Modern Antenna and Microwave Circuits

arXiv:1911.08484v3 [eess.
SP] 15 Sep 2020

Modern Antennas and Microwave Circuits
A complete master-level course
by
Prof. dr.ir. A.B. Smolders, Prof.dr.ir. H.J. Visser and dr. Ulf Johannsen
Electromagnetics Group
Center for Wireless Technology Eindhoven (CWTe)
Department of Electrical Engineering
Eindhoven University of Technology, The Netherlands
Version September 2020

Modern Antennas and Microwave Circuits - A complete master-level course
A.B. Smolders, H.J. Visser, U. Johannsen.
Eindhoven University of Technology, version September 2020.
A catalogue record is available from the Eindhoven University of Technology Library

ISBN: 978-90-386-4943-6
Subject headings: Antennas, microwave engineering, phased-arrays, wireless communications.
c 2020 by A.B. Smolders, H.J. Visser, U. Johannsen, Eindhoven, The Netherlands.

All rights reserved. No part of this publication may be reproduced or transmitted in any form or
by any means, electronic, mechanical, including photocopy, recording, or any information storage
and retrieval system, without the prior written permission of the copyright owner.
Contents
1 Introduction 5
1.1 Purpose of this textbook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Antenna functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 History of antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Electromagnetic spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 System-level antenna parameters 11

2.1 Coordinate system and time-harmonic fields . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Field regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Far field properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Radiation pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Beam width . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Directivity and antenna gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.7 Circuit representation of antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.8 Effective antenna aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.9 Polarization properties of antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.10 Link budget analysis: basic radio- and radar equation . . . . . . . . . . . . . . . . 21
3 Transmission line theory and microwave circuits 23

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 Telegrapher equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 The lossless transmission line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 The terminated lossless transmission line . . . . . . . . . . . . . . . . . . . . . . . . 28
3.5 The quarter-wave transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.6 The lossy terminated transmission line . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.7 Field analysis of transmission lines . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.8 The Smith chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.9 Microwave networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.10 Power Combiners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.11 Impedance matching and tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.12 Microwave filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.13 Microwave amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.13.1 Power gain, available gain and transducer gain . . . . . . . . . . . . . . . . 62
3.13.2 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
1
2 CONTENTS
3.13.3 Amplifier design using constant gain circles . . . . . . . . . . . . . . . . . . 67

3.14 Low-noise amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.14.1 Noise in microwave circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.14.2 Noise Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4 Antenna Theory 77
4.1 Maxwell’s equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2 Radiated fields and far-field approximation . . . . . . . . . . . . . . . . . . . . . . 83
4.3 Electric dipole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.4 Thin-wire antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.5 Wire antenna above a ground plane . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.6 Folded dipole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.7 Loop antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.7.1 Small loop antennas: magnetic dipole . . . . . . . . . . . . . . . . . . . . . 103
4.7.2 Large loop antennes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.8 Magnetic sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.9 Lorentz-Larmor theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.10 Horn antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.11 Reflector antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
4.12 Microstrip antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5 Numerical electromagnetic analysis 133

5.1 Method of moments (MoM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6 Phased Arrays and Smart Antennas 141

6.1 Linear phased arrays of isotropic antennas . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.1 Array factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.2 Uniform amplitude tapering . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.1.3 DFT/FFT relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.1.4 Grating lobes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.1.5 Amplitude tapering and pattern synthesis . . . . . . . . . . . . . . . . . . . 148
6.1.6 Directivity and taper efficiency . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.1.7 Beam broadening due to beam scanning . . . . . . . . . . . . . . . . . . . . 154
6.1.8 Phase shifters (PHS) versus true time delay (TTD) . . . . . . . . . . . . . . 155
6.2 Planar phased arrays of isotropic radiators . . . . . . . . . . . . . . . . . . . . . . . 156
6.3 Arrays of non-ideal antenna elements . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.4 Mutual coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
6.5 The effect of phase and amplitude errors . . . . . . . . . . . . . . . . . . . . . . . . 165
6.6 Noise Figure and Noise Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.7 Sparse arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
6.8 Calibration of phased-arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.9 Focal-plane arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
CONTENTS 3
A Vector calculus and coordinate systems 179

A.1 Vector algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
A.2 Vector operations in rectangular (Cartesian) coordinates . . . . . . . . . . . . . . . 179
A.3 Vector operations in cylindrical coordinates . . . . . . . . . . . . . . . . . . . . . . 180
A.4 Vector operations in spherical coordinates . . . . . . . . . . . . . . . . . . . . . . . 181
A.5 Commonly used vector products . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
A.6 Vector identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Bibliography 183
4 CONTENTS
Chapter 1
Introduction
1.1 Purpose of this textbook

This textbook provides all relevant material for Master-level courses in the domain of antenna
systems. For example, at Eindhoven University of Technology, we use this book for two courses
(total of 10 ECTS) in the master program of Electrical Engineering, distributed over a semester.
The book includes comprehensive material on antennas and provides a solid introduction into
microwave engineering, ranging from passive components to active circuits. We believe that this
is the perfect mixture of know-how for a junior antenna expert. The theoretical material in this
textbook can be supplemented by labs in which the students learn how to use state-of-the-art
antenna and microwave design tools and test equipment, such as a vector network analyser or a
near-field scanner.
Several chapters in this book can be used quite independent from each other. For example,
chapter 6 can be used in a dedicated phased-array antenna course without the need to use Maxwell-
based antenna theory from chapter 4. Similarly, chapter 3 does not require deep back-ground
knowledge in electromagnetics or antennas. The antenna theory presented in chapter 4 is inspired
by the original lecture notes of dr. Martin Jeuken [1].
1.2 Antenna functionality

According to the Webster dictionary, an antenna is:
i. one of a pair of slender, movable, segmented sensory organs on the head of insects, myriapods,
and crustaceans,
ii. a usually metallic device (such as a rod or wire) for radiating or receiving radio waves.
The second definition of an antenna is used within the domain of Electrical Engineering. An
antenna transforms an electromagnetic wave in free space in to a guided wave that propagates
along a transmission line and vice-versa. In addition, antennas often provide spatial selectivity,
which means that the antenna radiates more energy in a certain direction (transmit case) or
is more sensitive in a certain direction (receive case). Traditional antenna concepts, like wire
antennas, are passive electromagnetic devices which implies that the reciprocity concept holds.
5
6 CHAPTER 1. INTRODUCTION
As a result, the transmit and receive properties are identical. In integrated antenna concepts,
where the transmit and receive electronics are part of the antenna functionality, reciprocity cannot
always be applied.
1.3 History of antennas

One of the principal characteristics of human beings is that they almost continually send and
receive signals to and from one another, [3]. The exchange of meaningful signals is the heart of
what is called communication. In its simplest form, communication involves two people, namely
the signal transmitter and the signal receiver. These signals can take many forms. Words are
the most common form. They can be either written or spoken. Before the invention of technical
resources such as radio communication or telephone, long-distance communication was very dif-
ficult and usually took a lot of time. Proper long-distance communication was at that time only
possible by exchange of written words. Couriers were used to transport the message from the
sender to the receiver. Since the invention of radio and telephone, far-distance communication or
telecommunication is possible, not only of written words, but also of spoken words and almost
without any time delay between the transmission and the reception of the signal. Antennas have
played an important part in the development of our present telecommunication services. Antennas
have made it possible to communicate at far distances, without the need of a physical connection
between the sender and receiver. Apart from the sender and receiver, there is a third important
element in a communication system, namely the propagation channel. The transmitted signals
may deteriorate when they propagate through this channel.
In 1886, Heinrich Hertz, who was a professor of physics at the Technical Institute in Karl-
sruhe, was the first person who made a complete radio system [4]. When he produced sparks at
a gap of the transmitting antenna, sparking also occurred at the gap of the receiving antenna.
Hertz in fact visualised the theoretical postulations of James Clerk Maxwell. Hertz’s first exper-
iments used wavelengths of about 8 meters, as illustrated in Fig. 1.1. After Hertz the Italian
Guglielmo Marconi became the motor behind the development of practical radio systems [5]. He
was not a famous scientist like Hertz, but he was obsessed with the idea of sending messages
with a wireless communication system. He was the first who performed wireless communications
across the Atlantic. The antennas that Marconi used were very large wire antennas mounted onto
two 60-meter wooden poles. These antennas had a very poor efficiency, so a lot of input power
had to be used. Sometimes the antenna wires even glowed at night. In later years antennas were
also used for other purposes such as radar systems and radio astronomy. In 1953, Deschamps
[6] reported for the first time about planar microstrip antennas, also known as patch antennas.
It was only in the early eighties that microstrip antennas became an interesting topic among
scientists and antenna manufacturers. Microstrip antennas are an example of so-called printed
antennas which are metal structures printed on a substrate (e.g. printed-circuit-board (PCB)).
Figure 1.2 shows some well-known antenna configurations which are nowadays used in a variety
of applications. Reflector antennas, horn antennas, wire antennas and printed antennas all have
become mass products. Reflector antennas are often used for the reception of satellite television,
whereas wire antennas are commonly used to receive radio signals with a car or portable radio
receiver. Printed antennas are also widely used nowadays, for example in smart phones. Printed
1.3. HISTORY OF ANTENNAS 7
Figure 1.1: First radio system of Heinrich Hertz operating at a wavelength of about 8 meter [4].
Reflector antenna (Westerbork Synthesis Radio Telescope) Wire antenna
Horn antenna Microstrip antenna
Metallic patch
εr
ground plane
Figure 1.2: Some well-known antenna concepts.

antennas can be realized as two-dimensional structures on PCBs or as three-dimensional struc-

tures on plastic substrates. More recently, printed antennas have even been realized on chips in
integrated circuits (ICs)[7].
Reflector antennas provide high directivity (spatial selectivity), but have the disadvantage
that the main lobe of the antenna has to be steered in the desired direction by means of a
highly accurate mechanical steering mechanism. This means that simultaneous communication
with several points in space is not possible. Wire and printed antennas are usually more omni-
directional, resulting in a low directivity. There are certain applications where these conventional
antennas cannot be used. These applications often require a phased-array antenna. A phased-
array antenna has the capability to communicate with several targets which may be anywhere in
space, simultaneously and continuously, because the main beam of the antenna can be directed
electronically into a certain direction. Another advantage of phased-array antennas is the fact that
they are relatively flat. Figure 1.3 shows two examples of phased arrays used in radio astronomy
and in base stations for future mm-wave 5G and beyond-5G wireless communications, respectively.
In a phased-array system, three essential layers can be distinguished 1) an antenna layer, 2) a layer
(a) (b)
Figure 1.3: Examples of phased-array antennas. (a) The one-square-meter-array (OSMA) was
a test-bed of the square-kilometer-array (SKA) and consists of an active center of 64 bow-tie
antennas with electric beamforming. The active center is surrounded by two rows of passive
bow-tie antennas [8], [9], [10]. (b) Dual-polarized active phased array prototype for mm-wave
(26.5-29.5 GHz) 5G/beyond-5G wireless communications [11], [12].
with transmitter and receiver modules (T/R modules) and 3) a signal-processing and control layer
that controls the direction of the main beam of the array. The antenna layer consists of several
individual antenna elements which are placed on a rectangular or on a triangular grid. Arrays
of antennas can also be used to realize multiple-input-multiple-output (MIMO) communication
Primary
Secondary
SERVICE
1.4
FIXED
MOBILE
current status of U.S. allocations.

AMATEUR
SATELLITE
SATELLITE
Mobile
FIXED
ACTIVITY CODE
AERONAUTICAL
BROADCASTING
BROADCASTING
AERONAUTICAL
AERONAUTICAL
EXAMPLE
FIXED SATELLITE
FEDERAL EXCLUSIVE
RADIONAVIGATION
MOBILE SATELLITE
AMATEUR SATELLITE
EARTH EXPLORATION
NON-FEDERAL EXCLUSIVE
JANUARY 2016
Capital Letters
DESCRIPTION
MOBILE
MARITIME
Office of Spectrum Management

SATELLITE
SATELLITE
SATELLITE
ALLOCATION USAGE DESIGNATION

LAND MOBILE
LAND MOBILE
INTER-SATELLITE
METEOROLOGICAL
METEOROLOGICAL
MOBILE SATELLITE
RADIONAVIGATION
1st Capital with lower case letters

MARITIME MOBILE
MARITIME MOBILE
NTIA. As such, it may not completely reflect all aspects, i.e. footnotes and recent changes made to the Table
of Frequency Allocations. Therefore, for complete information, users should consult the Table to determine the
This chart is a graphic single-point-in-time portrayal of the Table of Frequency Allocations used by the FCC and
RADIO SERVICES COLOR LEGEND
capacity [13].
U.S. DEPARTMENT OF COMMERCE
Facsimile: (202) 512-2250 Mail: Stop SSOP, Washington, DC 20402-0001

For sale by the Superintendent of Documents, U.S. Government Printing Office
Internet: bookstore.gpo.gov Phone toll free (866) 512-1800; Washington, DC area (202) 512-1800
FEDERAL/NON-FEDERAL SHARED
SATELLITE
National Telecommunications and Information Administration

STATES
RADIOLOCATION
SPACE RESEARCH
AND TIME SIGNAL

SPACE OPERATION
RADIONAVIGATION
RADIO ASTRONOMY
UNITED
THE RADIO SPECTRUM
RADIODETERMINATION
STANDARD FREQUENCY
TIME SIGNAL SATELLITE

RADIOLOCATION SATELLITE
ALLOCATIONS
STANDARD FREQUENCY AND

FREQUENCY
RADIONAVIGATION SATELLITE
30.0 3.0 300.0 30.0 3.0 300 0
MARITIME FIXED MOBILE AERONAUTICAL
FIXED - MOBILE - RADIONAVIGATION RADIOLOCATION Radiolocation FIXED MOBILE 30.56 MOBILE (R)
SATELLITE SATELLITE 3.1 3.025
and
(Earth-to-space) (Earth-to-space) 328.6 AERONAUTICAL
Satellite
0 kHz
Space research FIXED
Time Signal
Earth
31.0 RADIOLOCATION Radiolocation LAND MOBILE
3MHz
(active)
AERONAUTICAL RADIONAVIGATION
satellite
MOBILE (OR)
(space-to-Earth)
exploration-
3 GHz
MARITIME
Standard Frequency
FIXED MOBILE (active) 335.4
30GHz
Aeronautical
3.155
300 kHz
(radiobeacons)
(radiobeacons)
30 MHz
Radionavigation
3.3 32.0
3GHz
SPACE EARTH
31.3 FIXED MOBILE MOBILE
300 MHz
RADIONAVIGATION
RADIO except aeronautical mobile (R)
FIXED
RESEARCH EXPLORATION -
BY
ASTRONOMY (passive) SATTELLITE (passive) Radiolocation Radiolocation 399.9 FIXED MOBILE 3.23 325
Amateur MOBILE SATELLITE Maritime AERONAUTICAL
31.8 RADIONAVIGATION SATELLITE Aeronautical
SPACE RESEARCH (Earth-to-space) 33.0 Radionavigation RADIONAVIGATION
(deep space) (space-to-Earth) RADIONAVIGATION 3.5 400.05 Mobile
32.3 AERONAUTICAL STANDARD FREQUECY AND TIME SIGNAL - SATELLITE (radiobeacons) (radiobeacons)
335
mobile
RADIONAVIGATION
FIXED
RADIOLOCATION Radiolocation (400.1 MHz) FIXED LAND MOBILE
** EXCEPT AERONAUTICAL MOBILE

(ground based)
MOBILE
400.15
Radiolocation
RADIONAVIGATION INTER-SATELLITE 3.6
except aeronautical
AERONAUTICAL MET. AIDS MOBILE SPACE RES. Space Opn. MET. SAT.
* EXCEPT AERONAUTICAL MOBILE (R)

FIXED-SATELLITE (Radiosonde) SAT (S-E) (S-E) (S-E) (S-E)
RADIONAVIGATION RADIOLOCATION 3.4
JUNE 1, 2011
33.0 (ground based) (space-to-Earth) Radiolocation 401.0 34.0
MET-SAT. EARTH Met-Satellite
RADIONAVIGATION 3.65 MET. AIDS SPACE OPN. EXPL Earth Expl Sat AERONAUTICAL
33.4 FIXED-SATELLITE (Radiosonde) (S-E) (E-S) SAT. (E-S) (E-S) (E-S) FIXED MOBILE MOBILE (R)
RADIOLOCATION Radiolocation MOBILE** (space-to-Earth) FIXED 402.0
3.7 MET. AIDS MET-SAT. EARTH EXPL Met-Satellite Earth Expl Sat 35.0 3.5
34.2 (Radiosonde) (E-S) SAT. (E-S) (E-S) (E-S)
Space research
(deep space) Radio- SPACE RESEARCH RADIO- 403.0
(Earth-to-space) location (deep space) (Earth-to-space) LOCATION METEOROLOGICAL AIDS (RADIOSONDE) FIXED
34.7 406.0
LAND MOBILE
DELMON C. MORRISON
RADIOLOCATION MOBILE SATELLITE (Earth-to-space)
Mobile
Radiolocation 406.1
35.5 36.0
(radiobeacons)
Earth EARTH RADIO
AERONAUTICAL
Aeronautical
FIXED MOBILE
AMATEUR
RADIONAVIGATION
exploration - EXPLORATION -
THIS CHART WAS CREATED

