gradual corona degradation. Should the gap completely fail, the
dielectric stress in the solid insulation would be 1200/20 ¼ 60 V/mil, which
is two orders of magnitude lower than the solid insulation’s short time
breakdown strength.
As seen in Figure 22.9, the corona inception voltage decreases with
increasing gap size and asymptotically approaches the minimum Paschen
value of 210 Vrms. This is because the Paschen minimum is a constant value,
although occurring at different pressures for different gap sizes. As the gap
increases, the corona inception voltage across the gap remains constant at
210 V, while the voltage contribution of the decreasing dielectric stress in
the solid insulation decreases. This voltage is based on the worst possible
combination of the gap pressure and the gap size (i.e., the minimum
Paschen value). It is likely that there is always some pressure in the void
due to continuous out-gassing of the insulation, giving a higher corona
inception voltage. The location of the void along the cable length, therefore,
matters. If the void is at mid-length of the cable, out-gassing may take a
long time due to the high resistance offered to the gas flow in a long narrow
duct. For increased reliability, it may be desirable to bleed some gas (such as
nitrogen) along the length of high-voltage cable insulation. This would
assure the desired gas pressure to maintain the required dielectric strength
of the cable.
There is an increasing push for mass reduction to reduce launch cost. For
this reason, many HV components do operate with some corona, which
FIGURE 22.9 Corona inception voltage versus laminar void size in a flat conductor cable.
must be controlled in manned spacecraft to protect the astronauts from

corona charging currents. The high value of mass minimization justifies a
sacrifice in the equipment life, provided reliability and safety are not
sacrificed. A life of 5 to 15 years has been typical for space components, but
reliability must be high since these components must operate without
servicing.
22.4.4 Atomic Oxygen

The interaction of atomic oxygen with the insulation surface is of particular
concern in low Earth orbit. This interaction is a reaction and/or removal of
mass from the surface during the mission life. The amount of interaction
depends on the total integrated dose of atomic oxygen impingement, which
is a strong function of the orbit altitude. Figure 22.10 shows such an
interaction in a typical polymer. The insulation design should allow such a
loss of material over the mission life in space.
Table 22.3 shows how various insulating materials degrade under
radiation and atomic oxygen. It shows that Teflon1 is bad under the effect
of either one. An estimated radiation dose on the ISS over 5 years is about
100 krads, so silicon rubber may be used in the design, provided that the
degradation under atomic oxygen is acceptable. The atomic oxygen fluence
rate on polar orbits at 400-km altitude is about 3 1021 atoms/cm2year,
which is less than that on the ISS.
22.4.5 Plasma and Charged Particles

The plasma in the mission orbit exerts the greatest influence on the voltage
that two bare conductors can withstand in space. The plasma density in
electrons/cm3 versus orbit altitude is depicted in Figure 22.11. It peaks
FIGURE 22.10 Mass loss due to atomic oxygen versus orbit altitude.
Table 22.3 Depletion rate of electrical insulation in space

environment
Radiation dose Reactivity, Depletion rate

before adverse 1 1024 in 3 1021
effect is noticed cm3/atomic atoms/cm2year
Insulation (Mrads) oxygen atom orbit, mil/year
Teflon 0.1 0.03 0.036

Silicon rubber 1 0.05-0.2 0.06-0.24
Kapton 100 3 3.6
Epoxies 1000 2.5 3.0
inLEO, and is negligible in GEO and beyond. The plasma acts as a shunt
resistance across the power components, such as the solar array, causing
power loss due to the resulting leakage current between bare conductors.
This loss becomes noticeable above 160 V. A voltage much higher than
160 V leads to voltage breakdown. The breakdown voltage versus plasma
density is shown in Figure 22.12. Combining Figure 22.11 and Figure 22.12
leads to the arcing voltage between bare conductors versus orbit altitude, as
plotted in Figure 22.13. The arcing voltage determines the maximum solar
array output voltage and also the minimum distance between printed
circuit board traces to withstand the voltage over the design life of the
mission. For this reason, NASA limited the ISS solar array output voltage to
160 V, with a 120-V distribution voltage and step-down converters where
required for existing 28-V hardware.
In plasma-free orbits such as GEO, the conductor in a near perfect
vacuum can withstand kilovolts or even higher voltages. However, the
FIGURE 22.11 Plasma density versus altitude in equatorial orbit.

FIGURE 22.12 Breakdown voltage threshold versus plasma density between bare conduc-
tors.
Paschen minimum breakdown voltage limits its operation only in GEO and
not during launch or transfer orbit. The out-gassing of the material inside
the spacecraft body and the flue plumes outside create an uncontrolled gas
pressure around the conductors. Moreover, the spacecraft interior during
launch undergoes a decreasing air pressrun as it attains the orbit, and may
take a long time after the launch to achieve a good vacuum. It is probable
that at some instant the p–d product may achieve a value corresponding to
the minimum breakdown voltage. Under such a condition, the bare
conductor would cause arcing and short circuit. For this reason, the system
designed to operate at higher than 160 V in GEO can be turned on only after
achieving the GEO orbit.
FIGURE 22.13 Arcing threshold voltage between bare conductors versus altitude.
As seen in Figure 22.13, the 300-km high orbit is the worst with the arcing
on-set voltage of only 160 V. If the spacecraft must go through this orbit, the
power distribution voltage must be below 160 V. If a solar array is
assembled by standard construction techniques — with coverglass not
completely shielding the metallic interconnects — is operated at voltage
higher than 160 V in LEO, the bare interconnects can act as plasma probes
and attract or repel charged particles. The generated voltages will be equal
to the space plasma potential at some location on the array. Since electrons
are more mobile than ions, the array will float at voltages that are mostly
negative with respect to space plasma potential. Cell interconnects at
voltages above this space plasma potential will collect electrons; those at
voltages below this will collect ions. The voltage distribution in inter-
connects relative to the space plasma potential must be such that these
electron and ion currents are equal (i.e., the net current collected is zero).
This flow of particles can be considered to be a current loop through space
that is in parallel with the operational system and hence is a power loss. In
addition, the coverglass used on the solar cells must also have a zero net
current collection. This interaction with space plasma forces the coverglass
to a small negative potential and can produce large voltage gradients in the
gap region between solar cells. This can subsequently give rise to arcing
conditions or transient breakdowns to space.
The severity of such plasma interactions depends on the array operating
voltage and the charged-particle environment. The operating voltage is
determined from power system trades, but will probably be less than
1000 V. At the projected operating voltage only the low-energy or thermal
plasma environment should be of concern since the array voltage is too low
to influence the higher energy environmental particles. This plasma
environment has particles with temperatures of about 1 eV and densities
that vary from a maximum of about 3 106 cm3 at 300 km to between 1
and 10 cm3 at geosynchronous altitudes. Hence, plasma interactions
should be more severe at the lower altitudes than at synchronous altitude.
A 100-kW space-worthy system operating at 500 V d.c. was designed by
NASA. It was made of ten arrays in parallel each with 10-kW capacity. It
was found that, under normal quiescent conditions, the power drain due to
the electron coupling current would be negligible. However, the arcing in
the negative voltage regions could seriously disrupt system operations
either by introducing a ripple on the output or by terminating operations
depending on the severity and location of the breakdowns. The orbital
oscillations range through ram, wake, and eclipse conditions. The shuttle
experiments that indicated that large space structures could create their
own plasma environment tend to make plasma interactions even more
critical.
Over the past decade, the advantages and disadvantages of large space
power system operations at high voltages have been argued and discussed.
There are obvious advantages to using high voltages in space. Possible
hazards to such operations with standard array technology have been

addressed. These interactions are not insurmountable but can be overcome
given adequate understanding of the phenomena. What is needed is a
systematic investigation to determine why discharges occur and how to
prevent them. This would require test programs involving large arrays with
self-generated voltages and flight experiments to prove that all of the
interactions can be minimized.
Tests in ground simulation facilities in which small solar array segments
were biased to positive and negative voltages in a plasma environment
have shown that interactions can be detrimental. When positive voltages
are applied, electron currents can be collected that become proportional to
the panel area at voltages greater than 100 V. Arcing or breakdown can
occur under negative bias voltages. This arcing threshold depends on the
plasma density and can be as low as 300 V in simulated 300-km orbit
plasma environment as shown in Figure 22.14(a) and (b).
FIGURE 22.14 Plasma coupling current versus applied voltage.

Relatively few tests have been conducted in which an array capable

of generating high voltages under solar simulation conditions was
operated in a plasma environment. PIX-I and PIX-II space experiments
conducted by NASA have shown that a 9-block array would float
electrically such that one block would have an average positive voltage
that would be 10% of the overall voltage, with the other eight blocks
progressively more negative. This test indicated that array behavior
could be approximated by considering the interaction with separate
blocks at an average voltage.
As for the effects of radiation, Bouquet and Winslow6 present their
study results on the lowest doses for which radiation effects on material
properties are noticeable. The report compiles data on the effects of
radiation on the physical properties of polymers and synthetic organic
materials. Emphasis is on materials of interest to nuclear equipment and
nuclear reactor designers. The data cover five categories of polymeric
materials: insulators, elastomeric seals and gaskets, lubricants, adhesives,
and coatings. More than 250 materials are represented. The document
reflects a continuation of the work in progress. It gives available data
on physical, electrical, and optical properties for each material in terms
of the lowest reported threshold dose. It indicates the radiation level at
which radiation effects become apparent and the dose for a 25% change
in the specified property. The literature source for the data is also
indicated. Both the threshold and the 25% change dose vary widely
within each of the five categories of polymer. Most of the polymeric
materials have thresholds above 105 rads. Some materials — fluorocar-
bons, nylons, epoxy resins, polyethylene oxides, and polyalphamethyl
chloroacrylates — have thresholds below 105 rads. The threshold data
are useful for setting a lower limit at which changes can be expected.
They may also be used as a guide for deciding when the expected
radiation is low enough to be disregarded for a given polymer. The
25%-change dose is a valuable radiation statistic because changes are
real at this dose level and are neither artifacts of the experiments nor
results of sample variations.
22.4.6 Temperature Extremes

The spacecraft component temperature can swing over a wide range
during sunlight and eclipse durations. Such temperature extremes must
be considered in the insulation selection. The components exposed to
space temperature (200 C) without survival heat may experience
damage. For example, most polymer insulation may shatter (crack) at
temperatures below 75 C. After such an exposure, corona may start in
the cracked insulation, thus shortening the component life in supporting
high voltage.
22.4.7 Design Guidelines

The corona, out-gassing, plasma, attitude control exhaust, permeation from
pressurized chambers, bubbles in liquid dielectrics in zero gravity, etc.
make high-voltage space insulation a different science than on Earth. More
experiments are needed to understand these factors.
Two types of insulation breakdown dominate in vacuum or low-pressure
gaps between bare conductors. Breakdown in small gaps is dominated by
the electrode surface, whereas in large gaps it is dominated by particle
contamination in the gap volume. The surface breakdown generally occurs
at lower stress levels than the volume breakdown. The pressure around out-
gassing surfaces can be orders of magnitude higher than the ambient
pressure, and can be a function of external stimuli, such as temperature and
incident charged particles.
If the spacecraft must operate or pass through the worst orbit in an
electrically ON state, the spacecraft power distribution voltage must be
lower than the breakdown voltage in plasma and the Paschen minimum
voltage. For voltages higher than that, several possible design approaches
can be considered:
Avoid the effect of plasma by placing the high-voltage component in a

metal enclosure. The enclosure drains the plasma to the spacecraft
ground, thus keeping it from reaching the electronic component (Figure
22.15).
Avoid the worst pressure for the Paschen minimum by venting a gas
(nitrogen) around the bare electrodes until the spacecraft reaches a safe
orbit.
Encapsulate the high-voltage component in solid insulation. The
material selected for encapsulation must have low out-gassing proper-
ties and the manufacturing process must be void-free to minimize
corona.
Place the high-voltage component in liquid or gaseous insulation. The
high fluid leakage rate in a hard vacuum must be accounted for in space
designs.
FIGURE 22.15 Space plasma shield configuration.

Turn ON the electrical system only after reaching the altitude where the
arcing onset voltage is higher than the distribution voltage. Until then,
the electrical power requirement (for survival heat, navigation or
controls) must be met by batteries operating at low voltage.
22.5 High-Voltage Direct Current

The automobile industry used 6-V power in the first 50 years. It has been
using 12-V d.c. since the early 1950s, but has now started moving towards
42-V d.c. to meet the growing power needs in modern automobiles. Military
tanks, electric guns, and turret drives in the 45 to 60-kW range use 270-V
d.c. power distribution. The aircraft industry has been using 400-Hz power
systems, but new standards have been developed for the industry to adopt
270 V d.c. The U.S. Navy has been using 155-V d.c. that meets MIL-STD-
1399.8
Major factors that influence the high-voltage design in space are:
Availability of space-qualified parts, such as tantalum capacitors,

semiconductors and power devices rated for high voltage may be
limited.
High-voltage d.c. contactors present a significant challenge in designing
switching relays.
Electrical discharge may occur in the 300 V range even under space
vacuum conditions.
Plasma discharge power loss in the array, which is 1% at 400 V d.c.
The interaction of plasma environment with a high-voltage d.c. system is

minimized by connecting the solar array wings at voltage with center
ground (Figure 22.16). This results in a power system operating at twice
the plasma interaction voltage. The EMI considerations in high voltage
d.c. systems are similar to those in the low voltage d.c. systems. In par-
ticular, MIL-STD-461 on the conducted ripple current should be achievable.
FIGURE 22.16 Plasma interaction on high-voltage solar array wings.

Direct-current systems are also expected to have low levels of radiated B-

and E-fields. Power conditioners operating at high switching frequency
would result in minimal EMI filters. Shielding techniques are similar to
those used in low voltage systems. Solid-state switches have reduced the
rate of rise and fall times in current and thereby reduced system transients.
The principal safety hazards in an electrical system are fire, shock, and
audible noise. Direct-current systems are typically immune from audible
noise. Fire hazards are common to all distribution systems regardless of the
voltage level. Ground faults in HV d.c. are cleared by RPC and RBI to
protect personnel and reduce fire hazards. No unusual safety problems
with HV d.c. systems have been reported in the literature
HV d.c. components developed for aircraft applications are suitable for
further development for space applications. The extensive space experience
in 28 to 160 V d.c. heritage allows development of similar designs for HV
d.c. systems. A number of HV d.c. components have been developed for use
with HV solar array and battery assemblies. A large number of vendors
have the technology and background to produce space-worthy components
for HV d.c. systems competitively. Compared to HV a.c., spacecraft design
engineers are much more familiar with design techniques and applications
of d.c. power equipments such as d.c.–d.c. converters, d.c. power
controllers, etc.
22.6 Alternating Current Versus Direct Current

The d.c. system is generally economical if power can be generated,
distributed and utilized at the same voltage, or the d.c.–d.c. power
conversion efficiency can match with that of the a.c. power transformer.
Moreover, d.c. generates minimum EMI, but is difficult to switch off. The
a.c. system, on the other hand, is generally economical where the power
generation, distribution, and utilization are done at different voltages, and
there are numerous users utilizing power at many voltage levels as in a
ground-based utility. The high efficiency of an a.c. power transformer in
converting one voltage into another is a dominant factor in favor of a.c.
Moreover, switching a large a.c. current OFF when it is passing through a
natural zero is easy, as is often done.
Since the d.c.–d.c. converter efficiency is rising with low-loss semicon-
ducting devices and efficient switch mode power converters, d.c. is
becoming attractive where only a.c. was considered practical a few years
ago. The history of a.c. versus d.c. power systems in the aerospace industry
can be examined in terms of the evolution of present day aircraft power
systems, which has progressed as follows:
1920s: 14 V d.c., 500 W for engine ignition

1930s: 28 V d.c., with power in kilowatts
1950s to 1980s: 208 V, 400 Hz, 3-phase Y, 4-wire grounded, variable

speed, constant frequency high power a.c. via mechanical gear
1990s: 208 V, 400 Hz, 3-phase Y, 4-wire grounded, with variable speed,
constant frequency a.c. via power electronics (no mechanical gear)
Future: 270 V d.c., or 400 Hz a.c. with variable frequency, constant
voltage d.c. (no gear or power electronic inverter).
The 28-V d.c. system used in early aircraft, both military and commercial,
contributed to the selection of 28 V d.c. for the first spacecraft. The 28 V d.c.
remained in use until the 1980s. Modern aircraft power systems, however,
have changed to a 400-Hz, 3-phase, 200 Y/115-V variable speed constant
frequency system shown in Figure 22.17. The space shuttle fleet generates
d.c. power using a fuel cell, but converts and distributes the power at 120 V,
400 Hz, 3-phase. An advanced system proposed for the aircraft industry
centers around 270 V d.c. as shown in Figure 22.18. The spacecraft
secondary power distribution can be 208 V, 3-phase Y, such that 120 V is
available for single-phase utilization. The primary voltage can be any, such
as 440 V 3-phase Y or i. The Japanese and European space industry
preferred 120 V d.c. for secondary distribution on the space station even
when NASA strongly preferred 3-phase, 208-V Y/120-V, 20-kHz system in
an early phase of the ISS design. Arguments made in favor of 120 V d.c. over
120 V 20 kHz a.c. were:
Decades-long experience
Extensive design data base
Low cost
Low EMI
High reliability
Synergistic development using heritage
The simplicity, low parts count, and previous space qualification history of
d.c. systems offer advantages over a.c. systems for high reliability in space
environments.
FIGURE 22.17 Variable speed constant frequency ac power system for aircraft.
FIGURE 22.18 High-voltage d.c. power system for aircraft.
22.7 High-Frequency Alternating Current

The power system in the hundreds of kilowatts to the megawatt rating is
likely to have numerous items of equipment working at different voltages
much lower than the generation and transmission voltages. The distribution
system for such high power platform has to be efficient, lightweight, and
flexible to accommodate a variety of electrical sources and loads. It would
require voltage converters for stepping down the generation voltage to suit
the equipment to be powered at various locations. A very large spacecraft
such as a 300- to 500-kW space station would therefore need a power
system that resembles the power utility on Earth, where each user draws
power at a voltage that is the optimum for a given application. The a.c.
voltage conversion from one level to another can be done most efficiently
using a transformer. A typical transformer efficiency is 97 to 99%, whereas
that of the d.c.–d.c. converter is 80 to 95%. Also, the a.c. equipment is
smaller in size and lighter in mass at high operating frequencies. For these
reasons, a very high-power system is likely to be designed to operate at
high-voltage and high-frequency a.c. The power system for the initial
version of the ISS in the 1980s was to generate 300 kW by a combination of
photovoltaic arrays generating d.c., and a solar dynamic system generating
high-frequency a.c. The power requirements for the crew living areas,
research projects, computers, and manufacturing facilities greatly varied, as
in the ground-based utility. Studies made by NASA indicated that high-
frequency a.c. distribution system would offer significant advantages in
such systems. Subsequently, NASA Glenn Research Center funded devel-
opment of a 440-V, 20-kHz a.c. power system that is flexible, versatile, and
transparent to user equipment, while maintaining high efficiency and low
mass. The output power could be either fixed or variable voltage and
frequency. Any electrical source, d.c. or a.c., was to be converted to
440-Vrms, 20-kHz, single-phase a.c. Power was to be distributed through
low-inductance cables.
At 20 kHz operating frequency, the energy transferred per cycle is
reduced by factor of 50 as compared to a 400-Hz power system. The number
of parts is reduced by a factor of about 5 relative to the conventional 3-phase
converters. The power loss is reduced by two thirds. These factors result
FIGURE 22.19 Load power converter mass versus input frequency of 3-phase a.c. power.
directly in mass and power savings. NASA trade studies compared the
mass per kilowatt processed by various power converters. They are
summarized in Figure 22.19, Figure 22.20, and Figure 22.21. Frequencies
higher than 60 not only reduce the mass of magnetic component, they also
offer a higher degree of safety to humans. Table 22.4 shows the relative
effect of various frequencies on human cells and tissues. Frequencies above
10 kHz are safer than d.c., while 60 and 400 Hz systems cause a higher
perception and paralysis effect than d.c.
As for the load power converter in high-frequency systems, the d.c.
power for loads can be converted from 20 kHz a.c. using a center tap
transformer with two rectifying diodes, an L-C filter and a free-wheeling
diode as was shown in Figure 10.21. The free-wheeling diode carries the
current when the voltage reverses, thus eliminating phase shift and a poor
power factor. This approach has one serious drawback in that the reflected
load current becomes a 20 kHz square wave loaded with high-frequency
harmonics. The power quality requirement for the proposed NASA system
required the reflected load currents to be high power factor sine waves.
Therefore, additional smoothing of the output and a post-regulation may be
required depending on the EMI sensitivity of the load.
FIGURE 22.20 Central power converter mass versus output frequency.

FIGURE 22.21 Power transformer mass versus operating frequency.
22.7.1 20-kHz Cable

NASA Glenn developed a flat virtually concentric (coaxial) low-inductance
cable for distributing 440 V/20 kHz a.c. power in a 300-kW space station
design in the late 1980s. The cable was designed and developed by
Induction General Inc., initially with flat Litz braids, which were later
replaced with flat solid copper conductors (Figure 22.22). The thickness of
the flat conductor is less than the skin depth given by
rffiffiffiffiffiffiffiffi

¼ ð22:5Þ
f
where ¼ resistivity of the conductor, ¼ permeability of medium around

the conductors, and f ¼ frequency. For a copper conductor, is 9 mm at 60
Hz, 3.5 mm at 400 Hz, and 0.5 mm (20 mils) at 20 kHz. With the thickness
limited by the skin depth, the ampacity of the conductor is based on the
width of the conductor. In a 440-V, 20-kHz, 75-kW cable design, the center
conductor is 0.5 mm thick 5 cm wide. The two outer conductors form the
return, and are 0.25 mm thick and slightly wider than the center conductor.
Each conductor is insulated and the inner and outer conductors are
separated by an insulating spacer to control the capacitance and inductance
of the cable. The gas permeable insulating spacers incorporated by W.L.
Gore & Associates allow out-gassing along the entire length of the cable,
which is required to avoid voltage breakdown above the Paschen
minimum.
Table 22.4 Relative effects of electric current on human cells and tissues
Frequency, Hz 0 (d.c.) 10 60 400 1k 10 k RF
Perception and paralyses, relative 0.2 0.9 1.0 0.7 0.6 0.2 0.01
FIGURE 22.22 Virtually concentric flat cable construction.
A 30-kW, 440-V, 20-kHz flat virtually concentric cable is estimated to be,

overall, 5 cm wide 1.25 cm thick and have the following electrical
parameters per meter length: R ¼ 2 m, L ¼ 15 nH, C ¼ 2 nF, mass ¼ 1 kg.
The 20 kHz cable connectors were also developed. One such design is
shown in Figure 22.23.
22.7.2 EMI in 20-kHz Alternating Current

The EMI comparison between d.c. and 20-kHz power systems includes a
study of the harness as well as the power converter noise. The d.c.
information is based on actual data for conducted and radiated emissions in
existing d.c. power systems. An EMI/EMC study for the ISS indicated that
the d.c. power distribution is superior to 20 kHz a.c.. The following
conclusions were reached against the use of a.c. on the ISS in the final
analysis:
An a.c. system is noisier than d.c., in the EMI sense.

EMI reduction in d.c. is lighter and easier to implement than those
required for a 20 kHz a.c. system.
FIGURE 22.23 Power connector for a 20-kHz flat cable.

(Source: NASA Glenn Research Center.)
Transients in a d.c. system are much less severe than those in an a.c.
system.
The grounding system in d.c. allows less structure ground current than
in a comparable a.c. system.
A d.c. harness is lighter than an a.c. harness, even without the extra
shielding that an a.c. system would require to meet the space station
radiated emission limits identified in JSC-30237.
The a.c. magnetic fields in the 10 kHz to 1 MHz frequency range, which
are of great concern to the plasma experiments, would be about 60 dB
higher from a 20 kHz power harness than from a comparable d.c.
harness.
NASA’s overall conclusions from the ISS trade studies on a.c. versus d.c.
in the late 1980s are summarized in Table 22.5. The NASA team debated
between 400 Hz and 20 kHz for a long time, and at one time finally
recommended 20 kHz. Significant developments for 20 kHz components
were under way before the program priorities changed to much lower
power levels for which 160 V/120 V d.c. was finally adopted for the ISS.
22.8 High-Power Components

NASA Glenn Research Center has been continuously developing high-
power components for space. These components include a family of bipolar
power switching transistors, fast switching power diodes, heat pipe cooled
high-frequency transformers and inductors, high-frequency conduction
cooled transformers, high-power high-frequency capacitors, remote power
controllers, and rotary power transfer devices. Many of these components
— the power switching transistors, power diodes, and the high-frequency
capacitor — are commercially available for space applications. Other
Table 22.5 Frequency trade of high-power space platforms
DC 400 Hz 20 kHz
Mature power converter and Mature technology used in New technology to be

switchgear in low voltage aircraft and submarines developed
Heavy power conversion Competitive with 20 kHz High efficiency, and lower
system mass per kW
Fewer series component, Transformer coupling to Loads transformer coupled
hence more efficient loads can improve efficiency to 20 kHz, simplifies LPC
Safer for crew Simpler protection Simpler protection
Low EMI Manageable noise Inaudible noise
No skin effect Skin effect loss Skin effect and dielectric
loss
components have been developed to the prototype level. Series resonant

d.c.–d.c. converters up to the 25 kW power level have been built and tested.
The power components that take advantage of both high voltage and
high frequency have also been developed. These components would enable
the building of large space power systems. Component technology is
continuing with goals such as high-power rotary power transfer, 10-kV
semiconductor switching device, high-temperature semiconductors, non-
impregnant capacitors, and high-voltage a.c.–d.c. remote power controllers.
The objective of this section is to summarize these new space power
component technology.
22.8.1 Rotary Power Transfer

Brush and slip rings assembly is needed at the joint of the satellite for
rotary power transfer. Their design generally has the following perfor-
mance limitations:
1 A/mm2 with solid carbon brushes used in conventional electrical

machines
10 A/mm2 with copper or silver coated graphite fibers in carbon matrix
30 A/mm2 with liquid metal current collectors (experimental for
ground-based systems)
The roll ring is a new rotary power transfer device developed for use in
the ISS. It transfers electrical power across the joint through rotating
flexures compressed between concentric conductors (Figure 22.24). Its
advantages are the elimination of sliding friction and the low torque needed
FIGURE 22.24 Roll rings for rotary power transfer in high-power d.c. or a.c. system.
to rotate the device. The roll ring can transfer both a.c. and d.c. power and is
lighter and more efficient (>99.6%). It consists of two concentric conductive
rings and many small rolling flexible conductive rings. The flexure is fitted
and captured in the annular space between the two rings. When the rings
are suitably attached to two structures on each side of the rotating axis, the
flexure provides a precise electrical coupling between the structures.
Performance and life advantages result from the optimum interfaces
where only micro-sliding occurs. NASA Glenn with Honeywell and
Sperry Flight Systems has developed a 400-kW roll ring assembly. Each
power circuit is capable of conducting 200 A at 500 V d.c., i.e., 100 kW. Five
power circuits may be used in parallel in a 500-kW assembly and 10 in a
1-MW assembly.
An alternative to the roll ring is the rotary power transformers at the
joint (Figure 22.25). The first design analysis performed was for 100 kW, 20
kHz unit consisting of four 25 kW modules. A 25-kW rotary transformer
with 300-V input and 1000-V output operating at 20 kHz was designed
based on the configuration shown in Figure 22.25. It has pancake geometry
and a radial axi-symmetric gap. The study found the rotary transformer
efficiency, mass and size favorable compared to those of the roll rings for
the same performance requirements and input characteristics. A 2-kW,
2-kHz square wave demonstration model was built at NASA Glenn and
performance characterized.
22.8.2 Switching and Protection

The ability to add and remove loads and to switch in spare units is
accomplished on the ISS with remote power controllers (RPCs). In the early
1970s, NASA Glenn started an RPC program with Westinghouse to develop
FIGURE 22.25 Rotary transformer for power transfer in high-power a.c. system.
RPCs that operate at 120 V d.c. instead of 28 V d.c. Prototypes were

developed and built in 5- and 30-A ratings. A 1000 V, 25 A d.c. prototype
RPC was completed with optical isolation, programmable transient
protection, and ability to withstand shorts. A transistorized version of
this RPC was also built. The d.c. RPC development drew from high voltage
d.c. contactors used extensively and successfully in motor speed controllers,
locomotion systems and submarines power switching. Large switchgear
type contactors are replaced by small contactors using rare-Earth magnets
to extinguish the arc generated during load interruption. This program
produced prototype units that are hermetically sealed, optically isolated,
computer controllable with over-current protection or I2t trip control that
are used in the 160/120 V d.c. system on the ISS.
The functions of RPCs developed for the ISS are:
Replace electromechanical relays

Connection or disconnection of loads and components
Interruption of fault and overload current
Interface with the PMAD system
Remote operation
Control the rate of rise and fall of current
The basic features of the ISS RPCs are:
Controlled circuit protection

Current limiting to prevent load transients
Wide temperature range
Electrical isolation of control and status signals from the power bus
High-speed trip-out response characteristics
Diagnostic capability
Load capacitor charge capability
Programmable trip characteristics
For much larger currents at the main bus level, remote bus isolator are
developed for ISS with the following basic features:
Short circuit protection

Current limit
Extended switching life
Built-in diagnostics
Status feedback
Bi-directional current limit

Bi-directional operation
dI/dt limit
For a higher voltage, such as 270 V d.c., contactors are available from
Hartman and Eaton Cutler for rated currents of a few hundred amperes
continuous with several hundred amperes interruption capability.
22.8.3 Semiconducting Devices

High-power spacecraft require high-voltage, high-current semiconducting
devices to control and condition the power. Unfortunately, the devices that
are presently available lack in-field service experience, and/or substantial
inventory of high-voltage d.c. components. In many cases, the available
devices are not capable of providing high voltage, high current, and fast
switching, while at the same time providing a low forward voltage drop.
The SCRs are capable of high voltages and currents, but are slow, have large
voltage drops, and require special circuitry to turn them on and off. Power
MOSFETs are capable of high voltages and fast switching, but the ON
resistance limits their current capability and forward voltage drop. The
bipolar transistors require special drive circuitry. In the early 1980s,
Glenn started developing high-voltage, high-power switching devices for
either a.c. or d.c. distribution system components. Many of them are
now commercially available.
For example, a 25 kW switching device program with Westinghouse
developed the D60T/D62T transistor capable of handling 50 A continuous
(200 A peak) collector currents with a gain of 10 at 400 to 500 V and a
switching frequency of 20 to 50 kHz. It found many terrestrial applications
as well as space applications. Subsequent development of the D7ST
doubled the power handling capability of the D60T from 25 kW to 50 kW,
with a 100 to 150 A current at a gain of 10, 400 to 500 V operating voltages
and 20 to 50 kHz switching frequency. Next came the high voltage D7ST
that raised the operating voltage to 1000–1200 V. The high-voltage D7ST is
capable of 25 to 50 A of continuous current (200 A peak) with a gain of 10.
This device has 30 kW power handling capability at switching frequencies
up to 50 kHz. The next device is the augmented power transistor which
raised the current capability to 70–115 A of continuous current (400 A peak)
with a gain of 10. This last device has the power handling capability of
75 kW at frequencies of 20 to 50 kHz. Most of these devices can be used in 25
to 50 kW high frequency inverters, high power d.c.–d.c. converters, electric
motor controllers, and power controllers.
NASA Glenn Research Center also funded, at Power Transistor
Company, a fast recovery, high-voltage power diode with a recovery time
compatible with the D60T and D7ST. The PTC 900 diode has a peak reverse
voltage of 900 to 1200 V with an average forward current of 50 A. The diode

can withstand peak surge currents of 1000 A and has a reverse recovery
time of 200 ns at 50 A. A follow-on program increased the current
capabilities of the diode to 150 A. These diodes are commercially available
and can be used in high-voltage, high-frequency inverters, high-power d.c.–
d.c. converters, high-power supplies, and motor controllers. Also, a gate-
assisted turnoff thyristor (GATT) was developed to handle 200 A at 1000 V
and frequencies up to 20 kHz.
For megawatts power systems, high-power semiconducting device such
as the MOS Controlled Thyristor (MCT) developed at GE research center
may be considered. The MCT is a power semiconductor device, which has
the voltage and current capabilities of the SCR and the fast switching and
low forward voltage drop of the bipolar transistor, while being controlled
like the power MOSFET. These attributes make the MCT ideally suited for
high-voltage, high-power applications in spacecraft power systems. The
MCT is available in ratings of a few thousand volts and a few hundred
amperes. If a 2500-V, 100-A MCT is qualified for space use, derated to 25%
in voltage and 50% in current, it can still operate at 625 V and 50 A, thus
delivering 31 kW per device.
Other components have been developed to the prototype stage that
demonstrated the technology readiness. These components would enable
the construction of large space power systems, high-voltage a.c. or d.c..
22.8.4 Capacitor
In the 1980s Maxwell Laboratories developed for NASA a 600-V space-
worthy capacitor with a maximum current capability of 125 A at 40 kHz.
The dielectric for the capacitor is polypropylene. This capacitor also has the
capability to operate with a 600 V d.c. bias. The calculated loss at full load
operation is 22 W, which may be compared to the 75-kVA rating of the
capacitor. This capacitor has a specific mass of 0.042 kg/kVAR, which
represents a decrease in size and mass by a factor of 7 and is commercially
available. It can be used in high-frequency d.c.–d.c. converters, d.c.–a.c.
inverters, tank circuits, and filtering and power factor correction.
22.8.5 Power Converters

The most commonly used buck converter topology is not optimal for power
converters rated in tens or hundreds of kilowatts power and operating at
high voltage (300 V d.c. or higher). Full utilization of the power capability of
switching semiconductors in these ratings favors a topology incorporating a
transformer with an appropriate turn ratio. Even where voltage differences
do not require a transformer, the input-to-output isolation may still be
required for system compatibility. Another potential system requirement in
high power systems is to minimize EMI resulting from switching high
currents. These requirements can be met by a transformer-coupled

converter using Cuk integrated magnetics. The power stage topology that
was shown in Figure 10.22 uses three sets of power semiconductors and
integrated magnetics with common input/output filter capacitors and a
shared Cuk capacitor bank. Each power stage operates at a fixed frequency
of 10 kHz with the beginning of each stage’s power cycle synchronized by a
ring counter. Separate control circuits determine the duty ratio for each
power stage. The integrated magnetics utilize a powdered iron toroid rather
than a gapped core to minimize flux radiation. The assembly sequence for
the integrated ‘‘lumps’’ requires winding the input and output inductors
separately, then stacking these assemblies and having the transformer
windings use the cores of both inductors. Since adjustable core gaps had
already been eliminated, other ways were explored to give the zero ripple
current condition in the input and output inductors. The simplest
alternative is to adjust the inductor/transformer turns ratio slightly.
However, the lump power requirements called for the windings to be a
few turns of large wires, with the result that any one-turn adjustment has
too large an effect. Another approach adjusts the ratio of the values of the
input and output Cuk capacitors (normally proportional to the square of the
transformer turns ratio) to give slightly different voltage waveforms on
each transformer winding. This technique can be utilized to eliminate the
fundamental ripple current on both input and output with only a small
residual ripple, primarily at the second harmonic of the main switching
frequency. However, this approach requires that each lump have its own
set of Cuk capacitors. When the Cuk capacitors are shared in a stagger-
phased mode, certain power stage duty ratios (33% and 67% with three
phases) exist with almost no power ripple in the Cuk capacitors. The value
of the capacitor bank then becomes almost irrelevant. Only in a single
power stage setup can a disproportionate voltage ripple on the Cuk
capacitors be utilized over a wide operating range to steer the ripple
currents from the inductor windings into the transformer windings.
22.8.6 Transformer and Inductor

NASA Glenn with TRW Corporation developed a heat pipe cooled 2.2-kW,
20-kHz transformer with 30% reduction in mass and volume to yield a
specific mass of 0.6 kg/kVA. The transformer was designed so that the
decrease in transformer mass would only increase the thermal radiator and
solar array mass by 50% of the decrease. Its size could replace the one used
in a 30-cm ion propulsion engine power processor. Around the same time,
Thermal Technology developed a space-worthy high frequency, 25-kVA,
20-kHz hypercooled aluminum transformer with efficiency greater than
99% and a specific mass of 0.13 kg/kVA. Power systems using such
transformers should show a marked improvement in system mass, size, and
efficiency. In conjunction with the heat pipe cooled transformer, a heat pipe
cooled inductor was also developed. The mass of the inductor was reduced
by 40% compared to the conventional inductor, resulting in a smaller
capacitor bank in the input filter. This inductor was also designed to fit into
the 30-cm ion propulsion engine power processor. As for a scaled-up
design, a 1-MW transformer design using heat pipes and aluminum
conductor is estimated to have a specific mass of 0.14 kg/kVA at 400 Hz,
0.080 kg/kVA at 1 kHz, and 0.04 kg/kVA at 20 kHz.
22.9 Very-High-Voltage System

Voltages ranging from 200 V to 100 kV may be needed for the future high-
power civil and military space missions. The state of space power
technologies is changing, and high-voltage systems are evolving rapidly.
Therefore, space-qualified high-voltage power components will assume an
increasingly greater role in the coming decades. Key design considerations
for such components and systems are presented below.
The plasma arcing inception voltage in a space environment varies from
200 V in LEO to 1000 V in GEO. It is therefore obvious that high-voltage
equipment rated in kilovolts cannot have bare conductors exposed to space
plasma. This requires positive insulation systems for the equipment
consisting one or more of the following design features:
Insulating gas, typically at high pressure

Insulating liquid
Perfect plasma-free vacuum with minimum out-gassing materials used
in the construction
Encapsulation with solid insulation, typically an epoxy-resin com-
pound
The incoming and outgoing insulated cables cannot be terminated in
bare exposed terminals. The insulated cable must enter inside a sealed
casing containing the insulating medium
The gas insulation, being mass effective, is likely to find applications in

very high voltage spacecraft. A 3-phase gas insulated transmission line
concept is shown in Figure 22.26 and Table 22.6 list the relative dielectric
strengths of commonly used insulating gases at atmospheric pressure. This
strength increases with increasing pressure. Notice that FreonTM, which is
commonly used in 2-phase cooling systems, has the best electrical strength,
followed by SF6, which is inert, nonflammable, nontoxic, and odorless.
Hydrogen is the least preferred electrical insulation, in addition to it being a
safety hazard.
The liquid insulation, being 3 to 4 orders of magnitude denser than gases,
possesses much higher dielectric strength and heat removal capacity.
FIGURE 22.26 Gas insulated 3-phase transmission line for very high voltage.
Moreover, like gas but unlike solid insulation, liquid tends to be self-healing
should a discharge occur in it. The commonly used liquid insulations with
their dielectric breakdown stresses are listed in Table 22.7. The design stress
margin depends on the dielectric stress concentration factors and other
uncertainties in the design calculations.
Major factors that cause the insulation to degrade are the operating
temperature and the corona discharge. Temperature degradation is well
understood. The insulation life reduces to one half for every 7 to 10 C rise in
the operating temperature. As for the corona, it degrades different materials
at different rates. The material which has high dielectric withstand
capability when new generally degrades faster under corona. The material
selection must therefore match with the mission duration. Yet another
degradation factor, applicable only to solid encapsulated components, is
thermal cycling. The differential coefficient of thermal expansion between
the conductor and the polymer encapsulant induces thermo-elastic stress
cycles. These may lead to internal cracks in the solid insulation, thus
degrading the dielectric properties of the insulation system.
As the power transmission distance increases, the voltage drop and
power loss can be kept reasonable only by using high voltage. For this
Table 22.6 The relative

breakdown strength of
commonly used insulating
gases at atmospheric
pressure
Gas Relative strength
Air 1
Nitrogen 1.05
CO2 0.95
SF6 2.5
C5F8 (Freon) 5.5
Hydrogen 0.6
Table 22.7 The breakdown strength

of commonly used insulating liquids
Design stress
Liquid (kV/mm)
Petroleum oil 2–6