(active)
SPACE
RADIO
ASTRONOMY
Radio
Space
LOCATION
(active)
location
RESEARCH
SATTELLITE (active)
FIXED MOBILE
research
sattellite (active) 410.0
SPACE EARTH EXPLORATION -
36.0 SPACE RESEARCH
FIXED MOBILE (space-to-space) 37.0
FIXED MOBILE RESEARCH SATELLITE 420.0
FIXED
(passive) (passive)
LAND MOBILE
SPACE RESEARCH
37.0 37.5
MOBILE FIXED Radio astronomy LAND MOBILE
(space-to-Earth)
(space-to-Earth) RADIOLOCATION Amateur 38.0
FIXED-SATELLITE
37.5 RADIO ASTRONOMY
SPACE RESEARCH FIXED-SATELLITE MOBILE FIXED 4.0
MOBILE FIXED (space-to-Earth) 38.25
(space-to-Earth) 450.0 FIXED MARITIME MOBILE 405
38.0 LAND MOBILE FIXED MOBILE 4.063 Aeronautical Mobile RADIONAVIGATION
39.0
NOT ALLOCATED
FIXED-SATELLITE 454.0 415
MOBILE FIXED FIXED LAND MOBILE LAND MOBILE
(space-to-Earth) 455.0 40.0
38.6 4.2 LAND MOBILE AERONAUTICAL
456.0 MARITIME
MOBILE FIXED-SATELLITE FIXED FIXED LAND MOBILE
(space-to-Earth) 460.0 FIXED MOBILE RADIONAVIGATION MOBILE
39.5 AERONAUTICAL LAND MOBILE FIXED
FIXED-SATELLITE
462.5375 435
MOBIL-ESATELLITE
MOBILE
FIXED RADIONAVIGATION LAND MOBILE
MARITIME
(space-to-Earth) (space-to-Earth)
MOBILE 462.7375 42.0
40.0 LAND MOBILE FIXED
467.5375
Earth exploration EARTH EXPLORATION
Satellite
satellite SATELLITE
4.4 LAND MOBILE
FIXED-
SPACE
MOBILE
SATELLITE
SATELLITE
467.7375
RESEARCH
4.438
Meteorological
(Earth-to-space)
(space-to-Earth)
(space-to-Earth)
(space-to-Earth)
(Earth-to-space)
(space-to-Earth)
FIXED MOBILE LAND MOBILE FIXED FIXED LAND MOBILE
40.5 470.0
BROADCASTING- 4.5 MOBILE FIXED
BROADCASTING
MOBILE-
except aeronautical mobile (R)
SATELLITE
SATELLITE 43.69
(space-to-Earth)
(space-to-Earth)
FIXED-SATELLITE
Mobile
41.0 4.65
ISM - 40.68 ± .02 MHz

Maritime
Aeronautical
FIXED
(TELEVISION)
BROADCASTING
FIXED- BROADCASTING- AERONAUTICAL MOBILE (R)
Radionavigation
LAND MOBILE
BROADCASTING
FIXED
SATELLITE SATELLITE LAND MOBILE 4.7
Fixed FIXED
MOBILE
Mobile MOBILE
(space-to-Earth)
(space-to-Earth)
512.0
FIXED-SATELLITE
42.0 AERONAUTICAL MOBILE (OR)
BROADCASTING SATELLITE BROADCASTING MOBILE FIXED 4.8 4.75
42.5 46.6 MOBILE
FIXED-SATELLITE FIXED
RADIO ASTRONOMY (Earth-to-space)
MOBILE** FIXED FIXED MOBILE FIXED MOBILE except aeronautical mobile (R) 495
43.5 4.94 47.0 4.85
MOBILE-SATELLITE (Earth-to-space)
MOBILE (distress and calling)
FIXED-SATELLITE (Earth-to-space) FIXED MOBILE**
45.5 FIXED MOBILE 505
RADIONAVIGATION- 4.99
SATELLITE MOBILE-SATELLITE (Earth-to-space) MOBILE MARITIME MOBILE
46.9 RADIO ASTRONOMY Space Research (passive) LAND MOBILE 4.995 510
RADIO- 5.0
NAVIGATION- MOBILE-SATELLITE RADIONAVIGATION-SATELLITE STANDARD FREQUENCY AND TIME SIGNAL (5 MHz)
MOBILE FIXED AERONAUTICAL RADIONAVIGATION 5.005 AERONAUTICAL 9
SATELLITE (Earth-to-space) (Earth-to-space) MARITIME MOBILE
47.0 5.01 49.6 FIXED RADIONAVIGATION
AMATEUR AMATEUR-SATELLITE AERONAUTICAL RADIONAVIGATION
RADIONAVIGATION-SATELLITE (ships only)
47.2 (space-to-Earth)(space-to-space) FIXED MOBILE 5.06 (radiobeacons)
FIXED-SATELLITE (Earth-to-space) MOBILE FIXED 5.03 50.0
48.2 525
(TELEVISION)
FIXED-SATELLITE (Earth-to-space)
BROADCASTING
MOBILE FIXED AERONAUTICAL RADIONAVIGATION AERONAUTICAL
50.2 MOBILE
mobile
SPACE RESEARCH EARTH EXPLORATION- 5.15 RADIONAVIGATION
FIXED
MOBILE
(passive) SATELLITE (passive) AERONAUTICAL FIXED-SATELLITE (radiobeacons)
except aeronautical
50.4 RADIONAVIGATION (Earth-to-space)
5.25 535
MOBILE-SATELLITE (Earth-to-space) 5.45
AMATEUR
FIXED
MOBILE
FIXED-SATELLITE (Earth-to-space) 608.0
Earth
RADIOLOCATION LAND MOBILE 54.0
(active)
(active)
EARTH
51.4 (medical telemetry and
AERONAUTICAL MOBILE (R)
SATELLITE
exploration-
MOBILE FIXED RADIO ASTRONOMY
Radiolocation
medical telecommand)
satellite (active)
Space research
SPACE RESEARCH
52.6 614.0
EARTH EXPLORATION-SATELLITE (passive) SPACE RESEARCH (passive) 5.255 5.68
SPACE RESEARCH
(passive) INTER- SATELLITE EARTH EXPLORATION-SATELLITE (passive)
5.73
1.4. ELECTROMAGNETIC SPECTRUM
55.78
Earth
RADIOLOCATION
(active)
EARTH EXPLORATION-SATELLITE (passive)
(active)
(active)
EARTH
INTER- MOBILE
exploration-
RADIONAVIGATION
SATELLITE
MOBILE FIXED
Radiolocation
Space research
satellite (active)
SPACE RESEARCH (passive)
except aeronautical mobile (R)
SATELLITE 5.35 5.59
56.9
INTER-
Earth
SATELLITE
(active)
(active)
BROADCASTING
(active)
EARTH
exploration-
SATELLITE
57.0
AERONAUTICAL
RADIOLOCATION
Space research
satellite (active)
Radiolocation
RADIONAVIGATION
(TELEVISION)
INTER- 5.46
SATELLITE SPACE RESEARCH (passive)
6.2 14
BROADCASTING
SPACE
58.2
Earth
EARTH EXPLORATION-
(active)
(active)
(active)
EARTH
RESEARCH
exploration-
SATELLITE
SATELLITE (passive)
FIXED FIXED FIXED FIXED
MOBILE MOBILE MOBILE

RADIOLOCATION
Space research
satellite (active)
Radiolocation
RADIONAVIGATION
SPACE RESEARCH SPACE RESEARCH SPACE RESEARCH
(passive)
EXPLORATION- EXPLORATION- EXPLORATION- EXPLORATION-

MARITIME MOBILE
Electromagnetic spectrum
59.0 5.47 698.0

RADIO- INTER- MARITIME BROADCASTING
LOCATION SATELLITE RADIONAVIGATION FIXED MOBILE
MOBILE
FIXED
(TELEVISION)
6.525
Earth
(active)
(active)
EARTH
(active)
59.3
exploration-
SATELLITE
763.0
ISM – 5.8 ± .075 GHz

RADIOLOCATION
Space research
INTER-
satellite (active)
SPACE RESEARCH
EXPLORATION-
MOBILE FIXED RADIOLOCATION
AERONAUTICAL MOBILE (R)
SATELLITE 5.57 FIXED MOBILE 6.685
(TELEVISION )
64.0 RADIOLOCATION MARITIME RADIONAVIGATION
BROADCASTING
Fixed
MARITIME METEOROLOGICAL 775.0
FIXED
INTER- RADIOLOCATION 6.765
MOBILE** FIXED RADIONAVIGATION AIDS
SATELLITE 5.65 BROADCASTING
RADIOLOCATION Amateur FIXED MOBILE MOBILE
65.0 FIXED
EARTH EXPLORATION-SATELLITE 5.83 except aeronautical mobile (R)
INTER- Amateur-satellite 793.0
MOBILE **
ISM - 61.25± 0.25 GHz

FIXED SATELLITE
RADIOLOCATION Amateur
MARITIME MOBILE
SPACE RESEARCH (space-to-Earth) 7.0
66.0 FIXED-SATELLITE
5.85 FIXED MOBILE
RADIO RADIOLOCATION MOBILE Amateur AMATEUR AMATEUR SATELLITE
MOBILE- RADIO INTER- (Earth-to-space)
ISM - 6.78 ± .015 MHz

NAVIGATION- 5.925 805.0 7.1
MOBILE SATELLITE NAVIGATION SATELLITE FIXED SATELLITE
SATELLITE FIXED
71.0 (Earth-to-space) FIXED MOBILE BROADCASTING
6.425 19.95
FIXED-SATELLITE AMATEUR
(Earth-to-space) MOBILE 806.0
LAND MOBILE
Earth)
STANDARD FREQUENCY AND TIME SIGNAL (20 kHz)
MOBILE
FIXED
Earth)
6.525
(space-to-
FIXED-
809.0
MOBILE
SATELLITE
FIXED-SATELLITE
(space-to-
FIXED LAND MOBILE 72.0
74.0 (Earth-to-space) FIXED 849.0 FIXED MOBILE 7.3
6.7 AERONAUTICAL MOBILE 73.0 BROADCASTING
20.05
BROADCASTING FIXED-SATELLITE (Earth-to-space)(space-to-Earth) FIXED
6.875 851.0 7.4
Earth)
Space
BROADCASTING
Earth)
FIXED-SATELLITE (Earth-to-space)(space-to-Earth) LAND MOBILE
research
FIXED-
MOBILE FIXED
(space-to-
SATELLITE FIXED
RADIO ASTRONOMY
MOBILE
(space-to-
854.0
SATELLITE
SATELLITE 7.025 74.6
76.0 MOBILE FIXED-SATELLITE (Earth-to-space) FIXED FIXED LAND MOBILE FIXED MOBILE
Space research RADIO 7.075 894.0
(space-to-Earth)
RADIOLOCATION Amateur AERONAUTICAL MOBILE 74.8
ASTRONOMY MOBILE FIXED
77.0 896.0 AERONAUTICAL RADIONAVIGATION
mobile (R)
FIXED
Space research RADIO 7.125 75.2
MOBILE
RADIOLOCATION Amateur-satellite Amateur FIXED LAND MOBILE
(space-to-Earth) ASTRONOMY FIXED MOBILE
except aeronautical
7.145 901.0 FIXED
RADIO
77.5 FIXED MOBILE 75.4
Space research FIXED SPACE RESEARCH (deep space)(Earth-to-space) FIXED MOBILE
(space-to-Earth) ASTRONOMY Amateur-satellite AMATEUR 7.19 902.0 8.1
78.0 RADIOLOCATION 76.0 MARITIME MOBILE FIXED
Space research RADIO FIXED SPACE RESEARCH (Earth-to-space)
RADIOLOCATION Amateur- Amateur 7.235 928.0 8.195
(space-to-Earth) ASTRONOMY satellite
81.0 FIXED FIXED
7.25 929.0
(AM RADIO)
FIXED-SATELLITE (space-to-Earth) MOBILE-SATELLITE (space-to-Earth) Fixed FIXED LAND MOBILE MARITIME MOBILE
930.0
Space
RADIO
BROADCASTING
FIXED-
7.3
research
FIXED
MOBILE-
FIXED MOBILE
MOBILE
SATELLITE
SATELLITE
8.815
(Earth-to-space)
ASTRONOMY
FIXED-SATELLITE (space-to-Earth) Mobile-satellite (space-to-Earth)
(space-to-Earth)
FIXED
ISM - 915.0± .13 MHz

931.0 AERONAUTICAL MOBILE (R)
84.0 METEOROLOGICAL Mobile-satellite
7.45 FIXED LAND MOBILE
FIXED-SATELLITE (space-to-Earth)
SATELLITE (space-to-Earth) FIXED 932.0 8.965
(space-to-Earth) FIXED AERONAUTICAL MOBILE (OR)
RADIO 7.55 9.04
935.0
FIXED-
FIXED-SATELLITE (space-to-Earth) Mobile-satellite (space-to-Earth) FIXED
FIXED
SATELLITE
ASTRONOMY FIXED LAND MOBILE FIXED
MOBILE
(Earth-to-space) (Earth-to-space)
(TELEVISION)
7.75 940.0 9.4
BROADCASTING
86.0 METEOROLOGICAL- FIXED MOBILE
SATELLITE (space-to-Earth) FIXED 941.0
7.85 FIXED
FIXED 944.0 BROADCASTING
7.9 FIXED
FIXED-SATELLITE (Earth-to-space) MOBILE-SATELLITE (Earth-to-space) Fixed
mation can be found at: www.ntia.doc.gov and www.itu.int

RADIO
SPACE
EARTH
960.0
(passive)
(passive)
Mobile-satellite
8.025 88.0
SATELLITE
RESEARCH
ASTRONOMY
EARTH EXPLORATION- FIXED SATELLITE (Earth-to-space)
9.9
EXPLORATION-
SATELLITE (space-to-Earth) (Earth-to-space) FIXED FIXED
92.0 (no airborne) 9.995
METEOROLOGICAL- Mobile-satellite
8.175 STANDARD FREQUENCY AND TIME SIGNAL (10 MHz)
FIXED-SATELLITE EARTH EXPLORATION- 10.005
RADIO- RADIO SATELLITE FIXED (Earth-to-space)
FIXED MOBILE (Earth-to-space) SATELLITE (space-to-Earth) AERONAUTICAL MOBILE (R)
(space-to-Earth) (no airborne) 10.1
LOCATION ASTRONOMY 8.215 AMATEUR
Mobile-satellite
94.0 EARTH EXPLORATION-SATELLITE FIXED-SATELLITE (Earth-to-space)
10.15
(space-to-Earth) (Earth-to-space) FIXED
(no airborne)
FIXED
FIXED Space research (deep space)(space-to-Earth)

8.4
RADIO SPACE RESEARCH (deep space)(space-to-Earth)
AERONAUTICAL
(R)
(active)
EARTH
(active)
SPACE
8.45
RADIONAVIGATION
except
(FM RADIO)
RADIO-
ASTRONOMY
SATELLITE
SPACE RESEARCH (space-to-Earth)
RESEARCH
FIXED
FIXED
Mobile
LOCATION
EXPLORATION-
8.5
BROADCASTING
aeronautical mobile
94.1 RADIOLOCATION Radiolocation
8.55 1164.0 11.175
MARITIME MOBILE
RADIO- RADIO RADIONAVIGATION-SATELLITE AERONAUTICAL

FIXED MOBILE
Earth
AERONAUTICAL MOBILE (OR)
EARTH
(active)
(active)
satellite
Space
RADIO-
Radio-
SPACE
LOCATION
SATELLITE
(active)
(active)
ASTRONOMY
research
location
exploration -
(space-to-Earth)(space-to-space) RADIONAVIGATION
LOCATION
EXPLORATION-
RESEARCH
95.0 1215.0 11.275
8.65 RADIONAVIGATION-
RADIOLOCATION Radiolocation SATELLITE AERONAUTICAL MOBILE (R)
Earth
(active)
EARTH
satellite
(active)
(active)
(active)
SPACE
RADIO-
9.0 (space-to-Earth)
SATELLITE
exploration-
RESEARCH
LOCATION
Space research
EXPLORATION-
AERONAUTICAL RADIONAVIGATION Radiolocation (space-to-space) 11.4
RADIO
RADIO-
RADIO-
FIXED
RADIO-
MOBILE
9.2
SATELLITE
108.0 FIXED
ASTRONOMY
NAVIGATION-
LOCATION
1240.0
NAVIGATION
MARITIME RADIONAVIGATION Radiolocation 11.6
100.0 9.3
Earth
(active)
(active)
EARTH
satellite
RADIO -
(active)
RADIONAVIGATION Meteorological Aids Radiolocation
RADIO-
SPACE
Amateur
(active)
SATELLITE
exploration-
EARTH
NAVIGATION
SPACE
LOCATION
Space research
AERONAUTICAL
EXPLORATION-
BROADCASTING
RESEARCH
RADIO EXPLORATION- 9.5
RESEARCH 1300.0 12.1
ASTRONOMY SATELLITE
Earth
EARTH
(active)
(active)
satellite
Space
RADIO-
Radio-
(passive)
SPACE
SATELLITE
(active)
(active)
location
exploration -
research
AERONAUTICAL RADIONAVIGATION
LOCATION
(passive) Radiolocation
EXPLORATION
FIXED
RESEARCH
102.0 1350.0
9.8 12.23
RADIO RADIOLOCATION Radiolocation FIXED MOBILE RADIOLOCATION
MOBILE FIXED 10.0 1390.0
ASTRONOMY RADIOLOCATION Radiolocation Amateur MOBILE ** Fixed-satellite (Earth-to-space)
10.45 FIXED MARITIME
105.0 1392.0
AERONAUTICAL
RADIOLOCATION Radiolocation Amateur Amateur-satellite MOBILE ** MOBILE
RADIONAVIGATION
SPACE 10.5 FIXED 1395.0
RESEARCH RADIO MOBILE RADIOLOCATION LAND MOBILE (medical telemetry and medical telecommand)
ASTRONOMY
FIXED 10.55 117.975 13.2
(passive) FIXED EARTH EXPLORATION - SATELLITE SPACE RESEARCH
109.5 10.6 (passive) RADIO ASTRONOMY (passive)
AERONAUTICAL
EARTH SPACE RESEARCH (passive) EARTH EXPLORATION-SATELLITE (passive) 1427.0 13.26
SPACE EXPLORATION-
FIXED MOBILE (R)
RADIO LAND MOBILE LAND MOBILE Fixed AERONAUTICAL MOBILE (R) Non-Federal Travelers Information Stations (TIS), a mobile service, are authorized in the 535-1705 kHz band.
RESEARCH SATELLITE RADIO ASTRONOMY SPACE RESEARCH (passive) EARTH EXPLORATION-SATELLITE (passive) 10.68 (medical telemetry and 121.9375 13.36
ASTRONOMY 10.7 medical telecommand (telemetry and telecommand) (telemetry) AERONAUTICAL MOBILE
(passive) (passive) 123.0875 RADIO ASTRONOMY
111.8 FIXED-SATELLITE (space-to-Earth) FIXED (telemetry and
1429.5 13.41
FIXED LAND MOBILE (telemetry & telecommand) AERONAUTICAL MOBILE Mobile
SPACE 11.7 telecommand) 123.5875 FIXED
RADIO FIXED (telemetry and Fixed-satellite
1430.0 except aeronautical mobile (R)
FIXED MOBILE RESEARCH FIXED-SATELLITE (space-to-Earth) LAND MOBILE
ASTRONOMY telecommand) (telemetry & telecommand) (space-to-Earth) 13.57
(passive) 12.2 1432.0 AERONAUTICAL
114.25 BROADCASTING-SATELLITE FIXED FIXED MOBILE **
BROADCASTING
SPACE EARTH 12.7 1435.0
MOBILE (R) 13.87
RADIO EXPLORATION- FIXED-SATELLITE (Earth-to-space) MOBILE FIXED 59
RESEARCH 13.25 MOBILE (aeronautical telemetry) 128.8125 FIXED Mobile
ASTRONOMY SATELLITE 1525.0
(passive) AERONAUTICAL except aeronautical mobile (R) STANDARD FREQUENCY AND TIME SIGNAL (60 kHz)
(passive) Aeronatuical
Earth
(active)
(active)
EARTH
MOBILE SATELLITE (space-to-Earth)
satellite
116.0
Space
14.0
SPACE
(active)
(active)
SATELLITE
exploration -
research
EARTH MOBILE (R) 61
EXPLORATION -
RESEARCH
SPACE Radionavigation 1559.0 132.0125 AMATEUR AMATEUR SATELLITE
INTER- EXPLORATION- 13.4 RADIONAVIGATION-SATELLITE AERONAUTICAL
RESEARCH SATELLITE AERONAUTICAL 14.25
SATELLITE (passive) (space-to-Earth)(space-to-space) RADIONAVIGATION AERONAUTICAL MOBILE (R)
Earth
(active)
EARTH
(passive) AMATEUR
(active)
satellite
136.0
SPACE
Radio-
Space
SATELLITE
location
RADIO -
exploration -
1610.0
research
EXPLORATION
RESEARCH
LOCATION
122.25 RADIODETERMINATION- AERONAUTICAL MOBILE SATELLITE 14.35
INTER-SATELLITE Amateur
ISM - 122.5± 0.500 GHz

FIXED MOBILE 13.75
AERONAUTICAL MOBILE (R) Mobile
SATELLITE (Earth-to-space)
ISM - 13.560 ± .007 MHz

123.0 RADIONAVIGATION (Earth-to-space) 137.0 FIXED
MOBILE-SATELLITE SPACE RESEARCH SPACE OPERATION MET. SATELLITE
satellite
1610.6 except aeronautical mobile (R)
(space-to-Earth) (space-to-Earth) (space-to-Earth) (space-to-Earth)
FIXED
RADIO RADIODETERMINATION- AERONAUTICAL MOBILE SATELLITE

MOBILE
14.99
and time signal

Space
RADIO -
FIXED-
RADIONAVIGATION
(Earth-to-space)
Space
(Earth-to-space) 137.025
research
Radio -
SATELLITE (Earth-to-space)
MARITIME
LOCATION
STANDARD FREQUENCY AND TIME SIGNAL (15 MHz)
location
research
ASTRONOMY MET. SATELLITE
SATELLITE
SPACE RESEARCH SPACE OPERATION
MOBILE-
Mobile-satellite
FIXED-
Standard frequency
Radio
RADIO-
RADIO-
SATELLITE
(Earth-to-space)
15.01
SATELLITE
SATELLITE
(space-to-Earth)
1613.8 (space-to-Earth) (space-to-Earth)
NAVIGATION
astronomy
NAVIGATION-
(space-to-Earth)
14.0 Mobile-satellite RADIODETERMINATION- AERONAUTICAL MOBILE SATELLITE 70
Space RADIONAVIGATION
130.0 Mobile-satellite (Earth-to-space) FIXED-SATELLITE (Earth-to-space) (space-to-Earth) SATELLITE (Earth-to-space) (Earth-to-space) MOBILE-SATELLITE SPACE RESEARCH SPACE OPERATION MET. SATELLITE 15.1
EARTH research (space-to-Earth) (space-to-Earth) (space-to-Earth) (space-to-Earth)
EXPLORATION- RADIO 14.2 1626.5 137.825
Mobile-satellite (space-to-Earth) FIXED-SATELLITE (Earth-to-space) MOBILE SATELLITE(Earth-to-space) Mobile-satellite SPACE RESEARCH SPACE OPERATION MET. SATELLITE 1605
SATELLITE
INTER-
ASTRONOMY
FIXED
(space-to-Earth) (space-to-Earth) (space-to-Earth) (space-to-Earth) BROADCASTING
MOBILE
SATELLITE
(active) 14.4 1660.0 138.0 MOBILE BROADCASTING
134.0 Mobile-satellite FIXED-SATELLITE 1615
Fixed Mobile (Earth-to-space) (Earth-to-space) MOBILE SATELLITE
Radio astronomy AMATEUR AMATEUR - SATELLITE 14.5 RADIO ASTRONOMY (Earth-to-space) FIXED MOBILE 15.8
136.0 FIXED Mobile Space research 1660.5
RADIO RADIO 14.7145 144.0
Amateur Amateur - satellite MOBILE Fixed Space research
RADIO ASTRONOMY SPACE RESEARCH (passive) AMATEUR AMATEUR- SATELLITE
LOCATION ASTRONOMY 146.0 FIXED
FIXED
141.0 14.8 1668.4 AMATEUR

FIXED MOBILE RADIO ASTRONOMY RADIOLOCATION Fixed SPACE RESEARCH MOBILE METEOROLOGICAL AIDS 148.0
EARTH EXPLORATION-
148.5 15.1365 (radiosonde) RADIO ASTRONOMY MOBILE-SATELLITE
SPACE RESEARCH FIXED SPACE RESEARCH Mobile (Earth-to-space) FIXED MOBILE 16.36
Radiolocation
RADIO ASTRONOMY SATELLITE 15.35 1670.0 149.9

(passive) (passive) RESEARCH EARTH EXPLORATION - FIXED MOBILE **
Federal TIS operates at 1610 kHz.
RADIO ASTRONOMY SPACE(passive) SATELLITE (passive) 1675.0 RADIONAVIGATION- MOBILE-SATELLITE MARITIME

MARITIME MOBILE
BROADCASTING
RADIOLOCATION
151.5 15.4 METEOROLOGICAL AIDS SATELLITE (Earth-to-space)
FIXED MOBILE RADIO ASTRONOMY METEOROLOGICAL MOBILE
155.5 AERONAUTICAL RADIONAVIGATION SATELLITE (space-to-Earth) (radiosonde) 150.05
EARTH EXPLORATION- SPACE 15.43 17.41 90
RADIO SATELLITE AERONAUTICAL FIXED-SATELLITE 1695.0 FIXED MOBILE FIXED
MOBILE RESEARCH METEOROLOGICAL 150.8
FIXED
ASTRONOMY (passive) (passive) RADIONAVIGATION (Earth-to-space) SATELLITE (space-to-Earth) Fixed MOBILE ** FIXED LAND MOBILE 17.48 1705
158.5 15.63 1710.0 152.855
MOBILE- FIXED- AERONAUTICAL RADIONAVIGATION LAND MOBILE
BROADCASTING
15.7 FIXED MOBILE 17.9
FIXED MOBILE SATELLITE SATELLITE Radiolocation RADIOLOCATION 1761.0 154.0 AERONAUTICAL MOBILE (R)
16.6 SPACE OPERATION (Earth-to-space) FIXED LAND MOBILE
Space research (deep space)(Earth-to-space)
FIXED
MOBILE FIXED
RADIO-
Radiolocation RADIOLOCATION 17.97

MOBILE
1780.0 156.2475 AERONAUTICAL MOBILE (OR)

LOCATION
164.0 17.1 MARITIME MOBILE RADIONAVIGATION

EARTH RADIO SPACE RESEARCH Radiolocation RADIOLOCATION SPACE OPERATION (Earth-to-space) 156.7625 FIXED
18.03
EXPLORATION- 17.2 1850.0 MARITIME MOBILE (distress, urgency, safety and calling) 1800
SATELLITE (passive) ASTRONOMY (passive) MOBILE FIXED 18.068
2000.0 156.8375 AMATEUR AMATEUR SATELLITE
Earth
167.0 MARITIME MOBILE
(active)
EARTH
Space
MOBILE SATELLITE (Earth-to-space)
RADIO-
SPACE
(active)
(active)
SATELLITE
Radio-
research
exploration-
MOBILE FIXED
location
FIXED-
LOCATION
157.0375
EXPLORATION-
satellite (active)
RESEARCH
INTER- 2020.0 MARITIME MOBILE FIXED 18.168
FIXED MOBILE SATELLITE 17.3 Mobile AMATEUR
SATELLITE FIXED-SATELLITE FIXED MOBILE 157.1875 110
(space-to-Earth) BROADCASTING-SATELLITE Radiolocation 2025.0 MARITIME MOBILE 18.78
(Earth-to-space) MOBILE except aeronautical mobile
174.5 17.7 157.45 18.9
INTER- FIXED-SATELLITE (Earth-to-space) FIXED BROADCASTING
FIXED MOBILE 17.8 SPACE OPERATION FIXED LAND MOBILE 19.02 1900
SATELLITE FIXED FIXED-SATELLITE (space-to-Earth) (Earth-to-space) 161.575 FIXED
EARTH
SPACE
SATELLITE
18.3
FIXED
174.8 (space-to-space)
RESEARCH
MARITIME MOBILE
MOBILE
(Earth-to-space)
EXPLORATION-
(space-to-space)
(Earth-to-space)
19.68
(space-to-space)
INTER- SPACE RESEARCH EARTH FIXED-SATELLITE (space-to-Earth) 161.625 MARITIME MOBILE
EXPLORATION- 18.6 2110.0 LAND MOBILE
SATELLITE (passive) SATELLITE (passive) SPACE RESEARCH FIXED-SATELLITE EARTH EXPLORATION - FIXED
19.8 RADIOLOCATION
MOBILE FIXED 161.775
FIXED
182.0 (passive) (space-to-Earth) SATELLITE (passive) 19.99

MOBILE
EARTH 18.8 2180.0 MOBILE except aeronautical mobile STANDARD FREQUENCY AND TIME SIGNAL (20 MHz)
SPACE RESEARCH 161.9625
MARITIME
RADIO EXPLORATION- FIXED-SATELLITE (space-to-Earth) MOBILE SATELLITE MARITIME MOBILE (AIS) 20.01
Radiolocation
ASTRONOMY SATELLITE (passive) (passive) 19.3 MOBILE FIXED (space-to-Earth) FIXED Mobile 2000
185.0 FIXED-SATELLITE (space-to-Earth) FIXED 2200.0 161.9875
EARTH 19.7 MOBILE except aeronautical mobile 21.0 MARITIME 130
SPACE RESEARCH INTER- FIXED-SATELLITE (space-to-Earth) MOBILE-SATELLITE (space-to-Earth) 162.0125 AMATEUR AMATEUR SATELLITE MOBILE FIXED
EXPLORATION- MOBILE
(passive) SATELLITE SATELLITE (passive) MARITIME MOBILE (AIS) 21.45
aircraft)
20.2
EARTH
SPACE
SPACE
163.0375 BROADCASTING 2065
SATELLITE
190.0 FIXED-SATELLITE (space-to-Earth)
FIXED
RESEARCH
(ling of sight only
(space-to-Earth)
MOBILE
EXPLORATION-
OPERATION
(space-to-space)
(space-to-Earth)
(space-to-space)
21.85
including aeronautical
(line of sight only)
flight testing of manned

MARITIME MOBILE
telemetry, but excluding
(space-to-Earth)
SPACE RESEARCH
and
(space-to-space)
EARTH EXPLORATION- FIXED
Earth)
satellite
Standard
frequency
(space-to-
time signal
(passive) SATELLITE (passive) MOBILE-SATELLITE (space-to-Earth) SPACE RESEARCH
2290.0 MOBILE 2107
FIXED 21.924 MOBILE
191.8 21.2 (space-to-Earth) FIXED AERONAUTICAL MOBILE (R)
RADIONAVIGATION SPACE RESEARCH EARTH EXPLORATION - (deep space)
MOBILE** FIXED MOBILE
MOBILE FIXED 22.0 except aeronautical mobile
INTER- MOBILE (passive) SATELLITE (passive) 2300.0 173.2 MARITIME MOBILE
FIXED
RADIONAVIGATION- 2170
MOBILE
SATELLITE 21.4 Amateur FIXED Land mobile 22.855
FIXED
SATELLITE 173.4 FIXED
MARITIME
FIXED
MOBILE
SATELLITE MOBILE 2305.0 MARITIME MOBILE
200.0 22.0 RADIOLOCATION FIXED MOBILE 23.0 MARITIME MOBILE
FIXED MOBILE** Amateur MOBILE** FIXED 174.0 Mobile (telephony)
RADIO EARTH SPACE RESEARCH 2310.0 FIXED
22.21 BROADCASTING Radio- except aeronautical mobile (R)
EXPLORATION- RADIOLOCATION MOBILE FIXED SATELLITE location Mobile Fixed 2173.5
ASTRONOMY SATELLITE (passive) (passive) SPACE 2320.0 23.2 160
RADIO
EARTH
RESEARCH AERONAUTICAL MOBILE (OR) MOBILE (distress and calling)
(passive)
209.0 BROADCASTING - SATELLITE
SATELLITE
ASTRONOMY Radiolocation Fixed
FIXED
EXPLORATION-
MOBILE**
FIXED- (passive) 2345.0 MOBILE 23.35
BROADCASTING Radio- 2190.5
RADIO RADIOLOCATION MOBILE FIXED SATELLITE location Mobile Fixed FIXED
FIXED MOBILE SATELLITE 22.5 except aeronautical mobile
(Earth-to-space) ASTRONOMY FIXED MOBILE 2360.0 24.89
22.55 MOBILE Fixed RADIOLOC ATION MARITIME MOBILE
217.0 FIXED MOBILE INTER-SATELLITE 2390.0 AMATEUR AMATEUR SATELLITE MARITIME MOBILE
FIXED-SATELLITE (Earth-to-space) MOBILE AMATEUR 24.99 (telephony)
23.55
FIXED
RADIO FIXED MOBILE 2395.0 STANDARD FREQ. AND TIME SIGNAL (25 MHz)
MOBILE
FIXED
ASTRONOMY 23.6 25.01 2194
MOBILE
SPACE RESEARCH (passive) LAND MOBILE
MARITIME
RADIO SPACE RESEARCH EARTH EXPLORATION - AMATEUR

(TELEVISION)
226.0 ASTRONOMY SATELLITE - (passive) 25.07

BROADCASTING
(passive) 2417.0 MARITIME MOBILE

24.0 25.21 190
RADIO AMATEUR AMATEUR-SATELLITE Amateur Radiolocation LAND MOBILE
24.05 2450.0
SPACE
EARTH
ASTRONOMY 25.33
(Passive)
(Passive)
SATELLITE
Radiolocation MOBILE except aeronautical mobile
RESEARCH
Radio- MOBILE FIXED 216.0 FIXED
EXPLORATION-
RADIO- 2483.5 MOBILE
AERONAUTICAL
Earth
231.5 Amateur 25.55
(active)
satellite
location RADIODETERMINATION- MOBILE SATELLITE
MOBILE
LOCATION except aeronautical

FIXED
FIXED Fixed RADIO ASTRONOMY RADIONAVIGATION
exploration -
MOBILE
FIXED MOBILE SATELLITE (space-to-Earth) (space-to-Earth) mobile

Land mobile 25.67
Radiolocation 24.25 2495.0 BROADCASTING 200
232.0 FIXED RADIODETERMINATION- MOBILE SATELLITE MOBILE except
217.0 26.1
except aeronautical mobile
24.45 SATELLITE (space-to-Earth) MOBILE** FIXED FIXED MARITIME MOBILE

FIXED- RADIONAVIGATION (space-to-Earth) Land mobile aeronautical mobile FIXED
INTER-SATELLITE 2500.0 219.0 26.175
FIXED MOBILE SATELLITE Radiolocation 24.65 MOBILE except
SPECTRUM OCCUPIED.
INTER-SATELLITE RADIOLOCATION-SATELLITE (Earth-to-space) MOBILE** FIXED LAND MOBILE
(space-to-Earth) 24.75 Fixed Mobile FIXED aeronautical mobile Amateur 26.48 2495
FIXED-SATELLITE 2655.0 MOBILE
ISM - 2450.0± .50 MHz

235.0 RADIONAVIGATION 220.0 FIXED except aeronautical mobile
FIXED-SATELLITE SPACE RESEARCH EARTH (Earth-to-space) FIXED LAND MOBILE 26.95
EXPLORATION- FIXED-SATELLITE
25.05 222.0 STANDARD FREQ. AND TIME SIGNAL (2500kHz)
(space-to-Earth) FIXED
satellite
(passive)
(passive)
(passive)
SATELLITE (passive) AMATEUR

Radio
FIXED
FIXED
(Earth-to-space) 26.96
Space research
225.0
Earth exploration-
MOBILE**
238.0 MOBILE
astronomy
25.25
FIXED INTER-SATELLITE MOBILE 2690.0 except aeronautical mobile 2505
RADIO- RADIO- 25.5 27.23
Mobile
FIXED-
FIXED MOBILE except aeronautical mobile
RADIO-
ISM –FIXED
ISM - 245.0± 1 GHz

NAVIGATION NAVIGATION- SPACE RESEARCH
SATELLITE
LOCATION
MOBILE
SATELLITE
(space-to-Earth)
EARTH
RADIO
(passive)
(passive) 27.41
ISM - 27.12 ± .163 MHz
SATELLITE
Aeronautical
EXPLORATION-
240.0
ASTRONOMY
EARTH
SPACE
FIXED
Inter-satellite
LAND MOBILE
FIXED
SATELLITE
RESEARCH
(Earth-to-space)
MOBILE
AERONAUTICAL
(space-to-Earth)
2700.0
time signal satellite

(space-to-Earth)
EXPLORATION -
FIXED MOBILE RADIOLOCATION 27.54
Standard frequency and

INTER-SATELLITE
valid in the United States, but is very representative for the world-wide spectrum allocation.
FIXED
MOBILE
RADIONAVIGATION
METEOROLOGICAL AERONAUTICAL MOBILE
24.125 ± 0.125
Amateur-satellite
241.0 28.0
FIXED
RADIOLOCATION RADIOASTRONOMY Amateur 27.0 Radiolocation AIDS RADIONAVIGATION

MOBILE
248.0 Inter-satellite FIXED MOBILE INTER-SATELLITE AMATEUR AMATEUR SATELLITE

Radioastronomy AMATEUR-SATELLITE AMATEUR 2900.0
except aeronautical mobile
SPACE RESEARCH 250.0 27.5 29.7 275

EARTH EXPLORATION- LAND MOBILE
FIXED
RADIOASTRONOMY FIXED-SATELLITE 29.8 AERONAUTICAL Maritime

MOBILE
(passive) SATELLITE (passive) Aeronautical

252.0 FIXED MOBILE FIXED Radionavigation
(Earth-to-space) RADIONAVIGATION Mobile (radiobeacons)
RADIONAVIGATION-SATELLITE RADIO MOBILE-SATELLITE 29.89 2850
29.5 FIXED 285
RADIO-
FIXED
ASTRONOMY (Earth-to-space) MOBILE
MOBILE
MARITIME
RADIO NAVIGATION 29.91 AERONAUTICAL MARITIME RADIONAVIGATION Aeronautical Radionavigation