Chlorinated hydrocarbons 4–12
Silicone oils 4–12
Synthetic hydrocarbons 10–25
reason, several kilometer long lines would require several kilovolts if the
drop were to be kept below 5%. At that voltage and distance, the use of SF6
to reduce the line mass by as much as 50% may be considered. As an
example, some kind of high dielectric strength gas insulation may be
necessary for connecting the SP-100 power source with a space station by a
several kilometers long tether cable operating at very high voltage. For such
a system, it is envisioned that the SP-100 power output may be stepped up
to about 10 kV and transmitted to a space station where it would be stepped
down to the user’s level. The electrical and mechanical performance and
mass characteristics of such a transmission line are compared using both
SF6 and vacuum. A 5-km-long transmission line mass can be significantly
reduced to about one half by substituting pressurized SF6 gas for vacuum
insulation. The improvements basically result from the higher dielectric
field strength that can be allowed in SF6. The insulating distance between
the several kilovolt line conductors and the grounded outer tube
substantially decreases with SF6, thus decreasing the outer tube diameter
and the support mass by a significant amount. There is an added mass of
the pressurized SF6 gas management system (reservoir, pumping, and
pressure monitoring systems) in a sealed line construction. However, this
added mass would be small compared to the mass of several kilometers
long line. On the other hand, the insulation mass reduction potential in the
50-m long SP-100 boom cable is much smaller than the added mass that
may be required for the gas management system. For comparison purposes,
the mass of a 100-kW, 10-kV, 5-km line is estimated to be 1600 kg with SF6
gas insulation versus 3200 kg with vacuum insulation, a net reduction of
1600 kg. The 200-V boom cable in the present SP-100 concept would weigh
7200 kg for a 100-kW 5-km line.
22.10 Repetitive Pulse Power

The smart bombs in the two Gulf Wars fell from U.S. fighter aircraft
precisely onto laser-illuminated targets. The carbon dioxide lasers in those
systems were driven by pulsed-power sources. Such specialized applica-
tions demand very high power for a very short time, i.e., pulsed power. The
pulsed laser is one such application. Others include medical, particle

accelerators, and even strategic defense projects, such as simulation of the
electromagnetic pulses (EMPs) that result from nuclear explosions. The
concept of pulsed power is not new. It was used in radar systems in World
War II and its new applications are expanding. Pulsed-power sources are
fully integrated systems, rather than single-power supplies. In such
systems, a high-voltage constant current source (rather than a constant
voltage source) is used to charge a bank of capacitors. The power supply is
not directly connected to the load; its sole use is to charge the capacitor
bank. A switch is added to control the discharge of the capacitor, and a
resistor limits the current. Fuses protect the equipment under adverse
conditions.
High-voltage power supplies can deliver up to 8000 J of energy per
second, with voltage ratings up to 50,000 V. Since capacitor charging varies
from zero to peak voltage, the energy measured is an average delivered to
the load. Though both linear and switching supplies are used in pulsed-
power systems, the trend is towards switch mode technology. Most units
are designed to fit in standard 19-inch racks. The switching frequency is
limited to 50 kHz, since the component cost for high-voltage applications
rises with the frequency.
In pulsed-power applications, the capacitor capability is measured in
terms of the specific energy. A good capacitor needs a high-capacitance,
low-temperature coefficient, and relatively low inductance. The capacitor
dielectric has to withstand very high voltage and current, offer temperature
stability, and have a fast discharge frequency. Pulse requirements vary with
the application. Rail-gun applications need a 5 to 40 kV pulse with pulse
widths of a few milliseconds. EMP simulation needs multi-megawatt pulses
of a few nanoseconds duration.
The trend in repetitive pulse-power system design is toward higher
energies, larger average power levels, and faster pulse repetition rates. This
trend results in requirements for components with higher reliability, lower
losses, and longer life times. Emphasis is placed on those components and
material requiring significant development efforts. Switches are by far the
weakest active components in pulse-power systems. Capacitors, for
example, are one of the weakest elements in high-power pulsed systems,
especially when operated at high repetition frequencies for extended
periods of time. Improved insulation in systems and components is
essential to increasing energy density and reliability. The pulse transformer
and inductor designs require special cooling and thermal time-constant
considerations.
The development of long-life, fast-discharge components is essential to
many future high-energy applications. Many of these systems are needed
where compactness, light mass, high reliability, and low loss are prime
requirements. Capacitors and switches presently available cannot meet this
combination of projected requirements in high repetition rate power
systems. At present, the energy density of capacitors operated at continuous

high pulse rates is typically a factor of three to ten less than those used in
single pulse systems.
The basic characteristics and problems associated with spark gaps and
thyratron switches used in pulsed power systems need to be addressed. Gas
dynamic spark gaps in the thyratron switches often require more power to
furnish compressed insulating gas than the average power they can pass.
The corona characteristics of liquid and solid insulation limit the
permissible dielectric stresses in the materials, and consequently the energy
density of the system. This following brief discussion is intended to
illustrate the major areas of development necessary for pulsed power
components.
Capacitors are the basic electrostatic energy storage elements in pulsed-
power systems. The rationale in evaluating capacitor technology is to
determine and understand the failure mechanisms and then work with
manufacturers to develop improved units. Recent programs have sought to
determine the capacitor’s weak points, the characteristic failure mechan-
isms, and the required dielectric characteristics compatible with high dI/dt
discharge capacitors operating at 1-kHz pulse rate frequency and 100 ns
discharge time.
The performance evaluation of low-inductance type of capacitors in the
100 ns, 1-kA, 1-kHz regime has led to the following conclusions:
Partial discharge (corona) is the main cause of failure in the pulse

power capacitors and are detectable during energy discharge.
The partial discharge inception stress level during discharge is about
257 V d.c., and is dependent on time duration of discharge and the
pulse repetition rate.
A high-quality, high-frequency diagnostic system is essential to observe
transients within capacitors during discharge. The high-frequency,
high-voltage probes developed for the system are required as diag-
nostic tools.
Preliminary tests indicate that Teflon–silicone oil and mica-paper
capacitors have longer life in the regime tested than the polypropy-
lene-silicone oil units, but with lower energy density. The latter is
attributed to the impregnation characteristic of Teflon units and
fabrication methods of the mica units, and may not represent intrinsic
limitations.
The major components of such power systems are the high voltage
capacitor, thyratron switch and the air-core transformer. The capacitor is
best suited for repetition rates in kilohertz. A battery may be used for a few
hertz rate. For a still lower rate, superconducting coils may find applica-
tions.
The following are the rough guidelines for selecting the energy storage
technologies for a pulse-power system:
Capacitor for high pulse rate in kilohertz

Battery for medium pulse rate
Flywheel or superconducting coil for low pulse rate
22.11 Multi-Megawatt Burst-Power

The power requirements for some strategic defense spacecraft range in tens
to hundreds of megawatts of burst power. Directed energy weapons,
typically placed in LEO, are examples. Such a power system is required to
deliver high power for a short time of engagement, after which it must
recharge. Important considerations in selecting the system architecture and
the subsequent design detail include the following:
Primary energy source, energy conversion, and energy storage to meet

the peak power requirement and rapid startup.
Power conditioning and distribution
Thermal management during engagements
Optimum voltage selection for mass minimization
Low vibration and low torque on the weapon platform
Long stand-by with low quiescent power requirements
Low mass, low cost, and high reliability
A future reusable directed energy weapon in LEO is predicted to require

power over 10 MW at 270 V d.c. for several minutes occasionally. For such a
short duration occasional, very high burst power, large-scale energy storage
in a flywheel or superconducting coil may be considered. However, the fuel
cell may be an ideal candidate. For the Air Force Research Laboratory,
Allen9 evaluated 10-300 MW burst power system design. He considered
several top-level options — chemical-dynamic, nuclear-dynamic, chemical-
fuel cell, and battery and flywheel which could be recharged using a solar
array power.
The study concluded that a system using regenerative solid oxide fuel cell
(SOFC) with hydrogen and oxygen reactants could deliver the required
power with the lowest power system mass, and is the best approach with
the presently available technologies. The PEM fuel cell system results in
higher mass, primarily from the extra cooling fluid and the associated
storage tank, which was significantly higher than the mass of the high
temperature radiator used with the SOFC. The fuel cell is the lowest mass
system primarily because it produces the power only when needed and at
270 V d.c., which does not require additional power conversion and
condition. The flywheel generally results in heavier power system that is
not used continuously. Cooling of the power system and the weapon
components in the study was achieved by a circulating ammonia loop.
The study found the following system optimization at the 10 MW power
level:
Power during engagement 10 MW

Primary energy source Chemical
Energy conversion Fuel cells
Energy storage None major
Voltage level(s) 270 V d.c.
The 10-MW SOFC system features are summarized in Table 22.8, and the
study results are summarized in Figure 22.27 for 5 min of the weapon run
time.
The solar array-SOFC system mass breakdown in percentage of the total
power system mass was reported as follows: 60% in the fuel cell stack, 10 in
fuel and steam tanks, 8 in reactants, 6 in radiators and plumbing, 14 in solar
array, and 2% in all other components.
Recharging the fuel cell requires electrolyzing the by-product water back
into hydrogen and oxygen. The SOFC can electrolyze water using the
recharging process. The electrolyzer is generally a separate unit from the
stack and the two cannot operate simultaneously. Recharging the fuel cell in
1 day in Allen’s study required a 213-m2 solar array generating 70 kW.
Extending the recharge time to 5 days cuts the array to 55 m2 generating
18 kW power (Figure 22.28). However, this may conflict with some mission
scenarios due to condensed periods of conflict.
The dynamic system may be required for sustained high power. Some
strategic defense weapons platforms may require burst powers up to 300
Table 22.8 Ten megawatt solid oxide fuel cell burst mode power
systems features
Burst power 10 MWe, scalable to 300 MWe

Burst mode duration 100–1000 s
Test duration 60–240 s/year
Mission life 10–15 years
Fuel cell efficiency 70%
Operating temperature 1250 C
Working pressure 135 atmosphere
Oxygen consumption rate 1.75 kg per MWsHydrogen consumption rate
0.015 kg per MWs
Effluents Negligible
FIGURE 22.27 10-MWe power system mass with alternate design concepts.
(Data source: D.M. Allen, Proc. 36th IECEC, pp. 243–249, ASME, 2001.)
MW. Such high power requirements are expected to consider nuclear or

chemical sources for the primary energy, and rotating electrical machines
for energy conversion. In a baseline concept, the reactor heats a fluid,
typically liquid metal such as mercury or sodium potassium. The vapor
then drives a turbine-generator using the Brayton or Rankine cycle. The
space nuclear system is similar to a ground-based nuclear power station.
The major difference in space is that the waste heat must be rejected only by
radiation back to space. Both systems use nuclear fission of a heavy metal
atom, which releases a large amount of energy, primarily in the kinetic
energy of the fission fragmentation. With uranium-238, the energy released
per fission amounts to the following:
166 MeV in kinetic energy of fission fragments

10 MeV in gamma radiation
17 MeV in neutrons and neutrinos
10 MeV in further fragmentation
Thus a total of 200 MeV thermal energy is released per fission. The
conversion between million electron volts and watts gives 3.1 1010 fissions
per second needed to release 1 thermal watt. This energy is used to generate
steam and drive a turbine-generator to convert to electrical power. In a
FIGURE 22.28 Solar array versus recharge time after 5 min of weapons run with 10 MWe
load. (Data source: D.M. Allen, Proc. 36th IECEC, pp. 243–249, ASME, 2001.)
large-scale power system in the megawatt range, the overall thermal to

electrical conversion efficiency is 40 to 50%.
The nuclear reaction must be contained in a well-shielded enclosure. In
principle, the chain reaction of the fission, once started, continues as long as
the critical mass is maintained, and the energy can be released at any
desired rate. However, in practice, the release rate is limited by temperature
problems, so that a given reactor has a maximum achievable release rate.
In using the synchronous generator, the steady-state torque on the
platform is a nonpulsating constant torque (of d.c. nature), and does not
affect the platform pointing accuracy. However, the transient torque during
a fault (e.g., short circuit on the output terminals of the generator) is of a
pulsating nature, and constitutes a major dynamic disturbance on the
platform. Such a torque is the product of the alternating short-circuit
current and the constant magnetic flux in the air gap, which becomes frozen
on the instant of a short circuit, in accordance with the constant flux-linkage
theorem.
Spacecraft momentum and vibration issues are important integration
considerations in any system using large rotating machines. However, as
cited above, rotating machines produce large dynamic torque on the
platform, particularly during faults, which may hinder the pointing
accuracy of the spacecraft. The disturbing torque is minimized by using
such machines in one or more pairs of counter-rotating machines to cancel
the torsion effects of each other. Figure 22.2910 shows one such arrangement
of three pairs of counter-rotating turbo-generators on a space platform.
There always remains a net torque on the platform, as two machines in a
pair cannot be located exactly at the same place. However, the momentum
FIGURE 22.29 The placement of multiple pairs of large rotating machines on a space
platform.
cancellation of 0.999999 by using counter-rotating machines is required to

minimize the spacecraft uncontrolled spin and/or tumble. In a MMW
power system, a mismatch of even less than 0.01% could gyrate the
spacecraft out of control, and a mismatch in the order of 0.001% could
introduce significant pointing errors.
In the case of a chemical-dynamic system, the Brayton engine is lighter
than the Stirling. Both engines may be used in counter-rotating pairs,
although the concern of mismatch is less severe due to their inertial energy
being lower by several orders of magnitude than in high-speed flywheel
energy storage. However, the vibration concerns remain. Here, a more
important integration issue is the fluid loop, which can reverse direction
under a high slew rate. Both heat engine and fuel cell reactants can be
stored at high pressure (3500 psi) or as a liquid at cryogenic temperatures.
Cryogenic storage results in lower system mass.
A major design challenge in the MMW weapons spacecraft is the thermal
management to remove heat from the electrical power source and then
remove it from the spacecraft while maintaining a narrow thermal gradient
between the components. A 100-MW system with 35 to 40% thermal-to-
electrical conversion efficiency engaged for 10 min needs 36 GJ of heat to be
dissipated eventually into space. The heat may be removed from the
spacecraft as generated, or stored on the spacecraft for gradual removal.
The thermal system options are direct radiation, liquid heat capacity
storage, and latent heat phase change storage with spray cooling. The
former would have the greatest mass, whereas the latter would result in the
least mass system. However, a major concern with spray cooling is the large
pressure differential required to pump the liquid and the vapor. A second
concern is the momentum of the fluid caused by high mass flow rate. A
thermal system split into at least two counter-directional fluid loops can
reduce the net momentum to a tolerable level.
Many aspects of MMW power system design have to be considered in

conjunction with the requirements of an actual space mission. In order to
take advantage of the dynamic power system characteristics, the system
must be thoroughly integrated into the spacecraft. The synchronous
generator is a more likely candidate in such high-power missions due to
the high air-gap torque characteristic. Among such machines, super-
conducting or hyper-conducting generators could be advantageous for
their low mass, low volume, and high efficiency. Superconducting
machines use a superconducting d.c. field coil on the rotor, which
completely loses its electrical resistance below a critical temperature, Tcr.
Until recently, the Tcr of superconductors has been around that of liquid
helium, but is now closer to that of liquid nitrogen, with the discovery of
some higher temperature superconductors. The a.c. stator winding in this
machine still uses conventional conductors, as a.c. superconductors do not
exist — not yet, at least. The hyper-conducting machine, on the other hand,
uses a high-purity aluminum conductor in both the rotor and the stator at
liquid nitrogen temperature. It exhibits extremely low electrical resistance,
but not precisely zero as in the superconducting machine. Both the
superconducting and the hyper-conducting generators are likely candidates
for high-power space missions.11
The design characteristics of a 300-MVA (270 MW) superconducting
generator are summarized in Table 22.9. Although a higher frequency
machine may be advantageous on space platforms for their lower mass and
higher efficiency, a two-pole, 60-Hz machine is considered here for two
reasons:
Table 22.9 Three-hundred megavolt ampere

superconducting generator design
parameters
Rating 300 MVA. 0.9 power factor

Voltage 24 kV, 60 Hz, 3-phase
Rotor speed 3600 rpm
Rotor cooling Liquid helium
Stator cooling gaseous hydrogen
Efficiency 99.5%
Active length 78 inches
Rotor outer radius 18.6 inches
Armature inner radius 23.5 inches
Armature outer radius 32.5 inches
Core inner radius 35.5 inches
Core outer radius 52.0 inches
Rotor inertia constant 0.62 s
(Data source: M.R. Patel, Proc. 25th IECEC, p. 990-

X, AIAA, 1990.)
The centrifugal stress in the rotor of high-power machine limits the

rotor speed to around 3600 rpm, and
Some applicable design data are available in published literature on the
300 MVA superconducting generator development program funded by
the Electric Power Research Institute12
The platform’s dynamic-control and the pointing system must account for
the net effect of all the rotating machines (generators, turbines and pumps)
on the platform, which depends on their exact placement. The net effect of
all such machines is platform-specific and varies from one platform to
another.
22.12 High-Temperature Components

High-temperature components are required in high-power solar dynamic,
SP-100, pulse-power, and electromagnetic pumps for liquid metal. For
energy storage at high temperature, a sodium–nickel chloride secondary
battery may be used up to 275 C, sodiumsulfur up to 350 C, and lithium-
ion sulfide up to 450 C. A very large sodium–sulfur battery with cylindrical
cells of total capacity 48 MWh has been built in Japan for terrestrial use.
Premium grade silicone rubber or polyamide fibers such as Nomex and
Kevlar are widely used insulation for service temperature up to 250 C and
voltages of 10 kV a.c. or 25 kV d.c.
High-voltage, high-temperature wires rated up to 600 V and 450 C are
commercially available for use in furnace ignition leads, thermocouples,
aircraft, and missiles. They are generally made from nickel-clad copper wire
insulated with glass reinforced mica tape or coated with a glass–ceramic
compound. Tungsten or nichrome wires with 3M NextelTM ceramic fiber or
Rockbestos PhosrocTM insulation can operate up to 1200 C continuously
and 1400 C in the short term. Such wires have also been considered for use
in submarine fireproof areas. Their radiation resistance has been tested up
to 107 rads.
The life of high-temperature conductors commonly used in heating
elements is shown in Figure 22.30. On the insulation side, it is always
important to choose a dielectric with thermal stability. One high-tempera-
ture wire candidate is cuprous oxide thin film insulation on a copper
substrate. It is available from Permalustetr Inc. in rectangular wires and thin
strips. The insulation is impervious to nuclear radiation, gamma rays,
corrosion, corona, high vacuum, and cryogenic temperatures. It can operate
at 1000 C without additional corrosion or oxidation of the copper. The
insulation film thickness is 3–4 micron for 100-V insulation. It is highly
flexible, does not out-gas or craze, and allows other over-coatings. It is
ceramic in nature. It is not a deposition, only a conversion coating. It would
allow compact coils in inductors and transformers.
FIGURE 22.30 High-temperature heater element life characteristics.
Nickel connectors are good up to 650 C. The bare connectors may be

coated with some sort of insulation, such as ceramic sprays, potting
compound, and ceramic fiber tapes.
The following is a brief summary of NASA/Westinghouse Lima research
on the development and evaluation of magnetic and electrical materials
capable of operating in the temperature range of 800 to 1600 F. Advanced
space power systems are the area of eventual application of such materials.
High-temperature magnetic materials with satisfactory strength are
needed for use in solid rotors of electrical generators. Screening tests
have been conducted on martensitic alloys, using additions of tantalum
plus tungsten and tantalum plus silicon to a base composition of iron with
15% nickel and 25% cobalt. They indicate that these alloys show promise for
meeting the desired creep stress exceeding 60,000 psi at 0.4% creep strain,
and magnetic saturation exceeding 13 kilogauss at 1100 F. In the 1200 to
1600 F range where dispersion strengthening is applied to cobalt and
cobalt–iron base alloys, results show the saturation magnetization goal of 12
kilogauss can be met by 27 cobalt–iron base alloys containing as much as 20
volumepercent dispersoid. No problem appears to exist in meeting the goal
of a coercive force of less than 25 oersteds at 1200 to 1600 F in either 27
cobalt–iron base or cobalt base compositions. Creep tests on Nivco alloy at
1100 F were conducted at pressures in the 3 to 5 109 torr range at 1100 F.
Design analysis suggests creep strains in excess of 1% should be run above
1050 F. This would provide more reliable design data for rotors in inductor
alternators even though the design is based on only a total creep strain of
0.4%, because this represents the integrated value of a wide range of
incremental creep strains.
The tests also determined the feasibility of high-temperature capacitors
using high-quality dielectric materials. Comparison of electrical data on
boron nitride (boralloy), polycrystalline alumina (Lucalox), polycrystalline
beryllia, and single-crystal alumina (Linde sapphire) showed that pyrolytic

boron nitride has significantly lower a.c. power loss, higher d.c. resistance,
and exhibits less change in capacitance with temperature than any other
material investigated in the temperature range from room temperature to
1100 F. In addition, the d.c. breakdown strength of pyrolytic boron nitride
is 7000 v/mil at 1100 F, which is several times greater than that obtained
with sapphire or beryllia.
Alkali-metal compatible ceramic-to-metal seals and combinations of
material designed into generator stator with a bore seal, a transformer,
and a solenoid were investigated of compatibility under electrical and
magnetic stresses at elevated temperature and under high vacuum. Ceramic
out-gassing studies indicated that the firing cycle applied to a ceramic using
nitrogen–hydrogen atmosphere followed by the vacuum treatment, which
they undergo during brazing is sufficient for minimizing oxygen bearing
out-gassing products. These would add only a few parts of oxygen to the
alkali-metal charge in a typical advanced electric power system. The stator,
transformer, and solenoid to provide data on a stability of materials applied
to typical electrical designs were installed in the thermal vacuum chamber
and stability testing was done at 1100 F hot spot temperature. The chamber
pressures were approximately 4 107 torr at the beginning of the 5000-h
tests and decreasing. The minimum cold chamber pressure obtained after
system bake-out was 1 1010 torr.
Magnetics for use at temperatures up to 600 C and voltages up to 600 V
d.c. may consider nickel clad copper coated with ceramic insulation or
electrochemically formed aluminum oxide film. Such a coating is primarily
a refractory-glass compound that can withstand flexibility and braiding
requirements consistent with normal coil winding operation.
To keep the surface from oxidation films and to enhance the mechanical
strength of copper at high temperature, nickel cladding is required. The
word cladding, as used by the copper wire industry, actually means
electroplating 27% of pure nickel by mass (or area, as both nickel and
copper have about the same mass density). The cladding reduces the
effective conductivity of the composite wire to about 75% of the core copper
because of the lower conductivity of nickel. With the core copper
conductivity of 85%, the composite wire would then have the conductivity
of 0.85 0.75 ¼ 0.64 times that of pure copper at room temperature. At
higher temperature, this value may be different. In magnetic environment,
nickel is slightly ferromagnetic at room temperature, but loses its
magnetism at about 700 F (370 C).
The SP-100 boom cable design requires electrical conductor and insulat-
ing materials that can continuously operate in 570 C environment. If we
add approximately 30 C for the internal temperature gradient, we need
conductor and insulating materials that can withstand about 600 C
continuous temperature at the reactor end. The cable temperature at 30 m
away from the reactor would be around 100 to 200 C depending on the wire
gage used, i.e., on the internal power loss in the wire. The total insulation
mass of a 200-V cable is less than 1.5 kg per 100 m. The 48 pairs of 50-m
boom cable have a total cable length of 4800 m with total insulation mass of
about 72 kg.
The conductor grade, which has a high electrical conductivity (close to
copper) and at the same time retains the necessary mechanical strength at
600 C suitable for the SP-100 boom cable application is CdCr copper. This
precipitation hardenable alloy has high mechanical strength, resistance to
softening at elevated temperatures, high flex life, and freedom from
embattlement. The boom cable is required to be flexible for ease in
deployment. Such flexibility is achieved by making the cable from
numerous thin strands (10 to 20 mils diameter). However, the thinner the
strands, the quicker they melt, should the temperature approach the
melting point even for a brief period (such as under a laser beam attack).
The threats and the resulting over-temperatures determine the final
conductor strand size. In the absence of a specific threat, adopting the
industry and military standards appears to be reasonable, which calls for
133 strands of 11 mils diameter in AWG No. 8 wire and 133 strands of 14
mils diameter in No. 6 wire. From the high-temperature insulation
candidates shown in Figure 22.31, glass ceramic or asbestos fibers are
potential candidates, although Nextel of 3M or Phosroc would surpass the
requirement.
FIGURE 22.31 High-temperature insulating material operating characteristics.

References
1. Barthelemy, R.R., Messie, L.D., and Borger, W., Military space power
system technology for the 21st century, Proceedings of the Intersociety Energy
Conversion Engineering Conference, 1986, Vol. Ill, pp. 1401–1410.
2. Brandhorst, H.W. and Best, S.R., Effects of hypervelocity impacts on solar
cell modules at high voltage, Proceedings of the 36th Intersociety Energy
Conversion Engineering Conference, ASME, 2001, Vol. I, pp. 353–358.
3. Binz, E. and Hartung, J., Solar dynamic power for a space station,
Proceedings of the 21st Intersociety Energy Conversion Engineering Conference,
pp. 2072–2076.
4. Merryman, S.A., Bandy, A.J., and Gordon, L.B., The breakdown character-
istics of an out-gassing dominated vacuum region, Proceedings of the
Intersociety Energy Conversion Engineering Conference, 1987, Vol. Ill, pp.
763–765.
5. Patel, M.R., High Frequency Power Distribution System, NASA Report No.
CR-175071, 1986.
6. Bouquet, F.L. and Winslow, J.W., Radiation Data Definitions and Compilation
for Equipment Qualification Data Bank, NASA Tech Briefs, January 1987.
7. SAE STD-1031, Aircraft 270 V dc power distribution bus, electrical power, HV
dc, aircraft, characteristics and utilization of.
8. MIL-STD-1399, Submarines 155 Vdc power distribution (100 kW range),
interface standards for shipboard systems.
9. Allen, D.M., Multi megawatt space powers technology program,
Conference, ASME, 2001, Vol. I, pp. 243–249.
10. Patel, M.R., Dynamics of high power rotating machines on space
platforms, Proceedings of the 25th Intersociety Energy Conversion
Engineering Conference, AIAA, 1990, Paper No. 990-X.
11. Gamota, G., One two three - zero resistance in defense applications, IEEE
Transactions on Magnetics, Vol. MAG-17, January 1981.
12. Patel, M.R., Nathenson, R.D., and Ahmed, M.E., ‘‘Designing pancake coils
of large 300 MW superconducting generator for electromagnetic and
thermal loads, IEEE Transactions on Power Apparatus, Vol. PAS-102, August
1983, pp. 2710–2716.
Chapter 23
Electric Propulsion
23.1 Introduction
Electric propulsion continues its expansion in space. More than 160
spacecraft in the Earth orbit and interplanetary space now use it in some
form. Many new programs have begun to develop the next generation of
electric propulsion for solar and nuclear powered spacecraft. Although it is
another load on the bus, some of its unique features that influence the
power system design and operation are discussed in this chapter.
In the geosynchronous orbit, the satellite must maintain the assigned
orbit altitude and inclination to stay locked with a fixed ground station. But
the orbit slowly degrades due to various drags in the space. The propulsion
system maintains the satellite in the desired orbit and controls its attitude
towards the Earth. Propulsion is also needed for orbit transfer and to
deorbit the satellite for disposal at the end of a mission. The requirements
for these functions are listed in Table 23.1.
The propulsion thrust in a typical satellite is produced by chemical fuels
carried on board. In a geosynchronous satellite, over 90% of the on-board
fuel is used for N–S station keeping, and the rest for E–W station keeping,
attitude control, and orbit transfer. The resultant thrust vector of all station-
keeping thrusters must pass through the center of the satellite mass. The
thrust vector for the attitude control, on the other hand, must be at a right
angle to the center of mass in order to generate the maximum moment.
Chemical propulsion typically uses a catalytic monopropellant hydrazine
(N2H4) fuel. No oxidizer is needed. Heat or a catalyst decomposes the fuel,
and the chemical reaction produces heat and thrust of the exhaust gas.
Some satellites use fuel and oxidizer in separate tanks to achieve higher
Table 23.1 Propulsion requirements for three key functions
Function Propulsion requirement
Orbit achievement High thrust, high efficiency, low precision

Orbit adjustments High precision, efficiency not an issue
Disposal High efficiency, high dormant reliability, low precision
566
Electric Propulsion 567
thrust levels. Such bi-propellant fuel systems generally uses hydrazine and
nitrogen tetroxide (N2O4). This yields specific impulse around 300 s.
At present, electric propulsion is used in low-thrust maneuvers.
However, attractive mass saving results in applications requiring high-
velocity increments, such as for N–S station keeping (50 m/s per year), and
for orbit raising (1 to 5 km/s). Full orbit transfer from low to geosynchro-
nous orbit requires about 6 km/s velocity. Electric propulsion has several
times the efficiency of a chemical system, and is attractive for attitude
control in LEO satellites. In GEO satellites, it is often used for N–S station
keeping, and sometimes for orbit raising, in which the transfer time can be
long. Although in theory, the thrust-to-drag ratio must be greater than 1 to
raise an orbit, it must be greater than 3 for practical control stability
requirements.
For the N–S station keeping (NSSK), electric propulsion can be powered
by the satellite battery without additional mass except for a dedicated load
power converter. Orbit raising requires much higher power, which must be
derived from the solar array. Small orbit raising using low specific impulse
devices may be considered using the existing solar array and battery. The
propulsion operation is performed during sunlight in the non-eclipse
season. During this period, the extra power available from the solar array
after meeting the load power requirement is used to fire the propulsion.
Additional power, if needed, can come from the battery up to the maximum
allowable DOD at the end of station keeping. Electric propulsion that
requires large amounts of power significantly impacts the power system
architecture and design. For example, full orbit raising would definitely
need an additional solar array or some other power source, which would
add mass.
Many electric propulsion technologies require a high-voltage power
source to generate arcs and plasma that is electrically conductive. The
power system control circuit must provide protection against damage to
either the thruster or power system components in the event of inevitable
arcing between high-voltage elements. The most frequent place for an arc
to occur is in the thruster. The high-voltage power source is designed in
two ways to provide protection against such arcing. One way is to
incorporate an over-current sensing circuit, which turns off the high
voltage in case of an arcing fault. Alternatively, the source can be
designed with high output impedance so that an arc merely results in
deep voltage sag. In the cesium thruster, discussed later in this chapter,
the thruster electrodes are subjected to excessive cesium atom deposition
when the high voltage is turned off. The cesium continues to flow from
the ion source, and the evaporation from the ionizer is neutral cesium
since no electrostatic field exists to draw off the ions. The voltage should
therefore be reapplied quickly. Under a loss of high voltage, electrode and
insulator contamination leading to excessive drain current and arcing is
possible at any time the cesium is flowing from the ion source. The power
system design must provide a method for handling an arc or loss of

voltage for a specified period of time.
23.2 Specific Impulse

The function of propulsion system is to apply impulse to change velocity of
the spacecraft. When the mass is ejected through nozzles, the reaction
propels the spacecraft in the opposite direction. Since the ejected mass must
be carried on board at the time of launch, the impulse produced per
kilogram of the ejected mass is taken as a figure of merit of the propulsion
system. The impulse is defined as the time integral of force. A constant mass
flow rate of dm/dt exhausting at an effective velocity Ve through a nozzle
creates a steady force F ¼ (dm/dt) Ve. Such a force applied for it seconds
results in impulse I ¼ Fit. The impulse created per kilogram mass of
propellant, called the specific impulse, is then given by
Isp ¼ I=m ¼ F=ðdm=dtÞ ð23:1Þ
The specific impulse is also known as the propellant efficiency or the

thrust to mass flow rate. Its unit is second. The total impulse a fuel of mass
im can create is then Ispim. This figure of merit, traditionally used for
comparing chemical propulsion systems, is less meaningful for comparing
electric propulsions. It is therefore modified to be the impulse created per
kilogram of total additional mass that must be carried on-board for electric
propulsion. The additional mass is mostly in the power system, and to a
lesser extent in the propellant. Most of the added mass is in the solar array,
and some in the power conditioning equipment. A modified figure of merit
for electric propulsion is developed below using the following notations:
Pe ¼ additional power needed for electric propulsion, watts

Mp ¼ mass to generate additional power for propulsion, kg
Psp ¼ specific power of the power source ¼ Pe/Mp, watts per kg
Me ¼ mass of the expellant needed on board, kg
Msc ¼ spacecraft mass without additional masses, kg
Mt ¼ total spacecraft mass ¼ Msc þ Mp þ Me, kg
Ve ¼ expellant exhaust velocity, m/s
Tb ¼ propulsion burn time, seconds
Assuming that the expellant mass flow rate is constant during the
propulsion burn time, it is Me/Tb. We also assume a constant expellant
velocity during the burn. Then, the energy conservation requires the
balance of power, i.e.,
Pe ¼ 1=2ðMe =Tb ÞVe2
which, by using classical rocket equations, leads to the following two

expressions:
expellant mass,
Mt Msc
Me ¼ ! ð23:2Þ
Ve2
1þ
2Tb Psp
power system mass,
Mt Msc
Mp ¼ ð23:3Þ
2Tb Psp
1þ
Ve2
Both Me and Mp rise and then fall with changing parameters. An

optimization study is therefore needed to minimize Me for a given mass
ratio of Msc/Mt. The study would also show the power related mass Mp is
lower at higher exhaust velocity up to a certain point, beyond which Mp
rises.
For electric propulsion, one figure of merit is the impulse per kilogram of
total additional mass (Me þ Mp), and another is the thrust-to-power ratio
(TPR) given by
dMe
Ve
TPR ¼ dt ð23:4Þ
Pe
With specific power achievable with the present power system technol-
ogies, a long burn time and low exhaust velocity are inherent characteristics
of electric prolusion. Significant power improvements of several fold are
needed to make electric propulsion viable for lift-off from a planetary
surface or orbit raising. However, electric propulsion for on-orbit maneu-
vering and interplanetary missions is practical and is being used at present.
In designing any type of electric propulsion, the exhaust velocity must be
selected to minimize the total mass added in the power system and the
propellant. That would maximize the available payload mass in a given
spacecraft with a given launch vehicle capacity. This trade is depicted in
Figure 23.1.
FIGURE 23.1 Payload maximization with electric propulsion.
23.3 Types of Electric Propulsion

Various types of electric propulsion are described below.
23.3.1 Electrothermal Hydrazine Thruster

The EHT, also known as Resistojet is the simplest of all in concept and
is depicted in Figure 23.2. A propellant is heated by a tungsten heating
element. The energy (enthalpy) of the propellant gas is increased by
heating, which is then converted into kinetic energy. The accelerated gas is
directed through nozzles and propels the spacecraft. The exhaust velocity is
a function of temperature, and is generally kept below 10 km/s from the
material life considerations. The achievable thrust level is in a fraction of
newton. The propulsion efficiency is around 75%. The propellant can be
hydrogen, nitrogen, or ammonia. Hydrogen, having high specific heat and
thermal conductivity, works better than nitrogen or ammonia. However, it
is corrosive, and requires cryogenic temperature for storage. Ammonia, on
the other hand, is easy to store without refrigeration, but is also corrosive,
limiting the life of both the heater and the nozzle.
23.3.2 Hydrazine Arcjet

The HAJ, in principle, is similar to EHT, except that the propellant is heated
by its own ohmic heating by passing through arcing electrodes, instead of
using a separate heating element. Thus, the propellant is directly heated. It
uses battery power during non-eclipse seasons. The specific impulse is high
in excess of 1500 s, but the excessive electrode wear is life limiting. A typical
GEO communications satellite carries four arcjet thrusters for N–S station
keeping maneuvers. Two arcjets are fired at a time for NSSK, one from each
FIGURE 23.2 Electrothermal hydrazine thruster.
battery, 2 weeks before or after an eclipse season. The arcjets are always
fired during sunlight, nominally at 1200 and 2400 hours with respect to
local midnight. Each HAJ consumes 1 to 2 kW power depending on the
spacecraft size. The arcjet power circuits are always fused. Since this is
usually the heaviest single load on the spacecraft, the arcjet fuse-blow
voltage transient is generally the most severe on the bus. The computer
simulation of this transient is therefore required for all new designs. The
hydrazine arcjet thruster typical envelope and electrical performance are
shown in Figure 23.3.
23.3.3 Ion Thruster

In the ion propulsion system (IPS), charged particles are accelerated in an
electrostatic field. The thrusters are classified according to how charged
ions are created from the neutral propellant. In a simple method, vapor of a
material having low ionization potential is ionized by passing through a
heated porous tungsten plate. Cesium is often used for the vapor material.
High heat loss due to thermal radiation in the process limits the overall
efficiency.
The schematic of ion thruster in Deepspace-1 spacecraft is depicted in
Figure 23.4. In this scheme, electrons emitted from thermionic cathode pass
through concentric cylindrical anodes. An externally applied weak
magnetic field sets the electrons in a spiral motion. The collision between
FIGURE 23.3 Hydrazine arcjet thruster typical envelope and electrical performance.
these accelerated electrons and the propellant ionizes the propellant. An

inert gas, such as argon or xenon, is used as the propellant. Thrust levels in
excess of 1 N are possible in large units. Good mass saving is achieved
using such propulsion for the N–S station keeping in geosynchronous
satellites. A typical chemical thruster emits particles at 3 km/s. A typical ion
thruster emits ions at 30 km/s, and needs only 10% propellant. The ion
propulsion thus saves launch cost significantly.
In the working principles, a strong electrostatic field is created between
two plates held at high voltage, generally at several kilovolts. A charged
particle injected into the E-field is accelerated along the electric field lines
with force equal to qE, where q is the particle charge and E is the electric
FIGURE 23.4 Ion thruster used in a Deepspace-1 spacecraft.