NAVIGATION
MOBILE-SATELLITE
Radiolocation
FIXED-SATELLITE 265.0 FIXED-SATELLITE

RADIOLOCATION
RADIO ASTRONOMY (Earth-to-space) MOBILE (R)
300 GHz
FIXED MOBILE (Earth-to-space) (Earth-to-space) FIXED (radiobeacons) (radiobeacons)
275.0
30 GHz
300 kHz
3 GHz
300 MHz
3 MHz
30 MHz
PLEASE NOTE: THE SPACING ALLOTTED THE SERVICES IN THE SPECTRUM

SEGMENTS SHOWN IS NOT PROPORTIONAL TO THE ACTUAL AMOUNT OF
NOT ALLOCATED 300.0 30.0 3000.0 300.0 30.0 3000 300
radar, radio astronomy and license-free Industrial-Scientific-Medical (ISM) bands. More infor-
Applications include wireless communication, satellite communications, military and commercial
Figure 1.4: Example of spectrum allocation, frequency range 0-300 GHz, United States, 2016.
allocation chart of the radio-frequency spectrum is shown in Fig. 1.4. This particular map is
treaty which defines the use of the world-wide radio-frequency spectrum. An example of such an
Conferences (WRCs) to determine the international Radio Regulations, which is the international
International Telecommunication Union (ITU). The ITU organizes World Radiocommunication
range which can be used for a specific application is restricted and regulated worldwide by the
Antenna systems radiate electromagnetic energy at a particular frequency. The specific frequency
between two devices (e.g. a base station and mobile user), resulting in a much higher channel
systems. In such a system several non-correlated spatial communication channels can be realized
9
Chapter 2
System-level antenna parameters
2.1 Coordinate system and time-harmonic fields

Radiation properties of antennas are generally expressed in terms of spherical coordinates. The
spherical coordinate system is shown in Figure 2.1. In this spherical coordinate system we will
use unit vectors along the r, θ and φ directions which are denoted by ~ur , ~uθ and ~uφ , respec-
tively. A particular point P (x, y, z) is indicated by the position vector ~r = x~ux + y~uy + z~uz =
p
x2 + y 2 + z 2 ~ur = |~r|~ur .
P(x,y,z)
dS
θ r
y
φ
Figure 2.1: Spherical coordinate system with dS = r2 sin θdθdφ = r2 dΩ, where dΩ = sin θdθdφ
and Ω is the solid angle.
In this book we will assume that all currents, voltages and electric and magnetic field
components have a sinusoidal time variation. The instantaneous time-domain electric field E~ is
~ according to:
now related to a complex electric field E
~ r, t) = Re[E(~
E(~ ~ r)eωt ] = |E(~
~ r)| cos (ωt + ϕ0 ), (2.1)
11
12 CHAPTER 2. SYSTEM-LEVEL ANTENNA PARAMETERS
~ r), ω = 2πf is the angular frequency, f is

where ϕ0 is the phase of the complex electric field E(~
the frequency of operation and Re[] denotes the real part of the complex variable. The electric
~ r) in the space-frequency domain can now be expressed in terms of spherical coordinates:
field E(~
~ r) = Er (~r)~ur + Eθ (~r)~uθ + Eφ (~r)~uφ .

E(~ (2.2)
2.2 Field regions

An antenna transforms a guided electromagnetic (EM) wave that propagates along a transmission
line into radiated waves that propagate in free space. These radiated waves can, in turn, be
received by another antenna. Due to reciprocity, the receiving antenna will transform the incident
EM waves into a guided EM wave along a transmission line. One could also interpret this as the
transition of electrons in conductors to photons in free space. The transition from a guided wave
along a transmission line to radiated waves from the antenna is illustrated in Fig. 2.2.
E-field direction at t=t0
~
Generator Transmission line
with a guided TEM wave
Antenna
Free-space spherical wave
Figure 2.2: Transition of a guided wave along a transmission line into radiated spherical waves
in free space using an antenna.
The electromagnetic field radiated by an antenna can be divided into three main regions: 1)
the near-field region, 2) transition or Fresnel region and 3) the far-field region. Fig. 2.3 shows an
example of an horn antenna and the corresponding field regions. In the far-field region, the electric
and magnetic field components are orthogonal to each other and to the direction of propagation,
resulting in transverse electromagnetic waves, also known as plane waves. For an antenna located
in the origin of the coordinate system of Fig. 2.1, the electromagnetic energy will flow in the
~ur direction. There is not a very clear transition between the three field regions. However, it is
common to use the so-called Fraunhofer distance to define the start of the far-field region:
2L2
R= , (2.3)
λ0
2.3. FAR FIELD PROPERTIES 13
Near Field Fresnel Region Far Field Region
+ L
-
Antenna
L3 2 L2
R = 0.62 R=
λ0 λ0
Figure 2.3: Definition of three field regions around an antenna.
c
where R is the radial distance from the antenna, L is the largest dimension of the antenna, λ0 =
f
is the free-space wavelength of the radiated electromagnetic wave and c = 3·108 [m/s] is the speed
of light. In chapter 4 we will show how criterium (2.3) can be derived directly from the theoretical
antenna framework that we will introduce.
2.3 Far field properties
Consider a transmit antenna located in the origin of the coordinate system of Fig. 2.1 and a
2L2
receive antenna in the far-field region at a distance R > . In this case, the incident field at
λ0
the receiving antenna will be locally flat, similar to an incident plane wave with an equi-phase
plane perpendicular to the direction of propagation. These plane waves in free space are transverse
electromagnetic waves (TEM) with the following general form in the frequency domain:
Ey (z) = Ae−k0 z , (2.4)
where k0 = ωc is the free-space wavenumber and A the amplitude. Note that in (2.4), we have
assumed that the plane wave propagates in the +z-direction. For the sake of simplicity we only
~ is perpendicular
considered a component in the ~uy direction. The corresponding magnetic field H
~
both to the E field and the direction of propagation.
The transmit antenna will radiate spherical waves. In chapter 4 and 6 we will investigate
the radiated fields in more detail. It will be shown that the far field in spherical coordinates
~ r) = Eθ (~r)~uθ + Eφ (~r)~uφ + Er (~r)~ur due to an antenna placed at the origin of our coordinate
E(~
system at ~r = (0, 0, 0) can be expressed in the following form:
e−k0 r
Eθ (~r) = Eθ (θ, φ) ,
r
e−k0 r (2.5)
Eφ (~r) = Eφ (θ, φ) ,
r
Er (~r) = 0.
Since EM waves in the far-field locally behave as TEM waves, we can write the corresponding
~ in terms of the electric field:
magnetic field H
~ r) = 1 ~ur × E(~
H(~ ~ r), (2.6)
Z0
q
µ0
where Z0 = 0 = 377Ω is the intrinsic free-space impedance. The radiated power density
expressed in Watts per square meter [W/m2 ], is known as the Poynting vector and describes the
directional energy flux (the energy transfer per unit area per unit time) of an electromagnetic
field. The Pointing vector in the far-field region only has a component in the radial direction
~ur . The time-average Poynting vector S~p (~r) over a period T = 2π
ω at a specific position ~
r in the
far-field is now given by:
ZT
1
S~p (~r) = S~p (~r, t)dt
T
0
ZT
1 ~ r, t) × H(~
~ r, t)dt
= E(~
T
0
ZT
1 ~ r)eωt ] × Re[H(~
~ r)eωt ]dt
= Re[E(~
T (2.7)
0
ZT
1 1 ~ ~ ∗ (~r)e−ωt ) × 1 (H(~ ~ ∗ (~r)e−ωt )dt
~ r)eωt + H
= (E(~r)eωt + E
T 2 2
0
1 ~ ~ ∗ (~r) + E
~ ∗ (~r) × H(~
~ r)]
= [E(~r) × H
4
1 ~ ∗ (~r)].
~ r) × H
= Re[E(~
2
In the far-field region we can use relation (2.6). As a result, expression (2.7) can be written in
the following form:
S~p (~r) = 1 ~ r) ~ ∗ (~r)] = 1 Z −1 Re[E(~

×H ~ r) × (~ur × E
~ ∗ (~r))]
2 Re[E(~ 2 0
= 1 −1 ~ r) ~ ∗ (~r))~ur − (E(~
·E ~ r) · ~ur )E
~ ∗ (~r)] (2.8)
2 Z0 Re[(E(~
1 −1 ~
= r)|2 ~ur ,
2 Z0 |E(~
2.4. RADIATION PATTERN 15
~ r) · ~ur = 0. We can conclude that the electromagnetic energy propagates in the radial
since E(~
direction ~ur .
2.4 Radiation pattern

The radiation pattern is a graphical illustration of the radiated power in a certain direction in the
~ has a r−1
far field region of the antenna. From (2.5) we can observe that the electric field vector E
dependence in the far-field region. The radiated power per element of solid angle dΩ = sinθdθdφ
can be determined from (see also Fig. 2.1):
P (~r) = P (θ, φ) = |r2 S~p (~r)|, (2.9)
where Poynting’s vector is given by (2.8). Expression (2.9) is only valid in the far-field region
where the radiated power per element of solid angle is independent of r. Let us assume that the
maximum value of P (θ, φ) occurs at the angle (θ, φ) = (0, 0). The normalized radiation pattern
F (θ, φ) is found from:
P (θ, φ)
F (θ, φ) = . (2.10)
P (0, 0)
Radiation patterns are always expressed in decibels [dB], that is using 10 log10 (F (θ, φ)). Some
examples are shown in Fig. 2.4 and Fig. 2.5. The main beam, also known as main lobe, of the
antenna is directed towards the direction of maximum radiation, in this case towards θ = 0. The
other local maxima in the radiation pattern are called sidelobes. The maximum sidelobe level is
often one of the design parameters of an antenna, since the maximum sidelobe level determines
the sensitivity of the antenna for (unwanted) interference from other directions. The height of
the sidelobes is determined by a number of factors, including the size, shape and type of antenna.
Some antennas, like small dipoles, do not have sidelobes: they only have a main lobe. In chapter
6 we will show in more detail how you can design array antennas with low sidelobes. From the
normalized radiation pattern we cannot determine all quality measures of an antenna. Therefore,
we will introduce some additional scalar antenna parameters, including beam width, directivity,
antenna gain, effective aperture, aperture efficiency and input impedance. These parameters will
be explained in more detail in the next sections.
2.5 Beam width

The beam width defines the beam area for which the radiated or received power is larger than
half of the maximum power, the so-called Half-Power Beam Width (HPBW). The beam width in
the principle planes (θ = 0 and φ = 0 plane) is expressed by θHP or as φHP .
2.6 Directivity and antenna gain

The directivity describes the beamforming capabilities of an antenna. It describes the concentra-
tion of radiated power in the main lobe w.r.t. all other directions. Let us consider an antenna
that generates a total radiated power Pt [W]. When this antenna would be an isotropic radiator,
Figure 2.4: Two-dimensional normalized radiation pattern. A cut in the φ = 00 -plane is shown.
The antenna has a uniformly-illuminated square-shaped aperture of size A = (8λ0 )2 . The first
sidelobe has a maximum level of -13.2 dB
the radiated power would be equally radiated over all directions with an uniform radiated power
Pt
density of [W per unit of solid angle]. Note that an isotropic radiator cannot exist in practise,
4π
it is only a theoretical reference that is used to define the directivity of a real antenna. The
directivity function D(θ, φ) of a real physical antenna is defined as the power density per unit of
solid angle in the direction (θ, φ) relative to an isotropic radiator:
P (θ, φ)
D(θ, φ) = . (2.11)
Pt /4π
The maximum of the directivity-function is the directivity D = max[D(θ, φ)]. Antennas always
have some losses, for example due to resistive losses in metal structures, dielectric losses in dielec-
tric substrates or impedance-matching losses between antenna and the connected source. Due to
these losses not all the power that is provided by the source will be radiated. When we substitute
the total radiated power Pt in (2.11) by the total input power Pin to the antenna, we get the
well-known antenna gain function G(θ, φ), which is given by:
P (θ, φ)
G(θ, φ) = . (2.12)
Pin /4π
Similar to the directivity function, the antenna gain function will have a maximum, which is called
the antenna gain G, with G = max[G(θ, φ)]. The antenna efficiency η is a quantity that descibes
2.7. CIRCUIT REPRESENTATION OF ANTENNAS 17
Figure 2.5: Three-dimensional normalized radiation pattern of the antenna described in Fig.2.4
how much of the input power towards the antenna is transformed into radiated power:
Pt
η= . (2.13)
Pin
and provides a relation between the directivity and antenna gain:
G = ηD. (2.14)
2.7 Circuit representation of antennas

Typically, antennas are connected to a generator (transmit mode) or detector (receive mode)
by means of a transmission line. More details on transmission line theory can be found in the
next chapter. Sometimes, antennas are directly connected to an amplifier, in which case power
matching to a transmission line is not perse required. The input impedance of the antenna is an
important measure that determines how much of the power transported along the transmission
line is actually radiated by the antenna. The input impedance of an antenna will be complex in
general:
Za = Ra + Xa
(2.15)
= Rr + RL + Xa ,
where Ra is the antenna resistance, Rr represents the radiated losses and RL the real losses like
Ohmic or dielectric losses. Furthermore, Xa is the antenna reactance which is related to the
stored reactive energy in the close vicinity of the antenna. Note that the radiation resistance
Rr is directly related to the total radiated power by the antenna and is in fact a measure that
defines how well the antenna works. Fig. 2.6 shows the equivalent circuit of an antenna which
is connected to a generator with an internal impedance Zg = Rg . We can define the radiation
Rg Ip
+
Antenna
~
+
Vp Vg Vp Za= Ra+jXa
-
-
Figure 2.6: Equivalent circuit model of an antenna. The antenna with impedance Za is connected
to a generator with internal impedance Rg .
resistance Rr by considering the antenna in transmit mode. Let I p be the current flowing through
the antenna and Pt the total radiated power. The radiation resistance Rr is now defined as a
resistance in which the power Pt is dissipated when the current I p is flowing through the antenna.
In other words:
Pt
Rr = 1 p 2 . (2.16)
2 |I |
The total radiated power Pt can be found by integrating the radiated power density over a sphere.
Using relation (2.8) and (2.5), we obtain the following expression for Pt :
Z Z
Pt = S~p (~r) · ~ur r2 dΩ
Z2πZπ (2.17)
1 −1 2 2
= Z [|Eθ (θ, φ)| + |Eφ (θ, φ)| ] sin θdθdφ,
2 0
0 0
At microwave frequencies it is, in general, not possible to measure voltages and currents
directly. Therefore, we cannot directly measure the input impedance Za . We will show in chapter
3 that it is possible to measure the (complex) amplitude of incident and reflected waves along
transmission lines. In this way we can experimentally determine the input reflection coefficient
of the antenna using a vector network analyser (VNA). Now let us assume that the antenna is
connected to a transmission line with a characteristic impedance equal to Z0t . Usually Z0t = 50Ω.
Furthermore, we will assume that an incident TEM wave propagates along the transmission line
with complex amplitude a and a corresponding reflected wave (in the opposite direction) with
2.8. EFFECTIVE ANTENNA APERTURE 19
complexe amplitude b. The reflection coefficient is now defined as the ratio between both complex
amplitudes:
b
Γ= . (2.18)
a
From transmission line theory (see chapter 3 for more details), it is well known that the relation
between the reflection coefficient and the input impedance at the input of the antenna is given
by:
Za − Z0t
Γ= . (2.19)
Za + Z0t
When the antenna is properly matched to a transmission line with Za = Z0t , we will obtain a
reflection coefficient Γ = 0.
2.8 Effective antenna aperture

Up to now, we have investigated the antenna parameters mainly by considering the antenna
in transmit mode. The receive properties of the antenna are, of course, also very important.
One of the key parameters that describes the antenna properties in receive mode is the effective
antenna aperture Ae . Ae defines the equivalent surface in which the antenna absorbs the incident
electromagnetic field that is dissipated by the load impedance ZL which is connected to the
terminals of the antenna. Now let Sp represent the power density (W/m2 ) of an incident plane
wave. The effective antenna aperture is now defined as the ratio between the power Pr delivered
to the load impedance and the power density Sp :
Pr
Ae = . (2.20)
Sp
Where we have assumed that the antenna is optimally matched to the load impedance, according
to Za = ZL∗ . In chapter 4, we will introduce the reciprocity theorem. The reciprocity theorem
defines the relation between the receive and transmit properties of an antenna. We will show that
there is an unique relation between the antenna gain G and the effective antenna aperture Ae
according to:
4πAe
G= . (2.21)
λ20
2.9 Polarization properties of antennas

The radiated electric field vector from an antenna will, in general, have two components, Eθ and
Eφ , both perpendicular to the direction of propagation ~ur in the far-field region. Both components
(when described as phasors in the frequency domain) will usually have a relative phase difference
with respect to each other. With a phase difference equal to 0, the resulting field will be linearly
polarized. As a result, the direction of the electric field vector in the time domain will be constant.
When a phase difference between the Eθ and Eφ components exists, the direction of the realized
~ r, t) will vary over time with a period of T = 2π/ω. The time-

total time-domain electric field E(~
domain electric field vector will now describe an ellipse, corresponding to elliptical polarization.
Now let us take a closer look at the time-domain electric field components:
Eθ (~r, t) = E1 (~r) cos (ωt − k0 r),

(2.22)
Eφ (~r, t) = E2 (~r) cos (ωt − k0 r + ϕ),
where ϕ is the phase difference between both electric-field components. Let us consider the case
that r = 0. We can find the elliptical trajectory of the electric field vector E = (Eθ , Eφ ) by
eliminating the time-dependence. We can rewrite (2.22) as:
Eφ = E2 (cos ωt cos ϕ − sin ωt sin ϕ)

(2.23)
Eθ
= E2 ( cos ϕ − sin ωt sin ϕ),
E1
resulting in:
2 2 !
Eφ Eθ Eθ

2 2 2 2
− cos ϕ = sin ωt sin ϕ = 1 − cos ωt sin ϕ = 1− sin2 ϕ. (2.24)
E2 E1 E1
or:
2 2
Eθ Eφ 2Eθ Eφ

+ − cos ϕ = sin2 ϕ. (2.25)
E1 E2 E1 E2
This equation describes an ellipse in the plane perpendicular to the direction of propagation ~ur .
The ratio between the major and minor axis of the ellipse is defined as the axial ratio (AR), with
AR ≥ 1. We can distinguish two special cases
i. Circular polarization (E1 = E2 = E and ϕ = ±π/2).