field intensity. Under this force, negatively charged particles move along
the rising field, while positively charged particles move along the falling
field. If the expellant mass Me having a total charge of Q is exhausted at
velocity Ve through the electric potential difference of U volts (not to be
confused with velocity V), the energy balance (in joules) gives
1=2M 2
e Ve ¼ QU ð23:5Þ
The propulsion power (in watts) required to create the propulsion thrust F
then follows the balance of power, i.e.,
Pe ¼ 1=2FVe ð23:6Þ
The above two equations lead to the propulsion thrust to electric power
ratio
sffiffiffiffiffiffiffiffiffi
F 2Me
¼ ð23:7Þ
Pe QU
The thrust-to-power ratio, a figure of merit, is maximized by using a

propellant having large mass to charge ratio (Me/Q). Argon, xenon, and
cesium fall in this category.
A version of this propulsion is being developed in Europe, and uses ions
created directly from a liquid metal surface by means of strong electric field.
Such propulsion gives a high force over short pulse time, and has high
efficiency. The gridded (conventional) ion thruster has a high exhaust
velocity. It has been developed extensively in the United States, Japan,
Germany, England, and France. The preferred version in this family is the
xenon ion propulsion system (XIPS) developed and tested at NASA/JPL.
The unique feature of the gridded ion thruster is the separation of the
ionization process from the accelerated beam by a series of electrostatic
grids. Neutral xenon is fed into the ionization chamber, where a low voltage
discharge is struck. A high ionization efficiency is achieved by hindering
the electrons from reaching the anode by a peripheral magnetic field. Ions
are extracted from the plasma, accelerated, and focused into tight beamlets
by the grid system. The negative accelerating grid prevents transmission of
electrons in either direction. Electron-rich plasma is emitted from the
neutralizer downstream of the grids, thus neutralizing space charge of the
ion beam.
The gridded xenon ion thruster differs significantly from the Hall thruster
discussed latter. XIPS has by far the highest exhaust velocity of all the
electric propulsion devices, but also gives a low thrust per unit power and
longer trip time. The ion flow through the grid is limited by space charge
buildup and the thrust density is low. On the positive side, the exhaust
beam is very well focused. A larger model of XIPS is being used in a four-
thruster system designed to do both the N–S station keeping and a
significant amount of orbit raising in large commercial satellites. Boeing’s
BSS-702 and BSS-601HP busses are such examples. The mass savings could
be up to three times as great as that associated with NSSK alone. For low-
and mid-Earth orbits, the entire orbit raising and de-orbit functions could
be performed by an ion propulsion system.
NASA’s Deepspace-1 interplanetary probe uses a gridded ion propulsion
system using PV power. It produces thrust with specific impulse
(propellant efficiency) 10 times greater than a typical chemical propulsion
system used during interplanetary missions. It uses a hollow cathode to
produce electrons that ionize xenon gas through collisions, yielding
positively charged atoms of xenon. The xenon ions are then electrostatically
accelerated by a potential of more than 1000 V and emitted from a 30-cm
thruster nozzle through an electrically charged grid. The power available to
the IPS during the Deepspace-1 mission comes from solar arrays.
The ion propulsion degrades the solar array in one way. It generates ions
that are discharged into space at high velocity. A large number of low-
energy electrons are also emitted to prevent the satellite from building up a
negative ion slowing down the charge. A few non-ionized atoms drift out of
the thruster and get ionized in collision with ions in the high-velocity
stream and form plasma. Through the drifted plasma, the low-energy
electrons can flow as the leakage current, draining power from the solar
array.
23.3.4 Stationary Plasma Thruster

The stationary plasma thruster (SPT), also known as the Hall-effect thruster,
falls into the mid-range of the exhaust velocity spectrum. Plasma (an
ionized gas that is electrically conductive) is made by stripping atoms of
their electrons, confining the resulting products, and preventing them from
recombining. Usually this means heating an appropriate gas mixture to a
temperature at which collisions break up the atoms and then keeping the
resulting product in equilibrium as long as needed. Forcing a large electric
current through a gas mixture can create a plasma. The SPT was originally
developed in the former Soviet Union, but recently has been improved and
tested extensively by a Russian, French, and American consortium. The SPT
is essentially an electrostatic ion thruster without grids. An axial electro-
static field and a radial magnetic field are established in the annular flow
chamber. An equilibrium discharge is set up when electron-rich plasma is
fed from the cathode while neutral xenon atoms are emitted through the
anode. Some of the electrons are trapped in the crossed electrostatic–
magnetic field and forms a virtual negative electrode, which accelerates the
ions. They ionize the xenon propellant by electron impact in the discharge
region and also neutralize the space charge of the accelerated ion beam. The
magnetic field produces a small azimuthal torque to the ions, which does
not change the thrust.
SPTs with improved performance are in various stages of development in
the 3 to 5 kW power range. A high power model SPT-100 has been used for
N–S and E–W station keeping operations on many satellites by Russia, and
is being used by Astrium, the French–German–British satellite builder on
Inmarsat and Intelsat communications satellites. It gives about 16 km/s and
generates about 80 mN of thrust using 1.5 kW of input power. This is a good
combination of thrust to power ratio of about 50 mN/kW and a fairly high
specific impulse. Its main advantage over the gridded ion thruster is the
high thrust density, which is possible with gridless acceleration. The thrust
output to power input ratio for the SPT is in the 50 to 70 mN/kW range. Its
drawback is a wide beam angle due to not having the focusing grids. This
may interfere with nearby spacecraft components. However, various other
alternatives for narrowing the exhaust plume are being developed.
23.3.5 Magneto-Hydrodynamic
The magneto-hydrodynamic (MHD) propulsion — also known as the
electro-hydrodynamic (EHD) — uses plasma. Neutral plasma is accelerated
under both the ohmic heating and the electrodynamic force generated by
the interaction of the electric field E and the magnetic field B. The thrust
that propels the spacecraft in MHD comes from the Lorentz force
F ¼ E B, where ¼ electric conductivity of the plasma. The force is
perpendicular to both the E and B fields. The accelerating force is imparted
to neutral particles in the gas by collision with ions, thus providing a direct
and efficient increase in the propellant velocity that propels the vehicle. The
electric conductivity of the plasma-mixed exhaust is an important
performance factor. It varies directly with the temperature and inversely
with the pressure. Its value of 0.10 m is considered adequate for
acceptable performance in the MHD accelerator. Since the self-induced
magnetic field accelerates ions, the thrust is proportional to the current
squared. The overall efficiency ranges from 10% in the kilowatt range to
40% in the megawatt range. A pulse mode of operation is essential in high
power ion thruster. Significant development in this area has been under
way in Japan. A thrust power of a few newtons with a specific impulse of
2000 s is achievable. Cathode erosion is a major life-limiting factor.
A constant E-field applied to plasma is not of interest here, as it adjusts
itself by developing a thin sheet of space charge, which shields the main
body of plasma from the E-field (as in a conductor body). On the other
hand, a constant B-field causes a Lorentz force on the charged particle in
plasma. The force is given by F ¼ Q (E þ V B), which is always
perpendicular to the velocity, V, of the particle. It causes the charged
particle to gyrate about the field lines without altering the internal space-
charge neutrality of the plasma. This force provides acceleration, keeping

the kinetic energy constant, as Lorentz force can do no work.
In the MHD propulsion, a trace of conducting fluid such as potassium or
cesium is added in a chemical rocket engine exhaust to ionize it. The
thermal plasma in the exhaust is propelled under a strong magnetic field
created by superconducting electromagnets. MHD accelerators have been
built in the USA and Russia, although not much work is currently under
way.
23.3.6 Pulsed Plasma Thruster

The pulsed plasma thruster (PPT) is a new system that has been
successfully tested for the first time on NASA’s Earth Observing Spacecraft
managed by Goddard Space Flight Center. Launched in November 2000, it
is a testbed for new instruments and technologies. In January 2002, the
reaction wheel that controls the craft’s pitch (up and down motion), was
turned off and the new system took over for a 4-h test. Millisecond pulses of
electricity fired across solid bars of Teflon turned portions of the Teflon into
an electrically charged gas that is forced out of the nozzle to generate thrust.
Tests showed that the system was about three times more efficient than
conventional chemical thrusters. This first test also eased concerns that the
system would produce electromagnetic interference that could affect other
systems on the spacecraft.
23.4 Performance Comparison

EHT and HAJ are both electrothermal hydrazine-based thrusters in which a
monopropellant is fed through a conventional thrust chamber and nozzle,
usually via a preliminary catalytic decomposition chamber. The gain in
exhaust velocity is derived from the electrical power added to the thrust
chamber. The propellant exhaust velocity in these two devices is relatively
low.
In EHT, the electric power is added using a resistance element, which is
the hottest component in the system. The maximum power that can be
added to the propellant is restricted by the melting point of the resistance
element. This, in turn, limits the rise in exhaust velocity to 2 to 3 km/s,
which is competitive with storable bipropellants giving 3 km/s. The EHT
has been used for NSSK on many satellites, and to raise small satellites from
the parking orbit to the final orbit.
In HAJ, the electric power is added directly to the propellant in the form
of a co-axial arc. The temperature in the arc region is several tens of
thousand kelvin, but most of the propellant near the walls is much cooler.
The HAJ has been used for NSSK on many large commercial satellites since
the early 1990s. It produces an average exhaust velocity of 5 to 6 km/s;
Table 23.2 Specific impulse of various

propulsion systems
Propulsion class Specific Impulse
Chemical (liquid and solid) 200–-400 s

Electrothermal hydrazine (EHT) 300–400 s
Hydrazine arcjet thruster (HAJ) 500–-1500 s
Pulsed plasma thruster (PPT) 1000–4000 s
Ion propulsion system (IPS) 2000–8000 s
about twice that of storable bipropellant, and the thrust of about 200 mN
using an input power just under 2 kW. Its thrust-to-power ratio of about 100
enables relatively fast maneuvers. However, its low exhaust velocity limits
the mass benefit relative to chemical propulsion. HAJ lends itself well to
spacecraft that already use hydrazine as a chemical propellant. HAJ used
for NSSK in many large commercial communication satellites gives an Isp of
about 500 s, compared to 200 s in traditional monopropellant thrusters, and
300 s in bi-propellant and EHT thrusters. This reduces the propellant mass
required for station keeping.
A gridless electrostatic ion thruster, such as the Russian SPT, has been
used on short duration missions with a specific impulse approaching 2000 s.
The gridded ion thruster can achieve a specific impulse in the 3000 to 4000 s
range.
The specific impulses of major types of propulsion systems are compared
in Table 23.2, and the power-to-thrust ratio versus exhaust velocity is
plotted in Figure 23.5. As for the voltage requirement, the HAJ generally
FIGURE 23.5 Power-to-thrust ratio versus exhaust velocity.

works at the bus voltage, the SPT thruster works around 300 V, and the IPS
works at voltages exceeding 1000 V.
The HAJ, SPT, and XIPS, each have their own inherent advantages
and disadvantages. The development continues in order to incorporate
these systems in full orbit raising. When achieved, it would significantly
change the Earth-orbiting satellite designs.
23.5 Solar PV Propulsion

Many new developments are under way in electric propulsion, which will
require high power to reach new frontiers. In particular, electric propulsion
is being studied for interplanetary missions since it gives mass efficient
architecture. However, multi-megawatt power levels are needed to obtain
reasonable transfer times. The power requirement drops to less than 1 MW
if cryogenic upper stage transportation technology is utilized with electric
propulsion. Solar electric propulsion becomes an attractive alternative to
the nuclear dynamic power system for a 1-MW system. It is applicable to a
wide range of NASA missions, such as human Mars exploration and robotic
exploration of the outer planets. For human Mars exploration, Kerslake and
Gefert1 have developed a 1-MW class conceptual design using various PV
technologies (Figure 23.6). The time required to travel to Mars, as estimated
by a NASA GRC computer code, by using a xenon Hall thruster, is about
230 days.2 The Kerslake study summarized in Table 23.3 shows that using
the present CuInS2 thin-film PV technology, it is feasible to design a 1-MW
power system with a mass under 10 metric tons and deployed solar array of
5800 m2. The EPS in this study consists of a 500-V d.c. distribution system
with 16 channels feeding eight 100-kWe thrusters. Since thrusters do not
operate during eclipse, the battery requirement is modest, about 13 kWh for
housekeeping and payloads. The 500 V d.c. permits direct drive thrusters
operation that greatly reduces mass of both the harness and the power-
processing unit, thus increasing the system efficiency. Copper wires with
FIGURE 23.6 Power system block diagram for 1-MW class electric propulsion.
Table 23.3 One-megawatt class solar electric propulsion system for Mars
exploration
Parameter ISS cr-Si Linear concen. a-SiGe thin film CuInS2 thin film
BOL power 1270 kW 950 kW 1150 kW 1050 kW

EPS Mass 46 MT 21 MT 12 MT 10 MT
PV array 14,000 m2 3500 m2 9000 m2 5800 m2
Cells/string 1450 270 340 880
Strings 990 910 9500 2370
Fluence MeV/cm2
for current 1.0 1016 2.2 1014 5.9 1014 5.9 1014
for voltage 2.6 1016 3.7 1014 1.3 1015 1.3 1014
Maximum leakage 260 A 1A 0 0
MT ¼ metric ton (1000 kg).

(Data source: T.W. Kerslake and L.P. Gefert, in Proc. 34th IECEC, p. 2449. SAE, 1999.)
Teflon insulation are used. NASA data on the space plasma effect at 500 V
d.c. operating voltage were used to estimate the parasitic leakage current.
The necessary 600-V silicon and silicon carbide based technology develop-
ments are well under way at NASA for switchgear components and remote
power controllers at this voltage.
The PV cell technologies considered by Kerslake and Gefert were:
The ISS PV array with an 8 8 cm crystalline silicon PV array design

Linear concentrator array based on SCARLET array design, using a 2
8 cm crystalline triple-junction GaInP2/GaAs/Ge cells operating at 7.5
sun
Thin film triple junction 5 5 cm amorphous aSiGe cells on folded thin
2-mil polymer membranes, and
5 5 cm CuInS2 thin film cells on thin 2-mil polymer membranes
In the conceptual design cited, the thin film cells are encapsulated with 1.5-
mil-thick FEP Teflon for isolation from the space plasma. The PV array
membranes are deployed using an inflatable rigidified longitudinal column
and a flexible composite lateral member. The array strings are negatively
grounded, and bypass diodes are incorporated every 10 cells to reduce
long-term degradation. The harness is multi-ribbon flat copper sized for 3%
voltage drop. The PV array surface is coated with transparent indium tin
oxide to prevent space charge buildup and arcing at high orbit altitudes.
In another study, Bailey, Hepp, and Raffaelle3 discuss the current state of
the art and address some of the technology development issues required to
make thin film PV viable for future solar–electric propulsion from Earth to
Moon or Mars. They have developed a conceptual design for a Mars solar
electric propulsion vehicle using thin film multi-junction PV array that fans
out (Figure 23.7). The study has concluded that the specific power around
FIGURE 23.7 The solar electric propulsion vehicle concept for human Mars mission.
(Source: NASA Glenn/Power and Propulsion/J. Hojnicki.)
1 kW/kg would be required, which is almost 40 to 50 times the present state

of the art.
23.6 Solar Thermal Propulsion

This propulsion concept uses concentrated solar heat. The concentrator is
generated geometrically from a right circular cone. The circular area
symmetric about the axis of rotation gives an on-axis paraboloid. A primary
off-axis concentrator is generated from a cone axis away from the axis of
rotation. In Figure 23.8, the solid line is the on-axis paraboloid and the
vertical line passes through point B, the bottom of the bowl. If a thruster
were close to point B near the paraboloid focus, the rocket engine would
FIGURE 23.8 Solar concentrator propulsion with paraboloid collector.

block the sunlight if it were in front or behind the direction of travel. Also,
the exhaust plume would impinge on the concentrator and degrade its
performance. Laug and Holmes4 have proposed a solution to this in using
an on-axis concentrator, which takes two elliptical sections of the
paraboloid away from the axis of symmetry. The required concentration
ratio for solar thermal propulsion is 10,000:1, which is much higher than
that for solar PV propulsion. This means a very small surface accuracy error
tolerance, about 1 mm rms, which is comparable with that for a 15-GHz
antenna.
23.7 Nuclear Power Propulsion

Humans cannot reach the outer limits of the solar system without nuclear
power. Most current spacecraft rely on chemical rockets for reaching an
orbit and on solar panels to generate power for on-board equipment. But
solar panels have a limited capability and their strength is weakened when
the spacecraft is far from the sun. For almost 50 years, NASA has been
doing planetary science in the same way. That is, to accelerate for 5 to 10
min and then stop and coast along. That is an ineffective way of exploring
deep space. We have reached a limit of what we can do with chemical
propulsion. Although a couple of probes have left our solar system, it takes
a very long time to get there. For making space travel faster, NASA is
developing nuclear-powered rockets that could halve the time it takes to
reach planets such as Pluto.
NASA has studied the use of nuclear-powered spacecraft in the past.
During the 1990s, nuclear energy was, politically, an unsound topic and
new developments on radioisotope generators and nuclear-fission reactors
were shelved. However, they have been revived and now receive funding.
The nuclear system initiative proposed by NASA in 2002 amounts to over a
billion dollars worth of new developments over 5 years.
Radioisotope propulsion relies on heat produced by the natural radio-
active decay of plutonium to generate electricity or to create a propulsive
thrust using nonnuclear xenon fuel. NASA has only one basic model of a
radioisotope power generator, but plans to use the new funding to develop
a more advanced system using nuclear reactor. The proposed system would
use a conventional chemical rocket launching a spacecraft equipped with
nuclear components. At a safe distance from Earth, the spacecraft would
detach from the rocket, the reactor would go online and the mission would
begin. Undoubtedly, the major obstacle is the issue of safety.
All space systems we build must safely survive the worse possible
scenario — a rocket blowing up on the pad. To aid safety, the nuclear
hardware on board the spacecraft would be launched in a cold non-
operating state. The reactor would be switched on when the vehicle reaches
2500 km from Earth, a distance chosen in accordance with NASA’s orbital
debris guidelines for preventing hazardous materials ending up in orbits

near the Earth.
This may make a nuclear power propelled craft possible in near future. It
is likely that the first beneficiary of the new system could be Mars Smart
Lander to be launched in 2009. With nuclear power, it may be able to operate
for years, rather than for months. Solar-powered sources last only for
months on Mars due to dust in the atmosphere that coats the solar panels.
23.8 Microwave Beam Propulsion

Electric power is traditionally generated on board the spacecraft. An
alternative approach studied by NASA is to generate a large amount of
power on another satellite and transmit it by a microwave beam,
eliminating the mass of the power generation system on the user spacecraft.
NASA’s interest in such a system is to reduce deep space mission time, such
as manned missions to Mars.
A study by Lindberry and Chapman5 focused on using an MHD
accelerator powered by a microwave beam for augmenting the specific
thrust of a chemical rocket engine from 450 s to 2500–4000 s with a
corresponding thrust of about 300 to 400 N and acceleration of 0.002
to 0.02 m/s2. No other all-electric or all-chemical propulsion can achieve
such performance alone, but it would require a strong magnetic field.
Superconducting magnet technology is well established and is commer-
cially used in numerous ground-based applications for generating strong
magnetic fields in the 10 to 20 T range. The old liquid helium-based
superconductors (<10 K) are being replaced by newer high-temperature
superconductors (around 100 K). Room temperature superconducting
technology is on the horizon, which is assumed to be available in the
NASA study of this concept. The major design challenge would be a
magnet coil with a current density of 1000 MA/m2, and corresponding
electrode erosion, and a mechanical force containment structure.
Thermal management of the equipment that handles such high currents
is also a challenge.
It is estimated that the MHD accelerator would have an efficiency in the
80 to 90% range, and only a couple of percent of the system mass. The
remaining 98% mass would be primarily in the receiving rectenna. A low
mass rectenna is therefore a critical technology for deploying this concept.
The frequency considered for such a system is from 3 GHz to 300 GHz
with the corresponding wavelengths from 0.10 m to 0.001 m. Brandhorst, in
a previously cited study, considered two power beaming options, micro-
wave and laser beams. Because of on-orbit aperture limitation, the
microwave option was dismissed. For the laser option, 10.6 and 1.06 mm
wavelengths were considered. The energy conversion system for a 10.6-mm
laser was either a thermal engine or a thermo–PV. For the 1.06-mm laser,
suitable energy conversion would be direct bandgap PV cells giving

excellent conversion efficiency around 50%.
Electrothermal propulsion using microwave power is another possible
concept that has been studied.6
The microwave power beam and superconducting energy storage are
further covered in Chapter 26 and Chapter 27, respectively.
23.9 Tether Power Propulsion

Space tether is being developed for a wide range of applications, such as
electrodynamic propulsion, momentum transfer, formation flying satellites,
remote sensing, and space structures. In 2004, NASA Marshall Space Center
plans to test a tether that will passively de-orbit a satellite back to Earth. The
proposed system will deploy 5 km of bare wire plus 10 km of special
insulated tether from a Delta II upper stage to achieve 0.4 N of drag thrust,
for de-orbiting. The experiment will use a bare tether for current collection
in lieu of the end-mass collector and insulated tether used on the TSS-1
mission.
The tether is a distinctly different propulsion system that is propellant-
free. It is electrodynamic in principle, and uses a long conducting wire
moving in the Earth’s magnetic field to generate electrical power for
propulsion. NASA has experimented with this concept a few times in space.
If future development proves to be successful, it can, one day, transfer
satellites from one orbit to another, and even power missions to the outer
planets.
In its working principle,7,8 electrons floating in space are made to flow in
a long conductor. This constitutes an electric current. A conductor carrying
current I moving in a magnetic field of density B generates mechanical force
per unit length equal to I B in the direction that is perpendicular to both B
and I. This force, known as the Lorentz force, works as a motoring force
which can raise or lower the payload orbit depending on the current
flowing towards or away from the Earth (Figure 23.9).
The tether can also be used for power generation. Experimental and
theoretical work using very large high-voltage structures in a simulated
LEO plasma has shown that substantial currents may couple between such
structures and the surrounding ionospheric plasma. A long tethered
satellite system would behave as a power generator if the induced V B
voltage in a long conducting tether moving in the Earth’s magnetic field is
allowed to drive a negative current through an on-board load, providing
electrical power at the expense of orbital energy. NASA’s tether power
generation concept utilizes such electrodynamic characteristic of a con-
ducting tether in the Earth’s orbit to generate useful electric power for a
number of applications. It has several advantages over conventional power
sources. It is 2.5 to 3 times as efficient as fuel cells in terms of fuel utilization,
FIGURE 23.9 Tether propulsion concept.

(From L. Johnson, IEEE Spectrum, p. 39, July 2000. With permission.)
and is not subject to day/night cycles as are solar arrays. The magnitude of
the power generated depends on the tether length. As an example, it has
been estimated that a current of 5 A through a 100-km tether can produce
upwards of 80 kW of power. Accounting for drag and other losses, the
system could operate at a net efficiency of 75%.
The concept has been around since the 1960s. In 1993, NASA’s Plasma
Motor Generator Mission used a tether equipped with a hollow cathode end
mass to collect electrons. In 1996, NASA flew another 20-km long, 2.5-mm-
diameter conducting tether between the space shuttle Orbiter and a 500-kg
satellite with scientific instruments provided by the Italian Space Agency.
This tethered satellite system (TSS) generated over 2 kW of power and 1 A
of current. While not a complete success, it provided valuable data for
future research and development on this unique concept of bleeding energy
from space charge.
The most critical requirement for the tether power system is the
availability of tether materials with the required strength, conductivity,
insulation, and protection from adverse effects of the space environment

such as radiation. Even with the availability of such materials, a number of
design issues need to be studied in the near term for gaining a clear
understanding of the technologies necessary for power generation. These
issues include advances in high-voltage technology and power processing
equipment. In addition, there are basic questions on the phenomenology
and design, which must be investigated. They are:
The plasma–electrodynamic interactions affecting return current losses

High-voltage electronic components and insulation materials
Hollow cathode current emission technology
D.c. tether impedance controller technology
Passive tether control techniques
The primary focus of the tether research is the plasma interactions for
power generation. It also considers plasma interactions related to the
system’s use as a source of thrust and as an extremely low-frequency (ELF)
antenna system. The concept of the tethered satellite system as a generator
or motor envisions the ejection of charge from one end of the system, and
the collection of a compensating current at the other end. The effect of the
magnetic field on charged particle collection is considered and is shown to
be substantial. Specific calculations of charged particle collection utilize
computer codes for geometrically complex objects and well-developed
analytical theories of spherical probes, both for the case of unmagnetized
plasmas. Without the aid of plasma contactors or large area collectors,
quiescent plasma theory indicates that the effect of the magnetic field would
limit the useable power developed in an initial electrodynamic tether
mission to about 1 kW. Since the plasma may be turbulent, however, this
estimate may be pessimistic. In any case, uncertainties in present knowl-
edge clearly point to a need for a better understanding of probes in
magnetized plasmas, including plasmas that are strongly perturbed by the
passage of electron beams.
The first electrodynamic tether experiment will involve electron beams.
The electromotive force (EMF) generated by motion across the geomagnetic
field will be used to operate an electron gun, which ejects a beam from the
shuttle into the surrounding plasma. Some fraction of the electrons
generated at the gun’s cathode may not escape the vehicle, instead leaking
current to the anode (shuttle ground) and causing a negative potential
buildup. This potential buildup, a sensitive function of the ratio of leakage
current to neutralizing current, is studied for the case where the
neutralization is affected by ram ions from the ambient plasma.
Estimates of the effect of leakage currents on shuttle potential ignore the
effect of interactions between the beam and ambient plasma. Strong
perturbations of the ambient plasma are frequently observed, but experi-
mental results do not yet permit firm statements on the magnitudes of

vehicle potentials that develop during beam ejection. One experiment,
conducted on the Space Shuttle/Spacelab-I showed that the vehicle charge
buildup is neutralized by injecting a plasma plume simultaneously with an
electron beam. This tended credibility to the effectiveness of the plume as a
plasma contactor, the term that rests on the expectation that a spacecraft
generated plasma acts as a conductor contacting the ambient plasma over
an effective area much greater than the geometrical area of the satellite. A
theory developed by NASA indicates not only that plasma contactors
substantially enhance charged particle collection, but also act as an efficient
source for emission of current into the space plasma. However, the
theoretical basis for these assertions, like the experimental one, is not
adequate for confident predictions. Both require further development.
The electrodynamic tether system is subject to time variations induced by
load current pulsation and by fluctuations in induced voltage and plasma
properties as the system orbits the Earth. The magnitudes of these
variations have been examined and their effect determined by considering
the system as an end-loaded and distributed RC line. Although the
variations are significant, the relevant time constants of the system are
short, permitting it to react in a quasi-steady manner. This latter conclusion
pertains also to the electron gun current pulsation that enables the system to
function as an ELF antenna.
While the results of previous research indicate that the tether is an
efficient generator of ELF waves compared to a tuned antenna in vacuum,
and that the waves couple only weakly into the Earth’s ionosphere cavity,
the tether may, nevertheless, prove useful as an electrical source.
Compared to the ohmic resistance of the tether, the radiation resistance is
small, not only for wave modes with frequencies in the ELF range, but also
for other wave modes excited by the motion of the tether. The implication is
that radiation resistance is not an important factor when computing time
dependent currents in the tether. While many questions relating to wave
generation and current closure through the ionosphere are not fully
answered, there is yet definitive evidence that these issues will profoundly
affect the electrical response of the system.
Other issues are the drag produced by deflection of particles by the
tether’s magnetic field and the electro-mechanical stability of the tether
against kink modes. In both cases the associated physical effects are found
to be insignificant.
With reference to Figure 23.10, Johnson8 describes the tether’s working as
follows:
1. The conducting tether attached to a space object moving in the Earth

magnetic field generates voltage across the wire given by e ¼ BLV.
2. Space electrons are attracted to the positively biased far end of the
tether.
Electric Propulsion
FIGURE 23.10 Tether working principle. (From L. Johnson, IEEE Spectrum, p. 40, July 2000. With permission.)
587
3. The electrons flowing downwards through the tether constitute a

current.
4. The current-carrying tether moving in the Earth’s magnetic field exerts
a drag force given by F ¼ BIL, causing de-orbiting of the space object.
5. A hollow cathode at the near end of the tether emits electrons back into
space, completing the electric circuit.
The force will be reversed if a reverse current is applied by a higher voltage

than that self-generated in the tether, such as from a solar array. In this case,
the orbit of the space object attached to the tether would rise.
The tether forms a unique type of electric circuit, which NASA
demonstrated in space with the Plasma Motor Generator in 1993 and the
Tethered Satellite System in 1992 and 1996. These missions deployed long
conducting tethers from orbiting spacecraft and generated kilowatts of
power. The tether is dragged through the ionospheric plasma through
which the spacecraft and the tether are traveling at a speed of 7 to 8 km/s.
In so doing, a few kilometers long aluminum wire extracts electrons from
the plasma at the end farthest from the payload and carries them to the near
end. A specially designed device, known as a hollow-cathode emitter,
expels the electrons to ensure their return to space. Currents in the plasma
complete the circuit.
Ordinarily, a uniform magnetic field acting on a current-carrying loop of
wire yields a net force of zero, since the force on one side of the loop is
canceled by that on the other side, in which the current is flowing in the
opposite direction. However, since the tethered system is not mechanically
attached to the plasma, the magnetic forces on the plasma currents in space
do not cancel the forces on the tether, and so the tether experiences a net
force.
As the tether cuts across the magnetic field, its bias voltage is positive at
the end farthest from the Earth and negative at the near end. This
polarization is due to the action of the Lorentz force on the electrons in
the tether. Thus, the natural upward current flow is due to the (negatively
charged) electrons in the ionosphere being attracted to the tether’s far end
and then returned to the plasma at the near end, aided by the hollow
cathode emitter (steps 3 to 5 above). The hollow cathode is vital. Without it,
the wire’s charge distribution would quickly reach equilibrium and no
current would flow. The Earth’s magnetic field exerts a drag force on the
current-carrying tether, decelerating it and the payload, and rapidly
lowering their orbit. Eventually, they re-enter the Earth’s atmosphere
(de-orbit).
A key advantage of the electrodynamic tether is that it does not need the
spacecraft’s on-board power. If the tether were being used to raise rather
than lower the spacecraft orbit, the current would have to be forced in the
other direction. An addition of a small PV power source would do it.
Some of the near term applications of tether propulsion systems are as

follows.
23.9.1 Space Debris Cleanup

Over the past few decades, the space around Earth has become cluttered
with debris, which could take years, and in some cases a century, to fall
from orbit. The danger is that old satellites, rocket stages, and trash thrown
overboard by early space travelers could collide with working satellites,
space shuttles, and space stations. The International Space Station, for
example, has to maneuver several times a year to avoid hitting debris,
burning up precious fuel each time. NASA and ESA have recommended
that national governments require future spacecraft to be self-removed from
the orbit at the end of their life spans. To do that, a satellite must be loaded
with extra propellant that could add as much as 25% of the satellite’s
weight. Using a tether to de-orbit would be inherently more mass efficient
and reliable. A tether is much lighter and more compact than conventional
thrusters, and can account for as little as 2% of a satellite’s total weight. It
has no complex valves, plumbing or circuitry that must stay operational
and leak-free for years. At the end of a satellite’s useful life, the tether
would unreel and tether-driven orbital decay would begin. A satellite’s
orbit, which would otherwise take 1 year to decay naturally, could be de-
orbited in a few days using a tether.
23.9.2 Orbit Transfer Vehicle

The electrodynamic tether can be attached to an unmanned space tugboat
that would carry satellites to higher orbit. After being launched into low
Earth orbit, the orbital transfer vehicle would grapple the satellite and
maneuver it to a new altitude or inclination. The tug could then lower its
own orbit to rendezvous with another payload and repeat the process.
Conceivably, several such orbital reassignments could be performed
without the need for propellant, making the tug relatively inexpensive to
operate. Because plasma density diminishes rapidly with distance from the
Earth, the tug could only operate below 2500 km. The lowest usable altitude
would be about 250 km, below which the atmosphere would begin to exert
too much aerodynamic drag.
23.9.3 Keeping the Space Station Afloat

The International Space Station is estimated to require 0.5 to 1.5 tons of
propulsion fuel replenished from the Earth for every year of operation. This
is to overcome the atmospheric drag, which would cause its altitude to drop
by several kilometers per week. ISS with an electrodynamic tether could
drastically cut its dependency on fuel. Atmospheric drag exerts a force of
up to 1 N on the station. NASA has designed a 10-km long, 0.6 mm 10 mm

aluminum ribbon tether capable of delivering 0.5 to 0.8 N of thrust with less
than 10 kW of solar array power needed to reverse the direction of the
current in the tether. More research is needed before such a tether can be
actually deployed on the space station. For example, if micrometeoroid or
orbital debris accidentally severs the tether, a mechanism is needed to keep
it from recoiling around the station and damaging key components.
23.9.4 Exploring the Outer Planets

Electrodynamic tether technology may be used to propel and power
spacecraft exploring the outer planets. Jupiter has an environment
particularly favorable to the electrodynamic tether. It has a strong magnetic
field, and its mass dictates high orbital velocities. Present Jupiter missions
rely on RTGs, but the power available is typically less than 300 W. In theory,
the tether could power the craft instruments and generate thrust at the same
time. For a circular orbit close to the planet, tether propulsive forces have
been calculated to be as high as 50 N and power levels as high as 1 MW.
This level of power can enable high-power instruments and radar. But it
also means resolving the high-power conversion, energy dissipation, and
tether overheating problems.
References
1. Kerslake, T.W. and Gefert, L.P., Solar power system analysis for electric
propulsion missions, in Proceedings of the 34th Intersociety Energy Conversion
2. Brandhorst, H.W., Power without wires, a solar electric propulsion
concept for space exploration, IEEE Aerospace and Electronic System,
Voltage. 16, No. 2, February 2001, pp. 3–7.
3. Bailey, S.G., Hepp, A.F., and Raffaelle, R.P. Thin film photovoltaic for
space applications, in Proceedings of the 36th Intersociety Energy Conversion
Engineering Conference, ASME, 2001, Paper No. AT-36.
4. Laug, K.K. and Holmes, M.R., Paraboloidal thin film inflatable concen-
trators and their use for power applications, in Proceedings of the 34th
2552.
5. Lineberry, J.T. and Chapman, J.N., MHD Augmentation of rocket engines
for space propulsion, in Proceedings of the 35th Intersociety Energy Conversion
Engineering Conference, AIAA, 2000, Paper No. 3056.
6. Power, J.L., Microwave electrothermal propulsion for space, IEEE
Transactions on Propulsion in Space, Vol. 40, No. 6, June 1992.
7. Cosmo, M.L. and Lorenzini, E.G., Tethers in Space Handbook, NASA
Handbook, December 1997.
8. Johnson, L., The tether solution, IEEE Spectrum, July 2000, pp 38–43.
Chapter 24
Fuel Cell Power
24.1 Introduction
The fuel cell was developed as an intermediate-term power source for space
applications. It was first used in a moon buggy. At present, it is routinely
used to power NASA’s fleet of space shuttles (STS Orbiter) that carries
components and crew to the International Space Station and other space
service missions.1 The fuel cell resembles a battery in that it converts the
chemical energy of a fuel directly into d.c. electricity. However, unlike a
battery, it does not run down in energy and have to be recharged. It keeps
producing electricity as long as the fuel is supplied. One can call the fuel cell
‘the gas battery.’ A typical fuel gas is hydrogen or a hydrogen-rich mixture
and an oxidant.
A fuel cell finds applications in space missions lasting for a few days to a
few weeks where a battery is not practical. It also has a potential use as an
auxiliary power source for orbit transfer vehicles. The regenerative fuel cell
integrated with an electrolyzer unit presents an attractive mass saving for
LEO satellites requiring large energy storage. It was a serious candidate in
place of the battery for the ISS.
The working of a fuel cell is the reverse of the electrolysis of water. In the
electrolysis, electricity is passed between two electrodes in water to produce
hydrogen and oxygen. In the fuel cell, hydrogen and oxygen are combined
to produce electricity and water. The energy conversion is direct from
chemical to electrical. Since the process is isothermal, the conversion
efficiency is not limited by the Carnot efficiency. This is unlike many
chemical-to-thermal-to-mechanical-to-electrical energy converters using
steam or internal combustion engines. It skips the usual combustion step
of conventional power systems and converts a high percentage of the fuel’s
available free chemical energy directly into electricity. The fuel cell
efficiency, therefore, can be about twice that of the thermodynamic
converter. It is as high as 65% in some designs, and 75 to 80% in solid
metal oxide fuel cells developed for ground base power plants. Its superior
reliability with no moving parts is an additional benefit over thermo-
dynamic power generators.
591
24.2 Electrochemistry of a Fuel Cell

A fuel cell consists of anode and cathode electrodes separated by a liquid or
a solid electrolyte. The electrodes are electrically connected through an
external load circuit as shown in Figure 24.1. Hydrogen or a hydrogen-rich
mixture is fed to the anode. The hydrogen fuel is combined with oxygen of
the oxidant entering from the cathode port. The hydrogen, however, does
not burn as in the internal combustion engine. It splits into hydrogen ions
(Hþ) and electrons (e), and produces electricity by an electrochemical
reaction. Water and heat are the byproducts of this reaction if the fuel is
pure hydrogen. With natural gas, ethanol or methanol as the source of
hydrogen, as in some ground based fuel cells; the byproducts include
carbon dioxide, and traces of carbon monoxide, hydrocarbons, and nitrogen
oxides, all of which are negligible.
The hydrogen–oxygen fuel cell consumes hydrogen as fuel, oxygen as
oxidant, and an aqueous acid solution as electrolyte. Their net reaction is
2H2 þ O2 ¼ 2H2O, and, of course, energy is released in the process. In one
type of fuel cell, the electrons flow from the anode and travel through the
external circuit to the cathode, powering the electrical load en route.
Hydrogen ions migrate through the electrolyte to the cathode, closing the
loop. At the cathode, they combine with the oxygen and the incoming
electrons from the external circuit to produce water. The kind of ions and
the direction in which they migrate vary, depending on the type of
electrolyte.
Major components and their functions in the fuel cell are:
The anode (fuel electrode) provides a common interface for the fuel and
electrolyte, catalyzes the fuel oxidation reaction, and conducts electrons
from the reaction site to the external circuit.
FIGURE 24.1 Fuel cell construction.