Equation (2.25) now represents a circle. The field will be circularly-polarized. When ϕ = π/2,
we obtain Left-Hand Circular Polarization (LHCP):
E~L = E (~uθ cos ωt − ~uφ sin ωt) , (2.26)
In case ϕ = −π/2, we obtain Right-Hand Circular Polarization (RHCP):
E~R = E (~uθ cos ωt + ~uφ sin ωt) . (2.27)
The axial ratio describes the quality of the circularly-polarized wave and is given by:
|EL | + |ER |
AR = . (2.28)
|EL | − |ER |
where E~L = Re[E

~ L eωt ]. Perfect circular polarization is realized when AR = 1.
ii. Linear polarization (E1 6= E2 and ϕ = ±π). Equation (2.25) transforms into the equation of
a straight line, with corresponding axial ratio AR = ∞.
2.10. LINK BUDGET ANALYSIS: BASIC RADIO- AND RADAR EQUATION 21
2.10 Link budget analysis: basic radio- and radar equation

The antenna is one of the components of a complete system. Examples of such systems include
wireless communication systems (e.g. 4G/5G) and radar. The antenna parameters that we
have introduced in the previous sections can be used to quantify the quality of a radio link or to
determine the range of a radar. Fig. 2.7 shows a schematic illustration of a wireless link, consisting
of a transmit and receive antenna. The transmit antenna is connected to a transmitter that can
generate Pt Watts of power. We will assume that both antennas have the same polarization and
orientation with respect to each other. The power density in (W/m2 ) at the receive antenna is
Transmit antenna Receive antenna

Gt Gr , Aer
Transmitter Receiver
Pt Pr
Figure 2.7: Radio link between a transmit antenna and a receive antenna over a distance r.
now given by:

P t Gt
Sr = , (2.29)
4πr2
where Gt is the antenna gain of the transmit antenna as compared to an isotropic antenna. The
power received by the receive antenna with an effective aperture Aer now becomes:
Pt Gt Aer
Pr = . (2.30)
4πr2
λ20 Gr
This equation is also known as the radio equation or Friis equation. By using Aer = in
4π
(2.30), we finally obtain:
Pt Gt Gr λ20
Pr = , (2.31)
(4π)2 r2
!
λ20
where Gr is the antenna gain of the receiving antenna. The term is also known as the
(4π)2 r2
free-space path loss and depends on frequency, since λ0 = c/f . However, this term is not really
a loss, since we had assumed in our ideal radio link of Fig. 2.7 that the entire system, including
free-space, is lossless. The frequency dependence comes from the relation between the antenna
gain Gr and the effective area Aer . It tells us that for a constant antenna gain, the effective area
scales with 1/f 2 . This implies that at higher frequencies antennas with a much larger antenna
gain need to be used in order to maintain the same range as compared to lower frequencies.
Note that a large antenna gain also implies a more directive antenna with a narrow beam. This
complicates the implementation of omni-directional wireless communication at millimeter-wave
frequencies. Phased-arrays with electronic beam scanning offer a solution for this problem, see
chapter 6.
As an example of a link-budget analysis, consider the Bluetooth system operating in the

2.4 − 2.48 GHz band. Assume omni-directional antennas (Gt = Gr = 1), an output power of
Pt = 1 mW (0 dBm) and a receiver sensitivity Pr,min = 10−10 W (= −70 dBm). The maximum
range rmax of this system then becomes:
s
Pt Gt Gr λ20
rmax = ≈ 30 [m]. (2.32)
(4π)2 Pr,min
Let us now consider a radar system. In this case, we replace the receive antenna by an
object with a radar cross section σ[m2 ]. The radar cross section of an object depends on the size
and shape (e.g. car or airplane) and determines how much of the incident power is reflected back
to the radar. We will assume that the transmit antenna can also be used as receive antenna,
therefore Gt = Gr = G. The latter is due to the reciprocity theorem which applies to passive
microwave structures including antennas. The received power of the radar is now given by:
P t Gt σ Pt G2 σλ20
Pr = A er = . (2.33)
4πr2 4πr2 (4π)3 r4
This equation is known as the radar equation. It is a first-order estimation of the received power
and shows that the probability of detection of an object decreases according to r4 . In a real system,
the radar equation will be somewhat more complicated. In addition, the maximum detection range
can be increased by using advanced signal processing techniques. More background information
can be found in Skolnik [16]. Now assume that the receiver of the radar requires a minimum
power of Pr,min to detect an object. The maximum radar range rmax now becomes:
" #1/4
Pt G2 σλ20
rmax = . (2.34)
(4π)3 Pr,min
As an example, consider an X-band radar operating at f = 10 GHz with Ae = 1m2 , Pt = 10

kW, σ = 1m2 and Pmin = 10−13 W (10 log10 Pr,min = −100 dBm). This implies that G = 14000
(GdB = 41 dB). The maximum range now becomes rmax = 54 km.
Chapter 3
Transmission line theory and

microwave circuits
3.1 Introduction
In this chapter we will extend the well-known circuit theory to transmission line theory. Circuit
theory can be applied to electrical circuits in which the size of the individual (lumped) components,
like resistors, capacitors and inductors, are much smaller than the electrical wavelength (size
0.1λ). In case the physical dimension of a component or network becomes a significant fraction
of the wavelength (size > 0.1λ), we need to apply transmission line theory. We consider the
transmission line to be a distributed network in which the voltages and currents vary along the
location along the line. In this chapter, we will derive transmission line concepts by using circuit
theory. Therefore, we do not need to solve Maxwell’s equations explicitly to analyze networks
based on transmission lines.
3.2 Telegrapher equation

Fig. 3.1 illustrates some examples of transmission lines. Well-known transmission lines used in
many microwave applications are coaxial cables, microstrip lines and waveguides. In addition,
long-range three-phase electrical power lines operating at 50 Hz should also be considered as trans-
mission lines. From basic electromagnetics courses, it is known that a transverse electromagnetic
wave (TEM) with zero cut-off frequency can propagate along transmission lines consisting of at
least two metal conductors, e.g. a coaxial cable. These type of transmission lines also allow other
modes, e.g. transverse electric (TE) or transverse magnetic (TM), to propagate above their cut-off
frequencies. However, in this chapter we will assume that only a single TEM mode can propagate
along the transmission line. A plane wave is also an example of a TEM wave, see section 2. Note
that along a waveguide (see Fig. 3.1) a TEM mode cannot exist, only TE and TM modes can
propagate. However, the theory introduced in this chapter can also be applied to waveguides, as
long as only a single mode can propagate.
Now let us consider a general transmission line that carries a TEM wave that propagates
along the +z-direction, as illustrated in Fig. 3.2. The transmission line is connected to a source.
23
24 CHAPTER 3. TRANSMISSION LINE THEORY AND MICROWAVE CIRCUITS
GND
Coaxial cable Parallel wires (twin lead)
Microstrip lines
GND
Strip lines
GND
Microstrip lines
Multi-layer Printed Circuit Board (PCB) High-voltage lines
Figure 3.1: Examples of transmission lines used in various applications.
As a result of this, a voltage v(z, t) and current i(z, t) will exist at the location z.
i(z,t)
+
source v(z,t)
TEM wave
-
0 z
Figure 3.2: Two-wire representation of a transmission line. A TEM wave propagates in the
+z-direction.
We would like to apply Kirchhoff’s voltage and current laws instead of Maxwells’ equations
to analyze the characteristics of the transmission line. This can be done by considering a small
section between z and z + ∆z of the transmission line, as illustrated in Fig. 3.2. Since this section
is electrically very small (∆z λ), we can model this small section using lumped-element circuit
3.2. TELEGRAPHER EQUATION 25
components as illustrated in Fig. 3.3, where the lumped elements are defined per unit length:
L = series inductance per unit length [H/m],
C = shunt capacitance per unit length [F/m],

(3.1)
R = series resistance per unit length [Ω/m],
G = shunt conductance per unit length [S/m].
i(z,t)
+
source v(z,t)
-
z z
i(z,t) i(z+z,t)
+ Rz Lz +
v(z,t) Gz Cz v(z+z,t)
- -
z
Figure 3.3: Lumped element circuit representation of a small section ∆z of a transmission line.
The series inductance L∆z originates from the magnetic field that is induced by the current
i(z, t) flowing on the line and the series resistance R∆z accounts for the related losses in the
conductors. The shunt capacitor C∆z is created by the voltage v(z, t) which creates an electric
field between the two wires. Dielectric losses related to dielectric material between the two wires
are represented by the shunt conductance G∆z. We can now apply Kirchhoff’s voltage and current
laws to the equivalent circuit of a small section ∆z. Kirchhoff’s voltage law provides:
∂i(z, t)
v(z, t) − R∆zi(z, t) − L∆z − v(z + ∆z, t) = 0. (3.2)
∂t
Similarly, Kirchhoff’s current law gives:
∂v(z + ∆z, t)
i(z, t) − G∆zv(z + ∆z, t) − C∆z − i(z + ∆z, t) = 0. (3.3)
∂t
Now divide (3.2) and (3.3) by ∆z and take the limit of ∆z → 0, resulting in two differential
equations with two unknowns (v(z, t) and i(z, t)) which can be written in the following form:
∂v(z, t) ∂i(z, t)
= −Ri(z, t) − L ,
∂z ∂t (3.4)
∂i(z, t) ∂v(z, t)
= −Gv(z, t) − C .
∂z ∂t
Equation (3.4) is known as the Telegrapher equation in the time domain. Since we assume a
sinusoidal time variation of all field components in this book, the voltage and current can be
written as:
v(z, t) = Re[V (z)eωt ],

(3.5)
ωt
i(z, t) = Re[I(z)e ],
in which ω = 2πf and f is the frequency of operation. In this case the time derivative of v(z, t)
in (3.4) transforms into a simple multiplication in the frequency domain:
∂v(z, t)
↔ ωV (z). (3.6)
∂t
By applying this in (3.4) we obtain the frequency-domain Telegrapher equation:
∂V (z)
= − (R + ωL) I(z),
∂z (3.7)
∂I(z)
= − (G + ωC) V (z).
∂z
The general solution of the frequency-domain Telegrapher equation is now a combination of a
voltage and current wave propagating in the +z direction and a wave propagating in the −z
direction:
V (z) = V0+ e−γz + V0− eγz ,

(3.8)
I(z) = I0+ e−γz + I0− eγz ,
where γ is the complex propagation constant:

q
γ = α + β = (R + ωL)(G + ωC), (3.9)
in which α is the attenuation of the line and β is the phase constant. Note that the wavenumber
k0 used in chapter 2 and chapter 4 is strongly related to the phase constant β. However, we will
use different symbols, since it is more common to use k0 in the antenna community and β in the
microwave engineering community. The term e−γz corresponds to a wave propagating in the +z
direction and eγz to a wave propagating in the −z direction. Note that V (z) and I(z) in (3.8)
represent the total voltage and current, respectively, at the position z along the line. It can be
3.3. THE LOSSLESS TRANSMISSION LINE 27
easily verified that (3.8) describes the general solution by substituting (3.8) into the Telegrapher
equation (3.7). By doing this, we can also obtain a relation between the complex voltage and
current amplitudes:
γ
I0+ = V +,
R + ωL 0 (3.10)
−γ
I0− = V −.
R + ωL 0
In addition, the characteristic impedance Z0 of the transmission line is defined by:
V0+ R + ωL
Z0 = + = . (3.11)
I0 γ
The time-domain representation of the voltage along the transmission line can be obtained
by using the frequency domain solution (3.8) and relation (3.5). So for a wave travelling along a
transmission line in the positive z-direction we obtain the following time-domain representation:
v(z, t) = |V0+ |cos ωt − βz + φ+ e−αz ,

(3.12)
where it can be observed that α describes the exponential attenuation along the line and where
the phase constant β determines the wavelength on the line:
2π
λ= . (3.13)
β
Question: Why is there a − sign in the expression of I0− in (3.10)?
3.3 The lossless transmission line

In the previous section we have derived the solution for the voltage and current along a trans-
mission line with losses, resulting in a complex propagation constant γ = α + β and complex
characteristic impedance Z0 . In a lot of practical applications, the losses along the transmission
line are very small and can, therefore, be neglected for a first-order analysis. When the trans-
mission line is lossless, the series resistance R and the shunt conductance G in Fig. 3.3 can be
neglected. Therefore, the propagation constant and characteristic impedance now become real:
√
γ = β = ω LC,
√
β = ω LC,
(3.14)
α = 0,
s
L
Z0 = .
C
The corresponding wavelength and associated phase velocity of the wave along the line are given
by:
2π 2π
λ = = √
β ω LC (3.15)
ω 1
vp = =√ .
β LC
Example
A transmission line operating at f = 500 MHz has the following per unit length parameters:
L = 0.2 [µH/m],
C = 300 [pF/m],
(3.16)
R = 5 [Ω/m],
G = 0.01 [S/m].
Using (3.9), we find that the complex propagation coefficient is calculated by:
q p
γ = α + β = (R + ωL)(G + ωC) = −590 − 10.99 = 0.23 + 24.3 [1/m], (3.17)
in which the square root was evaluated according to Im(γ) > 0. Re-calculating this for a lossless
line with R = G = 0 gives:
√
γ = β = ω LC = 24.3 [1/m]. (3.18)
3.4 The terminated lossless transmission line

Consider the terminated lossless transmission of Fig. 3.4. The line is terminated at z = 0 with a
complex load impedance ZL = RL + XL . The total voltage and current at the position z along
the line are found by (3.8) and (3.10) and using γ = β:
V (z) = V0+ e−βz + V0− eβz ,

(3.19)
V0+ −βz V0− βz
I(z) = e − e .
Z0 Z0
This equation still has two unknown coefficients V0+ and V0− . A relation between both coefficients
can be found by using the relation between the voltage and current along the load impedance.
At z = 0 this relation is given by:
VL V (0) V + + V0− (3.20)

ZL = = = 0+ Z0 .
IL I(0) V0 − V0−
3.4. THE TERMINATED LOSSLESS TRANSMISSION LINE 29
IL
V(z), I(z)
+
Z 0,  VL ZL
-
z
-l 0
Figure 3.4: The terminated lossless transmission line, terminated with a complex load impedance
ZL
Rewriting (3.20) results in:
ZL − Z0 + (3.21)
V0− = = V .
ZL + Z0 0
We can now introduce the reflection coefficient Γ as:
V0− ZL − Z0 (3.22)
Γ = = + = .
V0 ZL + Z0
Some special terminations are:
ZL = Z0 , Γ = 0, matched load ,
ZL = 0, Γ = −1, short circuit termination, (3.23)
ZL = ∞, Γ = 1, open circuit termination.
In case of a matched load ZL = Z0 , the voltage along the line is constant |V (z)| = |V0+ |. This line
is now also called to be ”flat”. In other cases when Γ = |Γ|eθ 6= 0, the magnitude of the voltage
will not be constant over the line:
|V (z)| = = V0+ 1 + Γe2βz = V0+ 1 + |Γ| e(θ−2βl) , at z = −l. (3.24)
The voltage along the line fluctuates according to:
Vmax = V0+ |1 + |Γ|| ,
Vmin = V0+ |1 − |Γ|| , (3.25)
Vmax 1 + |Γ|
VSWR = = ,
Vmin 1 − |Γ|
where the VWSR is the voltage standing wave ratio which is a quantity that describes the fluc-
tuation of the voltage along the line. The VSWR is an important quantity that is often used to
specify the load conditions of active devices, like power amplifiers (PAs). PAs use semiconductor
transistors which have a maximum break-down voltage. By specifying a maximum VWSR, we
can avoid the maximum voltage along the line to exceed the break-down voltage of the transis-
tor. The reflection coefficient as defined in (3.22) can be generalized to any point z = −l on the
transmission line according to:
V0− e−βl
Γ(z = −l) = = Γe−2βl , (3.26)
V0+ eβl
where Γ = Γ(z = 0).

Now let us consider the configuration of Fig. 3.5 and investigate the input impedance of
the terminated line with length l. The input impedance Zin looking into the terminated line from
the location z = −l is now given by:
h i
V (z = −l) V0+ eβl + Γe−βl
Zin = = + βl Z0
I(z = −l) V0 [e − Γe−βl ] (3.27)
ZL + Z0 tan βl
= Z0 .
Z0 + ZL tan βl
Equation (3.27) shows that the input impedance of a terminated line can be tuned to a specific
value by changing the length l of the transmission line. This can be very useful in matching
circuits and filters, as we will see lateron.
IL
V(z), I(z)
+
Z 0,  VL ZL
-
z
Zin -l 0
Figure 3.5: The terminated lossless transmission line with characteristic impedance Z0 and length
l. We will assume that at the terminal z = −l the terminated line is connected to another very
long transmission line with characteristic impedance Z0 .
Example
Consider a lossless transmission line of length l which is short-circuited at the termination
(ZL = 0). As expected in case of a short-circuit, we find from (3.22) that the reflection at the
3.4. THE TERMINATED LOSSLESS TRANSMISSION LINE 31
termination Γ = −1. The voltage and current along the line is determined from (3.19) and takes
the form:
V (z) = −2V0+ sin βz,
2V0+ (3.28)
I(z) = cos βz,
Z0
Zin (z = −l) = Z0 tan βl.
The corresponding V SW R = ∞ since the minimum voltage on the line Vmin = 0. When a power
amplifier is connected to a short circuited transmission line, the output voltage at the output port
of the amplifier may become much larger than the breakdown voltage. In this case, the amplifier
will be destroyed.
When the terminated transmission line is not well matched to the connecting transmision
line, that is Zin 6= Z0 , part of the incident power will be reflected. Now let us assume that the
load impedance is purely resistive ZL = RL . The incident power Pin and reflected power Pr can
be related by using the magnitude of the voltage reflection coefficient:
|Vr |2
Pr 2RL
= = |Γ|2 . (3.29)
Pin |Vin |2
2RL
Note that in (3.29) we have used the time-average power which can be calculated from:
ZT
1
Pav (z) = P(z, t)dt
T
0
ZT
1 (3.30)
= v(z, t)i(z, t)dt
T
0
1
= Re [V (z)I ∗ (z)] ,
2
in which a similar derivation as used in (2.7) was applied. Note that the time domain voltage and
current can be expressed as:
h i h i
v(z, t) = Re V (z)eωt = Re |V (z)|eωt+φv
(3.31)
h i h i
i(z, t) = Re I(z)eωt = Re |I(z)|eωt+φi .
Exercise
Derive the expression for time-average power, given by (3.30), by using relation (3.31).
The return loss RL is now defined as the amount of power reflected by the load:
RL = −10 log10 |Γ|2 = −20 log10 |Γ|. (3.32)
Related to return loss is the insertion loss. The insertion loss describes the loss of a transition
between two transmission lines, as illustrated in Fig. 3.6. According to (3.19) the transmitted
wave for z > 0 takes the form:
V1 (z) = V0+ T e−βz , (3.33)
where T is the transmission coefficient:

Z1 − Z0 2Z1
T =1+Γ=1+ = . (3.34)
Z1 + Z0 Z1 + Z0
Let Pin be the power of the incident wave and P1 the power of the transmitted wave. The insertion
loss IL is now found as:
|V1 |2
 
P1

 2Z1 
IL = −10 log10 = −10 log10  
 |V + |2 
Pin 0
2Z0 (3.35)
Z1

= −20 log10 |T | + 10 log10 .
Z0
Z0 Z1
z
0
Figure 3.6: Interface between two transmission lines causing insertion loss.
3.5 The quarter-wave transformer

A very nice application of the terminated transmission line is the so-called λ/4 transformer as
illustrated in Fig. 3.7. A transmission line section of length l = λ/4 with a characteristic
impedance Z1 is terminated by a load impedance RL . In this section we will only consider
a resistive load. The transformer is connected to another transmission line with characteristic
impedance Z0 of which we will assume that the length is infinitely long or connected to a matched
source with impedance Zs = Z0 (no reflections from the source). The impedance looking into
3.5. THE QUARTER-WAVE TRANSFORMER 33
Z0 Z1 RL
z
Zin -/4 0
Figure 3.7: The λ/4 transformer in which a λ/4-transmission line is connected to a terminated
lossless transmission line, terminated with a load impedance RL
the transformer at z = −λ/4 can be calculated using (3.27), where it should be noted that the
quarterwave line has a characteristic impedance Z1 . Using l = λ0 /4 and ZL = RL we get:
RL + Z1 tan βl
Zin = Z1
Z1 + RL tan βl
RL + Z1 tan βλ/4 (3.36)
= Z1
Z1 + RL tan βλ/4
Z12
= ,
RL

where βl = 2π λ
λ π
4 = 2 . From (3.36) we find that the reflection coefficient at z = −λ/4 is zero
when the following condition is satisfied:
Zin − Z0 p
(3.37)
Γ(z = −λ/4) = = 0 −→ Zin = Z0 −→ Z1 = Z0 RL .
Zin + Z0
Example
Consider a load resistance RL = 100 Ω which needs to be matched to a long transmission line
with characteristic impedance Z0 = 50 Ω. Design a quarter-wave transformer at the frequency
f0 and plot the magnitude of the input reflection coefficient versus the normalized frequency
√
f /f0 . From (3.37) we find that Z1 = Z0 RL = 70.71 Ω. The magnitude of the input reflection
coefficient is found from:
Zin − Z0
|Γ| = ,
Zin + Z0
RL + Z1 tan βl (3.38)
Zin = Z1
Z1 + RL tan βl
πf
βl= .
2f0
Fig. 3.8 shows a plot of the amplitude of the reflection coefficient of this quarter-wave transformer.
We can observe that only at the design frequency an optimal matching is obtained. This is a
clear drawback of a single-stage quarter-wave transformer and limits the frequency bandwidth of
such a matching circuit. The bandwidth can be increased by using multiple stages. Note that the
transformer is also matched at l = nλ/4 with n = 3, 5, .....
0.35
0.3
0.25
abs(Gamma)
0.2
0.15
0.1
0.05
0
0 0.5 1 1.5 2 2.5 3 3.5 4
f/f0
Figure 3.8: Magnitude of the input reflection coefficient versus the normalized frequency f /f0 of
a λ/4 transformer terminated with a load impedance RL = 100 Ω
3.6 The lossy terminated transmission line

In practise, transmission lines will always be lossy, although the theory of lossless lines is generally
a very good first-order estimation. In case of a lossy terminated line, as illustrated in Fig. 3.9,
the voltage and current along the line are found using the general solution of the Telegrapher
equation (3.8) and can be written in the following form:
V (z) = V0+ e−γz + Γeγz ,

(3.39)
V0+ −γz
− Γeγz ,

I(z) = e
Z0
where Γ is the reflection coefficient at the termination z = 0. The input impedance at z = −l is

now found by:
h i
V (z = −l) V0+ eγl + Γe−γl
Zin = = + γl Z0
I(z = −l) V0 [e − Γe−γl ] (3.40)
ZL + Z0 tanh γl
= Z0 .
Z0 + ZL tanh γl
3.7. FIELD ANALYSIS OF TRANSMISSION LINES 35
IL
V(z), I(z)
+
Z 0,  V L ZL
-
z
Zin -l 0
Figure 3.9: The lossy terminated transmission line with γ = α + β.
The power delivered at z = −l is now:
1
Pin = Re (V (−l)I ∗ (−l))
2
2 (3.41)
V0+
= e2αl − |Γ|2 e−2αl .
2Z0
The corresponding power delivered to the load at z = 0 is found by:
1
PL = Re (V (0)I ∗ (0))
2
2 (3.42)
V0+
= 1 − |Γ|2 .
2Z0
The loss in the terminated line of length l is found as the difference between Pin and PL :
2
V0+
(3.43)
h i h i
Ploss = Pin − PL = e2αl − 1 + |Γ|2 1 − e−2αl .
2Z0
The first term in (3.43) is the loss of the incident wave along the line with length l. The second
term describes the loss of the reflected wave.
3.7 Field analysis of transmission lines

Real transmission lines may take various shapes. Fig. 3.10 shows the cross section of various
types of transmission lines that support TEM or quasi-TEM wave propagation. The physical pa-
rameters like width of a microstrip line and dielectric constant and loss tangent of the substrate
can be translated into transmission parameters Z0 , α, β and/or in terms of L, C, R, G. However,
for most transmission line types this cannot be done in an analytical way, since we need to solve
Maxwells’ equation using the boundary conditions on the metal structures. Approximate models
can be applied, but numerical methods need to be used in order to obtain very accurate results.
Nowadays, several open-source and commercial software packages, like QUCS and ADS are avail-
able with tools to determine the transmission line parameters for various types of transmission
GND
Coaxial cable, TEM mode
Parallel wires (twin lead), TEM mode
GND
Stripline, TEM mode

GND
Microstrip, quasi-TEM mode

GND
GND GND
Co-planar waveguide, quasi-TEM mode

GND
Metal waveguide, non-TEM mode

GND
Figure 3.10: Cross sections of various transmission lines that support TEM, quasi-TEM or non-
TEM wave propagation.
lines. A type of transmission line for which we can easily solve Maxwells’ equations is the parallel-
plate waveguide as illustrated in Fig. 3.11. Although it is not a very practical transmission line,
its characteristics are quite similar to more practical transmission lines, e.g. microstrip line. It
consists of two large metallic plates separated by a dielectric substrate with permittivity = r 0
and permeability µ0 . We will assume that the transmission line is lossless. In addition, it is
assumed that the width w d. The TEM wave has zero cut-off frequency and propagates along
the z direction, therefore, Ez = Hz = 0 and E ~ =E ~ t = (Ex ~ux + Ey ~uy )e−βz . In this case, it can be
shown that the transverse components of the electric and magnetic field can be found by solving
Laplace’s equation:
~t
∇2t E = 0,
(3.44)
~ t = 0,
∇2t H
~ t = Ex ~ux + Ey ~uy and ∇2t = ∂ 2 /∂x2 + ∂ 2 /∂y 2 resembles the

where the transverse electric field is E
Laplacian operator in the (x, y) dimensions. From (3.44) we can observe that the transverse fields
are the same as the static fields in a parallel plate configuration. From electrostatics, we know
3.7. FIELD ANALYSIS OF TRANSMISSION LINES 37
y
Φ(y=d)=V0
Φ(y=0)=0
ε,µ0 d
x
w
Figure 3.11: The parallel-plate transmission line that support a TEM wave.
that the electric field in (3.44) can be expressed as a gradient of a scalar potential Φ(x, y):
~ t = −∇t Φ(x, y),

E (3.45)
This scalar potential Φ(x, y) satisfies Laplace’s equation:
∇2t Φ(x, y) = 0. (3.46)
The boundary conditions of this differential equation are as indicated in Fig. 3.11:
Φ(x, 0) = 0,
(3.47)
Φ(x, d) = V0 .
A solution of Φ(x, y) cannot depend on x, since we have assumed that the width w is very large
(close to infinite). Therefore, Φ(x, y) takes the form:
V0 y (3.48)
Φ(x, y) = ,
d
which can be easily verified by substituting this in (3.46) and (3.47). The electric and magnetic
fields between the plates are now obtained using (3.45):
V0 −βz
~
E(x, y, z) = −∇t Φ(x, y)e−βz = −~uy e ,
d (3.49)
1 V0
~
H(x, y, z) = ~
~uz × E(x, y, z) = ~ux e−βz ,
η ηd
√ p
where the propagation constant β = ω µ0 and the intrinsic impedance η = µ0 /. The corre-
sponding voltage and current along the line are now given by:
Z d
V (z) = − Ey dy = V0 e−βz
0 (3.50)
Z w Z w Z w
~ · ~uz dx = wV0 −βz
I(z) = Jz dx = (−~uy × H) Hx dx = e .
0 0 0 ηd
The transmission line parameters are then found by (with R = G = 0):
V ηd
Z0 = = [Ω],
I w
µ0 ~ 2 dS = µ0 d H
Z Z
L = |H| ,
|I|2 w m (3.51)
~ 2 dS = w F
Z Z
C = 2
|E| ,
|V0 | d m
√ √

1

β = ω LC = ω µ0 .
m
3.8 The Smith chart

The Smith chart is a very useful graphical tool to visualize the behaviour of transmission-line
circuits. The Smith chart was developed by P. Smith in 1939, well before computer-based design
tools had become available. The Smith chart is still de-facto de standard representation tool in
microwave engineering. The Smith chart is basically built around the complex representation of
the reflection coefficient of a transmission line circuit, for example as described in (3.37) in case of
a quarter-wave transformer. The reflection coefficient is complex and can be expressed in terms
of amplitude and phase:
Zin − Z0 (3.52)
Γ= = Γr + Γi = |Γ|eφΓ .
Zin + Z0
When we would plot Γ in the complex plane, we find that all values are within the unit circle, since
|Γ| ≤ 1. In case Zin = Z0 , Γ = 0 and we would end up in the center of the unit circle. The most
common form of the Smith chart is shown in Fig. 3.12. Note that the normalized impedance values
zin = rin +xin = Zin /Z0 are listed in the Smith chart, where rin , xin are the normalized resistance
and reactance, respectively. Let us now investigate some special cases in more detail. First assume
that the transmission line circuit is purely resistive with Zin = Rin + Xin = Rin = rin Z0 . In this
case we find that:
Zin − Z0 rin − 1 (3.53)

Γ= = .
Zin + Z0 rin + 1
Now Γ is real and describes a line along the horizontal axis with Γ = −1 for rin = 0 (short),
Γ = 0 for rin = Z0 (matched load) and Γ = 1 for rin = ∞ (open). Usually, Z0 = 50 Ω is used.
Another special case occurs when the load is purely reactive with Zin = Xin = xin Z0 . It can
3.8. THE SMITH CHART 39
Figure 3.12: The Smith chart. The normalized impedance values zin = rin + xin = Zin /Z0 are
listed. Usually Z0 = 50 Ω.
be shown that the reflection coefficient Γ now describes a circle with amplitude |Γ| = 1. This can
be generalized to the case of a constant resistance and varying reactance. In this case we obtain
circles as can be seen in Fig. 3.12. The lines for a load with constant reactance and varying
resistance can also be observed in the Smith chart and correspond to parts of large circles that
extend the unit circle. Smith charts with both impedance and admittance values are also available
and are useful when designing serial or parallel matching circuits. Fig. 3.13 shows such a chart.
Exercise
Plot the following load conditions in the Smith chart:
i. Zin = 0,
ii. Zin = ∞,
iii. Zin = 50 Ω,
iv. Zin = 50 − 150 Ω.

Figure 3.13: Smith chart with normalized impedance and admittance coordinates.
3.9 Microwave networks
Consider a microwave network with K-ports as illustrated in Fig. 3.14. Each port is connected to
a transmission line on which TEM-modes can propagate. At port k, the total voltage and current
is given by:
Vk = Vk+ + Vk− ,
(3.54)
Ik = Ik+ − Ik− .
3.9. MICROWAVE NETWORKS 41
Microwave network
I1 Ik IK
+ - + - + -
V1 Vk VK
Figure 3.14: K-port microwave network with port voltages and currents.
The relation between the port voltages and port currents is defined by the impedance matrix [Z]:
    
 V   Z
 1   11 Z12 . . Z1K
 I 
 1 
    
    
    
 V2   Z21 Z22 . . Z2K   I2
    

= . (3.55)
    
 
    
 .   . . . . .  .
    

    
    
    
    
VK ZK1 ZK2 . . ZKK IK
In most practical case we will investigate two-port microwave networks with K = 2. The relation
between port voltages and currents then reduces to:
    
 V1   Z11 Z12   I1 
    
 =  , (3.56)
    
    
V2 Z21 Z22 I2
where the individual components of the Z-matrix are found by:
V1 V1
Z11 = Z12 =
I1 I2 =0 I2 I1 =0 (3.57)
V2 V2
Z21 = Z22 =
I1 I2 =0 I2 I1 =0
Example Consider the network of Fig. 3.15, where a 50 Ω resistor is connected in parallel.
The components of the impedance matrix are now found using (3.57):
V1 V1
Z11 = = 50 Ω Z12 = = 50 Ω
I1 I2 =0 I2 I1 =0 (3.58)
V2 V2
Z21 = = 50 Ω Z22 = = 50 Ω
I1 I2 =0 I2 I1 =0
I1 I2
→ ←
+ +
V1 50 Ω V2
− −
Figure 3.15: 2-port microwave network with a parallel 50 Ω resistor.
The voltage-current relation of a microwave network can also be expressed by means of an admit-
tance matrix [Y ]. In case of a two-port network, we obtain:
    
 I1   Y11 Y12   V1 
    
 =  , (3.59)
    
    
I2 Y21 Y22 V2
where the individual components of the Y -matrix are found by:
I1 I1
Y11 = Y12 =
V1 V2 =0 V2 V1 =0 (3.60)
I2 I2
Y21 = Y22 =
V1 V2 =0 V2 V1 =0
Although the impedance and admittance matrices describe the behaviour of a microwave network
completely, it appears that this description is not very practical at microwave frequencies. This is
due to the fact that we cannot measure the complex values of the total port voltages and currents
in an accurate way at these frequencies. However, we can measure the amplitude and phase of
the incident and reflected voltage waves in an accurate way by measuring the complex amplitude
of the incident and reflected electric field of the TEM wave. This is done with a vector network
analyzer (VNA). Let us consider again the K-port microwave network, but now described in terms
of the complex amplitude of the incident and reflected waves at each port. This is illustrated in
3.9. MICROWAVE NETWORKS 43
Fig. 3.16. The normalized complex incident and reflected wave amplitudes are given by:
V+
√k = Ik+ Z0 ,
p
ak =
Z0 (3.61)
V−
√k = Ik− Z0 .
p
bk =
Z0
Note that Z0 = 50 Ω is usually used in measurements. The relation between the incident waves [a]
Microwave network
a1 b1 ak bk aK bK
Figure 3.16: K-port microwave network described using normalized complex amplitudes of the
incident and reflected TEM waves
and reflected waves [b] is determined by the scattering matrix [S], which for a two-port microwave
network takes the following form:
    
 b1   S11 S12   a1 
    
 =  , (3.62)
    
    
b2 S21 S22 a2
where the individual scattering parameters are found by:
b1 b1
S11 = S12 =
a1 a2 =0 a2 a1 =0 (3.63)
b2 b2
S21 = S22 =
a1 a2 =0 a2 a1 =0
So the scattering parameter S11 can be found by properly matching output port 2. The other
parameters are found in a similar way. In case of a one-port network, we find that the input
reflection coefficient Γin = S11 . Now consider the two-port microwave network of Fig. 3.17 with
source impedance ZS at the input port and load impedance ZL at the output port. The parameter
|S21 |2 is known as the forward gain and |S12 |2 is the reverse gain. Note that in case of passive
Γin Forward Gain (|S21|2) Γout
ZS a1 a2
Two-port
ZL
+ b1 Microwave Network b2
VS
-
ΓS Reverse Gain (|S12|2) ΓL
Figure 3.17: Two-port microwave network with input and output terminated with ZS and ZL ,
respectively
microwave networks both the forward and reverse gain are smaller than 1 and S21 = S12 due to
the reciprocity principle. The input and output reflection coefficients of this network are found
by:
b1 S12 S21
Γin = = S11 +
a1 1
− S22
ΓL (3.64)
b2 S12 S21
Γout = = S22 + ,
a2 1
− S11
ΓS
where ΓS and ΓL are the reflection coefficients when looking into the source and load, respectively:
ZS − Z0
ΓS =
ZS + Z0 (3.65)
ZL − Z0
ΓL = .
ZL + Z0
Combining (3.64) and (3.65) we find that Γin = S11 only if ZL = Z0 . Similarly, we find that
Γout = S22 when ZS = Z0 .
Exercise
Consider the two-port network of Fig. 3.18, where a 50 Ω resistor is connected in series between
both ports. The source and load impedance are ZS = ZL = 50 Ω. Show that the scattering
3.10. POWER COMBINERS 45
matrix takes the following form:

   
 S11 S12   1 2 
   3 3 
 = . (3.66)
   
   
2 1
S21 S22 3 3
50 Ω
50 Ω a1 a2
50 Ω
+ b1 b2
VS
-
Figure 3.18: Two-port microwave network consisting of a 50 Ω lumped element connected in series
between input and output ports
3.10 Power Combiners