Fuel Cell Power 593
The cathode (oxygen electrode) provides a common interface for the

oxygen and the electrolyte, catalyzes the oxygen reduction reaction, and
conducts electrons from the external circuit.
The electrolyte transports one of the ionic species involved in the fuel
and oxygen electrode reactions, while preventing the conduction of
electrons to avoid a short circuit.
Other components may also be necessary to seal the cell, to provide for
gas compartments, and separate one cell from the next in a fuel cell
stack.
The fuel cell is thus a static electrochemical device that generates electricity
by using a chemical reaction without altering the electrodes or the
electrolyte materials. This distinguishes the fuel cell from the electroche-
mical battery. Unlike the conventional battery, the fuel cell has no electrical
energy storage capacity. Hence, it must continuously supply the reactant
and withdraw the reaction products during operation.
24.3 Electrical Performance

The fuel cell basically works as a voltage source with an internal resistance.
The electrical potential appears at the terminals of two electrodes involved
in the process. The theoretical value of the fuel cell potential is 1.25 V, which
matches that of the NiCd and NiH2 batteries. Multiple fuel cells stack up in
series–parallel combinations using heavy graphite pallets for the required
voltage and current, just as the electrochemical cells do in the battery.
However, as soon as the current is drawn, the voltage drops significantly
due to various losses. Because the primary loss mechanism is ohmic loss in
the electrodes, the voltage continues to drop with increasing current. The
voltage drop is given by
Vdrop ¼ þ loge J ð24:1Þ
where J is current density at the electrode surface, and and are constants
which depend on temperature and the electrode surface. The theoretical
potential difference of 1.25 V between the anode and cathode in the
hydrogen–oxygen fuel cell is determined by the difference of the free
energy of the reaction product and the fuel and oxidant. This potential is
different in different fuel cells depending on the reactions involved. The
electrical performance of a fuel cell is represented by the electrode voltage
versus surface current density, commonly known as the polarization curve
or the V–I curve as shown in Figure 24.2. Ideally, a single H2–O2 fuel cell
could produce 1.25 V d.c. under ambient conditions. Undesirable ions
and products of the intermediate irreversible reactions decrease the cell
FIGURE 24.2 Fuel cell polarization curve — voltage versus current.
potential, even at open circuit. Further voltage drop under load results from
various irreversible polarizations in the cell, such as:
Resistance polarization, i.e., the electrical resistance of the electrolyte

and electrodes
Concentration polarization, i.e., the accumulation of ions and reaction
products and the depletion of consumed ions and reactants in the
electrolyte near the electrode surfaces
Activation polarization, i.e., the reluctance of the fuel and oxidant to
undergo reaction at each electrode
Activation polarization causing energy losses associated with reactions
All ohmic losses within the cell, i.e., electrodes, current collectors,
contacts, and ionic impedance of the electrolyte
Concentration polarization causing the energy losses associated with
mass transport
The net result of these polarizations is that the practical fuel cell produces
between 0.5 and 1.0 V d.c. at currents of 100 to 400 mA/cm2 of cell area. Fuel
cell performances can increase by increasing cell temperature and reactant
partial pressure. A trade-off exists between achieving higher performance
by operating at higher temperature or pressure and confronting the
materials and hardware problems imposed at more severe conditions.
The practical operating range of the fuel cell is controlled by ohmic loss.
The V–I characteristic in this region is very similar to that of the battery,
except that the average discharge voltage is lower. The voltage drops
Fuel Cell Power 595
approximately linearly with increasing current and also with time, a shown
in Figure 24.3.2 At any given time, the terminal V–I relationship can be
expressed as
V ¼ Vo kI ð24:2Þ
where Vo ¼ open circuit voltage and k ¼ constant. The value of k increases

and Vo decreases with time. The power at any operating point is given by
P ¼ VI ¼ ðVo kIÞðVo VÞ=k ð24:3Þ
The maximum power is when dP/dI ¼ 0, which occurs at V ¼ 1=2 Vo,

leading to
Pmax ¼ 1=4Vo2 =k ð24:4Þ
Unlike the PV cell, the fuel cell does not work in the use the input energy or
lose it mode. It uses the on-board fuel to generate power. For this reason, the
fuel cell is not operated at Pmax until it approaches the end of life. It is rather
operated at the maximum fuel efficiency until the EOL.
Since Vo degrades with time, so does the Pmax. The open circuit voltage
can be expressed as a function of time as
Vo ðtÞ ¼ Vo ð0Þ Ko h ð24:5Þ
where h ¼ number of hours since the fuel cell is placed in operation. With a
voltage regulating converter between the fuel cell and the load, the life of
the fuel cell can be defined as the time it takes for the voltage to decay below
the required input voltage, or the Pmax to fall below the required output
FIGURE 24.3 Fuel cell operating characteristic degradation with time.

FIGURE 24.4 Pmax =Prated ratio versus time determines the life.
power. It can be predicted from the Vo versus time relation. The expected
life of the fuel cell is determined as shown in Figure 24.4.2
The transient electrical performance model of the fuel cell include
electrochemical, thermal and mass flow elements that affects the electrical
output.3 Of primary interest is the electrical response of the cell to a load
change. To design for the worst case, the performance is calculated under
both the constant reactant flow and the constant inlet temperature.
24.4 Types of Fuel Cells

A fuel cell is classified by the types of fuel and electrode materials used in
the construction. The space-qualified fuel cell in the future is likely to take
advantage of recent advances in commercial fuel cells using extensive
ground experience on advanced technologies. Various types available at
present are described below.
Solid polymer electrolyte (SPE) was used in the early manned missions
lasting less than 1 week. Since then, two parallel fuel cell technologies have
been developed: proton exchange membrane (PEM) for Gemini and alkaline
for Apollo missions to the moon. An improved aqueous alkaline technology
was also selected for NASA’s space shuttle fleet. A lightweight alkaline fuel
cell stack has been recently developed by United Technologies for NASA. It
is designed to use propellant grade H2 and O2 to generate a peak power of
3.5 kW for an average load of 2.2 kW. Specific powers of these units were
approximately 110 W/kg for the standard space shuttle fuel cell and
approximately 150 W/kg for the lightweight stack.
A high-power shuttle fuel cell stack operating on dedicated cryogenic
reactants H2 and O2 is designed for a peak power of 12 kW with brief
overloads of 15 kW and an average load limited to 4.5 kW for the life
consideration. The average load has now been increased to 7 kW. The
space shuttle does not have batteries. It obtains starting power from the
Fuel Cell Power 597
ground before launch, and then uses fuel cells after the launch. Each
shuttle has three fuel cells generating 30 to 36 V d.c. power. Some of
this power is used in d.c. form, and some is converted in 400 Hz a.c. for
aircraft-type equipment used in the shuttle. It has 118 kg of dry mass.
The output power capability of 12 to 15 kW gives a pecific power
around 100 to 120 W/kg based on the stack and accessories without
power conditioning and gas supplies. Although the electrical startup
time is 10 to 20 min and the shutdown is instantaneous, the whole
system requires about 2 h of warm-up time before it can be loaded. The
design life for the space shuttle power plant is 5000 h at rated power. It
is likely that the life of the shuttle power plant could be increased to
20,000 h (along with higher power capability) with the incorporation of
advanced technology items.
The alkaline fuel cell used in the space shuttle combines hydrogen
and oxygen to generate electricity, with water as a byproduct. Oxygen
enters the cell through the sintered-nickel cathode where a catalyst
produces OH ions and delivers them to the cell’s potassium hydroxide
electrolyte. The ions drift through the alkaline electrolyte to the anode,
where they combine with hydrogen atoms to form water molecules and
release electrons, and the electrical power is delivered to the external
load. The energy released by 1 kg of cryogenic hydrogen combines with
8 kg of oxygen is about 34,000 Wh. That is 3800 Wh/kg of fuel,
excluding the fuel tank and the supporting systems. This amounts to
the energy conversion efficiency of about 50%. The byproduct water is
used for the crew.
The PEM technology is found more suited for spacecraft applications at
present. NASA Glenn Research Center continues developing the PEM and
regenerative fuel cells for a wide variety of applications, including launch
vehicles, planetary and interplanetary spacecraft, planetary surface power,
aircraft propulsion, and terrestrial use. For spacecraft applications, the
PEM technology offers enhanced safety, longer life, lower weight, higher
reliability, higher peak power capability, compatibility with propulsion
fuel, and potential for lower cost. A prototype 5.25 kW modular PEM fuel
cell was tested in 2000.4 The goal of the current PEM programs at NASA5,6
are three times the power and four times the life compared with the alkaline
fuels cell of the same mass and volume; with the peak to average nominal
power ratio of six. Further developments will be needed to qualify for
manned flights.
The performance of various types of fuel cell for space applications is
compared in Table 24.1. Compared to the PV array, the fuel cell gives
much higher power per kilogram. Flexibility is another major advantage,
as it does not need sun pointing and provides the same power during
both day and night. Its disadvantage is that it needs to carry fuel on
board.
Table 24.1 Various fuel cell performance compared
Fuel cell technology Specific power (W/kg) Life, in hours
Alkaline 100–150 50,000

Solid polymer 100–150 50,000
Alkaline (space shuttle) 300–400 3000–5000
Lightweight cell under development 600–700 TBD
24.5 Regenerative Fuel Cells

The fuel cells presented above convert energy one way: from fuel to
electricity. They are not designed for the reverse recharging operation.
Recharging the fuel cell requires an electrolyzer to decompose the water
back into hydrogen and oxygen. The electrolyzer is generally a separate
unit from the fuel stack and the two cannot operate simultaneously. The
hydrogen–oxygen regenerative fuel cell (RFC) has been developed for
energy storage in short-term space missions, but not used in space as yet.
The RFC was considered for the ISS but dropped in favor of the NiH2
battery due to its poor round trip energy efficiency. However, a high-power
long-term manned mission to the moon or Mars may find the RFC practical.
It would use solar energy to produce power and independent electrolysis
units for hydrogen and oxygen production. This is advantageous not only
for the energy storage, but also for life support and for auxiliary thrusters as
well.
The RFC is based on a hydrogen–oxygen proton exchange membrane fuel
cell and electrolyzer technology. The fuel stacks consists of multiple cells,
each with certain active area operated at 60 psi, 80 C and nominal 50 A/m2
current density. Cells are series–parallel connected and include bypass
diodes. The electrolyzer stack consists of multiple cells with certain active
areas, operated at 315 psi and a nominal current density of 20 A/m2. The
hydrogen and oxygen reactants are stored in gaseous form at 3000 psi,
while water is stored at 14.7 psi. Hydrogen is contained in Kevlar tanks
with 10-mil titanium liners to minimize gas diffusion. Similar tanks are
used to contain oxygen and water. The fuel cell operating temperature is
maintained by a radiator using water as the working fluid. The electrical
energy efficiency for the electrolyzer is 90% and that of the fuel cell is 60%,
thus the overall round trip efficiency of the systems is about 54%. The life is
estimated to be 10,000 h.
The RFC has a good peak power capability, about 10 times the base
power. This may be useful for providing power pulses. The primary
disadvantage of RFC is the peripheral pumps and plumbing that can
handle fluids reliably over a long mission life in GEO. Providing
redundancy in such a system may incur significant mass penalty. Other
than the reliability issue, the fuel cell is a well-developed technology. It may
Fuel Cell Power 599
be possible to service the RFC system for a LEO mission of 5 years. In LEO,
it has a reasonable specific energy around 5 to 8 Wh/kg and round trip
energy efficiency of 60 to 70%. Because of the lower efficiency compared to
the battery, both the solar array charging power and the thermal cooling
requirement would be greater. In high power GEO applications, mass-
optimized RFC units may provide 20 to 35 Wh/kg. The high specific energy
is largely due to the need for only a small electrolyzer unit because of the
relatively long recharge time available in GEO. The specific energy,
however, is sensitive to various thermal and electrical design requirements.
Proton Energy Systems7 is currently developing 1 to 100 kW RFC for LEO
and GEO applications. The company has tested 25 to 35% d.c.–d.c. round
trip efficiency. The efficiency is low, but the specific energy and specific
power are high compared to the battery. At stack level, 300 to 400 Wh/kg
and 70 to 100 W/kg are possible, including 10% for hydrogen storage. This
compares with about 200 Wh/kg with the Li-ion battery.
24.6 RFCs for Space Colonies

When human exploration of space grows into colonies on the moon and
Mars, power systems that last for decades will be required. Small nuclear-
based power systems can provide long-term sustainable power for
interplanetary missions and for human habitation in space colonies.
However, the systems issues related to the use of large nuclear power
plants needed to provide large amounts of power for life support, are yet to
be resolved. These issues include the system integration of all functions of
the habitat, as well as safety concerns. The regenerative fuel cell provides an
alternative with significant advantage. It coincidentally furnishes a backup
supply of oxygen and water valuable to the inhabitants in the event of
emergencies and unforeseen circumstances. A simple architecture of such a
power system is shown in Figure 24.5. It includes the following:
PV array for power generation during the cyclic day time periods of
approximately 14 Earth days
Electrolyzer that uses d.c. electricity from the array and electro-
chemically converts water into hydrogen, oxygen, and low-grade heat
Fuel cell stack that electrochemically converts hydrogen and oxygen
into d.c. electricity, water, and low-grade heat during the cyclic night
period of approximately 14 Earth days
Storage tanks for pure water
High-pressure hydrogen and oxygen tanks
Automated controls
FIGURE 24.5 Power system architecture for human colony in space (Moon or Mars.)
A preliminary study of a 25-kW power system for a lunar habitat by

Voecks8 compared the KOH alkaline technology with the PEM technology
for use in both electrolysis and fuel cell units, each with 65% efficiency. The
study identified significant advantages associated with the PEM technology
for the following reasons:
PEM fuel cell and electrolyzer requires less pressure control complex-
ity.
Containment of the aqueous KOH electrolyte is much more difficult
than polymeric membrane.
The environment is very demanding, while some aspects are simpler for
operating an RFC on the moon compared to on the Earth. The moon
operation requires storage of hydrogen, oxygen, and water, whereas the
terrestrial operation does not need to store oxygen, since air can be fed
through the cathode on a one-pass basis. It is easy to maintain the PV power
generation schedules on the moon, as the sun time and dark are nearly
constant in the lunar month (also day, which is 29 days, 12 h, 24 min and
2.9 s of the Earth.) On the Earth, the seasonal variations and the clouds can
make the sunlight vary over a wide range.
24.7 RFCs for Satellites

The RFC may perhaps replace the battery in satellites, although not in very
near future, because even the efficiency-optimized RFC design has a much
lower round trip energy efficiency than the battery (Table 24.2). Moreover,
the radiators and heat pipes required in the RFC system are significant in
terms of both the mass and integration into the spacecraft design. The
protection against single-cell failure using a bypass diode is not practical in
case of the fuel cell. For a space-based radar power system as a hypothetical
example, the mass and volume of the fuel cell energy storage alone — not
counting the systems integration related mass — are roughly estimated and
Fuel Cell Power 601
Table 24.2 Round trip energy efficiency of RFC and

battery
Energy technology Round trip energy efficiency
Nickel cadmium battery 70–80

Nickel hydrogen battery 75–85
Hydrogen–oxygen RFC:
Mass optimized design 55–60
Efficiency optimized design 65–70
compared with that using two batteries in Figure 24.6 and Figure 24.7,
respectively. This study assumed a mission life of 5 years, orbit inclination
90 at various altitudes, power 20 kW, discharge time equal to charge time
giving the duty ratio of 50%, and the battery DOD 25% for NiCd and 35%
for NiH2. It is seen in the figures that the RFC power system results in low
mass in low orbits, comparable mass in mid orbits, and a very large volume
in both orbits. In GEO, the RFC would certainly result in much greater mass
in almost all cases.
However, when the designs are further developed, the RFC may find
applications in high-power spacecraft. For a 10-MW pulse power directed
energy weapon platforms in LEO as an example mission, Allen9 reports a
conceptual design study comparing the alternative systems, such as the
RFC, heat engines, flywheel, and batteries that could be recharged using PV
array. The study concluded that the solid-oxide RFC with PV charge power
could deliver the required power with the lowest mass. The details of the
Allen study are given in Chapter 22.
FIGURE 24.6 Alternative energy storage mass for 20-kW space-based radar.
FIGURE 24.7 An alternative energy storage volume for 20-kW space-based radar.
24.8 Commercial Fuel Cells

In terrestrial applications, the fuel cell has the potential of replacing the
combustion engines in automobiles, downtown buses, and hospital stand-
by power generators. Large cells for off-grid power market include some
military applications. The U.S. energy policy has placed significant
emphasis on the hydrogen-based economy. Some of these terrestrial fuel
cells research may be useful for future space applications.
The low temperature (250 C) fuel cell is now commercially available from
several sources. It uses phosphoric acid as the electrolytic solution between
the electrode plates. The high temperature fuel cell has higher energy
conversion capability per kilogram of weight, but cost relatively more. Solid
oxide, solid polymer, molten carbonate, and proton membrane exchange
fuel cells in this category are being developed. The Fuel Cell
Commercialization Group in the USA has recently field-tested a molten
carbonate direct-fuel cell for a 2 MW utility-scale power plant. The German
car manufacture Daimler-Benz and Ballard Power Systems of Canada are
developing solid polymer fuel cells for automobiles as an alternative to
battery-powered vehicles. Their target is to sell the first commercial fuel cell
powered car before the year 2005. Major automakers around the world have
accelerated research funding to be ready for consumers by 2010.
Howaldtswerke-Deutsche Werft AG of Germany has reported having
built four 212A-class submarines that are fuel cell powered. The design
is air-independent, and produces no noise or exhaust heat. It carries a
crew of 27 with displacement of about 1450 tons. It is scheduled of
commissioning in the German Navy in 2004. Large alkaline fuel cells have
been built for deep submergence rescue vehicles (DSRVs) with 20 to 35 kW
power levels.
Fuel Cell Power 603
Solid-oxide fuel cells of several different designs, consisting of

essentially similar materials for the electrolyte, the electrodes, and the
interconnections, are being investigated worldwide. Most success to date
has been achieved with the tubular geometry being developed by the
Westinghouse-Siemens Corporation in the USA and Mitsubishi Heavy
Industries in Japan. The cell element in this geometry consists of two porous
electrodes separated by a dense oxygen ion-conducting electrolyte. It uses
ceramic tube operating at 1000 C. The fuel cell is an assembly of numerous
such tubes. It is developed for multi-megawatt combined cycle gas turbine
and fuel cell plants up to 60 MW capacity. The cell voltage degrades less
then 0.1% per 1000 h of operation, and has life in tens of thousand hours of
operation. The prototype has been tested for over 1000 thermal cycles with
zero performance degradation, and 12,000 h of operation with less then 1%
performance degradation.
References
1. Oman, H., Fuel cells power for aerospace vehicles, IEEE Aerospace and
Electronics System Magazine, Vol. 17, No. 2, February 2002, pp. 35–41.
2. Babasaki, T., Take, T., and Yamashita, T., Diagnosis of fuel cell deteriora-
tion using fuel cell current–voltage characteristics, in Proceedings of the 34th
01-2575.
3. Hall, D.J. and Colclaser, R.G., Transient modeling and simulation of
tubular solid oxide fuel cell, IEEE Power Engineering Review, Paper No. PE-
100-EC-004, July 1998.
4. Perez-Davis, M.E. et al. Energy storage for aerospace applications, in
ASME, 2001, Vol. I, pp. 85–89.
5. Hoberecht, M. and Reaves, W., PEM fuel cell status and remaining
challenges for manned space flight applications, in Proceedings of the 1st
International Energy Conversion Engineering Conference, AIAA, 2003, Paper
No. 5963.
6. Burke, K., Fuel cells for space science applications, in Proceedings of the 1st
No. 5938.
7. Barbir, F., Dalton, L., and Molter, T., Regenerative fuel cells for energy
storage – Efficiency and weight trade off, in Proceedings of the 1st
No. 5937.
8. Voecks, G.E. et al. Operation of the 25 kW NASA Lewis Research Center Solar
Regenerative Fuel Cell Testbed Facility, NASA Technical Report No. 97295,
1997.
9. Allen, D.M., Multi-megawatts specifications power technology compar-
ison, in Proceedings of the 36th Intersociety Energy Conversion Engineering
Conference, ASME, 2001, Vol. I, pp. 243–249.
Chapter 25
Flywheel Energy Storage
25.1 Introduction
A flywheel stores kinetic energy in rotating inertia. The stored energy can be
converted into electricity using an electromagnetic machine in the generat-
ing mode. Recharging — spinning the wheel — is done using the same
machine in the motoring mode. The round trip energy efficiency of the
flywheel system is high — 85 to 90% compared to 70 to 75% with the
electrochemical battery. Flywheel energy storage is an old concept that has
now been developed for a variety of commercial applications. Automobiles,
utility load leveling, and uninterrupted power supplies (UPS) are a few
examples. At present, commercial UPS systems using flywheel–battery cost
$300 to $500 per kWh. The space-qualified NiH2 cells cost $80,000 to $100,000
per kWh. The flywheel system is being developed by NASA and industry
partners for replacing the battery in the spacecraft power system.1–4 One is
scheduled to fly for flight tests on the International Space Station (ISS) in
2006. It is estimated that flywheel energy storage can improve the depth of
discharge (DOD) and life by a factor of 2 compared to those of the battery.
An integrated energy and momentum storage flywheel system can be even
more advantageous.
The following developments have made flywheel technology more
attractive now than ever before:
Advances made in high-strength lightweight fibers. For example, fibers

with ultimate tensile strength of one million pounds per square inch are
commercially available now.
Improvements in the fiber–resin composite rotor design and manufac-
turing process5 and in understanding their performance characteristics.
Availability of high-speed magnetic bearings,6–8 eliminating the
mechanical contact between the stator and the rotor, thus making the
friction zero and the self-decay rate negligible.
The above technological advances when combined make it feasible to

design a flywheel to operate at up to 100,000 revolutions per minute,
yielding much higher specific energy compared to that with electrochemical
batteries.9
604
Flywheel Energy Storage 605
Battery performance is sensitive to many factors having highly nonlinear

relationships. Therefore, the design, analyses, and manufacturing, are time
consuming. Moreover, the battery depth of discharge is limited to a value
far below full capacity. The DOD is 30 to 35% in LEO satellites, where the
battery is often the heaviest components of the power system. The flywheel
can replace the battery with the following benefits:
Much higher specific energy, hence lower mass and volume

Deeper depth of discharge, up to 90% even in low Earth orbits, resulting
in lower Wh rating and mass
Lower hardware manufacturing and launch costs
Insensitive to temperature, hence lower thermal system mass and cost
Easy charge/discharge control as the stored energy depends on the
wheel speed alone
The charge/discharge ratio has no impact on life
No capacity degradation over life
No voltage sag over life
No trickle charging required
Shorter engineering design and analysis time
Longer cycle life under charge/discharge cycles
Flexibility in designing for a given voltage and current
Improved quality of power as the electrical machine design can be
made stiffer than the battery
Matches well with peak power tracking solar array
High peak to average power ratio (LEO communications satellites need
more power while traveling over highly populated areas)
These benefits have the potential of making the flywheel the least-cost
energy storage alternative per Wh delivered over the operating life on
ground or in space. However, it has the following negative aspects:
Higher output impedance because of the electrical machine (about the

same as PRU converters in a fully regulated bus)
Lower reliability, as a single failure would cause loss of significant
power capability
Most important issue is safety at speeds > 50,000 rpm
Vibration and transient torque tolerance on the spacecraft platform for
attitude control
No launch and ascent power, as the flywheel cannot be spinning during

launch
Gyroscopic torque tolerance on attitude control
25.2 Photovoltaic–Flywheel Power System

A functional block diagram of the spacecraft power system using a solar
array for power generation and a flywheel for energy storage is shown in
Figure 25.1. The configuration represents a direct energy transfer fully
regulated bus, which is used in most high-power satellites, except that the
flywheel now replaces the battery. Attached on the flywheel shaft is an
electromagnetic machine, which works in either mode, as a generator or as
a motor. The machine converts mechanical energy into electrical energy and
vice versa. Among several types of motor–generator designs available, the
brushless d.c. machine is the most suitable for use with the flywheel. In
operation, the PV array power during sunlight supplies the load and
charges (spins up) the flywheel using the electrical machine in the motoring
mode. During the eclipse, the flywheel is despun and the energy is fed to
the load via the electrical machine in the generating mode. The bus-sense
electronics senses the bus voltage, compares it with the reference voltage,
and feeds the amplified error signal to the mode controller. The mode
controller activates the buck or boost converter in the power regulator unit
(PRU). It also controls the charge (motor) mode, the discharge (generator)
mode and the shunt mode as required during the eclipse and sunlight
periods.
The machine voltage, being a linear function of the flywheel speed varies
over a wide range. The charge and discharge converters in the PRU provide
interface between a widely varying machine voltage and the closely
regulated bus voltage. It is possible to design PRU converters with input
FIGURE 25.1 Spacecraft power system architecture with flywheel energy storage.
FIGURE 25.2 Fatigue life of graphite fiber–epoxy composite compared with stainless steel.
voltages varying over the 1 to 3 range. This allows the machine speed to
vary over the same range. That is, a low rotor speed can be one third of the
full speed. Since the energy storage is proportional to the speed squared,
the flywheel state of charge at low speed can be as low as 10%. This means
90% of the flywheel energy can be discharged with no design difficulty in
the power electronics, or in any other components of the system.
The fatigue life of the composite rotor may limit the number of charge/
discharge cycles the flywheel can withstand. Experience indicates that
polymer fiber composites generally have longer fatigue life than solid
metals (Figure 25.2). A properly designed flywheel, therefore, can last much
longer than the battery and can discharge to a much deeper level. Flywheels
made of composite rotors have been fabricated and tested to demonstrate
more than 10,000 cycles of full charge and discharge.10
25.3 Flywheel System Components

The major components of the flywheel assembly are:
Rotor rim, where the energy is stored

Hub interfacing the rim with the shaft
Magnetic bearings for on-orbit operation
Mechanical bearings for emergency, launch, and high slew rates
Electrical machine (d.c. brushless motor–generator)
Sensors embedded in the wheel to sense physical variables, such as
position, vibration, temperature, and the overall health of the wheel
during all phase of the mission operations
Each component is briefly described below.
25.3.1 Rotor Rim

The rotor rim is the primary component of the flywheel that stores the
required energy and momentum. It is made of a high-strength fiber–resin
composite matrix, with fibers principally oriented in the hoop direction to
withstand high hoop stress under the rotational centrifugal force. The hoop
stress is a maximum at the outermost tip of the wheel. For economical two-
rim designs (Figure 25.3), the outer rim is made of high-strength fiber
composite and the inner rim is made of low-cost fiber composite.
25.3.2 Hub
The hub supports the rotor and allows it to adjust radially under
mechanical and/or thermo-elastic stress. It also is made of fiber–resin
composites, with fibers oriented at suitable angle(s) to support the multi-
dimensional stresses.
25.3.3 Magnetic Bearings

Conventional mechanical bearings are used up to speeds below 50,000 rpm.
At high speed, they have high friction and vibration. Speeds approaching
100,000 rpm can be possible only by using magnetic bearings, in which
the mechanical contact is eliminated, thus eliminating the friction. The
aerodynamic windage is eliminated by running the rotor in vacuum. Low-
flux pulsation as the rotor rotates around the discrete actuator coils produce
small electromagnetic loss in the metallic parts. This loss, however, is
negligible compared to the loss due to friction in mechanical bearings.
FIGURE 25.3 Two-rim flywheel rotor design.

The magnetic bearing raises and controls the shaft position using passive
magnetic and/or active electrodynamic forces. The passive design uses
permanent magnets of like polarities to keep them separate under the
repulsive force between them. The active magnetic bearings come in a
variety of configurations using permanent magnets and dynamic current
actuators to achieve the required restraints. A rigid body can have six
degrees of freedom. The bearings retain the rotor in five degrees of freedom,
leaving one for rotation. Among various alternatives, one configuration is
depicted in Figure 25.4. The dynamic actuator coils provide the horizontal
restraint. The currents in the coils are controlled in response to the feedback
loop controlling the rotor position. The permanent magnets are used to
provide free levitation support of the shaft. The permanent magnet bias also
helps stabilize the shaft under a rotor drop. The electromagnet coils provide
stabilization and control. The control coils operate at low duty cycle, and
only one servo controller loop is needed for each axis. The servo control
coils provide active control to maintain the shaft stability by providing
restoring forces as needed to maintain the shaft in the centered position.
The position and velocity sensors are used in the active feedback loop. The
electric current variation in the actuator coils compels the shaft to remain
centered in the air gaps.
FIGURE 25.4 Magnetic bearing in one configuration (U.S. Patent 4000929.)

25.3.4 Touchdown Mechanical Bearings

These bearings support the shaft before the magnetic bearings are activated,
or during high-g maneuvers and shocks, or in case of a magnetic bearings
failure. These bearings can be ball, roller, or any combination. They last a
long time for speeds up to 30,000 rpm (Figure 25.5), but can support
occasional use at higher speeds in case of magnetic bearings failure.
25.3.5 Electrical Machine

The electrical machine works as a generator or motor and converts
mechanical energy into electrical energy and vice versa. Among several
types of moto–-generator design available, the brushless d.c. type appears
to be the most suitable with the flywheel system for space applications.
25.3.6 Sensors
Various sensors, some embedded in the wheel, sense physical variables,
such as position, vibration, temperature, etc. They monitor the health of the
wheel during all phases of operation. They also detect failures to command
shutdown of the system if necessary.
Various sensors and bearing locations in one configuration of the
flywheel assembly are shown in Figure 25.6. In another configuration, the
rotor can be located radially outward, as shown in Figure 25.7, which forms
a volume-efficient packaging. Moreover, it minimizes the thermo-elastic or
rotational stress at the rotor–hub interface by providing radial resilience.
Figure 25.8 depicts a NASA concept for the flywheel system packaging. A
pre-stressed rotor, a strain-matched rotor, and mass-loaded designs are
other configurations that can minimize stresses.11
FIGURE 25.5 Radial load and speed limitations of ball and roller bearings.
FIGURE 25.6 Flywheel sensor and bearing locations.
FIGURE 25.7 Flywheel components and assembly packaging with stress-relieving hub.
FIGURE 25.8 NASA concept of flywheel packaging.

(From S. Stankiewicz, IEEE Spectrum, p. 49, April 2002. With permission.)
25.4 Power and Momentum Management Electronics

Although not a part of the wheel assembly, this is an important part of the
flywheel system. It is an independent box, separate from the wheel. It
senses the bus voltage and bucks or boosts the bus voltage to match with
the motor–generator voltage depending on the wheel speed at a given time.
It provides a total interface between the motor–generator and the regulated
voltage bus. It also contains the charge and discharge converters, the power
management function, the momentum management function and the
associated telemetry and commands. It controls the charge (motor) mode,
the discharge (generator) mode, and the shunt mode as required during
an eclipse and sunlight periods. Additionally, it controls the torque as
commanded by the attitude control system.
25.5 Energy and Stress Relations

The kinetic energy stored in a flywheel having the mass moment of inertia J
rotating at an angular speed is given by
E ¼ 1=2J!2 ð25:1Þ
The centrifugal body force is resisted by the hoop stress. The outermost
fibers on the wheel are stressed the most. This stress is kept below the
permissible limit. Stress analysis would show that the maximum hoop
stress is directly proportional to the linear speed of the outermost fiber (tip
speed of the wheel). For this reason, among wheels made from the same
material, a wheel of twice the diameter can be spun only at one half the
speed to keep the linear tip speed the same. Denoting this speed limit as
Vtipmax, the maximum energy that can be stored per unit mass in a wheel of
inner radius Ri and outer radius Ro can be shown to be
" 2 #
1 2 Ri
Emax ¼ Vtipmax 1 þ ð25:2Þ
4 Ro
Since the ratio Ri/Ro is always less than unity, the above equation clearly
shows that the thin rim flywheel design with Ri/Ro approaching unity
maximizes the specific energy storage for a given tip speed limitation. The
maximum design tip speed achievable in the modern fiber composite rotor
is 1150 m/s with the burst speed of 1500 m/s without preloaded design. A
30-cm (12-inch) rotor operating at 100,000 rpm would produce 1500 m/s tip
speed. Thus, a 100,000-rpm wheel has to be less than 25 cm in diameter.
For a thin rim flywheel of mass M and average radius R, the mass
moment of inertia J ¼ MR2. Therefore, Equation 25.1 can be written as
E ¼ 1=2MR2 !2 ð25:3Þ
It is the specific energy that matters the most in spacecraft power system
designs. The specific energy in Wh per unit mass, denoted by E1, is E
divided by M. That is, from Equation 25.3,
E1 ¼ 1=2R2 !2 ð25:4Þ
The hoop stress at the outermost fiber of wheel made with material of
mass density is given by
¼ R2 !2 ð25:5Þ
Combining Equations 25.4 and 25.5 leads to the specific energy
E1 ¼ 1=2ðmax =Þ ð25:6Þ
Thus, the specific energy of the flywheel is proportional to the maximum

hoop stress allowed in the outermost fiber of the rim divided by the mass
density of the material. Obviously, high specific energy can be achieved by
using high ultimate tensile strength materials, which are also light in mass
density, such that the ult/ ratio is high. Fiber composites far exceed most
metals in this respect. Therefore, they are used almost universally in all
flywheel designs being tested for potential applications. Figure 25.9 shows
the specific energy in rim flywheels of various materials. The figure shows
that the specific energy of composite fiber wheels is much higher than the
metal wheels, and the fused silica fibers have the theoretical maximum
specific energy 20 times that of the state-of-the-art NiH2 battery. Although
flywheel system overheads are much higher than those in a battery, such
high specific energy makes flywheel energy storage development attractive
for space applications.
A preliminary design of a 10-kWh flywheel using various fiber composite
materials is depicted in Figure 25.10. It shows what Equation 25.6 indicates.
That is, the specific energy increases linearly with the ( ult/) ratio of the
material. The maximum hoop stress max has been kept two thirds of the
ultimate tensile strength ult in this 10-kWh example design. The specific
FIGURE 25.9 Theoretical maximum specific energy of rim flywheel of various materials.
FIGURE 25.10 Specific energy and mass of 10-kWh flywheel versus ult/ ratio.
energy of 90 to 120 Wh/kg has been demonstrated at the rotor (rim þ hub)
level in a commercial design running at 60,000 rpm.
25.6 Flywheel Application Example

An LEO satellite requires a heavy battery due to low depth of discharge
dictated by the cycle life considerations. It would therefore benefit the most
by using the flywheel energy storage. Patel12 evaluated the benefits of
replacing a battery with an equivalent flywheel in NASA GSFC’s EOS-AM
satellite as an example. The study was a point design based on reasonable
possibilities. Table 25.1 gives the power system performance requirements
and key design parameters using the NiH2 battery.
Table 25.1 Power system requirements and key

design parameters for an EOS-AM mission using
an NiH2 battery
Parameter Requirement
Spacecraft life 5 years

Load power during eclipse 2530 W
Load power during sunlight 2530 W
Orbit period 1.65 h
Eclipse duration 38 min
Bus voltage 120 V
Energy storage batteries Two 54 cells
50-Ah NiH2
Battery depth of discharge 30% (max)
Photovoltaic array output after sliprings 5000 W
The first task for the system sizing is to perform an energy balance
analysis over one orbit period. Such analysis using a flywheel in the
example mission was performed using a spreadsheet program shown in
Figure 25.11. The analysis determines the component ratings, as seen in the
bottom half of the figure. It is noteworthy that the PV array output required
with the PV–flywheel system is lower, 4662 W versus 5000 W in the PV–
battery system. This reduction is because the round trip energy efficiency in
the flywheel system with the motor–generator machine is higher than that
in the battery system. The result is a 6.75% smaller PV array, which is a
significant side benefit of the flywheel system, in addition to its primary
benefit of reducing the energy storage mass.
The ratings derived in Figure 25.11 are for the total power system. A 20%
margin was added in all component ratings on and above those derived in
FIGURE 25.11 Power system configuration and energy balance analysis for determining
power system component ratings.
the figure. The mass and volume estimates were then developed for the
flywheel system assuming the following performance parameters:
Three equal wheels, two counter-rotating and one extra for reliability.
The number of wheels, however, may be greater than three depending
on the overall mission design
Rotor design speed 100,000 rpm at full spin-up (operating tip speed
<1.0 to 1.2 km/s)
Ultimate tensile strength of the fiber–resin composite 300,000 psi
Maximum allowable hoop stress in the outermost fiber of the wheel
200,000 psi
Rotor length-to-diameter ratio 0.75. This ratio influences the vibration
modes and the critical speeds
Minimum speed at the end of eclipse 35% of the full spin-up speed.
This gives the maximum depth of discharge of 88%
Efficiency of the charge and discharge converters in the PRU is 94%
each
Efficiency of the motor-generator machine is 95% in either direction
The mass estimate of various components using the NiH2 battery and
also the flywheel for this example spacecraft are given in Table 25.2. Table
25.3 summarizes the total mass and volume of the power system using the
NiH2 battery and also the flywheel. Table 25.4 highlights differences
between the two. It indicates that for LEO satellites with a 2500-W load, the
flywheel system can save 458 pounds mass (35%), 13.8 cubic feet volume
(55%), and 32 square feet solar array area (6.7%). Although the cost benefit
of using flywheel energy storage is not predictable at this stage, the
reduction in hardware and launch costs are expected to be several million
dollars per spacecraft.
25.7 Integrated Energy and Momentum Storage

The momentum storage and control requirements of the attitude control
system are traditionally met by using flywheels such as momentum wheels
and reaction wheels. Significant savings are possible by using dual function
flywheels for storing energy for the power system and momentum for the
attitude control system of the satellite.13–15 In the integrated energy and
momentum storage system (IEMS), at least one energy-momentum-wheel
on each axis controls the momentum in that axis and stores the
corresponding energy at the same time (Figure 25.12). Placing an additional
wheel on each axis would give a degree of freedom by simultaneously
storing the required energy and the required momentum. One more
skewed wheel on multi-axis gimbals is needed for redundancy. Thus, a
Table 25.2 Mass of EPS components using NiH2

battery and flywheel (pounds)
EPS with NiH2 EPS with

Component battery flywheel
Solar array assembly 396 370

Array drive electronics 15 15
Power equip battery assembly 310
Hex bay battery assembly 316
Battery power conditioner 47
Sequential shunt unit 98 90
Power distribution unit 102 102
Power switching unit 14 14
Rotor rim 45
Rotor hub 10
Motor–generator 10
Power management electronics 80
Magnetic bearings 18
Magnetic bearing electronics 10
Magnetic bearing sensors 5
Touchdown bearings 10
Auxiliary power supply 5
Shaft (steel) 15
Housing 20
Launch battery (primary) 21
Total EPS mass (pounds) 1298 840
(Data source: M.R. Patel, Proc. IECEC, SAE, 1999.)
Table 25.3 Total EPS mass and volumes using NiH2 battery
and flywheel in the example mission
Design parameter EPS using NiH2 battery EPS using flywheel
EPS mass 1298 lb 840 lb

EPS volume 23.4 cu ft 10.8 cu ft
Solar array area 483 sq ft 449 sq ft
Table 25.4 Overall benefits of using

flywheel in the example mission
EPS using flywheel

Benefit versus battery
Mass reduction 458 lb (35%)

Volume reduction 13.8 cu ft (55%)
Solar array area reduction 32 sq ft (6.7%)
FIGURE 25.12 Energy–momentum wheel placement on three axes.
typical single fault-tolerant IEMS would require at least seven energy-

momentum-wheels.
25.7.1 Energy and Momentum Relations

A flywheel having a mass moment of inertia J spinning at speed ! stores the
energy E ¼ 1=2 J!2, which is a scalar quantity. It also stores the momentum M
¼ J!, which is a vector quantity. If two wheels spinning on a given axis have
mass moments of inertia J1 and J2 and angular speeds !1 and !2,
respectively, then the total energy storage is
E ¼ 1=2J1 !21 þ J2 !22 ð25:7Þ
which is a scalar quantity describing an equation of ellipse in the !1–!2

plane. The total momentum of such two wheels is
M ¼ J1 !1 þ J2 !2
Rearranging the above, we have
!2 ¼ M=J2 ðJ1 J2 Þ!1 ð25:8Þ
which is a vector quantity describing an equation of straight line with slope