Power combiners and power splitters are one of the most commonly used microwave circuits. For
example, in basestations for wireless communications, where a high effective isotropic radiated
power (EIRP) is required to provide coverage of large macro-cells. A high EIRP can be achieved
by combining the output power of several power amplifiers using a power combiner. Another
application is in array antennas, where received signals from the individual antennas can be
combined into a single output signal. An example of a 4-channel power combiner is shown in
Fig. 3.19. Since power combiners are passive transmission line circuits, reciprocity holds. As a
result, power combiners can also be used as power dividers. The analysis of power combiners in
this section is based on the description from David Pozar [17]. Fig. 3.20 shows the basic power
divider with unequal power division ratio α and a power combiner with equal power division ratio
with α = 0.5. It can be shown that the three-port network of Fig. 3.20 cannot be at the same
time lossless, reciprocal and matched at all ports at the same time [17]. Therefore, any practical
realization of a combiner or divider will be a compromise. In the rest of this section we will
investigate several well-known power combiner/divider types. Since the passive combiner/divider
is a reciprocal device, we will only consider dividers in the remaining part of this section.
Let us start by taking a closer look at the T-junction divider as illustrated in Fig. 3.21. An
input transmission line with characteristic impedance Z0 is connected at the junction with two
other transmission lines with characteristic impedance of Z1 and Z2 , respectively. The additional
susceptance jB represents the fringing fields and higher-order modes that might exist at the T-
junction. For the sake of simplicity, we will assume that B = 0. The input admittance at the
Figure 3.19: Example of a power combiner/divider with an unequal power division ratio, a 4:1
Wilkinson combiner used in base-station antennas, courtesy Rob Mestrom TU Eindhoven.
P2=αP1
P1
Unequal
divider
P3=(1-α) P1
P2
P1 =P2+P3 Equal
combiner
P3
Figure 3.20: Power divider with unequal power division ratio α (top) and a power combiner with
equal power division ratio (bottom).
input of the T-junction is now simply:
1 1 (3.67)
Yin = + .
Z1 Z2
1
For a matched input port it is required that Yin = . Now suppose that we would like to design
Z0
an equal-split divider with α = 0.5 and Z0 = 50 Ω. The input power delivered to the matched
divider is:
V02 (3.68)
Pin = .
2Z0
Z2
+
Z0 V0 jB
-
Z3
Yin
Figure 3.21: T-junction power divider with unequal power division ratio.
The corresponding output powers are given by:
V02 1
P1 = = αPin = Pin ,
2Z1 2 (3.69)
V2 1
P2 = 0 = (1 − α)Pin = Pin .
2Z2 2
Both equations can be satisfied with the following choice for the characteristic impedances:
Z1 = 2Z0 = 100 Ω,
(3.70)
Z2 = 2Z0 = 100 Ω.
The input impedance is now indeed Zin = 50 Ω because of the parallel connection of two 100 Ω
impedances. However, the output ports are not matched and have a quite poor reflection coeffi-
cient:
Zout1 − Z1 33.3 − 100 (3.71)

Γ1 = Γ2 = = = −0.5,
Zout1 + Z1 33.3 + 100
Note that poor output matching is not always a problem in an application. Another limitation
of the simple T-junction divider is the poor isolation between the output ports, which could be a
major issue in phased-array antennas.
A way to improve the poor output matching and poor isolation is by adding a star connection
of lumped-element resistors at the junction, as illustrated in Fig. 3.22. In this case, an equal-split
divider is created. The impedance Z looking via the Z0 /3 resistor into port 2 is given by:
Z0 4Z0 (3.72)
Z= + Z0 = .
3 3
port 2
P2
Z0/3 Z0
Z0/3 +
port 1 V2
Z0/3 -
+
P1 Z0 V1 Z
- +
V3
-
Z0
Zin P3
port 3
Figure 3.22: Resistive power divider with equal power division ratio α = 0.5
From the input port we then obtain a Z0 /3 resistor in series with a parallel connection of Z at
both output ports:
Z0 1 4Z0

Zin = + = Z0 . (3.73)
3 2 3
As a result, the input is well matched. Due to symmetry, the two output ports are also matched.
Therefore, all diagonal components of the scattering matrix [S] are zero: S11 = S22 = S33 = 0.
The other components of the scattering matrix can be found by determining the relation between
the input voltage V1 and the output voltages V2 and V3 . Since all ports are matched, the total
voltage is equal to the voltage amplitude of the TEM-wave travelling in the +z direction, see also
(3.61). The voltage at the central node is now easily found by using circuit theory:
2Z0 /3 2V1 (3.74)

V = V1 = .
Z0 /3 + 2Z0 /3 3
The corresponding output votages at port 2 and 3 then become:
Z0 3V V1 (3.75)
V2 = V3 = V = = .
Z0 + Z0 /3 4 2
As a result, the three-port scattering matrix takes the following form:
   
 S11 S21 S31   0 1 1 
   
   
  1 
[S] =  S21 =  1 . (3.76)
   
 S22 S23 
 2  0 1 

   
   
   
S31 S32 S33 1 1 0
From (3.76) we observe that S21 = S31 = S23 = 1/2, which corresponds to a power level of
−6 dB with respect to the input power level. The output power at each of the output ports
1
P2 = P3 = Pin . Therefore, 50% of the input power is dissipated in the resistors. This is a major
4
drawback in most microwave applications. However, in integrated circuits (ICs) resistive dividers
could be useful, due to their small size. The power loss in such a circuit is not always a big issue
since it can be compensated by adding additional amplification stages. Note that the isolation
between the output ports of the resistive divider is also quite poor.
So far we have investigated power dividers with major limitations in terms of matching,
isolation or power dissipation. A power divider which combines excellent matching properties,
high isolation and low loss is the Wilkinson power divider. Wilkinson dividers can be made for
any division ratio, but we will only investigate the equal split case α = 0.5 in more detail in this
section. Fig. 3.23 shows the transmission line circuit of the equal-split Wilkinson power divider
and a realization in microstrip technology. It consists of a T-junction with two quarter-wave
√
transmission lines with Z1 = Z2 = 2Z0 which are connected via a lumped-element resistor
R = 2Z0 . The Wilkinson divider can be analyzed by using the even-odd mode technique. In fact,
port 2
λ/4 Z0
Z1
port 1
R
Z0
port 3
Z2
Z0
λ/4
√
Figure 3.23: Equal split Wilkinson power divider, Z1 = Z2 = 2Z0 and R = 2Z0
we will excite the output ports with either the same voltage 2Vs (even mode) and with opposite
voltage 2Vs and −2Vs (odd mode). By using superposition, we then obtain the total voltage at
the input port. From the voltages, we can determine the scattering matrix. The even-odd mode
analysis starts by creating a new drawing that illustrates the symmetry of the Wilkinson power
divider as shown in Fig. 3.24. All ports are terminated with a load impedance of Z0 . Furthermore,
√
we will assume that Z = 2Z0 and R = 2Z0 . Lateron, we will show that with these particular
values the best performance is obtained.
Even-mode analysis
In this case the source voltages at port 2 and port 3 are equal with Vs2 = Vs3 = 2Vs . Due to
the symmetry in Fig. 3.24, no current will flow through the resistors R/2, which means that the
symmetry line can be represented as an open-circuit. As a result, we can limit our analysis to the
circuit shown in Fig. 3.25. The resistor R/2 can now be neglected. The input impedance looking
-
V2 Z0
+
port 2
Z +
Vs2
2Z0 R/2 -
λ/4
port 1
Symmetry line
+
2Z0 V1 R/2
-
Z port 3
+ +
V3 Z0 Vs3
- -
Figure 3.24: Symmetrical representation of the equal-split Wilkinson power divider.
-
V2e Z0
port 2 +
Z +
- 2Vs
V1e R/2 -
2Z0
+ λ/4
port 1
open open
circuit circuit Zin,even
z 0 −λ/4
Figure 3.25: Even-mode equivalent circuit of the equal-split Wilkinson power divider.
into port 2 is now easily found using the quarter-wave transformer equation (3.36):
√
Z2 ( 2Z0 )2 (3.77)
Zin,even = = = Z0 .
2Z0 2Z0
The voltage along the lossless λ/4 line with phase constant β is found using:

V (z) = V0+ e−βz + Γeβz . (3.78)
As a result, the port voltages are given by:
V2e = V (z = −λ/4) = V0+ (1 − Γ) = Vs ,

(3.79)
Γ+1
V1e = V (0) = V0+ (1 + Γ) = Vs ,
Γ−1
where Γ is the reflection coefficient looking at port 1 into the load impedance 2Z0 :
√ √
2Z0 − 2Z0 2− 2 (3.80)
Γ = √ = √ .
2Z0 + 2Z0 2+ 2
The voltage at port 1 now becomes
√
V1e = −Vs 2. (3.81)
Odd-mode analysis
The source voltages at port 2 and port 3 now have equal amplitudes but opposite sign, Vs2 = 2Vs
and Vs3 = −2Vs . Due to the asymmetry, the voltage along the symmetry line is zero as illustrated
in the equivalent circuit of Fig. 3.26. The quarter-wave transformer now transforms the short at
-
V2o Z0
port 2 +
Z +
- 2Vs
V1o R/2 -
2Z0
+ λ/4
port 1
Figure 3.26: Odd-mode equivalent circuit of the equal-split Wilkinson power divider.
the end of the line to an open circuit at the start of the line at port 2. The remaining circuit is a
voltage divider. Therefore, the voltages at port 1 and port 2 are given by:
V1o = 0,
(3.82)
1
V2o = (2Vs ) = Vs .
2
Total solution: superposition of even-odd modes

By applying superposition of the even- and odd-mode analysis we obtain the total solution when
exciting port 2. In a similar way, the solution when exciting port 3 can be found. In both the
even and odd mode, port 2 is matched. Therefore S22 = S33 = 0. In addition, it can be shown
√
that with Z = 2Z0 port 1 is also matched S11 = 0. The transfer functions when all ports are
matched are now found by superimposing the port voltages:
V1e + V1o −
S12 = S21 = =√ ,
V2e + V2o 2
− (3.83)
S13 = S31 =√ ,
2
S23 = S32 = 0.
The ideal Wilkinson divider has no losses since |S21 |2 = |S12 |2 = 1/2, so when used as a divider,
50% of the power is directed to each of the output ports. Furthermore, the isolation between the
output ports is perfect since S23 = S32 = 0. When we use it as a combiner to combine input
signals from port 2 and port 3, the ideal Wilkinson combiner will also be lossless as long as both
signals are identical, both in phase and amplitude. Any unbalance between the input signals
will be dissipated in the resistor. In a practical realization of the Wilkinson divider, there will
be losses due to metal losses, dielectric losses and spurious radiation. In addition, the isolation
(20 log10 |S32 |) will be limited, typically to a value between -20 and -30 dB.
3.11 Impedance matching and tuning

Consider the circuit of Fig. 3.27 where a source with complex source impedance ZS = RS + XS
is connected to a complex load ZL = RL + XL . The source could be a power amplifier and the
load could be an antenna. The time-average power delivered to the load is found by:
1 (3.84)
PL = Re {VL IL∗ } .
2
where
VS ZL
VL = ,
ZS + ZL (3.85)
VS
IL = .
ZS + ZL
IL
ZS
+
+
VS ZL VL
-
-
Figure 3.27: Conjugate matching of a source impedance to a complex load impedance.
When substituting (3.85) in (3.84) we obtain:

( )
1 |VS |2 ZL 1 |VS |2 RL (3.86)
PL = Re = .
2 |ZS + ZL |2 2 (RS + RL )2 + (XS + XL )2
The maximum delivered power is now found by determining the value of (RL , XL ) for which the
partial derivatives are equal to zero:
∂PL ∂PL (3.87)

= = 0.
∂RL ∂XL
We find that this occurs when RL = RS and XL = −XS , in other words when ZL = ZS∗ . The
maximum power transfer to the load occurs when the source and load impedances are conjugate
3.11. IMPEDANCE MATCHING AND TUNING 53
ZS
+
Matching
VS ZL
circuit
-
Figure 3.28: Matching circuit to obtain a good matching between the source and load.
matched to each other. In a lot of situations direct conjugate matching of the source to load is not
possible. In these cases, we can use a matching circuit as illustrated in Fig. 3.28 The matching
circuit can be created using either lumped elements, like inductors (with inductance L [H]) and
capacitors (with capacitance C [F]), or by using distributed transmission-line components. Note
that lumped elements can only be used when the size of the component is much smaller as
compared to the wavelength. In PCB technologies, surface-mount devices (SMD) are commonly
used, with sizes down to SMD0201 (0.25 mm × 0.125 mm) and even SMD01005 (0.125 mm ×
0.0675 mm). SMD components are typically used at frequencies up to 6 GHz. In semiconductor
technologies, much smaller inductors and capacitors can be realized. As a result, lumped-element
matching can be used up to much higher frequencies in integrated circuits.
Lumped-element matching can be realized by placing inductors and capacitors in series or
parallel to the load impedance. The impedances of these components in the frequency domain
are given by:
1
capacitor : Z = ,
ωC (3.88)
inductor : Z = ωL,
where ω = 2πf . The Smith chart is a useful tool to design such a network. Several commercial
and open-source tools are available to design lumped-element matching circuits. Suppose that we
want to match a particular load impedance ZL to a source impedance ZS . Fig. 3.29 shows the
degrees of freedom when placing an inductor or capacitor in series or in shunt (parallel) to the
load impedance. By using several of these combinations (shunt-series) or (series-shunt), we can
reach any point in the Smith chart. In addition, transmission line sections could be added. Note
that ideal capacitors and inductors do not exist in practise at microwave frequencies. One has to
take parasitics and losses into account. This will deteriorate the performance of a lumped-element
matching circuit.
Exercise
Design a lumped-element matching circuit to match a complex load impedance ZL = 200−100 Ω
to a real load ZS = 100 Ω. The frequency of operation is 1 GHz. You can use the design tools
Series C ZL
Shunt L
Series L
ZL
Series C Shunt C ZL
Shunt C C
Series L ZL
Shunt L L ZL
Figure 3.29: Lumped-element matching illustrated in a Smith chart.
ADS or QUCS (open-source, http://qucs.sourceforge.net/). Check you final result in a Smith

chart using the strategy as shown in Fig. 3.29.
Due to practical limitations of lumped elements at microwave frequencies, we generally

prefer to use transmission-based matching circuits. When both the load and source impedance
are real, we can use the quarter-wave transformer as discussed in section 3.5. Another matching
strategy is to use stubs which are placed in parallel to a transmission-line transformer. An example
of such a circuit is shown in Fig. 3.30. The stub is either open ended or shorted. The imput
Z0 Z0 ZL
Z0
l YT=1/ZT
open or
shorted
stub
Figure 3.30: Single-stub matching circuit

3.12. MICROWAVE FILTERS 55
impedance is then obtained using (3.27):
shorted stub : Zs = Z0 tan βl = ΩZ0 = ΩL,

(3.89)
Z0 Z0 1
open stub : Zs = = = .
 tan βl Ω ΩC
Apparently, a shorted stub is the equivalent of an inductor and an open stub is the equivalent of
a capacitor, with l ≤ λ/4. This equivalence appears to be very useful when designing microwave
filters, as we will discuss in more detail in section 3.12.
Impedance matching with the single-stub tuner of Fig. 3.30 is now achieved as follows.
Let YT = 1/ZT be the admittance of the terminated transmission line with length d. We can
determine YT using equation (3.27):
ZL + Z0 tan βd
ZT = Z0 ,
Z0 + ZL tan βd (3.90)
1
YT = = GT + BT .
ZT
Now choose the length d in such a way that GT = Y0 . Furthermore, the shunt stub is designed to
cancel the imaginary part of YT , so Ys = −BT . Next to single-stub tuners, double stub tuners
can be used to increase the degrees of freedom. One of the additional advantages of a double-stub
tuner is that it can be designed in such a way that the line length between the two stubs can be
fixed, which simplifies the construction of such a tuner.
Exercise
Determine the parameters (l, d) of a single-stub tuner with a shorted stub to match a complex load
impedance ZL = 100 − 60 Ω to a 50 Ω transmission line. Use transmission lines with Z0 = 50 Ω.
Answer: d = 0.125λ and l = 0.118λ, where λ is the wavelength in the transmission line.
3.12 Microwave filters
Microwave filters can be used to manipulate the frequency response of a microwave system. Mi-
crowave filters are constructed using transmission line components. At lower frequencies (< 5
GHz) lumped elements (L, C) can be used as well. Well-known filter characteristics are low-pass,
high-pass, bandpass or bandstop. Fig. 3.31 illustrates these four types using lumped elements.
There is a lot of literature available for designing lumped-element filters [17], [18]. In this section,
we will assume that such an ideal lumped-element prototype filter is already known. We will show
how such a lumped-element filter can be transformed into a microwave filter using transmission
line components. In the previous section in (3.89) we have already shown that shorted or open
stubs are equivalent to inductors and capacitors. With the particular choice that the length of
Low-pass High-pass Band-pass Band-stop
Figure 3.31: Lumped-element filters.
2π λ π

the stub l = λ/8 we find that Ω = tan βl = tan = tan = 1 which results in:
λ 8 4
shorted stub : Zs = Z0 = L,

(3.91)
Z0 1
open stub : Zs = = ,
 C
which resembles the equivalence of an inductor with inductance L and capacitor with capacitance
C at ω = 1, respectively. Figure 3.32 illustrates the equivalence between the open/shorted stub
and L,C lumped elements for any value of the angular frequency ω = 2πf . Stubs can be used in
l=λ/8 at ω
Short
L Z0=ωL
Circuit
Zin=jωL Zin=jωL
l=λ/8 at ω
Open
C Y0=ωC
Yin=jωC Circuit
Yin=jωC
Figure 3.32: Equivalence of lumped-element L,C and shorted/open stubs with length of λ/8. This
is also known as Richard’s transformation.
parallel (see Fig. 3.30) or in series to a transmission line. A parallel stub can be easily realized in
a practical PCB technology, for example as a shorted or open-ended microstrip line. A series stub
cannot be easily realized. Therefore, it is most common to transform series stubs to parallel stubs
using the so-called Kuroda’s identities. Fig. 3.33 shows the four identities. We will now show
Z1/n2
l l
1/Z2 Z1 Z2/n2
Z1
l l
Z2 n2Z1 1/(n2Z2)
l l 1:n2
Z1 Z2 Z2/n2 Z1/n2
1/Z2 1/(n2Z2)
l l n2:1
Z1 n2Z1
Z2
Figure 3.33: Kuroda’s identities. n2 = 1 + , l = λ/8.
Z1
how Kuroda’s identity can be used to design a low-pass filter in microstrip technology. We will
start with a lumped-element prototype, corresponding to a third-order maximally flat low-pass
filter prototype (see [17], Table 8.3). The ideal circuit model and the corresponding steps towards
the final transmission-line equivalent is provided in Fig. 3.34. The filter has a -3 dB roll-off at
10 GHz. We have assumed that both the input and output of the circuit are connected to 50Ω
ports. The design starts with the ideal circuit model using two inductors and a parallel capacitor.
Next the lumped elements are transformed to transmission line sections, in line with Fig. 3.32.
Finally, by adding additional λ/8 line sections, we can apply Kuroda’s identity (see Fig. 3.33).
The final design can be easily realized in microstrip technology on a PCB. The corresponding
simulated performance of the original ideal circuit model and the final transmission-line low-pass
filter is shown in Fig. 3.35.
Bandpass filters are useful to suppress out-of-band interference. In this way, bandpass filters
can help to reduce the linearity requirements of a receiver or reduce the out-of-band spurious
emission generated by non-linearities of a transmitter. A very practical realization of a bandpass
filter can be created by using parallel-coupled line sections, for example implemented in microstrip
technology. An example of a two-section coupled-line filter is shown in Fig. 3.36, where the top
view of a microstrip implementation is shown. It uses two half-wavelength open-ended resonators
which are coupled over a quarter-wave length to each other and to the connecting input and
output transmission line. The characteristic impedance of the input and output transmission line
is assumed to be equal to Z0 . The width of the resonators and the spacing between the resonators
determine the characteristic impedances of the odd and even modes that can propagate along
the coupled-line sections. By changing the resonators width and spacing, a specific bandpass
characteristic can be realized. Note that short-circuited resonators can also be used. One could
Z0=50.3 Ω λ/8 λ/8 Z0=50.3 Ω