(J1/J2) and the !2-axis intercept of M/J2. Figure 25.13 shows a constant
energy ellipse and a constant momentum line of a pair of two unequal
wheels. Two unequal wheels required to store a given level of energy while
simultaneously maintaining a certain momentum level would operate at
either of the two intersection points, P1 or P2. Both points of operation are
dynamically stable; hence the system can indeed operate at any one point,
which is closest to the prior operating point.
If two wheels on each axis are equal, each with inertia J, then the total
energy stored is
FIGURE 25.13 Energy–momentum wheel operation with two unequal wheels on each axis.
E ¼ 1=2 J!21 þ 1=2 J!22 ð25:9Þ
Rearranging the above gives
!21 þ !22 ¼ 2E=J ð25:10Þ
This is the equation of a circle with a radius of

sffiffiffiffiffiffiffi
2E
J
The total momentum of two equal wheels on each axis is
M ¼ J!1 þ J!2 ¼ Jð!1 þ !2 Þ ð25:11Þ
Rearranging the above leads to
!2 ¼ ðM=JÞ !1 ð25:12Þ
which is a straight line with slope of 1 and !2-axis intercept of M/J. Thus,
with two equal wheels on each axis, the constant energy ellipse becomes a
circle and the momentum line takes the slope of 1.
25.7.2 Energy Depletion Constraints

As the spacecraft cannot discharge energy without discharging the
momentum, the maximum depth of energy discharge is limited by the
minimum momentum storage requirement on a given axis. Such an
operating constraint is analyzed below.14
FIGURE 25.14 Energy–momentum wheel operation during sunlight with two equal wheels
on each axis.
A pair of equal wheels that must maintain a given energy level, but
change its momentum level to meet the attitude control requirement, must
operate at any point on the circle, while changing the momentum line closer
or farther from the origin. This is shown in Figure 25.14. On the other hand,
if the system is required to deplete the energy without depleting the
momentum, the speed of the wheels must change to contract the energy
circles while maintaining the momentum line as shown in Figure 25.15. A
FIGURE 25.15 Energy–momentum wheel operation during eclipse with two equal wheels on
each axis.
mission constraint on depleting energy while meeting the momentum

storage requirement makes the momentum line tangent to the minimum
energy circle at the deepest DOD. Or, the radius of the minimum energy
circle must be equal to the momentum requirement at the end of eclipse.
This gives the necessary requirement for energy-momentum-wheel
operation as
sffiffiffiffiffiffiffiffiffiffiffiffi
2Emin 1
> M or Emin > J M2 ð25:13Þ
J 2
In all modes of spacecraft operation, the radius of the minimum energy

circle must be greater than the distance of the momentum line from the
origin. The circle and the line have two intersecting points of operation,
with the limit of the line being tangent to the circle of the minimum energy
level as shown in Figure 25.15. Hence, the maximum permissible DOD is
Emin
DODmax ¼ 1 ð25:14Þ
Emax
This is the upper limit on the DOD from the energy-momentum-wheel

operational point of view. It is necessary that the minimum energy circle at
the end of the eclipse must be no smaller than being tangent to the required
momentum line on each axis. This establishes the allowable depth of energy
discharge in the energy-momentum-wheel system.
25.8 Energy–Momentum Wheel Example

The EOS-AM power system is again taken as an example to estimate
potential benefit for replacing the battery and the momentum wheel with
IEMS.15 The primary design parameters, which influenced the IEMS
estimates summarized here are listed below.
1. The power system ratings derived in Figure 25.11 are for the total EPS,
i.e., for all operating wheels combined. The ratings of each wheel, the
motor-generator and the power and momentum management electro-
nics (PMME) would depend on one of the configuration options
applicable in the satellite under consideration.
2. If the flywheel is used only for the energy storage, the net zero
momentum requires two wheels counter-rotating on each axis. One
extra wheel is required for redundancy. Hence three wheels, each with
50% of system rating, are required.
3. In the IEMS configuration, however, one wheel is needed for the
momentum control on each of the three axes, which also stores energy
within its momentum limitation. One additional wheel is required for
the remaining energy storage requirement, and one more for redun-
dancy. Hence, five wheels are the theoretical minimum number of
wheels needed. Each wheel with 25% of system rating is used in the
IEMS mass and size estimates. The momentum constraint is not applied
in this study.
4. All wheels are assumed to be equal in size. However, it is recognized
that operational considerations in some satellites may require wheels of
different sizes.
5. A 20% margin is added in all component ratings on and above those
derived in Figure 25.11.
6. The rotor design was considered around three major design variables:
(a) the hoop stress, (b) the ultimate strength of the fiber–resin
composite, and (c) the length-to-diameter ratio, which influences the
vibration modes and the critical speeds. With the maximum hoop stress
to ultimate strength ratio of 0.7 and the length to diameter ratio of 0.75,
the rotor design speed of 100 krpm is used. This speed is considered
achievable. Chen et al.16 have reported a 100-Wh, 200 krpm flywheel on
magnetic bearings, where the rotor was made of fiber–epoxy composite
with the maximum operating hoop stress of 150 kpsi in the outermost
fibers.
7. The hub mass is taken as 20% of the rim mass. This is the best estimate
available at present.
8. The PMME provides a total interface between the motor–generator and
the regulated voltage bus. It contains the charge and discharge
converters, the power management (mode control) function, the
momentum management function and the associated telemetry and
commands.
25.8.1 Providing Launch and Ascent Power

One potential design issue with the IEMS is that the wheels may or may not
be spinning at the time of launch to provide the required power during
launch and accent. This issue must be resolved with the launch vehicle
prime contractor. As a conservative approach here, it is assumed that the
wheels cannot be spun until the PV array is exposed to the sun. At that time,
the first task would be to spin-up the wheels to full speed (100% state of
charge). Then onward, the system must be in energy balance in each orbit.
An effective alternative for providing the launch and ascend power is the
primary battery. Among available options, the lithium carbon monofluoride
(LiCFx) has high specific energy and a long flight history in launch vehicles.
LiCFx cells are available in several ampere-hour ratings. Two primary
batteries, each with 24 LiCFx cells of 40 Ah capacity, would provide an
equivalent of full charge of the EOS-AM batteries, and at the same time be
compatible with the 120-V bus. The mass of such batteries (inclusive of all
battery hardware) has been estimated and included in the IEMS designs for
comparison with the existing EPS.
25.8.2 Replacing Batteries and Reaction Wheels with IEMS

Based on the design parameters and assumptions described in above, the
IEMS power system component mass and volumes have been estimated
by Patel15 for EOS-AM satellite and are compared with those using the
existing and future batteries. Table 25.5 summarizes these results.
Various new secondary battery technologies are under development. The
Jet Propulsion Laboratory recently evaluated about 45 major and minor
technology development plans to increase the specific energy of batteries.
The study selected lithium-ion as the most promising technology for future
Table 25.5 IEMS vs. battery mass (kg) in an EOS-AM satellite
EPS with existing EPS with EPS with

Component NiH2 battery future battery IEMS
Solar array 180 180 167

Array drive electronics 6.7 6.7 6.7
Power equip battery assembly 140 76
Hex bay 144 77.5
battery assembly
Battery power conditioner 21.5 21.5
Sequential shunt unit 44.6 44.6 41.5
PRUa 46.5 46.5 46.5
Switching and fusing unit 6.5 6.5 6.5
Reaction wheel 57.5 51.7 Incl.b
Rotor rim 16.8
Rotor hub 3.6
Motor–generator 3.8
PMME c 37.3
Magnetic bearings 11.5
Magbearing electronics 5.7
Magbearing sensors 4.5
Touchdown bearings 5.7
Auxiliary power supply 4.5
Shaft (steel) 7.6
Housing 11.2
Launch battery (primary) 9.6
TOTAL MASS 647.3 kg 511 kg 390 kg
TOTAL VOLUMEd 780 liters 550 liters 315 liters
SOLAR ARRAY AREAd 44.8 m2 44.8 m2 41.6 m2
a
Power regulator unit.
b
Function included in flywheel.
c
Power and momentum management electronics.
d
Summary results of size and volume are given here for brevity.
(Data source: M.R. Patel, Proc. IECEC, p. 200-x, AIAA, 2002.)
spacecraft. The on-going NASA/JPL battery development programs are

targeted to achieve the following goals in 5 to 7 years:
Improve the specific energy from the present 50 Wh/kg to as high as

100 W/kg at the cell level.
Increase the permissible depth of discharge in LEO satellites from the
present 30% to as high as 50%.
Several parallel battery development programs are under way to meet

these goals. It is difficult to predict which particular one would come close
to the stated goals. Even after the predictions are achieved in the laboratory
over the next 5 to 7 years, electrical power systems engineers would
incorporate the results in their designs only after completing the time-
consuming life tests and accumulating adequate performance data on
actual flights. Experience indicates that this may take an additional 5 years.
Without predicting which particular technology would become flight-
qualified in the next 5 to 7 years, it is assumed here that the batteries coming
to the design stage after 5 to 7 years would have specific energy of 75 Wh/
kg at the cell level, and can be operated for 10 years in LEO at 40% depth of
discharge. With these improvements, the future battery mass and volume
would reduce to almost one-half of their present values. Since IEMS would
have to complete with these improved batteries, the EOS-AM power system
estimates with the future battery technology is also included in the present
study. Improvements in the reaction wheel assembly, however, are likely to
be small in view of its technical maturity. Therefore, it is assumed that the
mass and volume of the existing reaction wheel will shrink by 10% over the
next 5 years.
25.8.3 Summary of Point Design Example

The bottom of Table 25.5 shows the benefits of using IEMS in an LEO
satellite having 2500 W payload power requirement. As can be seen, the
IEMS could save 40% mass, 60% volume, and 7% solar array area when
compared to the present batteries. With future batteries, the IEMS could
save 24% mass, 43% volume, and 7% solar array area. Although the cost
benefits of using IEMS are not addressed in this study, the potential
reduction in the hardware and the launch costs could be in several million
dollars per spacecraft.
25.9 Electrical Machine Options

Various types of electrical machines and their suitability for use in the
flywheel energy and momentum storage system are examined.
25.9.1 Synchronous Machines

Synchronous machines are further classified as: (a) excited machine
(brushless version), (b) PM machines (always brushless), and (c) reluctance
machines (no excitation needed). Out of these three, the reluctance machine,
although the simplest and the least expensive of all, is unsuitable with a
flywheel, as it works only as a motor. The flywheel system needs bi-
directional energy conversion (both in the motor and the generator modes).
The excited synchronous machine with slip rings and brushes in any
configuration is not suitable either. The PM synchronous machine is widely
used in many applications. Chen et al.16 have reported a small 100-W,
2-pole, 15 kHz, radiation cooled, brushless, PM synchronous machine design
for 200-krpm flywheel on magnetic bearings. The neodymium–iron–boron
PM is common in such designs, since it offers high magnetic strength. It is
also strong in compression, although weak in tension which is minimized in
the design.
With f ¼ frequency and P ¼ number of poles, the machine speed (in rpm)
is given by
120 f
speed ¼ ð25:15Þ
P
In the discharge mode, the generated voltage and frequency magnitudes

are directly proportional to the flywheel speed. In the charge mode, the
motor speed is spun up by increasing the frequency and voltage such that
the V/f ratio is maintained constant. Maintaining the constant V/f ratio is
necessary to avoid the saturation in the magnetic core of the machine. This
is an open loop speed control.
Electrodynamically, the PM synchronous machine can become unstable if
the load angle exceeds its maximum permissible value determined by the
synchronous and transient reactances. This can happen under sudden load
swings or rapid frequency changes during acceleration and deceleration.
Two such machines working in parallel for redundancy or capacity boost
can also become unstable with each other under load swings and/or
uneven load sharing. With such complexities in operation, the synchronous
machines could be unsuitable with widely swinging loads.
When two or more machines are working in parallel, as is always needed
for reliability in the space application, merely maintaining a constant V/f
ratio in an open loop system is not enough. The synchronous operation
among all machines is necessary. Therefore, the speed of all machines must
be maintained exactly constant, otherwise they would fall out of step and
lose synchronism. The synchronism among all machines working in parallel
is maintained by a closed loop system, in which the shaft position is sensed
and fed back to control the frequency, and hence the speed. This is achieved
by exciting the winding whose magnetic axis aligns with the PM pole axis.
This, in essence, makes the machines work as synchronous machines.
The power electronic converters convert the variable voltage, variable
frequency a.c. power to d.c. and vice versa. The L-C filter is used between
the converter and the d.c. bus to minimize the harmonics generated by the
PWM converters. The 3-phase machine windings are connected in
ungrounded Y, which inherently eliminates all triple harmonics (3rd, 6th,
9th, . . . etc.)
The power transfer between the d.c. bus and flywheel during the charge
operation can be achieved in tapered power, constant power or constant
current mode. The first two may require high current in the beginning,
making it necessary to use a machine of high rating for a short time. The
constant current mode at three current levels, i.e., high, moderate, and low,
may be more mass efficient.
The synchronous machine can be electrically represented by a source emf
and series resistance and reactance as shown in Figure 25.16(a). The relation
between the internally generated emf and the terminal voltage is given by
the phasor diagram shown in Figure 25.16(b) and (c) for the generating and
the motoring operations, respectively. From the diagram, the power
transfer is derived as17
P ¼ Pmax sin ð25:16Þ
where Pmax ¼ EfVt/X, Ef ¼ internal field excitation voltage (emf), Vt ¼

terminal voltage, X ¼ synchronous reactance, and ¼ power angel.
Physically, is the angle between the magnetic axes of the stator phase coil
and the rotor magnet. The power transfer also depends on the terminal
voltage, the internal emf, and inversely with the synchronous reactance. It
can be increased by choosing a machine with low synchronous reactance,
which is made of the sum of the armature reaction and the leakage
reactance. An ironless machine would result in a low reactance machine.
25.9.2 Induction Machine

This is inherently a brushless a.c. machine. On the ground, it is the most
widely used machine as a motor because it is simple, of low cost, and has a
rugged construction. In principle, it can work as a generator under a
leading power factor, but its electrodynamic stability can be a difficult issue
to resolve under varying operating conditions. Moreover, it gives lower
specific power. Its speed can be changed by changing the frequency or by
changing the applied voltage. The latter method of speed control is very
inefficient as it produces losses in the rotor at high slip frequencies. Overall,
this machine is unsuitable with the flywheel in space.
FIGURE 25.16 Synchronous machine equivalent circuit and phasor diagram.
25.9.3 Direct Current Machine

The d.c. machine has two varieties — excited and permanent magnet (with
or without brushes). The speed is controlled by varying the applied voltage.
The speed would change and settle where the resulting back emf plus the
armature IR drop equals the applied terminal voltage, i.e.,
Vt Ia Ra
speed ¼ ð25:17Þ
Ke
where Vt ¼ applied voltage, Ia ¼ armature current, Ra ¼ armature resistance

and Ke ¼ electromagnetic constant of the machine.
The d.c. machine is inherently stable under load swings or voltage
variations. With PM in the rotor, it is brushless, and has no fundamental
I2R loss in the rotor (only eddy and hysteresis losses due to parasitic flux
pulsations).
Thus, the PM d.c. machine and the PM synchronous machine are the only
two candidates in flywheel technology for space. These two machines look
very much alike with almost the same specific power and no significant
difference in efficiency or cooling requirements. Sometimes the two are not
clearly distinguishable unless their control methods are examined. The
synchronous machine speed can be controlled only by varying the applied
frequency, whereas the d.c. machine speed is controlled by varying the
applied voltage. Therefore, these two machines significantly differ in their
speed control method and hence the power electronics design.
Being the heritage motor in the momentum and reaction wheels at
present, the permanent magnet (PM) d.c. machine would be the lowest
nonrecurring cost approach. In addition, the PM d.c. machine has certain
desired performance characteristics suitable for space applications.
Since the d.c. machine speed is controlled by voltage and it has linear
speed–torque–current relations, its controls and power electronics are
simpler and lighter than with the synchronous machine. For this reason,
the PM brushless d.c. machine is the most suitable candidate for electro-
mechanical power conversion to and from the flywheel. The selection of the
PM d.c. machine for flywheel technology in space is, therefore, a sound one
for the reason given here. There are further options in the PM d.c. machine,
namely the axial gap and the radial gap machines. Considering the flywheel
geometry, the conventional radial gap machine is the most natural fit,
which is also easily scaleable for different power levels.
25.10 Motor–Generator Design Issues

Some of the constructional features of the PM d.c. machine that may benefit
the design are the ironless armature to eliminate iron losses, Litz wire to
minimize the skin and proximity effects, and a composite carbon/graphite
bobbin for improved heat conduction from the coil. Drawing from industry
experience in designing high-frequency motors for aircraft and navy
submarines and the superconducting generator design, the ironless
armature of Litz wire in a graphite bobbin is a sound baseline approach.
However, the technical challenge of fully developing a high speed, efficient,
and entirely conduction cooled motor–generator design remains. Based on
the ironless airgap armature and Litz wire construction, the estimated mass
of a space worthy 1-kW, 20 kHz machine is 1.5 kg. A 5-kW, 2-pole PM
machine is estimated to weigh 4 to 6 kg/kW at 3600 rpm, 2 to 3 kg/kW at
12,000 rpm and 0.4 to 0.6 kg/kW at 100,000 rpm. The scaling law for
machine mass is as follows:
a
kW
mass ¼ ð25:18Þ
speed
where the exponent equal to 0.5 applies to 1 to 5 kW machines, and 0.7

applies to 100 to 1000 kW machines.
25.10.1 Constant Torque Versus Constant Power Design

In the motor–generator (M–G) design, one approach is to deliver constant
torque and the other is to deliver constant power. Table 25.6 shows that in
the constant power design, the motor has to be rated 3.0 times that in the
constant torque design. On the other hand, the charging power delivered to
the flywheel at minimum speed under constant torque operation is only one
third, the remaining two thirds must be diverted to the shunt dissipators.
Although the constant torque approach optimizes the machine perfor-
mance, it would lead to poor performance in the satellite power system as a
whole. The constant power M–G design approach saves mass in the
photovoltaic array several times the mass it adds in the motor-generator.
Since proportionally similar savings would be seen in other missions as
well, the constant power M–G design is the preferred design approach in
the PV–flywheel power system for space applications.
One of the critical design issues in the motor–generators design for
constant power would be the much higher (9 times) armature I2R losses in
the stator windings when operating at the minimum speed. Since the space
motors must be cooled by conduction only, removing such a high level of
heat may be a fundamental design issue not faced by their terrestrial
counterparts. In space motors, the cooling would most probably require use
of heat pipes. However, placing heat pipes in the extremely high frequency
magnetic field (rotating at 100,000 rpm) would pose new design issues. For
this reason, a motor–generator designed for terrestrial applications in
general would not be applicable for space use. The underlying reason for
this is that, in space, power not used in charging the flywheel is wasted,
which is not the case in an automobile or in UPS applications on the
ground. This system-level operating difference between the two applica-
tions introduces significant differences in motor–generator design.
Table 25.6 Motor–generator performance at one

third speed relative to full speed
M–G performance Constant torque Constant power

parameter design design
Voltage (%) 33 33
Current (%) 100 300
Power (%) 33 100
Armature loss 100 900
Eddy loss (%) 10 10
M–G rating (%) 100 300
25.11 Magnetic Bearings Design Issues

Conventional roller bearings in high-speed flywheels result in high
transmitted vibrations and high drag torque resulting in high self-discharge
rates. The magnetic bearings provide contactless suspension with far
reduced drag torque and vibrations. They have flown on the Russian
space station MIR for years. However, the design challenges with magnetic
bearings are in the sensors and complex control system.
Actively controlled bearings require complex controls systems. The
control current is proportional to the square of the gap. Passive magnetic
bearings, on the other hand, give high system efficiency and reliability, but
it typically gives lower stiffness and lower damping. A typical passive
bearing design for 60-krpm operations may have radial stiffness of 1800 lb/
inch and the modal damping of 5%. Earnshaw’s theorem states that in any
inverse square law system following the Laplace equation, such as the
magnetic field, it is impossible to have a completely 3-dimensional stable
passive magnetic bearing using only the forces of a static field of permanent
magnets. The theorem, however, does not apply in active feedback systems,
time-varying fields, ferrous liquids, and superconductors. The active
feedback control system is the most commonly used method for stabilizing
the magnetically suspended bearings. The operating speed, hence the
specific energy of the flywheel, increases with the ratio of the polar to the
transverse moment of inertia (Ip/It). The ratio also produces larger changes
in the momentum for more effective attitude control. For a stable active
bearing design, the Ip/It ratio should be preferably below 0.8 or above 1.2,
although its value of about 1.0 has been used in a 60 krpm design.7
Several concepts for designing magnetic bearings are possible. The
relative merit of any one configuration over another can be compared only
after their designs are developed in sufficient details and traded for
optimizing the performance, cost, and technical and schedule risks.
However, any configuration that is selected must have the following five
performance attributes:
Low eddy current and hysteresis losses (magnetic drag)

Sufficient restoring forces in all five degrees of freedom
Spring constants, both in translation and torsion, that assure the
mechanical stability and avoid resonance at the spacecraft excitation
frequencies
Electrical control loop stability
Sufficient dynamic damping to attenuate disturbances
The following paragraphs compare a few design concepts offered on the

market at present.
25.11.1 Low-Speed Bearings

Teldix, in Germany, and others have been offering magnetic bearings in
their momentum wheels for many years. The speed range of these wheels is
3000 to 6000 rpm. In their design, the magnetic flux travels through the pole
shoes on the stator and the magnetic feedback ring on the rotor. The
reluctance lock between the pole shoes and the magnetic feedback ring
provides the vertical restraint. The control coils are segmented, hence they
would suffer excessive magnetic drag at high speed, making them
unsuitable for the flywheel energy storage.
25.11.2 High-Speed Bearings

Among homopolar and other design concepts, one shown in Figure 25.4
uses a permanent magnet (PM) for passive restrain and electrical coils for
active control. However, since the coils are segmented, it would create
pulsating flux in the shaft as it rotates, thus generating magnetic drag. At
speeds above 60 krpm, drag loss would limit the speed. One can reduce
the drag by using thin laminations, but that would introduce noise and
vibrations. This issue needs to be satisfactorily resolved for speeds
approaching 100 krpm.
There are several other configurations of magnetic bearing commercially
available which may be suitable for high-speed operation. Regardless of the
configuration chosen, detailed electro-mechanical design for a magnetic
bearing would need the following tasks to be performed:
1. The magnetic bearing must be restrained in five degrees of freedom (all

except the one we want to rotate freely). The restoring forces and spring
constants in the five degrees of freedom (three translational and two
rotational) are computed from the finite element electromagnetic flux
plots and using either the pressure approach and/or the energy
approach. In free space, the magnetic pressure (force per unit surface)
and the magnetic energy stored per unit volume are given by is 1=2 B2/ o.
The change in energy per unit displacement is force, and the force per
unit displacement is the spring constant. This would give a translational
spring constant for a linear displacement, and a torsional spring
constant for a rotational displacement. A positive spring constant
would make the system naturally stable in that direction, while a
negative spring constant would need a feedback control to stabilize the
system.
2. The above spring constants, along with the dynamic damping and the
mass and angular inertia constants, are used to develop a complete
electro-mechanical model of the rotor-bearing system.
3. The resonance frequencies (square root of the spring constant over

inertia) are computed and kept away from the system excitation
frequencies.
4. The feedback control system model is then developed to study the
dynamic electro-mechanical response and stability. The gains and
damping are adjusted to fine-tune the design performance as required.
5. The finite element flux plots are used to calculate the magnetic drag at
rated speed. There would still be significant stray (parasitic) loss at high
speed that would defy all analyses. The magnetic loss components are
therefore estimated using engineering judgment at the design stage,
and then from carefully planned tests to separate eddy loss (varying
with speed squared) and hysteresis loss (varying with speed) for use in
subsequent designs.
25.12 Flywheel System Controller

The rotating flywheel has six degrees of freedom, the x and y positions at two
axial locations which keep the shaft straight, one in the Z direction at either
end, and one angular position. The angular position is unrestrained. That
leaves five degrees of freedom that must be restrained by the bearings —
mechanical or magnetic. The magnetic bearing works on magnetic forces —
attraction and/or repulsion — created by passive permanent magnets or
active electromagnets with feedback control systems. A typical clearance
gap in the radial direction is 0.25 to 0.50 mm, and is maintained within
0.025 mm. The shaft position in the five degrees must be sensed and the
errors amplified and fed back to the system for corrective response in terms
of the coil currents in the electromagnets. Figure 25.6 earlier showed the
location of the five restrains and the corresponding sensors.8
Eddy current sensors are generally used in magnetic bearings. The
theoretical basis for these sensors is Faraday’s law of electromagnetic
induction. The induced voltage in any coil is proportional to the rate of
change of the flux through the coil. A change in the shaft position in any
direction causes a proportional change in the magnetic flux in the gap in
that direction, which translates into the induced voltage in the eddy current
sensor. The sensitivity of the sensor is therefore measured in terms of the
induced voltage per micrometer of the position change. For greater
sensitivity, the sensors are operated at high frequency. The eddy current
sensor, therefore, has three major components: a radio frequency oscillator
to generates a signal, the sensor that radiates the signal through the probe
tip, and the demodulation circuit to convert the returned signal into usable
form. A conducting material placed in the proximity of the sensor tip would
generates eddy currents, which opposes the flux from the sensor tip in
accordance with the Lenz law. The resulting loss of magnetic strength in the
return signal is measured, demodulated, and converted into a voltage

signal. This signal is proportional to the gap between the sensor tip and the
conducting surface, which is fixed to the shaft. Such sensors are available
commercially with a sensitivity of 10 to 15 mV/mm. The major design issues
are the control system and the EMI cross-talk between the sensors, since
some sensors are located close to each other.
A simplified schematic of the control systems is shown in Figure 25.1718
The bearing control code generates command signal to allow shaft
levitation. These command signals are converted to drive currents by the
PWM amplifier, and passed on to the control system, which alters the shaft
position. The eddy current sensor signal is processed by the signal
conditioning system and fed back to the bearing controller. The mode
controller controls the pulsing of the converter switches as required for
operating the systems in the charge, discharge or idle mode. Truong, Wolff,
and Dravid18 have developed the modeling technique for a flywheel, a
3-phase PM synchronous motor-generator and a 3-phase bi-directional half-
bridge converters (Figure 25.18). The operation is described below starting
from the minimum DOD. The controllers set the system in the charge mode.
Initially the motors take in the maximum charging current within the set
limit. As the motor speed rises, the counter emf builds up and the current
drops. When the speed reaches the maximum value, the system enters idle
mode with zero power transfer. The wheel maintains its speed until
commanded into the discharge mode. The synchronous machines as the
generator feeds the bus with a constant current set by the maximum limit, or
using a constant power criteria. The latter is perhaps a better approach.
Significant work has been conducted in the flywheel test facility at NASA
Glenn. A successful operation with speed and position feedback control of
the unit has been demonstrated.19 A 4-pole permanent magnet synchronous
machine suspended on magnetic bearings was controlled with a field
orientation algorithm.
FIGURE 25.17 A simplified control system for active magnetic bearings.

FIGURE 25.18 Flywheel control system configuration.
25.13 NASA Flywheel Program

Flywheel momentum storage has been in use for spacecraft navigation and
attitude control system since the beginning of the space program. A
prototype flywheel energy storage combined with attitude control system
had been built and tested at Draper Laboratory at 30,000 rpm in the early
1990s (Figure 25.19). The concept is now under active developments at
NASA Glenn for replacing the electrochemical battery in spacecraft
electrical power systems. In the current NASA design, flywheel operating
speeds vary from 60,000 rpm maximum to 20,000 rpm minimum,
representing 90% DOD. Flywheel inertia is set by the energy storage
requirement, and safety and reliability considerations set the maximum
operating speed. The friction and windage losses are negligible. The voltage
magnitude and frequency are linearly proportional to the flywheel speed.
The phase current magnitude and the phase angle determine the magnitude
and direction of the power transfer between the flywheel and the bus. The
actual voltage and power controls are implemented by the PWM power
converters. The switching action is affected by the power angle of the rotor.
Capacitive filters are used on the bus side of the converters. A series
inductive filter is also used to reduce the harmonic current in the phase
coils.
The projected goals of the NASA and AFRL programs13 at Honeywell
and Lockheed Martin are:
Specific energy >200 Wh/kg

Cycle life >75,000
FIGURE 25.19 Energy storage and attitude control flywheel system for spacecraft under tests.
(ßThe Charles Stark Draper Laboratory, Inc. All rights reserved. Reprinted with permission.)
Round trip energy efficiency >90%

Systems cost reduction >25%
NASA Glenn has also been developing flywheel energy storage systems
(FESSs) for the International Space Station.20 At present, the ISS power
system consists of solar arrays and NiH2 batteries. The estimated life of a
battery is 5 years. However, the design life of the ISS is 15 years. NASA
plans to replace one battery along with the battery charge/discharge
converters with a flight-qualified FEES in 2006. The flywheel will replace
the battery at the same time, thus requiring no modifications to the ISS. The
projected life of the FESS is 15 years (92,000 C/D cycles and 1100 startup
and shut-down cycles), thus requiring no replacements during the ISS

mission life. NASA estimates the life cycle cost savings of about 150 million
dollars.
The design features of the ISS flywheel unit (Figure 25.20) are:11,20
Operating speed 53,000 rpm to 41,500 rpm, which results in a tip speed
of 916 m/s and 39% DOD (versus 35% for the NiH2 battery. The laminar
stress at the rotor-hub interface is not an issue
Contingency DOD 89% (53,000 rpm to 17,800 rpm) for 30 cycles over the
life
Specific energy of 27 Wh/kg versus 10 Wh/kg with the NiH2 battery
Radial and axial magnetic bearings in 106 torr vacuum and mechanical
touchdown bearing
Two counter-rotating rotors made of carbon fiber composite with
titanium shaft
Eclipse power 4.1 kW, peak power 5.5 kW (comparable with the NiH2
battery)
Energy storage 5.5 kWh, versus 4.6 kWh for the NiH2 battery
Round trip energy efficiency 83% versus 65% for the NiH2 battery
FIGURE 25.20 Cross section of NASA flywheel module.

(Source: E.B. Gietl et al. NASA Glenn TM 2000-210209.)
Life expectancy 15 years versus 5 years for the battery

Wheel size: 33 cm diameter 48 cm long with 5 kWh capacity, initially
in compression (near zero stress in operation)
Electrical machine: 3-kW, 4-pole, 3-phase, PM synchronous machine
Protection: RPC with less than 6 V dead-facing for crew safety
Each FESS assembly: 0.9 m wide 0.85 m long 0.625 m high and
205 kg mass
Each wheel assembly with signature for failure detection
Life tested to 60 years (4 operating life) as required by the manned
space flight fracture design for critical components
Ratio of polar to transverse moment of inertia Ip/It of 0.43
Operating temperature 22 to 82 C, actively cooled
Permanent magnet with homopolar magnetic bearings
Three-phase a.c. permanent magnet motor–generator
Housing to support the flywheel and protect from orbital debris (not
hermetically sealed, the space naturally provides vacuum for the rotor)
Wheel not spinning during launch
Tested up to 60,000 rpm with safety enclosure containing water in the
test bed.
In fully developed and space-qualified hardware, the specific energy of the

flywheel system is estimated to be around 5 times that of the NiH2 battery.
References
1. NASA/Aerospace flywheel development program document, Web site http://
space-power.grc.nasa.gov/ppo/project/flywheel/May 15, 2001.
2. Harris, C., Flywheels for spacecraft energy storage and attitude control, in
Flywheel Energy Storage Workshop, Oak Ridge, TN, National Technical
Information Service, Washington, D.C., November 1995, Conference
Records 9510242.
3. Wehmer, J., Spacecraft flywheel system requirements, in Flywheel Energy
Storage Workshop, Oak Ridge, TN, National Technical Information Service,
Washington, D.C., November 1995. Conference Records 9510242.
4. Uherka, K., Experimental test facility for flywheel energy storage utilizing
superconducting magnetic bearings at the Argonne National laboratory, in
Information Service, Washington, D.C., November 1995, Conference
Records 9510242.
5. Maass, D., Development in polar woven/RTM composite flywheel rotors,
in Flywheel Energy Storage Workshop, Oak Ridge National laboratory, TN,
National Technical Information Service, Washington, D.C., November
1995, Conference Records 9510242.
6. Siebert, M. et al. A passive magnetic bearing flywheel, in Proceedings of the

36th Intersociety Energy Conversion Engineering Conference, ASME, 2001,
Paper No. 100-x.
7. Palazzolo, A. R., Beach, R., et al. Magnetic Suspension for flywheel
batteries, in Proceedings of the 36th Intersociety Energy Conversion
Engineering Conference, ASME, 2001, Vol. 1, pp. 97–103.
8. Dever, T.P., Palazzolo, A.B., Thomas, E.M., and Jansen, R.H., Evaluation
and Implementation of Eddy Current Position Sensors in Magnetically
Suspended Flywheel Systems, in Proceedings of the 36th Intersociety Energy
Conversion Engineering Conference, ASME, 2001, Vol. I, pp. 689–694.
9. Harris, C., Flywheels for spacecraft energy storage and attitude control, in
Information Service, Washington, D.C., November 1999.
10. Aerospace Flywheel Workshop Records, U.S. Air Force Research
Laboratory and NASA Glenn Research Center, Cleveland, OH, October
1999.
11. Thomson, R., Flywheel energy storage system in Proceedings of the 35th
No. 100-x.
12. Patel, M.R., Flywheel energy storage for spacecraft power system, in
SAE, 1999, Paper Number 183.
13. McLallin, K.L., Fausz, J., Jansoen, R.H., and Bauer, R.D., Aerospace
flywheel development for IPACS applications, in Proceedings of the 36th
Intersociety Energy Conversion Engineering Conference, ASME, 2001, vol. I,
pp. 105–110.
14. Patel, M.R., Energy-momentum wheel for satellite power and attitude
control systems, in Proceedings of the 35th Intersociety Energy Conversion
Engineering Conference, AIAA, 2000, Paper No. 200-x.
15. Patel, M.R., Integrated energy-momentum wheels for spacecraft, in
Conference, AIAA, 2002. Paper No. 200-x.
16. Chen, H., Walter, T., and Lee, N., Recent development of magnetic
bearings for flywheel energy storage, in Proceedings of the 1st International
Energy Conversion Engineering Conference, AIAA, 2003. Paper No. 6105.
17. Say, M.G., Alternating Current Machines, John Wiley & Sons, New York,
1983.
18. Truong, L.V., Wolff, F.J., and Dravid, N.V., Simulation of flywheel energy
storage system controls, in Proceedings of the 36th Intersociety Energy
Conversion Engineering Conference, ASME, 2001, Paper No. 33-38.
19. Kenny, B. H. et al. Advanced motor control test facility for NASA GRC
flywheel energy storage systems technology developments unit, in
Conference, ASME, 2001, vol. 1, pp. 91–96.
20. Grantler, J.H., Wiegmann, B.M., and Wagner, R., Flywheel energy storage
system for the International Space Station, in Proceedings of the 36th
pp. 263–267.
Chapter 26
Superconductors in Space
26.1 Introduction
This chapter deals with a technology that is fully developed on the ground,
and naturally fits even better in the low temperature environment of space.
Working at the liquid nitrogen temperature prevailing in space, the newly
discovered high-temperature superconductors are expected to make rapid
inroads into a variety of space applications. For example, it has a high
potential for storing large amounts of energy for multi-megawatt burst
power systems for space defense. While some barriers are yet to be
overcome, these materials have enormous potential for new space and
ground-based applications, or merely to expand proven technologies that
currently use liquid helium superconductors. Cryogens carried on board for
propulsion may be judiciously routed to augment the cooling of new
superconductors at selected places on the spacecraft platform.
The energy density in the magnetic field is much higher than in an
electrical field. The air can support a uniform E-field of 3.1 kV/mm in a
standard atmosphere. Most liquids break down at 10 kV/mm. A liquid-
filled capacitor operating at 3.5 kV/mm has a stored energy density of U ¼
1=2 "E2 ¼ 1=2 (2 8.88 10 –12) (3.5 106) ¼ 108 J/m3 between the electrode
plates. The energy storage density in magnetic field is given by U ¼ 1=2 BH ¼

1=2 B2/mo. An air-cooled copper coil with a current density of 1.5 A/mm2
producing a 2 tesla B-field would store U ¼ 1=2 22/(4 10 –7) ¼ 1.6 MJ/
m3 in the hollow of the coil. This is four orders of magnitude higher than
the energy density in the E-field. A superconducting coil carrying much
higher current density would result in an even higher energy density in
the square proportion of the magnetic field density. Commercial super-
conducting magnets are available at present to produce a 5 to 15-T field
over 1/100 to 1 liter volume for a variety of applications. They yield an
energy density of 40 MJ/m3 at 10 T and 90 MJ/m3 at 15 T.
26.2 Magnetic Energy Storage

The energy stored in the magnetic field of a coil in air or in free space is
given by
U ¼ 1=2B2 =
640
Superconductors in Space 641
in joules/m3, or
U ¼ 1=2I 2 L ð26:1Þ
in joules, where B ¼ magnetic field density produced by the coil (T), ¼

magnetic permeability of free space ¼ 4 107 H/m, I ¼ current in the
coil (A), and L ¼ inductance of the coil (H).
The relation between the coil current I and the voltage V is
dI
V ¼ RI þ L ð26:2Þ
dt
where R and L are the coil resistance and inductance, respectively. For
steady-state energy storage at constant d.c. current, the second term in
Equation 26.2 becomes zero. The voltage required to circulate a d.c. current
is then V ¼ RI. The resistance of the coil is temperature dependent. The
electrical resistance of most conductors decreases with decreasing tempera-
ture as shown in Figure 26.1. In certain conductors, as the temperature is
reduced to the cryogenic range, the resistance abruptly drops to precise
zero at some critical temperature Tc. Below this temperature, no voltage is
required to circulate current in the coil, and the coil terminals can be
shorted to continue free circulation of the current. The circuit time constant
L/R is now infinite, meaning that the current will continue to flow in the
coil indefinitely. The coil is said to have attained the superconducting state
with zero resistance. The energy in the coil then freezes, and remains stored
indefinitely.
The superconductor requires a supercool operating temperature to attain
the critical superconducting state. Early superconductors operating around
4 K were required to be immersed in liquid helium at almost absolute zero.
Recently discovered materials that become superconducting at much higher
FIGURE 26.1 Conductor resistance versus temperature approaching absolute zero.

temperatures are leading to new technological breakthroughs. They can be

operated at liquid nitrogen temperature (80 K), or even at temperatures
exceeding 100 K. Raising the critical temperature of new materials to 100 K
allows the use of liquid nitrogen as the coolant. That has a number of
advantages, such as:
Liquid nitrogen costs much less than liquid helium

It has greater heat capacity, which means less is needed
It can be transported in a portable container, and
It has an ideal temperature range for optimum operation of semi-
conductor devices, making hybrids circuits of semiconductors and
superconductors attractive
Although the superconducting phenomenon was discovered decades

ago, industry interest in developing practical applications started in the
early 1970s. In the USA, the pioneering work was done in this field by the
General Electric Company, the Westinghouse Research Center and the
University of Wisconsin. During the 1980s, a grid-connected 8 kWh super-
conducting energy storage system was built with funding from the
Department of Energy, and was operated by the Bonneville Power
Administration in Portland, Oregon. The system demonstrated over one
million charge/discharge cycles, meeting its electrical, magnetic and
structural performance goals. Conceptual designs of large superconducting
energy storage systems up to 5000 MWh energy for utility applications have
been developed in the past.
On the ground, the superconducting coil is suitable for long-term energy
storage. It requires no electro-mechanical conversion, hence has high
efficiency. Coils that can store 4 to 6 MJ are commercially available in low
volume by American Superconductors. The Wisconsin Power Company has
a large superconducting ring used for stabilizing the grid. High-tempera-
ture superconductors are expected to further accelerate the following on-
going research and development.
Utility load leveling by storing electrical energy at night for use at times
of peak demand, eliminating brownouts and reducing the amount of
generating capacity needed. Magnetic bearings and levitating trains
suspended on magnetic fields could move from prototype to commercial
use. In medicine, diagnostic machinery that uses superconductive equip-
ment to probe the body’s soft tissue could become more sensitive and less
expensive.
Highly efficient generators and motors developed by companies like
General Electric, Westinghouse, Siemens, and American Superconductors
may become commercially viable products. The superconducting generator
gets rid of all magnetic steel, giving much higher specific power by 4- to
5-fold (excluding the cryo overhead). Toshiba Corporation of Japan and
GEC-Alsthom along with Electricité de France are actively persuing

developments in this field.1,2
Improved superconductors may be used in computers to replace the
wires between transistors and chips. This could increase computing speed
while reducing the amount of power that computers dissipate in heat.
Superconducting interconnects between semiconducting devices could
increase circuit speed as devices become smaller. Critical current densities
around 1 MA/cm2 are needed for interconnects between devices on chips,
and 100,000 A/cm2 densities are necessary on printed circuit boards
between chips.
In space, superconductive switches — being very sensitive to certain
kinds of radiation — could be used by the military to locate enemy satellites
by detecting minute amounts of infrared radiation. The superconducting
coil is also a potential candidate for energy storage for space-based missile
defense weapons to produce brief pulses of extremely high power needed
to drive antimissile lasers, particle beams, and electromagnetically pro-
pelled canons. The energy storage coil could be switched to a discharge
mode in 30 ms, compared to 15 min to prepare a conventional rotating
generator. The defense department has also funded programs at the utility
companies to develop a large superconducting energy storage ring. In
normal use, it would store energy generated at low cost during off-peak
hours for their own use during peak hours. However, the Pentagon could
turn to them in the event of a missile attack and instantly divert the stored
energy to the defense needs of the moment.
Superconducting converter technology is also being developed for Air
Force applications. It is for either the d.c.–d.c. converter or a.c.–d.c.
converter depending on the power source. This technology results in very
high conversion efficiency of about 99%, coupled with high specific power,
around 1 kW/kg for d.c.–d.c. converters and 5 kW/kg for a.c.–d.c.
converters in large power ratings typical of defense weapons. These
power densities are about double the value that can be achieved with
conventional designs.
And, of course, superconducting energy storage may one day be
considered as a replacement for the battery, but that is some time in the
future.
26.3 Critical J, B, and T

Three parameters, namely, the temperature, current density, and the
magnetic field intensity jointly determine when a superconductor attains
the precisely zero resistance. The critical value of each parameter —
marking the transition to superconductivity — varies with changes in the
other two. For example, as the temperature and magnetic field increase, the
critical current density declines. Hence, superconductivity thrives below
but vanishes above a three-dimensional critical surface, which is unique to

each superconductor. Among the early superconductors, niobium titanium
(NbTi) and niobium tin (NbSn) have been extensively used. Both have the
critical temperature around 9 K, which requires liquid helium cooling to 4 K
to allow some margin. The maximum flux density of around 9 T can be
produced using NbTi at 4 K, and around 16 T using NbSn at 4 K. The critical
current density is 1000 A/mm2 at 10 T, both being orders of magnitude
higher than those in the conventional copper coil magnets (5 A/mm2).
NbSn is mechanically weak and brittle compared to NbTi, posing some
difficulty in manufacturing.
The 1986 discovery of high-temperature superconductors has made
three types of new superconducting materials available now, all made
from bismuth or yttrium cuprate compounds. Current density and field
requirements for various superconductors are shown in Figure 26.2. In the
case of yttrium barium copper oxide, the critical surfaces for thin film and
FIGURE 26.2 Critical current density versus magnetic field density and temperature for
various superconductors.
bulk material made with AT&T’s melt-textured growth process extend past
80 K, the liquifying temperature of nitrogen. The critical surfaces of the
earlier superconductors, niobium titanium and niobium tin, are smaller,
which force them to operate near 4 K, the liquifying temperature of helium.
When metals (A and B), and ceramics (C and D) operating at 4 K and 80 K,
respectively, are compared for current density and field relationships, the
melt-textured bulk ceramic (C) is inferior, but if normally processed, would
barely show up on the graph.
The superconducting coil is wound with wires made of numerous
superconducting filaments in a copper or aluminum matrix. Copper and
aluminum have lower resistance than the new materials in their non-
superconducting state. Thus, if the critical current were accidentally
exceeded in the superconductor, electrical current would switch from the
ceramic to the surrounding metal matrix, reducing ohmic heating. The goal
of the present development programs funded by the DoE and DoD are to
produce such wires with engineering current carrying capacity of 1000 A/
mm2 based on the entire cross sectional area of the wire, including
the copper matrix and the insulation. This is about 70 to 100 times that of
the traditional coil in standard atmosphere. The manufacturing process
includes the coating of noble metals like silver to turn brittle high-
temperature superconducting ceramics into flexible wires. It applies not
only to the powder-based wires made of a bismuth strontium compound,
but also to the next generation of wires using yttrium barium copper oxide,
which many companies are developing in the USA, Europe, and Japan. The
current density and magnetic field requirement for various superconduct-
ing applications are shown in Figure 26.3.
FIGURE 26.3 Current density and magnetic field requirement for various superconducting
applications.
26.4 Magnet Coil Design

In spacecraft, a strong magnetic field emanating from a solenoid coil would
interfere with the attitude control and other equipment sensitive to the
magnetic field. The toroidal coil design eliminates this problem, as the flux
is self-confined to a closed ring-shape volume as shown in Figure 26.4. The
toroid with major radius R and the number of turns N carrying current I
produce an internal magnetic field B
¼ oNI/(2R), which varies inversely
with the major radius. The circular current produces hoop stress in the
conductor, and the circumferential current produces repulsive force
between the two sides of the toroid. The resulting mechanical stresses
vary around the circumference, thus causing strong hoop and bending
stresses. The bending stress needs a center support structure. Once it is
supported, the remaining force produces a pure hoop stress, which is
internally supported by the tensile strength of the conductor. The super-
conducting coil is generally manufactured with high pre-stress, so that
cooling to the operating cryogenic temperature does not produce damaging
stress. However, mechanical creep during room temperature storage and
high temperature excursions during forced warm up are of concern.
26.5 System Configuration

A typical superconducting energy storage system is shown in Figure 26.5.
The coil is charged by an a.c.–d.c. converter in the power supply. Once fully
charged with the required current and cooled below the critical tempera-
ture, a short is placed, thus making the solenoid coil a permanent magnet.
The converter continues providing the low voltage needed to overcome
losses in the room temperature parts of the circuit components. This keeps
the constant d.c. current flowing (frozen) in the superconducting coil. In the
storage mode, the current is circulated through the normally closed switch.
FIGURE 26.4 Self-contained magnetic field in toroidal coil.

FIGURE 26.5 A superconducting energy storage system configuration.
The power supply itself for charging can be a superconducting d.c.

generator, such as a Faraday disk.
The system controller shown in the figure has three main functions:
Control the solid state isolation switch

Monitor the load voltage and current
Interface with the voltage regulator that controls the d.c. power flow to
and from the coil
If the system controller senses the line voltage dropping, it interprets that
the system is incapable of meeting the load demand. The switch in the
voltage regulator opens in less than 1 ms. The current from the coil now
flows into the capacitor bank until the system voltage recovers to the rated
level. The capacitor power is fed directly to the load or after inverting into
a.c. if necessary. The bus voltage drops as the capacitor energy is depleted.
The switch opens again, and the process continues to supply energy to the
load continually. The system is sized to store sufficient energy to power the
load for specified duration.
Superconducting energy storage has several advantages over other

technologies:
The round trip energy efficiency of the charge/discharge cycle is high

around 95%. This is higher than that attainable by any other technology
Much greater charge/discharge cycle life
Much longer life, up to about 30 years
The charge and discharge times can be extremely short, making it
attractive for supplying large power for short time if needed
Have no moving parts in the main system, except in the refrigeration
components, hence high reliability
26.6 Cryogenic Temperature

In a superconducting energy storage system, the main cost is to keep the
coil below the critical superconducting temperature.
The NbTi and NbTi alloys are stable at high B and high J without
quenching (i.e. losing superconductivity). It usually operates in a 4 K bath of
liquid helium, which has a boiling point of 4 K at one atmospheric pressure.
At 4 K, cooling becomes difficult. Recovering 1 W of heat loss from 4 K to
room temperature cost about 500 W of refrigeration power, even at perfect
Carnot cycle efficiency of (Th Tc)/Th. Therefore, heat leakage from the
leads connected to the external circuit must be controlled by proper thermal
design, as they determine the capital and running cost of the cryogenic
equipment.
On the other hand, room temperature magnets lose more power in the
magnet itself, and little in the cooling power. The room temperature power
in conventional magnet amounts to roughly 100 times that for the
superconducting magnet’s room temperature power. Thus, the trade is
highly favorable even for the liquid helium superconducting magnet. The
new superconductors having critical temperatures around 100 K are cooled
by liquid nitrogen, which needs orders of magnitude less refrigeration
power. As a result, numerous programs around the world have started to
develop new commercial applications using liquid nitrogen cooled super-
conductors.
A copper matrix avoids local thermal transient heating and subsequent
quenching. Also, when the coil is grossly quenched, it provides an
additional path for the normal currents to flow without excessive heating.
During quenching, the stored magnetic energy is converted into heat.
Generally, quenching starts from a small local hot spot due to friction
under vibration or any other reason. The spot may become over-heated
more than the bulk surroundings. The hot spot temperature can be
controlled by rushing coolant to avoid local burning. If the coil is quenched,

a large potential of hundreds, or even thousand, of volts is induced in the
coil, which may cause arcing between the turns. Switching an external
dump resistor or a short-circuited coil across the magnet to inductively
transfer part of the stored energy can provide the quenching protection.
26.7 Cryogenic Behavior of Components

Semiconducting devices, inductors, transformers, capacitors, resistors, and
structural materials used in the power system components behaves
differently at cryogenic temperatures around 100 K (near liquid nitrogen)
than at the conventional temperature around 100 C.
26.7.1 Semiconductors
The electrical behavior of a semiconductor device is dominated by several
key parameters, as follows.
26.7.1.1 Carrier mobility

The ease with which conduction electrons and holes move through the
semiconductor crystal significantly varies at low temperatures. The carrier
mobility increases, peaking between 100 K and 200 K depending on the
doping concentration, and is typically a factor of 2 or 3 times that at the
room temperature.
26.7.1.2 Number of dopant atoms ionized

The percentage of dopant atoms ionized and contributing to the conduction
process begins to decrease at approximately 125 K. At 80 K the effective
dopant concentration is lower and some conduction processes are affected.
26.7.1.3 Minority carrier lifetime

The probability of a trapping center to capture both a hole and a conduction
electron increases, thus reducing the minority carrier lifetime. For all
semiconductor devices, the leakage current, being thermally induced, is an
exponential function of temperature, and therefore decreases rapidly as the
temperature is lowered. Since the leakage current and the breakdown
voltage vary in a similar fashion, the breakdown voltage also decreases as
temperature is reduced, although not as rapidly. The reduction is relatively
small, about 10% at 80 K.
26.7.1.4 PN junction diode

As the temperature is reduced, the bulk resistance due to the increase in
mobility decreases, but it also increases the offset voltage. At 80 K, the offset
voltage is approximately double the room temperature value (1.4 V versus

0.7 V). The reverse recovery time is reduced at lower temperature because
of the lower minority carrier lifetime. The net result is a higher conduction
loss and faster recovery. Even with these changes, the PN junction diodes
functions satisfactorily at 80 K.
26.7.1.5 Schottky barrier diode

As in the PN junction diode, the offset voltage of the Schottky diode
increases as the temperature is lowered. The ratio is about 2:1 from room
temperature to 80 K. The bulk resistance has a greater importance in
Schottky diode and it decreases with temperature. The net result is still an
increase in conduction loss, but not as severe as in the PN junction device.
26.7.1.6 Bipolar transistor

The bipolar transistor operates through the injection of the minority carriers
from the emitter region through the base to the collector region. The current
gain depends on both the injection efficiency of the emitter–base junction
and the lifetime of the injected carriers. Both these phenomena grow less
efficient as the temperature is lowered, so the performance of the transistor
decreases as it is cooled. The present generation of transistors is not
designed to operate at low temperatures. As a result, their performance
becomes relatively poor at 80 K. However, it is possible to design devices
that will operate at such low temperatures. Therefore, such devices could
become available when a significant market of low temperature electronics
develops.
26.7.1.7 Power MOSFETs

The MOSFET operation does not depend on the injection of carriers across a
junction, nor is the minority carrier lifetime a factor. As long as sufficient
majority carriers are present in the source region, drain current will flow
when a bias is present on the drain and a voltage is applied to the gate.
Carrier freezeout in the source and drain regions does not impact device
performance until the conduction electron (or hole) concentration is
reduced to a small percentage of its original value. The main factor
affecting MOSFET operation is the carrier mobility, which increases at low
temperature. As a result, device on resistance is reduced by a factor of 2 to 3.
With such a great improvement, the MOSFET actually works better at 80 K
than at room temperature.
26.7.2 Magnetics
Ferromagnetic metals generally behave poorly at low temperatures,
displaying reduced permeability and drastically increased losses. Nickel
iron in tape, laminations or powder form is much better. The saturation flux
density increases at low temperature in several magnetic materials.

Increased flux density is particularly useful in filter inductors with large
dc currents. For a given induced emf in the magnetic core, the e2/R eddy
loss increases at low temperature because the resistance decreases. The
actual increase in core loss, however, is small. The ohmic loss in the
winding conductor drastically decreases due to decreased resistance. Even
if the windings are not superconducting, the low temperature is still
advantageous. The resistivity of copper drops by a factor of 13 from þ100 C
to 80 K.
Some insulating materials used in magnetics show substantial improve-
ments in dielectric strength. The dielectric strength of 1-mil thick Kapton
film is about 6500 V/mil at room temperature and 11,000 V/mil at 80 K. The
life expectancy also rises dramatically at lower temperatures. High-voltage
power converters can benefit from both these trends. However, insulating
films become brittle at low temperatures, so sharp radius bends must be
avoided and adequate mechanical support provided. On balance, mag-
netics should demonstrate improved performance at 80 K, especially if the
winding is immersed in liquid nitrogen.
26.7.3 Capacitors
Some capacitors are usable at cryogenic temperatures, and some are not.
Film and mica capacitors maintain their capacitance values well at low
temperature. Solid tantalum types also hold up very well down to 4 K, but
wet tantalums are not usable at cryogenic temperatures due to large
capacitance change and very high ESR. Inadequate data are available on the
ESR of film capacitors at low temperatures, except that the dissipation
factor of the dielectric increases and the resistivity of the foil decrease. The
reduced ESR, coupled with the increased thermal conductivity of the
dielectric, translates into a much higher current rating for low-temperature
operation.
26.7.4 Resistors
Resistors are primarily used in the control circuits only, not in the power
trains. Most metallic resistance decreases at low temperature. The
performance of metal film resistors is more than adequate at 80 K. For
precision dividers or similar uses, the bulk metal resistors are generally
good and should be adequate even at 10 K (liquid helium temperature).
Wire-wound resistors also would work well. However, the ubiquitous
carbon composition resistor is not suitable for low-temperature operation. It
has a large temperature coefficient, and the resistance value displays
considerable hysteresis during temperature cycling.
26.7.5 Structural Materials

All materials become brittle at low temperature, including magnetic
stainless steel. Nonmagnetic stainless steel remains relatively ductile at
liquid helium temperature. Insulating materials, such as Kapton, KelF, and
glass–epoxy break at lower elongation. Resins when cooled approach the
glass transition temperature, below which they become brittle.
The thermal conductivity of most materials used in the power converter
construction increases as the temperature is reduced. For example, the
thermal conductivity of silicon increases by a factor of 10 from room
temperature to 8 K. One of the greatest limitations to increasing power
densities is the problem of getting heat out of the electronic parts. The heat
transfer is very efficient if the device is operated in a boiling liquid nitrogen
bath. Operating power converter circuits at 80 K reduces the total power
dissipation and improves the heat transfer. This should allow achieving
much higher power densities with no technical barriers.
To summarize this section, cryogenic power equipment can be designed
to operate at liquid nitrogen temperatures. In many cases the performance
can actually be improved over the room temperature operations. As the use
of 80 K superconductors becomes practical, higher efficiency, and compact
power converters could become possible with performance well beyond the
present state of the art.
References
1. DeWinkel, C.C. and Lamopree, J.D., Storing power for critical loads, IEEE
Spectrum, June 1993, pp. 38–42.
2. Balachandran, U., Super power, progress in developing the new super-
conductors, IEEE Spectrum, July 1997, pp. 18–25.
Chapter 27
Microwave Beam Power Satellite
27.1 Introduction
The traditional power system in space uses a solar array and battery for
each satellite. The concept of using a centralized large power-satellite
(powersat) has been receiving some seed funding at NASA for some time.
In this concept, depicted in Figure 27.1, one high-power satellite would
generate bulk power, which is then transmitted to multiple satellites by
laser or microwave beams. The electric propulsion of spacecraft using beam
power from a powersat was discussed in Section 23.8. The beam power can
also be used for a rescue operation on a remote Earth surface or anywhere
in the ocean. Advantages of the central power satellite transmitting beam
power to multiple user spacecraft are:
Replaces PV array by much smaller power receiving antenna in the user

spacecraft
Eliminates battery by proper orientation with the power satellite(s) that
avoids satellite-to-satellite eclipse
FIGURE 27.1 Power transmission to multiple missions from central power satellite.
653
Extends the mission life where the spacecraft life is solar array or
battery limited
Provides large peak power if and when needed, such as for electric
propulsion, thus significantly decreasing the on-board fuel mass
Orienting the spacecraft to the powersat beam may be simpler than
orienting to the sun
The power generated by the powersat can be transmitted to user satellites in

two ways: by a microwave beam or by a laser beam.
27.2 Microwave Beam

The power transmission efficiency in space by microwave beam is given by
the following expression,1
¼ 1 exp ðAt Ar =d2 2 Þ ð27:1Þ

where
¼ fraction of the transmitted power captured by the receiving antenna

At and Ar ¼ transmitter and receiver antenna area, respectively
d ¼ distance between the transmitter and receiver
¼ wavelength of the microwave power beam
The distance between the powersat and the user satellite may vary.
Microwave transmission may be effectively used over short distances.
The frequencies considered for such systems2 are from 3 GHz to 300 GHz
with the corresponding wavelengths from 100 mm to 1 mm.
In 2003, a NASA funded experiment was designed to transmit power by a
microwave beam from the Goldstone 100-m antenna to a space sail at 800-
km altitude. The Goldstone steerable dish radiates up to 1=2 MW power.
However, the sail, being outside the range, was estimated to receive only
1700 W. The resulting microwave pressure was estimated to accelerate the
sail by 107g and demonstrate the principle of beaming power to a
spacecraft. The acceleration of the sail depends only on power and not on
the frequency of the beam. In another experiment at JPL and the University
of California at Irvine, a 10-kW, 7-GHz microwave beam in a vacuum
chamber produced a power density of 1 kW/cm2 to heat a sail surface to
2000 K.
Microwave transmitters have been under development for much longer
than lasers. They are far more efficient and cost much less to build. They do
not damage the receiving surface as lasers can, and do not refract while
passing through air. However, microwaves require much larger antennae
for the same focusing ability.
Microwave Beam Power Satellite 655
27.3 Laser Beam

Laser beams may be more efficient over distances in the 10,000 to 50,000 km
range.3 The laser system consist of three parts:
First the solar energy collected on the powersat is converted into a laser
beam using a solid-state solar pumped laser. It consists of a crystal
placed in the focus of a parabolic solar concentrator.4
The monochromatic laser radiation is then focused into a beam using an
optical mirror of appropriate size. The wavelength of such laser is equal
to 1.06 mm, and the conversion efficiency 20 to 25%. A focused power
laser beam then transmits the power to the user satellite or a rescue site
on the Earth at 80 to 90% efficiency.
Finally, the received energy is converted into electrical power using the
conventional PV cell. The conversion efficiency of PV cell under laser
illumination is around 50%. The maximum electrical power output of
the silicon PV cell using such systems can be 3000 W/m2 from laser
radiation compared to 300 W/m2 from natural solar radiation. The solar
array requirement on the user satellite is thus greatly reduced, and the
battery can be eliminated by avoiding powersat-to-user satellite
eclipses.
The powersat, once developed and built, can make the user satellite much
lighter and less expensive, so that many can be launched at low cost per
launch. This could open up entirely new kinds of missions in the Earth orbit
and in interplanetary space at much less incremental cost. The concept is
like the 19th century railroad system: once the tracks are laid, the train itself
is a small added expense.
27.4 Space-to-Ground Power

Driven by the environmental and depletion concerns of the fossil energy
sources, both the DoE and NASA have funded space-based generation of
power for ground use. The equivalent mass density of air on Earth with
typical moisture and pollution is 1.5 times that of pure air. Solar radiation
on the normal Earth surface with air mass 1.5 (AM 1.5) is about 1000 W/m2.
With 12-h nights, on average, seasonal sun variations, and overcasts, the
annual average energy yield of a ground based PV system may be around
6 kWh/m2 per day. For a 400 MW power station on the ground using 20%
efficient PV cells would need 40 million square meters of PV array. Storing
sufficient energy to last 5 days without sun would require 60,000 MWh of
energy storage assuming 80% round trip energy efficiency. A space-based
system can reduce the collector area to one quarter and eliminate the need
of energy storage by making the beam power dispatchable on demand.
Performance of the solar array placed on the Earth’s surface versus being
in LEO and GEO is summarized in Table 27.1. It shows that the energy
collection per square meter in space is several fold higher than that on
Earth. Furthermore, it is 50% higher in GEO than in LEO. The conceptual
study made in the past has considered the medium Earth orbit around
10,000-km altitude for such powersats for ground use.
NASA and DoE have explored ambitious proposals for harnessing solar
energy for terrestrial use by deploying solar powersats in geosynchronous
orbits that could generate power 24 hours a day, 365 days a year. One
concept study has shown that one such satellite with 146-km2 solar array
could deliver power equivalent to 10 nuclear power plants on the Earth. Its
1 km2 antenna would transmit power in a sharp 2.4 GHz microwave beam
to an Earth-receiving station. Here, it would be converted into d.c. and
inverted into 10,000-MW 60-Hz a.c. and delivered to the distribution system
of the electrical power utility. The basic converter would use inductors —
perhaps superconducting — to store energy and boost the voltage.
In the proposed concept, the beamed power is converted to 60 or 50 Hz
utility power using high-voltage converters. Massive series–parallel con-
nections of numerous converters would be needed to beam gigawatts of
power to Earth from the outer space.5 The space-to-space power transmis-
sion would be in watts/m2, while the space–to-ground power transmission
would be needed in hundreds of watts/m2. However, U.S. government
regulations limits microwave radiation to 110 W/m2. Therefore, the
hundreds of watts/m2 beam intensity would require a change in the
regulations or special permission in a remote location from where it can be
transmitted to populated load centers. This high-beam intensity is still a
few-fold lower than the natural sun intensity of 1000 W/m2 on the ground.
Therefore, it may not be a cancer risk to humans, as some may fear.
Several innovative concepts are being studied for collecting solar energy
in space and transmitting microwave beams to another spacecraft that may
be orbiting the Earth, or on a planetary surface, or on its way to an
interplanetary mission. NASA’s Solar Space Power Exploratory Research
Table 27.1 The performance of solar array placed on the ground and in
LEO and GEO
Parameter Earth LEO GEO
Air mass 1.5 with average moisture and pollution 0 0

Solar radiation W/m2 1000 in full sun 1350 1350
500 in partial sun
Useful sunlight hours/day 6 on average 16 24
Incident energy Wh/year 1643 7884 11826
Launch and maintenance 0 Low High
cost in orbit
and Technology program is investigating systems at power levels ranging

from 100 kW to 1200 MW. The building blocks of such a system are:
A large sun-oriented solar array that tracks the sun and generates
power at high voltage in the 400 to 1000 V range
Rotating microwave transmitter in space that tracks a receiving antenna
(rectenna)/rectifier station on Earth
Microwave beam at several GHz frequency, using solid-state power
converters, magnetrons or klystrons
A rotary joint between the solar collector and the transmitter
High-voltage cables
d.c. to microwave power converters in space and microwave to d.c.
power converters and d.c. to 50 or 60 Hz a.c. inverters on the ground
A concept study for delivering 1200 MW from GEO orbit to the ground
grid has developed the following estimates:6
Solar array with 4 to 5 concentration ratio with futuristic conversion

efficiency of 39% generating at specific power of 1000 W/kg and 550
W/m2 at 1000 V
A 5.8 GHz microwave beam of Gaussian power density distribution
with 10 dB taper from the transmitter center to the edge
Solid-state transmitters operating at 80 V and the solar array at 1000 V
The power distribution from solar array to the SSPC is at 100 kV, so
converters are needed at both ends of the distribution lines. These
converters substantially add into the system mass. Using 6000 V
magnetrons and reducing the distribution voltage to 6000 V could
eliminate the transmission voltage converters
Rectenna size on Earth about 7450 m in diameter.
Overall wireless power transmitting efficiency around 35%.
A system for 1200 MW power to the grid requires a solar array of

7,300,000 m2 area, amounting to 2700 2700 m2 or about 3000 m diameter
array in space. The rectenna on the ground is estimated to be 44,000,000 m2
area or 7500 m diameter. The ground receiver would be about six times the
solar array area, because of: (a) power losses in various components, and (b)
the beam power density limited under the federal regulation on microwave
power.
The mass of the above concept satellite is estimated to be 22,500 to 30,000
metric tons at launch and 17,000 to 22,000 metric tons in orbit. A great many
technological developments and demonstrations are needed to bring the
cost per kWh energy delivered to the ground competitive with grid
power.7,8
27.5 Solar Dynamic Power Generation

Solar to electrical power conversion in the above schemes can be done using
two primary options:
Thin-film PV cells on a flexible array, and

Solar dynamic system using a concentrator and a Brayton, Stirling or
Rankine engine driving an electrical generator
Between the two options, the solar dynamic using the Brayton cycle offers
the following advantages:
Moderate technological developments

Higher conversion efficiency
No batteries are needed, as the heat can be stored as the latent heat of
working fluids
Minimal performance degradation with age for longer life and low
maintenance
Higher specific power to minimize the launch cost
Scalability to high power up to hundreds of megawatts.
The solar dynamic power system was discussed in Chapter 21. The solar
dynamic concept described there was developed and demonstrated for
space station design during the mid-1980s. The development work then was
centered on a 300-kW, 3-phase, 460-V, 20-kHz power system. At the
hundreds of megawatts power levels of interest here, the proposed
dynamic scheme uses a refractive Fresnel concentrator to collect solar
energy, a heat receiver, a Brayton cycle turbine, and a rotating generator to
generate electrical power. The waste heat is dissipated back to space via a
radiator.
The electrical generator is a 3-phase synchronous generator operating at
high frequency in the range from 400 Hz to 20,000 Hz. The higher the
frequency, the lower the mass of materials needed in the system magnetics,
such as the generator and the transformer. This is because lower magnetic
flux is required at higher frequency for a given voltage. The power can be
generated at a low voltage (200 to 1000 V) and transformed into a high
voltage (5 to 25 kV) for transmission and distribution to the transmitting
antennas. Alternatively, the generation and transmission both can be at the
same voltage around 15 kV.
Table 27.2 Performance projections for the 10-MW solar

dynamic module
Parameter State of the art Near term Far term
System specific power, W/kg 21 116 129

System specific area, W/m2 283 389 450
Brayton cycle efficiency, % 39 42 47
PV efficiency, % 30 33 35
Concentrator area, m2 24,800 22,300 20,800
Concentrator diameter, m 178 168 163
Radiator area, m2 10,135 3200 1400
(Data source: Mason, L.S., Proc. 34th IECEC, p. 2601, SAE, 1999.)
Based on this earlier groundwork, Mason4 of NASA Glenn Research

Center has developed a conceptual design for a 10 MW module for a solar
power satellite. The Mason study has taken into account a projected launch
vehicle capability (< 20,000 kg) and solar concentrator sizes (< 200 m
diameter). Table 27.2 shows the system level performance estimates that
could possibly be achieved in a couple of decades. It assumes certain new
enabling technologies identified in Table 27.3 that could be available
between 2015 and 2020. A 1.6-GW power generation can perhaps be met
using 160 modules in a 40 4 configuration. Mason estimates that such a
station would cover 9.5 km 0.65 km area with mass of 4.5 million kg.
Lynch9 as proposed an alternate approach of using sun tower architec-
ture for PV power generation as shown in Figure 27.2. It requires 100 kV
transmission. The antenna in the figure is the microwave transmitting
antenna.
Table 27.3 Solar dynamic component advances
Component State of the art Near term Far term
Concentrator Rigid panel, Rigid, Inflatable,

Al coating, 88% Ag coating, Ag coating,
reflectance 92% reflectance 92% reflectance
10 kg/m2 2 kg/m2 0.2 kg/m2
Power system 500 V, 1 kHz, 5 kV, 1 kHz, 10 kV, 1 kHz,
3-ph generator, 3-ph generator, 3-ph generator,
60 C Si electronics, 150 C Si electronics, 300 C Si electronics,
10-kV distribution, 10-kV distribution, 10-kV distribution,
100-m cable, 100-m cable, 100-m cable,
93% efficiency, 93% efficiency, 95% efficiency,
2 kg/kW 1 kg/kW 0.2 kg/kW
(Data source: Mason, L.S., Proc. 34th IECEC, p. 2601, SAE, 1999.)
FIGURE 27.2 Solar PV power generation using the sun tower concept.
27.6 Powersat Developments

The proponents of space-based solar power seek a big science response to a
problem that needs a more subtle solution. However, generating such
power with gigantic panels in space, transmitting by microwave or laser
beam, and distributing on Earth can be expensive and perhaps unsafe at
present. The on-going research may find answers, but may take decades.
However, the concept has captured the imagination of many engineers
around the world. The studies include multi-megawatts solar powersats
and microwave wireless power beaming to other spacecraft or to the Earth.
On the other hand, technological implementation of the concept is
exceptionally challenging. NASA estimates that tens of billions of dollars
may be needed before transmitting the first watt of power from a GEO
based powersat. The key technologies required before such concept can be
materialized are:
Low-cost thin film PV cells, concentrators, or high-temperature solar

dynamic power generation
Wireless power transmission by high-frequency power beams
High-voltage, high-current, high-frequency power converters
Structural, assembly, and thermal aspects of the massive station design
Space transportation to the powersat and robotic assembly, main-
tenance, and servicing
Resolution of the environmental, regulatory, and safety concerns

Resolving the complex system integration issues
For the powersat beam power from space to ground to be economical, the
cost at the 60 Hz side of the distribution station should be below 5 cents per
kWh. For this to happen, the following performance would be required:
PV efficiency of 45% with specific power of 1 kW/kg (about 40 times the

presents state of the art)
Thin film PV with vapor deposition on polymer substrate
Potential PV candidates are CuInGaSe2, CuInS2, and CdTe, all of which
work at voltages around 300 to 500 V
High-temperature (400 C) polymer substrate or low-temperature
deposition method.
d.c.-to microwave beam-to a.c. conversion efficiency of 80%.
Launch cost comparable with that of the airplane transportation at
present.
All studies at present conclude that such powersats could not be cost
effective in the near future, but may become economically viable in a few
decades.
References
1. Lineberry, J.T. and Chapman, J.N., MHD Augmentation of rocket engine
for space propulsion, in Proceedings of the 35th Intersociety Energy Conversion
Engineering Conference, AIAA, 2000, Paper No. 3056.
2. Koert, P. and Cha, J.T., Millimeter wave technology for spec power
beaming, IEEE Transactions on Microwave Theory and Technology, Vol. 40,
No. 6, June 1992, pp. 1251–1258.
3. Grechnev, A.B. et al. Centralized power as basis of new philosophy of
space power engineering, in Proceedings of the 34th Intersociety Energy
Conversion Engineering Conference, SAE, 1999, Paper No. 2436.
4. Mason, L.S., A solar dynamic power option for space solar power, in
SAE, 1999, Paper No. 2601.
5. Kusic, G., Conversion of beamed microwave power, in Proceedings of the
35th Intersociety Energy Conversion Engineering Conference, AIAA, 2000,
Paper No. 3071.
6. SAIC and Futron Corporation, Space Solar Power Concept Definition Study,
NASA Report No. SAIC-99/1016, February 1999.
7. Mankins, J.C. and Howell, J., Overview of the space solar power
exploratory research and technology program, in Proceedings of the 35th
No. 3060.
8. Carrington, C. et al. The abacus/reflector and integrated symmetrical

concentrator concept for space power collection and transmission, in
AIAA, 2000, Paper No. 3067.
9. Lynch, T.H., Sun tower PMAD architecture, in Proceedings of the 34th
2441.
Index
A altitude, 111
aluminum, conductor material, 298–299
abbreviations, xiii–xviii AM0, see Air mass zero (AM0)
acceleration, xxii ambient functional performance test,
acceptance integration and testing, 455 456
acoustic integration and testing, 457–458 American National Standards Institute
acoustic tests, 460 (ANSI), 146
acronyms, xiii–xviii American Superconductors, 642
active unit redundancies, 440–441, 443 American wire gage (AWG), 293
ADE, see Array drive electronics (ADE) amorphous silicon (PV cell), 180
ADEOS-11 (Japan), 36 amorphous technology, 140–141
advanced NiCd battery, 199–200 ampacity of wires, 293–294
Advanced PV and Electronics Ampere-hour ratings, 234
Experiment (APEX), 187 Ampere’s law, 395
Advanced Radioisotope Power System AMTEC, see Alkaline metal thermal
(ARPS), 501 to electric converter (AMTEC)
advanced radioisotope thermoelectric system
generator, 502 analog control, control circuit, 69
AE, see Autumnal equinox (AE) analog-digital hybrid control, control
Aerospace Corporation, 170 circuit, 70
aging effect, 497–498, 497–498 ancillary components
AgO-Zn, see Silver zinc (AgO-Zn) basics, 319
battery battery cell voltage monitor, 328–329
AgZn battery, 41 battery letdown unit, 330, 330
AH counter and sensor, 353, 354 battery pressure monitor, 328–329
Air Force Research Laboratory, 555 BCVM, 328–329
air mass zero (AM0), 137 capacitors, 332–333
albedo ceramic capacitor, 333
launch and transfer orbit current meters, 330–332
environment, 26 deployment components, 322–325,
power, 147 325
spacecraft level trades, 112 electro-explosive device, 322–323
thermal control system, 5 electromagnetic interference shield,
alkaline metal thermal to electric 340–344, 341–343
converter (AMTEC) system, 52, electronic packaging, 336–337
502–503 filters, 333–334, 333–334
Allen studies, 555–556, 601 Hall-effect transducer, 331, 331
A2100 (Lockheed Martin Space laser initiated deployment, 323,
Systems), 84, 86 323–324
alternating current vs. direct current, magnetic amplifier, 331, 331–332
537–538, 538–539 metalized polypropylene capacitor,
alternator, see Dynamic system with 332
alternator metallic resistance wire, 326
663
664 Index
ancillary components (continued) B

radiation shield, 337–340, 338–339
relays, 327–328, 328–329 back-surface illumination, 147–148
shape memory metal, 324 backward analysis, 127
shunt, 330–331 baffles, 15
solar array drive, 319–322, 321–322 Bailey, Hepp and Raffaelle studies, 579
tantalum capacitors, 332 Ballard Power Systems, 602
telemetry and commands, 334–336, Ballistic Defense Organization, 187
335 Banks studies, 101
thermal controller, 326–327 battery
thermistor, 326–327, 327 advances, 251–253, 253
thermocouple, 326 ascent power, 234–235
ANSI, see American National Standards assembly, 463, 463–464
Institute (ANSI) basics, 195
AO, see Atomic oxygen (AO) burst power capability, 227–228
APEX, see Advanced PV and Electronics charge/discharge, 60, 214–217,
Experiment (APEX) 215–217
apogee kick motor, 8, 16 charge efficiency, 217–218, 219
Apollo-11, 478 charge regulation, 240–243, 241
architecture trades, 78, 78–79 cycle life, 224–227, 226–227, 248–251,
arcing, 94, 422, 550 250
arc period, 311 design, 228–234, 233
area, conversion factors, xx dissipation, 351
ARFL, 635 dynamic model, 245–248, 246–247
Ariane, 26 electrical circuit model, 213–214, 214
ARPS, see Advanced Radioisotope electrochemical cell, 195–197, 196
Power System (ARPS) energy efficiency, 218–219
array construction, 148–153 flat pack, 237–238
array design, 166–178, 167 flywheel energy storage, 624
array drive electronics (ADE), 319, see fully-regulated bus, 64
also Solar array drive (SAD) health monitoring, 363–364, 364
array life, 461 integration and testing, 463
Arrhenius equation, 248 internal impedance, 217–218
ascent latch-up, 365, 365–367
flywheel energy storage, 623 launch power, 234–235
power, battery, 234–235 lithium-ion, 209–212
power system functions, 131 lithium polymer, 212, 213
simulation, 362–363, 363 management, 243–245
a-Si, see Amorphous silicon (PV cell) multiple charge rates, 242
atmosphere, 27 nickel cadmium, 198–200
atomic oxygen (AO) nickel hydrogen, 200–208, 201–202,
high-power high-voltage systems, 204
529, 529–530 nickel metal hydride, 208–209
mass erosion, 99–102 performance comparison, 228,
on-orbit environment, 29–30 228–231
Attia studies, 96 photovoltaic-battery system, 59–60, 60
attitude and orbit control system, 4 primary battery, 41–42, 185, 251
aurora, 31 properties and performance, 214–224
autumnal equinox (AE) propulsion system, 6
eclipse, 17 rechargeable, 195
operational orbit, 16, 19 replacement, 624, 624–625
solar array sizing, 122 safety considerations, 238–240
AWG, see American wire gage secondary, 195
(AWG) self-discharge, 219–220, 221
Index 665
self-healing, 221–224, 222–224 test points, 466

silver zinc, 213 body mounted array, 150, 151
single charge rate, 242 body stabilization, see Three-axis
sizing, 124 stabilization
solar PV-battery, 43–45 Boeing, 509
thermal design, 235–238, 236–237 BOL, see Beginning of life (BOL)
trades, 117 Boltzmann constant, xxi, 142
types, 197–213 bonding, electromagnetic interference,
unregulated charging, 243 410–411
vertical stack, 238, 238 Bonebright, Stern and, studies, 190
V-T characteristics, 219, 220 Bonneville Power Administration, 642
wine rack, 238, 239 booms, 5
battery, health monitoring, 363–364 boost converter, 97, 270, 272
battery bus, 66 Bouquet and Winslow studies, 534
battery cells, life test, 458 BPM, see Battery Pressure Monitor
battery cell voltage monitor (BCVM), (BPM); Battery pressure monitor
328–329 (BPM)
battery charge (buck) converter, Brandhorst and Chen studies, 480
266–270, 267–268 Brandhorst studies, 582
battery charge control mode, 67 Brayton cycle
battery charge/discharge, 442, 442–443 dynamic system with alternator, 505,
battery discharge (boost) converter, 270, 510, 510
270–271 multi-megawatt burst-power, 557,
battery discharge diode, 70 559
battery letdown unit (BLU), 199, 330, nuclear or chemical-dynamic power
330 system, 49
battery pressure monitor (BPM), solar concentrator-dynamic power
328–329 system, 47
BCVM, see Battery cell voltage monitor solar dynamic power, 658
(BCVM) Brill, Smith and, studies, 227
Becquerel studies, 135 British Aerospace, 86
beginning of life (BOL) British wire gage (BWG), 293
back-surface illumination, 148 broadband electromagnetic interference,
degradation factors, 169 412–413
offsetting from sun, 176 broadcast communications, 7
on-orbit simulation, 357–358 Bruce and Marcoux studies, 212
peak power tracking controller, 76 brushes exciter, 512–513, 514, 516
power system level trades, 114–115 BSS-702 bus, 83, 190
radiation damage, 174 buck-boost converter, 271–272, 271–272
shunt circuit design, 264 buck converter
solar array sizing, 123 flyback converter, 272
solar PV-battery, 44 inductor coupled buck converter, 274
beta angle, 23–24 power electronics damage, 97
biaxial-flexure test, 139 burst power capability, 227–228
biaxial gimbals, 322 bus
bipolar junction transistor (BJT), 255, basics, 3
650 battery, photovoltaic-battery system,
BJT, see Bipolar junction transistor (BJT) 66
black body temperature, effective, xxii fully regulated, photovoltaic-battery
bleeding load resistor, 273 system, 63, 63–67, 71–73
BLU, see Battery letdown unit (BLU) impedance, 368, 368–371, 370–371,
Bode plot 382–385, 383–386
bus impedance and stability, 464 integration and testing, 464
control system stability, 386 ISS 160 to 120-V bus, 79–82, 80–81
666 Index
bus (continued) CFR, see Code of Federal Regulations

large communications satellite busses, (CFR)
83–86 Chapman, Lindeberry and, studies, 582
micro-satellite, photovoltaic-battery charge cut-back mode, 66
system, 89 charge/discharge (C/D)
peak power tracking, photovoltaic- basics, 214–217, 215–217
battery system, 73–77, 74–78 battery design, 231
photovoltaic-battery system, 86–89 cycle life model, 248
small satellite, 86–89 end of life, 225
sun-regulated bus, 60, 70–73, 71, 365 NiH2 battery, 234
unregulated, 70–71 regulators, 391
70-V, 84–86, 85 self-heating battery, 222–223
100-V, 83, 83–84 charged particles, 30–31, 529–534,
50-V or less, 86, 87 530–531, 533
bus module, 350, 355 charge efficiency, 217–218, 219
bus ripple filter design, 265–266, charge mode, 69
265–266 charge regulation, 240–243, 241
Bussmann fuses, 307 chemical-dynamic power system, 49,
bus voltage control, 67–69, 68 559
bus voltage controller, 65 chemical fuel, 40
bus voltage regulator (BVR), 393 Chen, Brandhorst and, studies, 480
bus voltage trades, 117–118, 118 Chen studies, 626
BVR, see Bus voltage regulator (BVR) Choi studies, 481, 500
bypass diode, 70 CIGS, see Copper indium gallium
diselenide (CIGS)
cladding, 563
C cleanliness, single-crystal silicon, 139
CMOS, see Complementary metal oxide
cable, 20 kHz, 541–543, 542–543 semiconductor (CMOS)
cable coupling, 397–401, 398–402 coatings, thermal control system, 5
cable shield, 299, 409–410, 410 Code of Federal Regulations (CFR),
cadmium telluride (CdTe), 140 414–416
calibrated step shunt, 263, 264 Comet/Meteor spacecraft, 187
California Institute of Technology, 291 commercial equipment, 414–417, 416
capacitor commercial fuel cells, 602–603
cryogenic behavior, 651 commercial off-the-shelf (COTS)
high-power high-voltage systems, 548 equipment, 413–417
metalized polypropylene, 332–333 common-mode electromagnetic
Carnot cycle and efficiency interference (common-mode EMI),
dynamic system with alternator, 506 37, 411, 411–412
fuel cell, 43, 591 common pressure vessel (CPV)
radioisotope thermoelectric battery trades, 117
generator, 490 nickel hydrogen battery, 203–206,
solar concentrator-dynamic power 205–206
system, 47 temperature, controlling, 237
thermionic power system, 51 communications and data handling, 3–4
carrier mobility, 649 communications loads, 391
Cassini, 486, 500 communications satellites, 11
C/D, see Charge/discharge (C/D) complementary metal oxide
CdTe, see Cadmium telluride (CdTe) semiconductor (CMOS), 382, 429
CEASE sensor, 103 component models, 391, 392
cells, 139, 224, 363 computer program structure, 350–356,
ceramic capacitor, 333 352
Ceraola studies, 246 computers, voltage transients, 373
Index 667
concentrator array, 183–193, 184–185 c-Si cell, see Crystalline silicon (PV cell)
conducted EMI, 395 cell
conductor materials, 297–299, 298 Cuk capacitors, 549
connectors, distribution harness, Cuk converters, 97
299–301, 300 current, power flow analysis, 126
constants, space flight, xxii–xxiii current limiting fuses, 311
constant torque vs. constant power, 630, current meters, 330–332
630 current regression method, peak power
continuity checks, 460 extraction, 164
control circuit, photovoltaic-battery current regulators, 277–279, 278
system, 69–70 current vs. voltage (I-V)
control electronics, 5–6 battery latch-up, 365
control system, stability, 386–388, photovoltaic-battery system, 58
387–389 power regulation, 60
conversion factors, xx–xxi shadow effect, 165
cooler, thermoelectric, 503 solar array, 143–147, 144–146, 389, 461
copper, conductor material, 297–299 solar cell test, 458–459
copper indium gallium diselenide sun intensity, 154
(CIGS), 140 cut-off wavelength, 137
corona CV-1147 silicone, 101
degradation, 526–529, 527–528 cycle life
high-power high-voltage systems, basics, 224–227, 226–227
526–529, 527–528 electrochemical cell, 197
power system tests, 460 model, 248–251, 250
cosmic radiation, 31 cycloconverter, 515
cost Czochralski process, 138
battery design, 233
concentrator array, 184–185
conductor materials, 297 D
GaAs, 168
multi-junction cells, 141 Daimler-Benz, 602
multipliers, Rule of Ten, 129 Daimler-Benz Aerospace, 86
plutonium-238, 233 damage, power electronics, 95–98,
power system options, 41 97–98, 102–103, 103
Si cells, 168 damping, three-axis stabilization, 14
solar concentrator-dynamic power DC-93-500 silicone, 101
system, 48 d.c to d.c converter unit (DDCU), 81
solar PV-power system, 47 deactivation, 131
solar thermal power system, 47 dead band, 69
spacecraft level trades, 111 debris
sun-regulated bus, 72 electric propulsion, 589
unregulated charging, 243 electrostatic discharge, 423
COTS, see Commercial off-the-shelf impacts, environmental effects, 102
(COTS) equipment near-Earth space environment, 28–29,
coupling modes, 397–401, 398–402 29
CPV, see Common pressure vessel spacecraft level trades, 112
(CPV) Deepspace 1, 188, 571, 574
critical parameters, superconductors, deep space missions, 486–488, 487, see
643–645, 644–645 also Interplanetary and deep space
Crofton and Francis studies, 94 missions
Croydon and Reynolds, Sack, studies, degradation
243 corona, 526–529, 527–528
cryogenic temperature, 648–652 solar array, 123, 169, 170
crystalline silicon (PV cell) cell, 180 solar cell tests, 459
668 Index
density, conversion factors, xx power flow analysis, 125–126, 125–127

deorbiting, factory-to-orbit events, 131 power systems level trades, 113–119,
Department of Defense (DoD) 114
advanced NiCd battery, 199 PV cell trades, 114–116, 115
geostationary orbit, 9 pyro power trades, 118–119, 119
nuclear threat, 37 solar array, 116, 122–124, 123
power systems comparison, 53 spacecraft level trades, 111–113, 112
superconductors, 645 trades, 113–119, 114, 119
thermionic power system, 52 worst-case error margin, 127–129
Department of Energy (DoE) design process phases, 129, 130
space-to-ground power, 655 de-spinning operation, 16
Stirling engine, 509 DET, see Direct energy transfer (DET)
superconductors, 642, 645 dielectric breakdown, 423–424, 424
dependent pressure vessel (DPV), 117, dielectric materials, 94
206–208, 207 dielectric stress concentration, 525–526,
deployment 526
circuits, 5 digital control, control circuit, 69–70
controller, 324–325, 325 diode o-ring, 443, 443–444
devices, 322–324 diode tests, 462
factory-to-orbit events, 131 DIPS, see Dynamic isotopes power
rate, 5 system (DIPS)
depth of discharge (DOD) direct current, fuel cell, 43
advanced NiCd battery, 200 direct current machine, 628–629
battery, 124, 197 direct drive thrusters, 578
cycle life, 225–226, 248–249 direct energy transfer (DET)
electrochemical cell, 196 battery design, 233
energy depletion constraints, 622 peak power tracking controller, 76, 79
flywheel energy storage, 604–605 power system architectures, 62–63
ISS, 82 discharge mode, 66, 68–69
launch and ascent simulation, 362 displacement damage, 96
lithium-ion battery, 211–212 distribution harness and protection
lithium polymer battery, 212 ampacity of wires, 293–294
memory effect, 198 arc period, 311
nickel hydrogen battery, 201 basics, 3, 293, 293
NiMH battery, 208 battery packaging, 296
on-orbit simulation, 358 cable shield, 299
performance comparison, 228 conductor materials, 297–299, 298
power flow analysis, 125 connectors, 299–301, 300
regenerative fuel cell, 601 derating fuses, 314–315
self-heating battery, 222–223 design process, 295, 304–306, 305
small satellite bus, 88 early fault detection, 318
derating, 447–448, see also Reliability and flexible harness, 306
derating fuse characteristics, 312–314, 313–314
design analyses list, 127, 128 fuse derating, 314–315
design failure, 431 fuse protection, 306–317, 307–308
design process and trades fuse rating, 309–311, 310–311
basics, 111 fuse selection criteria, 315
battery, 117, 124 fuse testing, 316–317
bus voltage trades, 117–118, 118 fuse types, 311, 312
design analyses list, 127, 128 mass minimization, 301–304, 302–303
design process phases, 129, 130 parallel fuses, 316
factory-to-orbit events, 130–131 pre-arc period, 310–311
functions over life, 131–132 protection of fuses, 306–317, 307–308
load power profile, 119–122, 120–122 rating of fuses, 309–311, 310–311
Index 669
redundant fuses, 315–316, 316 electrical generator technologies,

remote power controllers, 317–318 512–517, 513
R-L-C parameters, 294–297, 295–297 electromechanical energy conversion,
sun-regulated bus, 72 510–512
unfused loads, 317 field excitation, 515–516
wire ampacity, 294 frequency limitations, 514, 516–517
wire insulation, 299 generators, synchronous, 512, 514,
diurnal (daily) variation, xxiii 515–516
DOD, see Depth of discharge (DOD) induction homopolar generator,
DoD, see Department of Defense (DoD) 513–514, 514, 516
DoE, see Department of Energy (DoE) overload capability, 515, 516
dopant atoms ionized, 649 parallel operations, 514, 515
dormant unit redundancies, 440–441 PM rotor, 512, 514, 515
dose, charged particles, 30–31 selection, 515
dose-rate damage, 96 solid rotor, 512, 514, 516
dosimeter, 92 speed limitations, 514, 516–517
DPV, see Dependent pressure vessel Stirling engine, 508, 508
(DPV) synchronous generators, 512, 514,
Draper Laboratory, 635 515–516
Dravid, Truong, Wolff and, studies, 634 thermodynamics, 508–510
dual-axis gimbals, 112 voltage regulation, 515–516
dynamic alternator energy conversion, wound rotor, 512–513, 514, 516
40
dynamic isotopes power system (DIPS),
48, 507 E
dynamic performance and stability
basics, 368 early fault detection, 318
bus impedance, 368, 368–371, Earnshaw’s theorem, 631
370–371, 382–385, 383–386 Earth annual mean albedo, xxii
component models, 391, 392 Earth equatorial radius (mean), xxii
control system, 386–388, 387–389 Earth observing Spacecraft (EOS-AM)
control system, stability, 386–388, electromagnetic compatibility, 407
387–389 electromagnetic interference, 407
high frequency ripples, 374, 374–376, energy-momentum wheel example,
375 622
major fuse blow transient, 380–382, flexible array, 152
380–383 flywheel energy storage, 615, 625
minor fuse blow cross-talk, 378–380, power flow analysis, 125
379 power system level trades, 115
quality of power, 378 shape memory metal, 324
ripple measurement, 376–377, Earth orbit classification, 6–12, 8
376–377 Earth’s effective black body
simulation, 388–393 temperature, xxii
solar array model, 389–391, 390 Earth’s long wavelength radiation, xxii
system stiffness, 368, 368–371, Earth-sun distance (AU), xxii
370–371 Eaton Cutler company, 546
voltage regulation and transients, 68, eclipse
371–373, 372–373 battery discharge (boost) converter,
dynamic system with alternator 270
basics, 505–506, 505–508 ISS, 79
Brayton cycle, 510, 510 peak power tracking bus, 73
brushes exciter, 512–513, 514, 516 power flow analysis, 125
comparison of machines, 514, satellites, 16–21, 18–21
514–515 solar array sizing, 122
670 Index
eclipse (continued) electrochemical cell, 195–197, 196

solar PV-battery, 44 electrochemical fuel, 40
spacecraft level trades, 112 electrochemistry, fuel cell power, 592,
stability and bus impedance, 384 592–593
sun-regulated bus, 70 electro-explosive device (EED)
eddy current, 633 basics, 322–323
EED, see Electro-explosive device (EED) pyro power trades, 118
effective black body temperature, xxii 70-V bus, 85
EHD, see Electro-hydro dynamic (EHD) electro-hydro dynamic (EHD)
propulsion propulsion, 575
EHT, see Electrothermal hydrazine electromagnetic compatibility (EMC)
thruster (EHT) COTS equipment, 413
electrical circuit model, 213–214, 214 Directive 89/336/EEC, 414–415
electrical generator technologies, fuses, 381
512–517, 513 integration and testing, 458
electrical machines, 610, 625–629 power system test, 460
electrical model, radioisotope electromagnetic interference (EMI)
thermoelectric generator, 495–496, ac current, 537
496 basics, 394
electrical power system (EPS), see also bonding, 410–411
Power system broadband, 412–413
computer program structure, 350 cable shielding, 409–410, 410
integration and testing, 465–466 commercial equipment, 414–417, 416
power flow analysis, 127 common mode, 411, 411–412
power system level trades, 113 conducted electromagnetic
satellite system, 5 interference, 395
spacecraft level trades, 111 control system stability, 386
spacecraft mass, 24 COTS equipment, 413–417
sun tracking, 161 coupling modes, 397–401, 398–402
telemetry and commands, 335–336 dc current, 537
Electricité de France, 643 EMP/nuclear threat, 417–420, 418, 420
Electric Power Research Institute, 561 flyback converter, 273
electric propulsion fuses, 381
basics, 566, 566–567 grounding, 408–409, 408–409
electrothermal hydrazine thruster, high-power high-voltage systems,
570, 571 542–543, 543
hydrazine arcjet, 570–571, 572 high-voltage direct current, 536
International Space Station, 589–590 integration and testing, 458
ion thruster, 571–574, 572 NASA test methods, 416
magneto-hydrodynamic type, power system test, 460
575–576 radiated electromagnetic interference,
microwave beam type, 582–583 395–397, 396
nuclear power type, 581–582 shape memory metal, 394
orbit transfer vehicle, 589 shield, 340–344, 341–343
outer planet exploration, 590 sources, 394–397
performance comparison, 576–578, 577 space compliance, 416–417
pulsed plasma thruster, 576 specifications, 403–407, 404–406
solar PV type, 578–579, 578–580 suppression methods, 407–411
solar thermal type, 580, 580–581 thermistor, 327
space debris cleanup, 589 twisting wires, 408
specific impulse, 568–569, 570 electromagnetic pulse (EMP), 417–420,
stationary plasma thruster, 574–575 418, 420, 553
tether power type, 583–590, 584, 587 electromechanical energy conversion,
types, 570–576 510–512
Index 671
electromotive force (EMF), 585 battery, 363–367, 364–365

electron fluence, 169 bus module, 350, 355
electronic packaging, 336–337 computer program structure, 350–356,
electrostatic discharge (ESD) 352
basics, 421, 421 health monitoring, battery, 364
body mounted array, 150 input variables, 355
dielectric breakdown, 423–424, 424 launch simulation, 362–363, 363
effects, 425–426 on-orbit simulation, 357–358
entry points, 424–425 output format, 355–356, 357
environmental effects, 93–95, 94–95 program verification, 356, 358
GEO, 421–422 simulation runs, 356–363
indirect interaction, 425 solar array module, 354
integration and testing, 458 sun-regulated bus, 365, 365–367
LEO, 422–423, 423 transfer orbit load budget, 359,
mitigation, 426–427 361–362
part sensitivity, 429, 429–430 transfer orbit simulation, 358–361,
puncture strength, 425 359
rigid panels, 148 energy density, 197
safety during tests, 430, 430 energy depletion constraints, 620–622,
solar array, 427–428, 428 621
solid insulation, 425 energy efficiency, 218–219
strike though gaps, 425 energy-momentum wheel (EMW), 616,
70-V bus, 84 622–625
electrothermal hydrazine thruster entry points, electrostatic discharge,
(EHT), 570, 571, 576 424–425
ELF, see Extremely (extra)-low environmental effects
frequency (ELF) atomic oxygen, 99–102
EMC, see Electromagnetic compatibility basics, 92
(EMC) damage, 95–98, 97–98, 102–103, 103
EMF, see Electromotive force (EMF) debris impacts, 102
EMI, see Electromagnetic interference electrostatic discharge, 93–95, 94–95
(EMI) mass erosion, 99–102
EMP, see Electromagnetic pulse (EMP) micrometeoroid and debris impacts,
EMW, see Energy-momentum wheel 102
(EMW) power electronics, 95–98, 97–98
end of life (EOL) solar array degradation, 45–46, 92–93,
battery, 225 93
degradation factors, 169 EOL, see End of life (EOL)
on-orbit simulation, 358 EOS-AM, see Earth observing Spacecraft
peak power tracking controller, 76 (EOS-AM)
power system level trades, 115 EPS, see Electrical power system (EPS)
radiation damage, 174 Epstein constant, 281, see also Steinmetz
solar array, 58 exponent
solar array degradation, 93 Equatorial radius (mean), Earth, xxii
solar array sizing, 122 equinox (EQX), 16, 22, 72
energy, conversion factors, xx equivalent electrical circuit, 141,
energy and momentum relations, 141–143, 143
619–620, 620 equivalent series resistance (ESR),
energy and stress relations, 613–614, 265–266
614–615 EQX, see Equinox (EQX)
energy balance and power management ESA, see European Space Agency (ESA)
analysis, 349 ESD, see Electrostatic discharge (ESD)
ascent simulation, 362–363, 363 ESR, see Equivalent series resistance
basics, 347–349, 348 (ESR)
672 Index
European Space Agency (ESA) flyback converter (buck or boost), 272,

Cassini, 486 272–273
flexible harness, 306 flywheel energy storage
Mercury mission, 479 advances, 253
Explorer, 500 application example, 615–616,
extremely (extra)-low frequency (ELF), 615–617, 618
585–586 ascent power, 623
basics, 604–606
battery replacement, 624, 624–625
F buck-boost converter, 272
components, 607–610
factory-to-orbit events, 130–131 constant torque vs. constant power,
failure mode and effect analysis 630, 630
(FMECA), 449 direct current machine, 628–629
failure rate statistics, 444 electrical machine, 610
Faraday’s constant and law electrical machines, 625–629
battery charge (buck) converter, 268 energy and momentum relations,
flexible harness, 306 619–620, 620
magnetics, 281 energy and stress relations, 613–614,
self-heating battery, 221 614–615
far field, radiated EMI, 397 energy depletion constraints, 620–622,
fast-blow fuses, 311 621
fault detection, 318 energy-momentum wheel example,
FCC, see Code of Federal Regulations 616, 622–625
(CFR) flywheel system controller, 611,
feedback control loop, 371–373 633–634, 634–635
Fellner studies, 211 high-speed bearings, 609, 632–633
Ferguson studies, 427 hub, 608
Fermi level, 136 induction machine, 627
FESS, see Flywheel energy storage integrated energy and momentum
system (FESS) storage, 617–622, 619, 624, 624–625
FET, see Field effect transistor (FET) ISS, 82
field effect transistor (FET), 429 launch power, 623
field excitation, 515–516 low-speed bearings, 632
field programmable gate array (FPGA), magnetic bearings, 608–609, 609,
70 631–633
filters, 333–334, 333–334 momentum and energy relations,
flash light go/no go tests, 460 619–620, 620
flat pack, 237–238 momentum management electronics,
flexible array, 151–153 612
flexible harness, 306 motor-generation design issues,
flexure test, 139 629–630
flight constants, space, xxii–xxiii NASA flywheel program, 635–638,
flight design analysis, 455 636–637
flight history, 500, 500–502 nuclear or chemical-dynamic power
fluence system, 49
calculation, solar array, 175–176, 177 photovoltaic-flywheel power system,
charged particles, 30 606–607, 606–607
meteoroids and debris, 29 power flow analysis, 125
near-Earth space environment, 37–39, power management electronics, 612
38 reaction wheel replacement, 624,
Van Allen radiation belts, 33 624–625
flux, 30 rotor rim, 608, 608
flux density, 30 sensors, 610, 611–612
Index 673
stress and energy relations, 613–614, remote power controllers, 317–318

614–615 selection criteria, 315
synchronous machines, 626–627, 628 testing, 316–317
touchdown mechanical bearings, 610, transient, 380–382, 380–383
610 types, 311, 312
flywheel energy storage system (FESS), unfused loads, 317
636 fuzzy logic, 353, 364
flywheel system controller, 611,
633–634, 634–635
FMECA, see Failure mode and effect G
analysis (FMECA)
foil (film) tantalum capacitor, 332 GaAs, see Gallium arsenide (GaAs)
force, conversion factors, xx GaAs/Ge cells, 143, 170
forward analysis, 127 GaInP/GaAs cells, 141, 187
Fourier series, 375 GaInP2/GaAs/Ge cells, 179
FPGA, see Field programmable gate Galileo, 484–485, 490, 500
array (FPGA) gallium arsenide (GaAs)
Francis, Crofton and, studies, 94 concentrator array, 183
free wheeling diode, 273, 277 cost, 168
Freon, 550 current regression peak power
frequency limitations, 514, 516–517 extraction, 164
Fresnel reflection, 147, 658 ISS, 82
Frohlich studies, 22 power regression peak power
Fuel Cell Commercialization Group, 602 extraction, 165
fuel cell power PV technologies, 137
basics, 591 radiation damage, 170
commercial, 602–603 solar array, 140
electrical performance, 593–596, thermo-photovoltaic (TPV) power
594–596 scheme, 51
electrochemistry, 592, 592–593 Ganymede (Jupiter moon), 484
power system options, 40, 42, 42–43 gate-assisted turnoff thyristor (GATT),
regenerative type, 598–599 548
satellites, 600–601, 601–602 GATT, see Gate-assisted turnoff
space colonies, 599–600, 600 thyristor (GATT)
types, 596–597, 598 Gaussian distribution function, 434
fuel movement, 15 GEC-Alsthom, 642–643
full shunts, 258–259, 258–259, 263 Gefert, Kerslake and, studies, 578–579
fully regulated bus, 63, 63–67, 71–73 General Atomic, 52
functional tests, solar array, 462 General Electric company, 291, 642
functions over life, 131–132 generators, synchronous, 512, 514,
fuses 515–516
arc period, 311 Genicom corporation, 327
blow speed, 370 GEO, see Geosynchronous Earth orbit
characteristics, 312–314, 313–314 (GEO)
crosstalk, 378–380, 379 geomagnetic storm, 36–37
derating, 314–315 Georgia Institute of Technology, 179
major blow transient, 380–382, geostationary orbit, 8–9, 10, 27
380–383 geosynchronous Earth orbit (GEO)
minor blow cross-talk, 378–380, 379 albedo, 147
parallel fuses, 315–316, 316 array design, 167
pre-arc period, 310–311 basics, 7–11, 10
protection, 306–317, 307–308 battery, 117, 231
rating, 309–311, 310–311 battery charge (buck) converter, 268
redundant fuses, 315–316 bus voltage trades, 118
674 Index
geosynchronous Earth orbit (GEO) system voltage options, 54

(continued) transformer and inductor, 549
charge/discharge, 215 Global Positioning System (GPS)
charge rate, 348 basics, 6, 7
cycle life, 225, 227 cable shield, 299
electric propulsion, 567 flexible harness, 306
electromagnetic pulse, 420 highly elliptical orbit, 11
electrostatic discharge, 94, 421–423, power system level trades, 115
427 propulsion system, 6
fuel cell power, 598–599 wire insulation, 299
fully regulated bus, 71 Gluck studies, 165
harness mass optimization, 304 Goodard Space Flight Center
high-voltage PV array, 519 (GFC-NASA), 576
hydrazine arcjet, 570 Govar and Squires studies, 240
lithium-ion battery, 211, 253 GPS, see Global Positioning System
lithium polymer battery, 212 (GPS)
load power profile, 121 gravity, lack of, 27
micro-satellite bus, 89 gravity gradient stabilization method,
nuclear threat, 420 13
offsetting from sun, 176 GRC, see Glenn Research Center
plasma and charged particles, (GRC-NASA)
530–531 grounding, EMI, 408–409, 408–409
powersat developments, 660 ground power cord, 67
power system, 106–107, 113 ground test, 130, 467, 468
regenerative fuel cell, 601 Guha studies, 180–181
self-heating battery, 222
solar array, 93, 116
spacecraft, 24, 113 H
space flight constants, xxii
space-to-ground power, 656–657 Haines studies, 479
sun-regulated bus, 70 HAJ, see Hydrazine arcjet (HAJ)
temperature, 158, 454 Hall-effect thruster, 573–574
trades, 113, 116–117 Hall-effect transducer, 330–331, 331
transfer orbit, 359 handover, 132
GFC, see Goodard Space Flight Center harmonic distortion, ripples, 376
(GFC-NASA) harness, see Distribution harness and
glass dielectric capacitors, 98 protection
glass-epoxy, 652 Hartman company, 546
Glenn Research Center (GRC-NASA) health monitoring, battery, 364
Brayton cycle, 510 heat, conversion factors, xx–xxi
cable, 20 kHz, 541 heaters, 5
flywheel energy storage, 634–636 heat radiation, 27
fuel cell power, 597 HEO, see Highly elliptical orbit (HEO)
high frequency alternating current, Hepp and Raffaelle, Bailey, studies, 579
539 high altitude burst test, 418–419
power system options, 40 high-frequency alternating current,
PV cell performance, 146 539–543, 540–541
PV technology advances, 180, 182 high frequency ripples, see Ripples
rotary power transfer, 544 highly elliptical orbit (HEO)
semiconducting devices, 547 basics, 7
solar dynamic power, 659 launch and transfer orbits, 16
Stirling engine, 509 power system requirements, 106
stretched lens array, 186–187 satellites, 11
switching and protection, 545 high-power components, 543–550
Index 675
high-power high-voltage systems Honeywell, 544, 635

alternating current vs. direct current, Howaldtswerke-Deutsche Werft AG,
537–538, 538–539 602
atomic oxygen, 529, 529–530 hub, flywheel energy storage, 608
basics, 518–519, 518–519 Hubble Space Telescope, 152, 469, 469
cable, 20 kHz, 541–543, 542–543 humidity, 458–459
capacitor, 548 hydrazine arcjet (HAJ), 570–571, 572,
charged particles, 529–534, 530–531, 576
533 hydrogen, 550
corona degradation, 526–529, 527–528 hysteresis loop, 281
design guidelines, 535, 535–536
dielectric stress concentration,
525–526, 526 I
electromagnetic interference, 542–543,
543 ICAP program, 386
high-frequency alternating current, IECEC, see Intersociety Energy
539–543, 540–541 Conversion Engineering
high-power components, 543–550 Conference (IECEC)
high-power nuclear TEC, 520–524, IEC regulations, 415, 417
521, 523 IEMS, see Integrated energy and
high-temperature components, momentum storage (IEMS)
561–564, 562, 564 IGBT, see Insulated gate bipolar
high-voltage design issues, 524–536 transistor (IGBT)
high-voltage direct current, 536, 536 illumination tests, 462
high-voltage PV array, 519–520, 520 impedance, source, 369
inductor, 549–550 impedance method, peak power
multi-megawatt burst-power, extraction, 164
555–561, 556–560 indirect interaction, 425
Paschen breakdown voltage, 524–525, indium tin oxide (ITO)
525 electrostatic discharge, 427
plasma and charged particles, rigid panels, 148
529–534, 530–531, 533 70-V bus, 84
power converters, 275, 548–549 individual pressure vessel (IPV)
repetitive pulse power, 552–555 battery trades, 117
rotary power transfer, 544–545, nickel hydrogen battery, 205
544–545 temperature, controlling, 237
selections, 56 wine rack, 238
semiconducting devices, 547–548 Induction General company, 541
switching and protection, 545–547 induction homopolar generator,
temperature extremes, 534 513–514, 514, 516
transformer, 549–550 induction machine, 627
very high voltage system, 550–552, inductor
551–552 design, 284, 284–286, 286
high-power nuclear TEC, 520–524, 521, high-power high-voltage systems,
523 549–550
high-power satellites, 14 inductor coupled buck converter, 274,
high-speed bearings, 609, 632–633 275
high-temperature components, 561–564, inflatable arrays, 153
562, 564 infrared (IR) thermal control system, 5
high-voltage design issues, 524–536 initial altitude spin, 360
high-voltage direct current, 536, 536 initial power-on, 465
high-voltage PV array, 519–520, 520 Inmarsat, 575
Hohmann transfer, 16 inner planets, 476
Holmes, Laug and, studies, 581 inner voltage control loop, 74
676 Index
input variables, 355 electric propulsion, 589–590

insulated gate bipolar transistor (IGBT), electromagnetic interference, 542
255 electrostatic discharge, 422, 424
insulating blankets, multilayer, 237 flywheel energy storage, 253, 604,
insulation, 460 636–638
integrated energy and momentum fuel cell power, 591
storage (IEMS), 617–622, 619, 624, fuel required, 589
624–625 high frequency alternating current,
integrated flux, 30, 33 539
integration and testing (I & T) 20 kHz alternating current, 542
acceptance, 455 lithium-ion battery, 253
acoustic, 457–458 low Earth orbit, 11
ambient functional performance test, maintenance, 589–590
456 photovoltaic-battery system, 79–82,
basics, 451, 452–454 80–81
battery assembly, 463, 463–464 plasma and charged particles, 530
bus impedance and stability, 464 remote power controller, 317
electrical power system, 465–466 rotary power transfer, 544
electromagnetic compatibility, 458 silicones, 101
electromagnetic interference, 458 solar PV-battery, 44
electrostatic discharge, 458 sun acquisition, 161
flight design analysis, 455 switching and protection, 545–546
ground test equipment, 467, 468 system voltage options, 56
initial power-on, 465 International Standards Organization
launch site, 467–469, 469 (ISO), 146
leak, 456 interplanetary and deep space missions,
life test, 458 see also Outer planets
limits of tests, 451–456 basics, 473, 473–475, 475–476
post-vibration performance, 457 deep space mission, 486–488, 487
power electronics, 464, 464 Jupiter mission, 484–486, 485
power system, 459–464 lunar mission, 478
pressure, 457 Mars mission, 481–484
qualification, 455 Mercury mission, 478–480, 479
random vibration, 457 near-sun mission, 480, 480–481,
sequence of tests, 456–457, 456–458 482–483
sine vibration, 457 Saturn mission, 484–486, 485
solar array, 460–462 temperature, deep space, 475–477,
solar array drive, 462–463 476–477
solar cell tests, 458–459 Intersociety Energy Conversion
spacecraft level tests, 465–466 Engineering Conference (IECEC),
survival, 455 190
temperature location, 455–456 ionization damage, power electronics,
test limits, 451–456 96
test points, 466–467 ion propulsion system (IPS), 571–574,
thermal vacuum cycles, 458 572
wires, 459 IPS, see Ion propulsion system (IPS)
inter-model transient problems, 393 IPV, see Individual pressure vessel (IPV)
internal impedance, 217–218 IR, see Infrared (IR) thermal control
International Space Station (ISS) system
a.c. vs. d.c., 538 Iridium, 322, 419
array construction, 152 ISO, see International Standards
COTS, 415 Organization (ISO)
debris and micrometeoroids, 102, 589 Isp, see Specific impulse (Isp)
dosimeter, 92 ISS, see International Space Station (ISS)
Index 677
I & T, see Integration and testing (I & T) KOH, see Potassium hydroxide (KOH)
Italian Space Agency, 486, 584, see also Kozlowski studies, 364
Cassini Krut and Karam studies, 187
ITO, see Indium tin oxide (ITO) Krut studies, 183
I-V, see Current vs. voltage (I-V) Kuiper belt, 474
J L
Jet Propulsion Laboratory (JPL) Lagragian points, 15
current regression peak power Laplace domain and properties
extraction, 164 battery model, 245–246
IEMS, 624–625 current and voltage regulators, 277
microwave beam, 654 magnetic bearings, 631
PC cell performance, 146 large, high-power satellites, 14
Pioneer-10, 486 large communications satellite busses,
power regression peak power 83–86
extraction, 164–165 large signal analysis, 393
PV technology advances, 180 laser beams, 655
JIMO, see Jupiter Icy Moon Orbiter (JIMO) laser initiated deployment, 323, 323–324
Johnson Island, South Pacific, 419 Laug and Holmes studies, 581
Johnson studies, 586 launch
Jones studies, 188 battery, 234–235
JPL, see Jet Propulsion Laboratory (JPL) factory-to-orbit events, 130
Jupiter flywheel energy storage, 623
basics, 474 integration and testing, site, 467–469,
magnetic field, 477 469
mission, power systems, 484–486, 485 orbit, satellites, 15–16
radiation effects, 98 power system, 105, 234–235
radioisotope thermoelectric pyro power trades, 118
generator, 500 simulation, energy balance and power
Jupiter Icy Moon Orbiter (JIMO), 520 management, 362–363, 363
site, 467–469
launch and transfer orbit environment,
K 26
launch vehicles, 8, 26
Kapton LCP, see Light concentrating panel
atomic oxygen, 30 (LCP)
cable shielding, 410 Leach corporation, 327
electronic packaging, 336 leak integration and testing, 456
ISS, 80 length, conversion factors, xx
radiation damage, 101–102 LEO, see Low Earth orbit (LEO)
superconductors, 652 Leonid meteor storm, 102
Karam. Krut and, studies, 187 Libration points, 15
KelF, 652 LiCFx battery, see Lithium/polycarbon
Kelly cosine, 147, 155 monofluoride (LiCFx) battery
Kepler’s laws, 12, 12 life-cycle cost, power system, 41
Kerslake and Gefert studies, 578–579 life test, 458
Kevlar light concentrating panel (LCP),
array construction, 148 188–191, 189, 191
flexible array, 151 limits of tests, 451–456
fuel cell power, 598 Lindeberry and Chapman studies, 582
high-temperature components, 561 Lindmyer studies, 390
King model, 34 linear sounds, 259–261, 260
678 Index
lithium-ion battery (Li-ion) flexible array, 153

advances, 253 flywheel energy storage, 605, 615, 617
basics, 209–212 fuel cell power, 591, 599
deep space missions, 487–488 gravity gradient method, 13
electrochemical cell, 197 harness mass optimization, 303
unregulated charging, 243 high-voltage PV array, 519–520
lithium/polycarbon monofluoride hydrazine arcjet, 570
(LiCFx) battery IEMS, 625
launch and ascent power, 623 ISS, 81
power system options, 41 launch and transfer orbits, 15–16
primary battery, 251 launch orbit, 15–16
lithium polymer (Li-poly) battery, 212, launch simulation, 362
213 lithium-ion battery, 211, 253
Little fuses, 307 lithium polymer battery, 212
Litton Polyscientific Corporation, 322 mass erosion, 99–100
load power conditioner (LPC), 73 multi-megawatt burst-power, 555
load power converter, 72, 276–277, nuclear threat, 417
277–278 offsetting from sun, 176
load power profile, 119–122, 120–122 orbit period, 347
loads, fully-regulated bus, 67 peak power tracking controller, 79
Lockheed Martin Space Systems plasma and charged particles, 530
dynamic system with alternator, 509 power system, 106–107, 113
flywheel energy storage, 635 regenerative fuel cell, 601
70-V bus, 84 satellites, 11
logic flow chart, energy balance self-heating battery, 222–223
software, 354 small satellite bus, 88
long wavelength radiation, xxiii solar array, 93, 116
Lorentz force solar concentrator-dynamic power
magneto-hydrodynamic propulsion, system, 47
576 spacecraft level trades, 113
tether power propulsion, 584, 588 space-to-ground power, 656
Van Allen radiation belts, 31 sputtering, 100–101
losses, 289–290 sun-regulated bus, 72
louvers, 5 system voltage options, 55
low Earth orbit (LEO) temperature effects, 158
advanced NiCd battery, 200 trades, 113, 116–117
albedo, 147 transfer orbit, 15–16
array design, 167 trickle charge, 352
ascent simulation, 362 very high voltage system, 550
basics, 7–8 low-speed bearings, 632
battery charge (buck) converter, 268 LPC, see Load power conditioner (LPC)
battery design, 231 lunar mission, 478
battery trades, 117 Lynch studies, 659
charge cut-back, 352
charge/discharge, 214–216
control system stability, 386 M
cycle life, 225
degradation, 93 Magnequench Corporation, 512
dynamic system with alternator, magnetic properties and components
506–507 amplifier, 331, 331–332
electric propulsion, 567 bearings, 608–609, 609, 631–633
electromagnetic pulse, 419 coil design, 646, 646
electrostatic discharge, 93, 422–423, conversion factors, xxi
423, 427 cryogenic behavior, 650–651
Index 679
damping method, 13 MCT, see MOS Controlled Thyristor

energy storage, 640–643, 641 (MCT)
field, 27–28, 28 mean lunar distance, xxii
magnetic moment, 177–178, 460 mean solar day, 10
materials, high-temperature, 562–564 mean time between failures (MTBF)
power electronics, 279–288, 280–283 electromechanical energy conversion,
magneto-hydrodynamic (MHD) 512
propulsion, 575–576 failure rate, quick estimate, 448
main bus switching unit (MBSU), 82 random failures, 434
main enabling plug (MEP), 85 redundant units, 441–442
major fuse blow transient, 380–382, mechanical tests, 460
380–383 Megellan, 477
management, battery, 243–245 memory effect, 198, 198–199, 203
maneuvers, energy balance, 360–361, MEO, see Mid Earth orbit (MEO)
361 MEP, see Main enabling plug (MEP)
manufacturing failure, 431 Mepcopal Corporation, 308
manufacturing magnetics, 288, 288 Mercury
Marcoux, Bruce and, studies, 212 basics, 473–474
Mariner-2, 477 flexible harness, 478–480
Mariner-5, 477 mission, power systems, 478–480, 479
Mariner-10, 477 PV cells, 477
Mars metalized polypropylene capacitor, 332
basics, 473–474 metallic resistance wire, 326
landing on, 478 metal-oxide semiconducting field effect
magnetic field, 477 transistor (MOSFET)
mission, power systems, 481–484 basics, 255–256
orbit, mass erosion, 99–100 cryogenic behavior, 650
PV power, 476–477 deep space missions, 487
Stirling engine, 509 power electronic damage, 96–98
wind turbines, 484 semiconducting devices, 547–548
Marshall Space Center, 583 meteoroids and debris, 28–29, 29
Mars Lander, 211–212 M-G, see Motor-generator (M-G)
Mars Smart Lander, 582 MHD, see Magneto-hydrodynamic
Mason studies, 507, 659 (MHD) propulsion
mass microelectronic circuits, voltage
battery design, 233 transients, 373
conversion factors, xx micrometeoroids and debris, see also
power system options, 41 Interplanetary and deep space
solar array trades, 116 missions
spacecraft level trades, 111 basics, 28–29
sun-regulated bus, 72 electrostatic discharge, 423
Massachusetts Institute of Technology, environmental effects, 102
mass erosion, 99–102 micro-satellite bus, 89–90, 89–91
mass minimization, 301–304, 302–303 Microsheet, 93, 148
maximum efficiency, design, 288–290, microwave beam power satellite
289–290 basics, 653, 653–654
maximum likelihood method, 249 laser beams, 655
maximum power transfer, 496–497, microwave beams, 654
496–497 powersat developments, 660–661
Maxwell Laboratories, 548 solar dynamic power generation,
Maxwell’s equations, 418 658–659, 659–660
MBSU, see Main bus switching unit space-to-ground power, 655–657, 656
(MBSU) microwave beams, 654
680 Index
microwave beam type, 582–583 inter-model transient problems, 393

mid Earth orbit (MEO) King model, 34
basics, 7–8 radioisotope thermoelectric
electromagnetic pulse, 418–420 generator, 495–496, 496
flexible harness, 306 simulation, 388–391
nuclear threat, 417, 420 simulation model, 388–391
offsetting from sun, 176 solar array model, 389–391, 390
power system level trades, 115 Molniya orbit, 11
power system requirements, 107 momentum, 558, 620
propulsion system, 6 momentum management electronics,
Mightysat II, 190 612
military handbook, specifications, and moon buggy, 42
standards MOS Controlled Thyristor (MCT), 548
MIL-C-11272, 98 MOSFET, see Metal-oxide
MIL-C-23269, 98 semiconducting field effect
MIL-HDBK-217, 251, 444–446, 446 transistor (MOSFET)
MIL-STD-263, 429 motor-generation design issues, 629–630
MIL-STD-461, 394, 403, 413, 536 motor-generator (M-G), 630, 630
MIL-STD-462, 394 MPG, see Multiple point ground (MPG)
MIL-STD-463, 394 MTBF, see Mean time between failures
MIL-STD-1399, 536 (MTBF)
MIL-STD-1522, 201 multi-axis gimbals, 617
MIL-STD-1540, 451 multi-junction cells
MIL-STD-1541, 451 advances, 179–180
MIL-STD-1553, 82 light concentrating panel, 190–191
MIL-STD-83576, 145 solar array, 141
Milner’s cumulative damage theory, multi-layer blankets, 5
249–251 multilayer insulating blankets, 237
minor fuse blow cross-talk, 378–380, 379 multimegawatts (MMW), 555–561,
minority carrier lifetime, 649 556–560
MIR space station, 101–103, 631, see also multiple charge rates, 242
Russia multiple output converter, 275–276, 276
missions, see also Interplanetary and multiple point ground (MPG), 407
deep space missions multistage multicouple, 494, 494
cost, 4 multistage PWM shunt, 261–262, 262
power system requirements, 108 Mylar, 410
voltage limitations, 56
Mitsubishi Heavy Industries, 603
MMW, see Multimegawatts (MMW) N
mode controller, 61, 65–66, 71
mode controller amplifier, 386 narrow orbit control, 116
models NASA, see National Aeronautics and
battery model, 245–246 Space Administration (NASA)
charge/discharge, 248 NaS battery, 41
components, 391, 392 National Aeronautics and Space
cycle life, 248–251, 250 Administration (NASA)
depth of discharge (DOD), 248–249 flywheel energy storage, 635–638,
dynamic, battery, 245–248, 246–247 636–637
dynamic model, 245–248, 246–247 JPL handbook PUB-82-69, 170
dynamic performance and stability, JSC-30237, 543
388–391, 390, 392 power systems comparison, 53
electrical circuit, battery, 213–214 space-to-ground power, 655
electrical circuit model, 213–214, 214 SSP30237A, 413–416
electrical model, 495–496, 496 SSP30238A, 413, 417
Index 681
testing, electromagnetic interference, basics, 208–209

416 BSS-702 bus, 83
National Oceanic and Atmospheric burst power capability, 227
Administration (NOAA), 103 capacity walk-down, 245
Naval Research Laboratory, 182 charge regulation, 240–241
near-Earth space environment comparison, 228
atmosphere, 27 design, 231, 233
atomic oxygen, 29–30 electrochemical cell, 197
basics, 26 flywheel energy storage, 615, 617
charged particles, 30–31 ISS, 80, 82
debris, 28–29, 29 radiation effects, 98
geomagnetic storm, 36–37 regenerative fuel cell, 601
gravity, lack of, 27 self-heating battery, 223
launch and transfer orbit sizing, 124
environment, 26 small satellite bus, 86
magnetic field, 27–28, 28 state of charge, 234
meteoroids and debris, 28–29, 29 trades, 117
nuclear threat, 37 70-V bus, 85
on-orbit environment, 27–31 100-V bus, 84
solar flare, 34–35, 35–36 NiH2, see Nickel metal hydride (NiH2)
solar winds, 34–35, 35–36 battery
total radiation fluence, 37–39, 38 NOAA, see National Oceanic and
Van Allen radiation belts, 31–34, 32–33 Atmospheric Administration
near field, radiated EMI, 397 (NOAA)
NEAR Shoemaker, 477 Nomex, 148, 561
near-sun mission, 480, 480–481, 482–483 non-operational on-orbit, 131
negative feedback control system, 386 non-redundant components, 449–450
Neptune, 473–474 nonreversible electrochemistry, 41
neural network, 364 normal operational orbit, 131
Newton’s laws, 12 northern lights, 31
Nextel, 561, 564 north-south keeping station (NSSK)
NiCd, see Nickel cadmium (NiCd) electric propulsion, 567, 576–577
battery geosynchronous orbit, 11
nickel cadmium (NiCd) battery hydrazine arcjet, 570
advances, 251 ion thruster, 572, 574
aging, 245 propulsion system, 6
basics, 198–200 NSSK, see North-south keeping station
burst power capability, 227 (NSSK)
capacity walk-down, 245 nuclear-dynamic power system, 49
charge regulation, 240–241 nuclear power electric propulsion,
comparison, 228 581–582
design, 231, 233 nuclear reactors, 40
electrochemical cell, 197 nuclear reactor-TEC power systems, 486
monitoring battery health, 364 nuclear-thermoelectric system, 48, 48–49
photovoltaic-battery system, 60 nuclear threat, 37, 417
regenerative fuel cell, 601
sizing, 124
trades, 117 O
100-V bus, 84
nickel hydrogen battery, 200–208, OBC, see On-board computer (OBC)
201–202, 204 objects, see Meteoroids and debris
nickel metal hydride (NiH2) battery obliquity of elliptic at Epoch 2000, xxii
advances, 251 offsetting from sun, 176–177
aging, 245 Ohm’s law, 326
682 Index
Olympus, 152 high-voltage design issues, 524

on-board computer (OBC) 20 kHz cable, 541
operational orbit, 349 Mars, 484
telemetry and commands, 335 plasma and charged particles, 531
transfer orbit, 348 Patel studies, 615, 624
one-axis gimbals, 161, 322 payload, 3
one million electron volts, 30 PDU, see Power distribution unit (PDU)
on-orbit environment, 27–31 peak power extraction, 162–163, 162–165
on-orbit simulation, 357–358 peak power tracking (PPT) components
operational orbit algorithm, 164
basics, 16, 17 photovoltaic-battery system, 73–77,
power system requirements, 106 74–78
state of charge, 349 power system architectures, 63
Opportunity rover, 253, 482 small satellite bus, 88
Orbiter peak solar flux, albedo, long wavelength
fuel cell power, 591 radiation, xxiii
launch and transfer orbit peel tests, 458–459
environment, 26 Pellon 2536, 198
low Earth orbit, 11 Peltier effect, 503
tether power propulsion, 584 PEM, see Proton exchange membrane
orbits, see also specific orbit (PEM)
classification, 6–12, 8 PEMS, see Power and Energy
mechanics, 12–13, 14 Management Software (PEMS)
operation, power system functions, penumbra, 17, 18, 154
132 performance test, power system, 460
transfer vehicle, 589 permanent magnet (PM), 512, 514, 515,
outer planets, 476, 590, see also 632
Interplanetary and deep space permittivity of free space, xxii
missions Phosroc (Rockbestos), 561, 564
outgassing photovoltaic-battery system
environmental effects, 99 analog control, 69
high-voltage design issues, 524 analog-digital hybrid control, 70
lack of gravity, 27 architecture trades, 78, 78–79
light concentrating panel, 190 basics, 58–61
output format, 355–356, 357 battery, 59–60, 60, 64
over-charge, 222 battery bus, 66
overload capability, 515, 516 bus voltage control, 67–69, 68
bus voltage controller, 65
control circuit, 69–70
P digital control, 69–70
direct energy transfer, 62–63
packaging, battery, 296 fully regulated bus, 63, 63–67, 71–73
parabolic concentrator, 191–193, 192 ground power cord, 67
parallel fuses, 316 International Space Station, 79–82,
parallel operations, 514, 515 80–81
parking orbit, 16, 106 large communications satellite busses,
partial shunts, 258, 258–259, 259, 263 83–86
parts-count method, 446–447 loads, 67
part sensitivity, 429, 429–430 micro-satellite bus, 89–90, 89–91
Paschen breakdown voltage mode controller, 65–66
corona degradation, 527–528 peak power tracker, 63
design guidelines, 535 peak power tracking bus, 73–77, 74–78
high-power high-voltage systems, power and energy management
524–525, 525 software, 67
Index 683
power distribution unit, 65 buck-boost converter, 271–272,

power regulation, 60–61, 60–62 271–272
power regulator unit, 65 bus ripple filter design, 265–266,
power system architectures, 62–63 265–266
shunt dissipator, 64 calibrated step shunt, 263, 264
small satellite bus, 86–89 current regulators, 277–279, 278
solar array, 58–59, 59, 64 flyback converter (buck or boost), 272,
solar array drive, 64 272–273
sun-regulated bus, 60, 70–73, 71 full shunts, 258–259, 258–259, 263
70-V bus, 84–86, 85 inductor coupled buck converter, 274,
100-V bus, 83, 83–84 275
50-V or less busses, 86, 87 inductor design, 284, 284–286, 286
photovoltaic cell, 135–137, 136–137 linear sounds, 259–261, 260
photovoltaic energy conversion, 40 load power converter, 276–277,
photovoltaic-flywheel power system, 277–278
606–607, 606–607 magnetics, 279–288, 280–283
Picofuse fuses, 307–308 manufacturing magnetics, 288, 288
pin diodes, 429 maximum efficiency, design, 288–290,
Pioneer, 35, 500 289–290
Pioneer-10, 485–486 multiple output converter, 275–276,
plasma and charged particles, 529–534, 276
530–531, 533 multistage PWM shunt, 261–262, 262
plasma arcing, 550 partial shunts, 258–259, 258–259, 263
plasma contactor, 586 polyphase PWM shunt, 262, 263
Plasma Motor Generator, 584, 588 power converters, 266–279
platinum, 326 power losses, core and coil, 282–284
Pluto, 473–474, 581 push-pull converter, 274, 274
plutonium-238, 233 PWM sounds, 259–261, 260
PM, see Permanent magnet (PM) resonant converter, 274–275
PN junction diode, 649–650 sequential linear shunt, 260–261, 261
point-to-multi-point communications, 7 shunt circuits, 263–265
point-to-point communications, 7 shunt regulator, 257–263, 258
Poisson’s probability distribution, 28, switching devices, 255–256, 256–257
433 transformer coupled forward
polycrystalline, 140 converter, 273, 273
polyphase PWM shunt, 262, 263 transformer design, 286–288, 288
popping, 242 trends, 291
post-vibration performance, 457 voltage regulators, 277–279, 278
potassium hydroxide (KOH), 201–203, power flow analysis, 125, 125–127, 126
600 power generation, 461
power, conversion factors, xx power gradient method, peak power
power and energy management extraction, 164
software (PEMS), 67 power level scaling, 56–57, 57
power converters, 266–279, 275, power losses, core and coil, 282–284
548–549 power management electronics, 612
power distribution unit (PDU), 62, 65 power regression method, peak power
power electronics, 3, 464, 464 extraction, 164–165
power electronics, damage, 95–98, 97–98 power regulation, 6, 60–61, 60–62
power electronics and magnetics power regulator unit (PRU)
basics, 255 charge rate, 348
battery charge (buck) converter, control system stability, 386
266–270, 267–268 flywheel energy storage, 606
battery discharge (boost) converter, fully-regulated bus, 65
270, 270–271 harness mass optimization, 301, 303
684 Index
power regulator unit (PRU) (continued) Proton Energy Systems, 599

mode controller, 66 proton exchange membrane (PEM), 555,
output source impedance, 375 597, 600
small satellite bus, 88 PRU, see Power regulator unit (PRU)
telemetry and commands, 336 P600L fuses, 308
powersat developments, 660–661 PSPICE program
power system conducted electromagnetic
basics, 3 interference, 395
integration and testing, 459–464 control system stability, 386
requirements, 105–106, 106 NASA vs. commercial test methods,
self-derived requirements, 106–107 417
system specifications, 108 simulation runs, 393
power system architectures, 62–63 pulsed plasma thruster (PPT), 576
power system functions over life, pulse width modulation (PWM)
131–132 converter, 269, 277
power system level trades, 113–119, 114 shunt regulators, 257, 260–261
power system options small satellite bus, 88
AMTEC system, 52 sounds, power electronics and
basics, 40–41, 40–41 magnetics, 259–261, 260
chemical-dynamic system, 49 puncture strength, 425
comparison of options, 52–53, 53–54 push-pull converter, 274, 274
fuel cell, 42, 42–43 PV, see Photovoltaic-battery system
nuclear-dynamic system, 49 P-V, see Power vs. voltage (P-V)
nuclear-thermoelectric system, 48, PV cell trades, 114–116, 115
48–49 PV technologies, 137–144, 138, 178–182,
power level scaling, 56–57, 57 179, 181–182
primary battery, 41–42, 42 PWM, see Pulse width modulation
scaling for power level, 56–57, 57 (PWM)
solar concentrator-dynamic system, PYRO, see Pyrotechnic (PYRO)
45–48, 46 properties
solar PV-battery, 43–45, 44–46 pyrotechnic (PYRO) properties
solar-thermoelectric, 51 bus (essential bus), 325
system voltage options, 53–56, 55 design process and trades, 118–119,
thermionic system, 51–52 119
thermo-photovoltaic scheme, 50, harness, 323
50–51 shock levels, 26
power transfer assembly, see Slip ring
assembly
Power Transistor Company, 547 Q
power vs. time curve, 119
power vs. voltage (P-V), 143–147, Qiu and White, Thieme, studies, 509
144–146 qualification integration and testing, 455
pre-arc period, 310–311 quality of power, 378
precession of the equinox, xxii quick estimate, 448
precision attitude control, 116
pre-handover, 131–132
prelaunch, 130–131 R
pre-operational drift orbit, 131
pressure, xx, 457 radiated EMI, 395–397, 396
pressure strain gage readings, 363 radiation, see also specific source
primary battery, 41–42, 42, 185, 251 damage, solar array, 169–174, 171–176
program verification, 356, 358 environmental effects, 98
Project Prometheus, 520 heat, launch and transfer orbit
propulsion system, 6, 7 environment, 26
Index 685
plasma and charged particles, 534 calculation example, 450, 450

shield, 337–340, 338–339 derating, 447–448
solar cell test, 458 diode o-ring, 443, 443–444
radiators, fixed, 5 dormant redundancies, 440–441
radio frequency (RF) amplifier, 73 estimates, 446–448
radioisotopes, 40 failure rate statistics, 444
radioisotope thermoelectric generator FMECA, 449
(RTG) MIL-HDBK-217, 445–446, 446
advanced type, 501–503, 502 non-redundant components, 449–450
aging effect, 497–498, 497–499 parts-count method, 446–447
assembly, 494–495, 495 quick estimate, 448
basics, 489–490, 490 random failures, 433, 433–434
cooler, thermoelectric, 503 redundancies, 439–443, 440
electrical model, 495–496, 496 series-parallel reliability, 435–439,
flight history, 500, 500–502 436–439
maximum power transfer, 496–497, theorems, fundamental, 434–435
496–497 wear-out failure, 434
multistage multicouple, 494, 494 remote bus isolator (RBI), 82
nuclear or chemical-dynamic power remote power controller (RPC)
system, 49 distribution harness and protection,
nuclear-thermoelectric power system, 317–318
48 ISS, 82
segmented TECs, 500–501 switching and protection, 545–546
single-stage multicouple, 493, 493 repetitive pulse power, 552–555
single-stage unicouple, 491, 492–493 Resistojet, 570
solar-thermoelectric power system, 51 resistors, 651
temperature effect, 497–498, 497–499 resonant converter, 274–275
thermoelectric cooler, 503 Reynolds, Sack, Croydon and, studies,
thermoelectric fundamentals, 243
490–495, 491 RFC, see Regenerative fuel cell (RFC)
Raffaelle, Bailey, Hepp and, studies, 579 rigid panels, 148–149, 149–150
random failures, 431, 433, 433–434 ripples, 374–377, 374–377
random vibration, 457 R-L-C parameters, 294–297, 295–297
Rankine cycle Rockbestos Phosroc, 561, 564
dynamic system with alternator, 505 rollover, 242
multi-megawatt burst-power, 557 roll spin, 359
nuclear or chemical-dynamic power root sum squared (rss), 128–129
system, 49 rotary power transfer, 544–545, 544–545
solar concentrator-dynamic power rotary vane damper, 5
system, 46 rotor rim, 608, 608
RBI, see Remote bus isolator (RBI) rovers, 253, 482
reacquisition, 131 RPC, see Remote power controller (RPC)
reaction wheel replacement, 624, 624-625 rss, see Root sum squared (rss)
rechargeable battery, 195 RTCA/DO-160C, 414–415
reconditioning, 199 RTG, see Radioisotope thermoelectric
redundancies, 439–443, 440 generator (RTG)
redundant fuses, 315–316, 316 Rule of Ten, design process, 129
regenerative fuel cell (RFC), 253, 598–599 Russia, see also MIR space station
relays, 327–328, 328–329 ISS, 80
reliability and derating nuclear reactor-TEC power systems,
active redundancies, 440–441, 443 486
basics, 431–432, 432 plutonium-238, 490
battery charger/discharge, 442 SP-100, 523
battery chargers, 442–443 thermionic power system, 52
686 Index
S scaling for power level, 56–57, 57

SCARLET, see Solar concentrator
SA, see Solar array (SA) refractive linear element technology
SABER program (SCARLET)
conducted electromagnetic Schottky barrier diode, 650
interference, 395 SDI, see Strategic Defense Initiative (SDI)
control system stability, 386 seasonal variations, 16–23
simulation runs, 393 ISS, 81
Sack, Croydon and Reynolds studies, solar PV-battery, 44
SAD, see Solar array drive (SAD) secondary battery, 195
safety considerations, 238–240 Seebeck properties, 326, 491–492,
safety during tests, 430, 430 494–495
Saft Corporation, 243 SEL, see Single event latch-up (SEL)
Sanyo company, 200, 208 self-derived requirements, 106–107
Sargent studies, 509 self-discharge battery, 219–220, 221
satellites self-heating battery, 221–224, 222–224
attitude and orbit control system, 4 semiconductors
basics, 3–4 cryogenic behavior, 649–650
beta angle, 23–24 high-power high-voltage systems,
communications and data handling, 4 547–548
Earth orbit classification, 6–12, 8 PV technologies, 137
eclipse, 16–22, 18–21 semicrystalline solar array, 140
electrical power system, 5 sensors, 610, 611–612
fuel cell power, 600–601, 601–602 sequence of tests, 456–457, 456–458
geostationary orbit, 8–9, 10 sequential linear shunt, 260–261, 261
geosynchronous orbit, 9–11, 10 sequential shunt unit (SSU), 81
gravity gradient stabilization method, series-parallel reliability, 435–439,
13 436–439
highly elliptical orbit, 11 SESB, see Single-event secondary
launch orbits, 15–16 breakdown (SESB)
low Earth orbit, 11 SEU, see Single event upset (SEU)
magnetic damping method, 13 shadow effect, 18, 165–166, 165–166
operational orbit, 16, 17 shape memory metal, 324
orbit, 6–13, 8, 14 shock spectrum, 26
propulsion system, 6, 7 shunt, 330–331
solar flux, 22–23, 22–23 shunt circuits, 263–265
spacecraft mass, 24, 24 shunt control, 461
spin stabilization method, 13–14, 14 shunt control loop, 390
stabilization, 13–15 shunt dissipator, 64, 462
structure and mechanisms system, 6 shunt mode, 66–67, 69
sunsynchronous orbit, 12 shunt regulator, 257–263, 258, 461
systems, 4–6 sidereal day, xxii, 10
thermal control system, 5 sidereal year, xxii
three-axis stabilization, 14–15, 15 Siemens, 642
transfer orbits, 15–16 signal analysis, 393
TT&C system, 5 silicon cells, 168
Saturn silicones, 101
basics, 473–474 silicon rubber, 30, 529
mission, power systems, 484–486, 485 silicon solar cells, 82
radioisotope thermoelectric Silver Teflon, 102, see also Teflon
generator, 500 silver zinc (AgO-Zn) battery, 213
Saturn, 26 simulation model, 388–391
SBR, see Space-based radar (SBR) simulation runs, 356–363, 392–393
Index 687
sine vibration, 457 light concentrating panel, 188–191,

single charge rate, 242 189, 191
single-crystal silicon, 138–139, 139 magnetic moment, 177–178
single event latch-up (SEL), 96 model, 389–391, 390
single-event secondary breakdown module, 354
(SESB), 97 multi-junction, 141
single event upset (SEU), 96 offsetting from sun, 176–177
single-point ground (SPG), 407 panels, 6
single pressure vessel (SPV), 117, parabolic concentrator, 191–193, 192
203–206, 205–206 peak power extraction, 162–163,
single-stage multicouple, 493, 493 162–165
single-stage unicouple, 491, 492–493 performance, 154–166
slip ring assembly, 320–322 photovoltaic cell, 135–137, 136–137
slow-blow fuses, 311 polycrystalline, 140
small satellite bus, 86–89 prototype, 53
small signal analysis, 393 PV technologies, 137–147, 138,
smart bombs, 552 144–146, 178–182, 179, 181–182
Smith and Brill studies, 227 radiation damage, 169–174, 171–176
SNAP-10, 486 rigid panels, 148–149, 149–150
SOC, see State of charge (SOC) rotation, 3
SOFC, see Solid oxide fuel cell (SOFC) semicrystalline, 140
solar array drive (SAD) shadow effect, 165–166, 165–166
basics, 319–322, 321–322 single-crystal silicon, 138–139, 139
fully-regulated bus, 64 sizing, 122–124, 123
integration and testing, 462–463 stretched lens array, 186–188, 186–188
life test, 458 structure and mechanisms system, 5
sun tracking, 162 sun acquisition, 158–161, 160
70-V bus, 84 sun angle, 154–155, 155
solar array (SA) sun intensity, 154, 154–155
albedo, power from, 147 sun tracking, 161, 161–162
amorphous, 140–141 temperature effect, 155–158, 156–160
back-surface illumination, 147–148 thin film, 140
basics, 3, 135 three or more wings, 151, 151–153
body mounted array, 150, 151 Van Allen radiation belts, 34
concentrator array, 183–193, 184–185 solar array trades, 116
construction, 148–153 solar cell tests, 458–459
degradation, 45–46, 92–93, 93, 169, solar concentrator-dynamic system,
170 45–48, 46
design, 166–178, 167 solar concentrator refractive linear
eclipse, 17 element technology (SCARLET),
electrical power system, 5 187
electrostatic discharge, 93–95, solar day, xxii, 10
427–428, 428 solar dynamic power generation,
equivalent electrical circuit, 141, 658–659, 659–660
141–143, 143 solar flare, 34–35, 35–36
flexible array, 151–153 solar flare maxima year, xxii
fluence, 39, 175–176, 177 solar flux
fully-regulated bus, 58–59, 59, 64 inner planetary missions, 473–474
gallium arsenide, 140 radiation, space flight constants, xxii
illumination test, 459 satellites, 22–23, 23–23
inflatable arrays, 153 spacecraft level trades, 112–113
integration and testing, 460–462 solar PV-battery, 43–45, 44–46
I-V characteristics, 143–147, 144–146 solar PV electric propulsion, 578–579,
launch and transfer orbits, 16 578–580
688 Index
solar radiation, 31, 40 stability, see also Dynamic performance

solar shunt regulators, 391 and stability
Solar Space power Exploratory Research bus impedance, 382–385, 383–386
and Technology Program, 656–657 control system, 386–388, 387–389
solar storms, 103 satellites, 13–15
solar thermal type, 580, 580–581 solar array, 461
solar-thermoelectric power system, 51 STARFISH, 419
solar winds, 31, 34–35, 35–36 state of charge (SOC)
solar year, xxii AH counter and sensor, 353
solid insulation, 425 battery voltage, 353
solid objects, see Meteoroids and debris BSS-702 bus, 83
solid oxide fuel cell (SOFC), 555–556 charge regulation, 240
solid rotor, 512, 514, 516 electrochemical cell, 196
solid sintered powder, tantalum launch and ascent simulation, 362
capacitor, 332 nickel hydrogen battery, 201
solid state power controller (SSPC), 82 NiH2 battery, 234
source impedance, 369 operational orbit, 349
southern lights, 31 photovoltaic-battery system, 59
SP-100, see Space Power-100 (SP-100) power flow analysis, 125
space-based radar (SBR), 48 telemetry and commands, 335
space colonies, 599–600, 600 transfer orbit, 348
space compliance, 416–417 stationary plasma thruster (SPT),
spacecraft level tests, 465–466 574–575, 578
spacecraft level trades, 111–113, 112 station keeping, 131, see also North-
spacecraft mass, 24, 24 south keeping station (NSSK)
space debris cleanup, 589 Stefan-Boltmann constant and law, xxi,
space flight constants, xxii–xxiii 475
Spacelab-I, 585 Steinmetz exponent, 281–282
Space Power-100 (SP-100) Stern and Bonebright studies, 190
boom cable design, 563–564 stiffness, system, 368, 368–371, 370–371
high-voltage PV array, 520–523 Stirling cycle, 49
nuclear-thermoelectric power system, Stirling engine
49 dynamic system with alternator, 508,
space shuttles, 42, 253 508
space-to-ground power, 655–657, 656 multi-megawatt burst-power, 559
specific energy, 197 radioisotope thermoelectric
specific impulse (Isp), 568–569, 570 generator, 502
specific power and power density, 197 stop-and-go operation, 361
Spectrolab, 141 storage, factory-to-orbit events, 130
speed limitations, 514, 516–517 Strategic Defense Initiative (SDI), 49, 518
Sperry Flight Systems, 544 stress and energy relations, 613–614,
SPG, see Single-point ground (SPG) 614–615
spinning, 16 stretched lens array, 186–187, 186–188
spin stabilization, 13–14, 14, 88 strike though gaps, 425
Spirit rover, 253, 482 structural materials, 652
SPT, see Stationary plasma thruster structure and mechanisms system, 6
(SPT) sublimation, 27
sputtering, 100–101, 422 substrate isolation test, 462
SPV, see Single pressure vessel (SPV) suffixes used, xix
Squires, Govar and, studies, 240 summer solstice (SS)
SS, see Summer solstice (SS) load power profile, 121
SSPC, see Solid state power controller operational orbit, 16
(SSPC) solar flux, 22
SSU, see Sequential shunt unit (SSU) spacecraft level trades, 112
Index 689
sun-regulated bus, 72 tapes, 5

sun acquisition, 158–161, 160 Taylor series, 159
sun angle, 154–155, 155 TCS, see Thermal control system (TCS)
sun intensity, 154, 154–155 TDMA, see Time division multiple
sun pointing error, 161 access (TDMA)
sun-regulated bus TDRS, see Tracking and data relay
battery latch-up, 365, 365–367 satellite (TDRS)
photovoltaic-battery system, 60, TEC, see Thermoelectric converter (TEC)
70–73, 71 system and cells
sunspots, 34 Teflon, see also Silver Teflon
sunsynchronous orbit, 12, 44 atomic oxygen, 529
sun trackers, 161, 161–162 NiCd battery, 198
superconducting magnets, 49 radiation damage, 101
superconductors, space environment Teldix, 632
basics, 640 Teledyne corporation, 327
bipolar transistor, 650 telemetry (TLM) and commands
capacitors, 651 basics, 3, 334–336, 335
carrier mobility, 649 solar arrive drive, 319
critical parameters, 643–645, 644–645 temperature
cryogenic temperature, 648–652 battery, controlling, 236–237
dopant atoms ionized, 649 compensation, 245
magnetic coil design, 646, 646 conversion factors, xx
magnetic energy storage, 640–643, 641 extremes, 534
magnetics, 650–651 location, integration and testing,
minority carrier lifetime, 649 455–456
PN junction diode, 649–650 missions, power systems, 475–477,
power MOSFETs, 650 476–477
resistors, 651 radioisotope thermoelectric
Schottky barrier diode, 650 generator, 497–498, 497–499
semiconductors, 649–650 sensors, thermal control system, 5
structural materials, 652 solar array, 123–124, 155–158,
system configuration, 646–648, 647 156–160
SuperNiCd, 98 temperature telemetry test, 462
suppression methods, 407–411 termination shock wave, 35
survival, 455 tethered satellite system (TSS), 584, 588
swelling, 242 tether power type, 583–590, 584, 587
switching and protection, 545–547 TF, see Transfer function (TF)
switching devices, 255–256, 256–257 theorems, 434–435
synchronous generators thermal controller, 326–327
dynamic system with alternator, 512, thermal control system (TCS), 5
514, 515–516 thermal cycling test, 461–462
multi-megawatt burst-power, 558 thermal design, 235–238, 236–237
synchronous machines, 626–627, 628 thermal equation, 158
SYSCAP program, 386 thermal shock, 458–459
system configuration, 646–648, 647 Thermal Technology company, 549
systems of units, xx–xxi thermal vacuum bake-out test, 460
system specifications, 107 thermal vacuum cycles, 458
system stiffness, 368, 368–371, 370–371 thermionic power system, 51–52
system voltage options, 53–56, 55 thermionic system, 51–52
thermistor, 5, 326–327, 327
thermocouple, 326
T thermodynamic energy converter,
45–48, 46
tantalum capacitors, 332 thermodynamics, 508–510
690 Index
thermoelectric converter (TEC) system satellites, 15–16

and cells simulation, 358–361, 359
near-sun mission, 481 state of charge, 348
segmented, radioisotope transformer, 286–288, 288, 549–550
thermoelectric generator, 500–501 transformer coupled forward converter,
solar-thermoelectric power system, 51 273, 273
thermoelectric cooler, 503 transition field, radiated EMI, 397
thermoelectric energy conversion, 40 traveling wave tube amplifier (TWTA),
thermoelectric fundamentals, 490–495, 72, 444
491 TRDF, see Total ripple distortion factor
thermoelectric material, 48 (TRDF)
thermo-photovoltaic (TPV) power trickle charge, 222
scheme Truong, Wolff and Dravid studies, 634
basics, 50, 50–51 TRW Corporation, 485, 549
near-sun mission, 481 TSS, see Tethered satellite system (TSS)
radioisotope thermoelectric twisting wires, 408
generator, 502 two-axis gimbals, 161
thermostats, 5 two-wing design, 113
Thévenin equivalent, 368–369, 385 TWTA, see Traveling wave tube
Thieme, Qiu and White studies, 509 amplifier (TWTA)
thin film, 140
three-axis stabilization, 14–15, 15
3M Nextel, 561, 564 U
three or more wings, 151, 151–153
thrusters, 14 UARS, see Upper Atmospheric Research
Thuraya (United Arab Emirates), 24 Satellite (UARS)
time-delay fuses, 311 Ulysses, 500
time division multiple access (TDMA), umbra, 17, 18
72 unfused loads, 317
TLM, see Telemetry (TLM) and United Arab Emirates (Thuraya), 24
commands units, systems of, xx–xxi
torque, conversion factors, xx University of California, 654
Toshiba Corporation, 642 University of Wisconsin, 642
total radiation fluence, 38, see also unregulated charging, 243
Fluence Upper Atmospheric Research Satellite
total ripple distortion factor (TRDF), (UARS), 422
375–376, see also Ripples Uranus, 473–474
touchdown mechanical bearings, 610,
610
TPV, see Thermo-photovoltaic (TPV) V
power scheme
tracking, telemetry, and command Van Allen radiation belts
(TT&C) system, 5, 466 Jupiter mission, 485
tracking and data relay satellite (TDRS) near-Earth space environment, 31–34,
geostationary orbit, 9 32–33
geosynchronous orbit, 11 power system requirements, 106, 477
meteoroid environment fluence, 29 radiation effects, 92, 98
meteoroids and debris, 102 radioisotope thermoelectric
trade names, xix generator, 490
transfer function (TF), 386 spacecraft level trades, 112
transfer orbit VE, see Vernal equinox (VE)
factory-to-orbit events, 130 velocity, conversion factors, xxi
load budget, 359, 361–362 velocity of light in free space V0, xxii
power system requirements, 105 Venus, 473–474, 477
Index 691
vernal equinox (VE) Westinghouse-Siemens Corporation,

eclipse, 17 603
operational orbit, 16, 19 White, Thieme, Qiu and, studies, 509
solar flux, 22 wine rack, 238, 239
vertical stack, 238, 238 Winslow, Bouquet and, studies, 534
very high voltage system, 550–552, winter solstice (WS)
551–552 operational orbit, 16
vibration, 457, 558 solar flux, 22
Virginia Polytechnic Institute and State spacecraft level trades, 112
University, 291 wire
Voecks studies, 599 ampacity, 294
voltage insulation, 299
power flow analysis, 126 integration and testing, 459
regulation and transients, 68, 371–373, losses, 461
372–373, 515–516 Wisconsin Power Company, 642
regulators, 277–279, 278 W.L. Gore & Associates, 541
transients, 373 Wolff and Dravid, Truong, studies, 634
voltage ratio method, 164 worst-case error margin, 127–129
voltage vs. temperature (V-T) wound rotor, 512–513, 514, 516
energy efficiency, 219, 220 WS, see Winter solstice (WS)
overcharge control, 351
small satellite bus, 88
volume, conversion factors, xx
Voyager X
Jupiter and Saturn mission, 486 xenon ion propulsion system (XIPS),
radioisotope thermoelectric 573–574, 578
generator, 500
termination shock wave, 35
V-T, see Voltage vs. temperature (V-T)
Y
yaw spin, 359–360
W
water, 43
wear-out failure, 431, 434 Z
Weibull distribution, 139, 249
Westinghouse Lima, 562 zinc, 27
Westinghouse Research Center, 547, 642 Zircar separators, 203

Patel - Spacecraft Power Systems PDF

Uploaded by

Copyright:

Available Formats

Patel - Spacecraft Power Systems PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Patel - Spacecraft Power Systems PDF

Uploaded by

Copyright:

Available Formats

SPACECRAFT

Library of Congress Cataloging-in-Publication Data

Patel, Mukund R., 1942-

Visit the CRC Press Web site at www.crcpress.com

# 2005 by CRC Press

navigation program, will involve many countries, possibly including China.

The emerging technologies in the spacecraft power systems cannot

Part B: Solar array, battery, power electronics and magnetics, distribution

Mukund R. Patel, Ph.D., PE, is a research engineer, educator and

ACS attitude control system

BCC battery charge controller

CASI Canadian Aeronautics and Space Institute

CDR critical design review

DAR double anti-reflective (coating)

EED electro-explosive device

FBA fuse board assembly

GN&C guidance, navigation and control

HAJ hydrazine arcjet

IECEC Intersociety Energy Conversion Engineering Conference

JPL Jet Propulsion Laboratory

KOH potassium hydroxide (electrolyte)

LCP light-concentrating panel

MEO mid Earth orbit

NASA National Aeronautics and Space Administration

OBC on-board computer

PCDU power control and distribution unit

PME propulsion module electronics

RBI remote bus isolator

SPE solid polymer electrolyte (fuel cell)

TCS thermal control system

USRA Universities Space Research Association

XIPS xenon ion propulsion system

WCBV worst case bus voltage in normal operation

SI unit = English unit Factor below

W/m2 Btu/(ft2 hr) 3.154 6

Other Units and Conversions

Acceleration due to Earth gravity 9.806 7 m/s2 (32.173 5 ft/s2)

e. Long wavelength radiation is a minimum when albedo is a

Part A: Power System Fundamentals

Part B: PV–Battery System

Part C: Power System Performance

Part D: Special Power Systems

22 High-Power High-Voltage Systems 518

Part A: Power System Fundamentals

Chapter 1 Satellite Overview 3

1.5.4 Three-axis stabilization 14

Chapter 2 Near-Earth Space Environment 26

Chapter 3 Power System Options 40

Chapter 4 Photovoltaic–Battery System 58

Chapter 5 Environmental Effects 92

5.5 Effects on other components 98

Chapter 6 Power System Requirements 105

Chapter 7 Design Process and Trades 111

Part B: PV–Battery Power System

Chapter 8 SolarArray 135

8.5 I–V and P–V characteristics 143

Chapter 9 Battery 195

9.5.4 Energy efficiency 218

Chapter 10 Power Electronics and Magnetics 255