L1=0.8 nH L2=0.8 nH
λ/8 λ/8
C1=0.64 pF
Z0=50 Ω Z0=50 Ω
Z0=24.9 Ω λ/8
Z0=50.3 Ω λ/8 λ/8 Z0=50.3 Ω

Z0=100.3 Ω Z0=100.3 Ω
λ/8 λ/8
Z0=24.9 Ω λ/8 Z0=99.7 Ω λ/8 Z0=24.9 Ω λ/8 Z0=99.7 Ω λ/8
Figure 3.34: Design of a transmission-line low-pass filter in 4 steps using Richard’s transformation
and Kuroda’s identities with n2 = 1.995. Maximally-flat filter with N = 3, R0 = 50Ω and
f−3dB = 10 GHz.
Figure 3.35: Simulated performance of a low-pass filter, ideal lumped-element model (red) and
transmission-line equivalent (blue).
argue that the half-wavelength resonators are in fact printed half-wavelength dipole antennas,
as introduced in section 4.4. However, when implemented in microstrip technology using an
electrically thin dielectric substrate, the electromagnetic field will be mainly concentrated between
the metal strip and ground plane, and as a result, a strong coupling with the adjacent coupled
lines will occur. Therefore, the (unwanted) spurious radiation will be very limited. A single pair
of coupled lines is shown in Fig. 3.37. The two lines which are coupled over an electrical length
λ/4 can be represented by the equivalent transmission line circuit as illustrated in Fig. 3.37 and
consists of two open stubs which are placed in series and are separated by a quarter-wave line
[19]. Now let us assume that Fig. 3.37 represents pair k of a coupled-line filter, where we have
a total of N resonators and N + 1 coupled pairs. Furthermore, let Zok and Zek represent the
odd and even-mode characteristic impedance of the k-th coupled pair, with k = 1..N + 1. The
corresponding characteristic impedances of the λ/4 stubs (indicated by Zks ) and of the λ/4 line
(indicated by Zkl ) of the equivalent circuit are provided in Fig. 3.37.
λ/4
Z0
in
λ/2
Z0
out
Figure 3.36: A two-section (N = 2) bandpass filter consisting of open-ended coupled lines.
In [19] design equations have been derived for the design of Chebyshev coupled-line bandpass
filters. The equations use a low-pass Chebyshev prototype as a starting point. Tables for such
low-pass prototypes can be found in [17], [18]. The design equations from [19] are summarized
below:
πω1 1
θ1 = , Q= ,
2ω0 tan θ1
Z0
K1 = KN +1 = √ ,
g0 g1
v 2
Q(Q2 + 1) P sin θ1
u
P = u , s = Z0 ,
u
t 1 K1 /Z0 (3.92)
Q+ 2
2(K1 /Z0 )
Ze1 = ZeN +1 = Z0 (1 + P sin θ1 ),
Zo1 = ZoN +1 = Z0 (1 − P sin θ1 ),

Z0
Z e1 , Z o1
Z e2 , Z o2
Z e3 , Z o3
Z0
k =1 k =2 k =3
Coupled pair k
Z ks Z ok −1 + Z ok
= Z ks=
+1 Z ok + Z ok +1
λ/4
Z ek − Z ok
Z kl =
2
λ/4
Figure 3.37: The equivalent transmission line circuit of the k-th coupled pair with k = 1..N + 1.
Note that the characteristics impedances of the first and last stub are Z1s = Zo1 and ZN
s N +1
+1 = Zo
.
and for the coupled pair with k = 2...N we have

s
2 2
Z0 Kk tan θ1

Kk = √ , Nk = + ,
gk−1 gk Z0 2
Kk

Zek = ZeN +1−k = s Nk + , (3.93)
Z0
Kk

Zok = ZoN +1−k = s Nk − ,
Z0
where gk are the component values of the low-pass prototype filter with cutoff radial frequency
ω1‘ = 1. Furthermore, ω0 = 2πf0 is the design radial frequency and ω1 = 2πf1 is the radial
frequency corresponding to the lower edge of the passband of the bandpass filter. The bandwidth
of the filter is 2(f0 − f1 )/f0 .
Let us now consider an example to illustrate the usage of design equations (3.92) and
(3.93). We will design a two-section Chebyshev coupled-line filter with a 20% bandwidth, 3 dB
ripple in the passband and a center frequency at 28 GHz. The first step is to determine the
low-pass prototype equivalent circuit. From [17] we find that the prototype values are: g0 = 1,
g1 = 3.1013, g2 = 0.5339, g3 = 5.8095 with a cutoff radial frequency ω1‘ = 1. For the given
bandwidth requirement we find that ω1 /ω0 = 0.9. When substituting these values into the design
3.13. MICROWAVE AMPLIFIERS 61
equations (3.92) and (3.93) we find the following values for the coupled-line filter:
Ze1 = 65.22 Ω, Zo1 = 34.78 Ω,

(3.94)
Ze2 = 57.89 Ω, Zo2 = 35.55 Ω.
The characteristic impedances of the stubs and λ/4 sections (see also Fig. 3.37) now become:
Z1s = 34.78 Ω, Z1l = 15.22 Ω,
Z2s = 70.33 Ω, Z2l = 11.17 Ω, (3.95)
Z3s = 70.33 Ω, Z3l = 15.22 Ω, Z4s = 34.78 Ω.
The final schematic of the bandpass filter is shown in Fig. 3.38 and the corresponding frequency
response is illustrated in Fig. 3.39.
TLOC TLOC TLOC TLOC

TL6 TL7 TL8 TL9
Z=34.7787 Ohm Z=70.3314 Ohm Z=70.3314 Ohm Z=34.7787 Ohm
E=90 E=90 E=90 E=90
F=28 GHz F=28 GHz F=28 GHz F=28 GHz
Ref
Ref
Ref
Ref
Term Term
Term1 TLIN TLIN TLIN TLIN TLIN Term2
Num=1 TL5 TL1 TL2 TL3 TL4 Num=2
Z=50 Ohm Z=50.0 Ohm Z=15.2213 Ohm Z=11.168 Ohm Z=15.2213 Ohm Z=50.0 Ohm Z=50 Ohm
E=90 E=90 E=90 E=90 E=90
F=28 GHz F=28 GHz F=28 GHz F=28 GHz F=28 GHz
Figure 3.38: Schematic ADS model of the two-section (N = 2) Chebyshev bandpass filter consisting
of open-ended coupled lines with a 3 dB passband ripple and 20% bandwidth around the center
frequency f0 = 28 GHz. Note that ideal transmission line models have been used.
Exercise
Design a two-section coupled-line bandpass filter with a 0.5 dB equal-ripple (Chebyshev) and a
30% bandwidth. The input and output are connected to 50Ω transmission lines. The low-pass
prototype values are: g0 = 1, g1 = 1.4029, g2 = 0.7071, g3 = 1.9841. Answer: Z1s = 24.75 Ω, Z1l =
25.25 Ω, Z2s = 48.15 Ω, Z2l = 17.96 Ω, Z3s = 48.15 Ω, Z3l = 25.25 Ω, Z4s = 24.75 Ω.
3.13 Microwave amplifiers

Antennas are almost always connected to amplifiers. In transmit mode, the antenna will be
connected to a power amplifier (PA), whereas in receive mode the antenna will be connected to
Figure 3.39: Simulated performance of the two-section bandpass filter of Fig. 3.38.
a low-noise amplifier (LNA). With the continuous increase of the operational frequency of new
wireless applications, highly integrated active antenna concepts will need to be developed in order
to combat the interconnect losses between antenna and amplifier. Therefore, it is crucial for the
future antenna engineer to have a good understanding of amplifiers.
Microwave amplifiers can be realized in several semiconductor technologies, such as silicon-
based technologies (CMOS and BiCMOS), and III-V technologies, such as gallium arsenide (GaAs)
or gallium nitride (GaN). Discrete transistors can be used to realize an amplifier, but most com-
monly the amplifiers will be integrated in a more complex RF integrated circuit (RF-IC). More
background material on microwave amplifiers can be found in [20].
3.13.1 Power gain, available gain and transducer gain
For the analysis and design of amplifiers it is convenient to use three definitions for power gain.
Consider the amplifier circuit of Fig. 3.40(a). In the center of this circuit we find the active device
(transistor) which is described in terms of a two-port network. The active device is connected to
the source via an input matching network and at the output it is connected to a load impedance
via an output matching network. Note that the combination of the output matching network and
load impedance could be the antenna impedance in case of a power amplifier which is directly
matched to an antenna. In a similar way, the source impedance and the input matching circuit
could be the antenna impedance in case of a low-noise amplifier. Fig. 3.40(b) shows the equivalent
circuit where the properties of the input and output matching networks are included in the source
impedance ZS and load impedance ZL , respectively. The three definitions that we will use are:
Pdel,L
i. Delivered power gain Gdel = , which is the ratio between the delivered power to the
Pdel,in
load Pdel,L and the delivered input power to the amplifier Pdel,in .
ΖS Ζin Ζout ΖL
Amplifier
Z0
Input Output
matching matching Z0
+
network network
VS ∼
- MS ML
ΓS Γin Γout ΓL
Pav,S Pdel,in Pav,out Pdel,L
(a)
I1 I2
+ Amplifier +
ZS
V1 V2 ZL
+
VS ∼
- - -
ΓS Γin Γout ΓL
(b)
Figure 3.40: (a) General amplifier circuit including an input and output matching network. (b)
Equivalent amplifier circuit.
Pav,out
ii. Available power gain Gav = , which is the ratio between the available power at the
Pav,S
output of the amplifier Pav,out and the available power from the source Pav,S .
Pdel,L
iii. Transducer power gain GT = , which is the ratio between the delivered power to the
Pav,S
load Pdel,L and the available input power from the source Pav,S .
In section 3.9 we already derived the expressions for the input and output reflection coefficients,
given by (3.64) and (3.65). Now let Zin be the input impedance looking into the amplifier as
indicated in Fig. 3.40. The voltage V1 can be written as:
VS Zin (3.96)
V1 = = V1+ (1 + Γin ) .
ZS + Zin
From (3.96) we can determine V1+ and calculate the time-average input power to the amplifier,
similar to the approach of section 3.6:
|V1+ |2 |VS |2 |1 − ΓS |2
Pdel,in = 1 − |Γin |2 = 1 − |Γ in |2
, (3.97)
2Z0 8Z0 |1 − ΓS Γin |2
where Γin , ΓS and ΓL are given by (3.64) and (3.65). In addition, we have expressed Zin in terms
of Γin by using
Zin − Z0 (3.98)
Γin = .
Zin + Z0
In a similar way we can determine the delivered power to the load:
|V2− |2 |VS |2 |S21 |2 1 − |ΓL |2 |1 − ΓS |2

2 (3.99)
Pdel,L = 1 − |ΓL | = ,
2Z0 8Z0 |1 − S22 ΓL |2 |1 − Γin ΓS |2
where we have written the incident and reflected voltage wave amplitudes in terms of the scattering
coefficients using (3.61) and (3.62). We then finally find the delivered power gain by using (3.97)
and (3.99):
|S21 |2 1 − |ΓL |2

Pdel,L (3.100)
Gdel = = .
Pdel,in |1 − S22 ΓL |2 (1 − |Γin |2 )
In a similar way we can determine the other two power gain definitions:
|S21 |2 1 − |ΓS |2

Pav,out
Gav = = ,
Pav,S |1 − S11 ΓS |2 (1 − |Γout |2 ) (3.101)
Pdel,L
GT = = Gav ML = Gdel MS ,
Pav,S
where MS and ML represent the source and load mismatch factors, respectively, and are given
by:
1 − |Γin |2 1 − |ΓS |2

Pdel,in
MS = = ,
Pav,S |1 − Γin ΓS |2 (3.102)
1 − |Γout |2 1 − |ΓL |2

Pdel,L
ML = = .
Pav,out |1 − Γout ΓL |2
Note that Γout is given in (3.64). Note that in case of a unilateral amplifier, e.g. S12 = 0,
the expressions for the input and output reflection coefficient (3.64) simplify to Γin = S11 and
Γout = S22 .
As an example, we will use the BFU730F transistor from NXP Semiconductors. A basic
circuit (excluding biasing) using this amplifier is shown in Fig. 3.41. The amplifier is used as
a low-noise amplifier which is connected at the input to a dipole antenna with an impedance of
73 Ω. The output of the amplifier is connected to a 50 Ω load. The complex amplitude of the
time-harmonic voltage source that represents the receiving antenna is equal to VS = 0.01 V. The
S-parameters of the BFU730F transistor in terms of magnitude and phase at 400 MHz are given
by
S11 = 0.87, −28◦ ,
S12 = 0.01, 76◦ ,

(3.103)
S21 = 26.73, 159◦ ,
S22 = 0.96, −16◦ .

BFU730
ZS=73 Ω
ZL =50 Ω
+
VS=0.01 V ∼
-
Figure 3.41: Amplifier based on the BFU730F transistor from NXP Semiconductors. The input
is connected to an antenna with an input impedance of 73 Ω which generates an equivalent time-
harmonic voltage of 0.01 V. The output is connected to a 50 Ω load.
By using (3.64) we find the input and output reflection coefficients:
Γin = 0.77 − 0.41,

(3.104)
Γout = 0.89 − 0.31,
The corresponding values for the three power gain definitions are found using (3.100) and (3.101):
Gdel = 34.7 dB,
Gav = 38.9 dB, (3.105)
GT = 29.7 dB,
and the corresponding source and load mismatch factors:
MS = 29.7 − 34.7 = −5.0 dB,

(3.106)
ML = 29.7 − 38.9 = −9.2 dB.
The power available from the source is
|VS |2 (3.107)
Pav,S = = 0.17 µW,
8ZS
which corresponds to -37.7 dBm. The available power at the output of the amplifier is then
Pav,out = Gav Pav,S = 1.3 mW which is equal to +1.2 dBm.
3.13.2 Stability
One of the key aspects in the design of amplifiers is the so-called stability of the amplifier circuit.
If |Γin | > 1 or |Γout | > 1 oscillation could be possible in the circuit of Fig. 3.40(a). The amplifier
is stable if the following two criteria are both satisfied:
S12 S21
|Γin | = S11 + < 1,
1
− S22
ΓL (3.108)
S12 S21
|Γout | = S22 + < 1.
1
− S11
ΓS
The amplifier is unconditionally stable if |Γin | < 1 and |Γout | < 1 for all passive source and load
impedances, e.g. with |ΓS | < 1 and |ΓL | < 1. The amplifier is conditionally stable if (3.108) is
satisfied only for certain source and load impedances. Conditional stability can be easily visualized
by using input and output stability circles in the Smith Chart. Let us consider the output stability
circle in more detail. It defines the boundary in the Smith chart between the stable and unstable
region for ΓL when |Γin | = 1. The radius rL and center cL of this output stability circle are given
by [20]:
S12 S21
rL = ,
|S22 |2 − |∆|2 (3.109)
(S22 − ∆S11 ∗ )∗
cL = ,
|S22 |2 − |∆|2
where ∆ is the determinant of the scattering matrix:
∆ = S11 S22 − S12 S21 . (3.110)
Fig. 3.42 illustrates the use of the output stability circle in a Smith Chart. In a similar way,
source stability circles can be drawn in a Smith Chart to indicate the boundary between the stable
and unstable region for ΓS when |Γout | = 1. The corresponding radius rS and center cS of the
input stability circle are given by:
S12 S21
rS = ,
|S11 |2 − |∆|2
(3.111)
(S11 − ∆S22 ∗ )∗
cS = .
|S11 |2 − |∆|2
Often we are only interested in knowing whether an amplifier is unconditionally stable
within a certain frequency range for all source and load impedances. To check this we can use the
Rollett stability condition [21], expressed in terms of a condition for two parameters, K and ∆:
1 − |S11 |2 − |S22 |2 + |∆|2

K = > 1,
2|S12 S21 | (3.112)
|∆| = |S11 S22 − S12 S21 | < 1.
Load stability circle
rL
cL
Amplifier
ZL
in L
Figure 3.42: Illustration of the output (load) stability circle in the Smith chart. It shows (green
region) which loads (represented by ZL or ΓL ) result in load stability with Γin < 1. Note that we
have assumed that |S11 | < 1.
As an alternative for the K-∆ condition, the single parameter (µ-test) introduced in [22] can also
be used to check if the transistor is unconditionally stable at a certain frequency:
1 − |S11 |2 (3.113)
µ = ∗ ∆| + |S S | > 1,
|S22 − S11 12 21
As an example, we can check the stability of the BFU730F transistor by substituting the scattering
parameters of (3.103) in to (3.112) or (3.113). We then find that K = 0.039 and |∆| = 0.836.
The corresponding µ-test value is µ = 0.38. This means that this device is not unconditionally
stable at this frequency.
3.13.3 Amplifier design using constant gain circles
The overall gain of the amplifier of Fig. 3.40(a) can be maximized by carefully choosing the input
and output matching networks, specified by MS and ML . The main drawback of this strategy
is that the overall circuit will become quite narrow band in general. Therefore, it is better to
compromise on the achievable gain in order to realize a more broadband amplifier. By plotting
so-called constant gain circles in the Smith chart a trade-off can be made between realized gain
and input and output matching. We will consider a unilateral amplifier with S12 = 0. The
unilateral transducer gain can now be written in the following form:
GT u = MS |S21 |2 ML , (3.114)
where the source and load mismatch factors are now given by:
1 − |ΓS |2

MS = ,
|1 − S11 ΓS |2
(3.115)
1 − |ΓL |2

ML = .
|1 − S22 ΓL |2
∗ with maximum value:
The factor MS is maximised when ΓS = S11
1 (3.116)
MS,max = .
1 − |S11 |2
We can now introduce a normalized value gS with
MS
(3.117)
gS = = MS 1 − |S11 |2 .
MS,max
For a constant value of gS , expression (3.117) will be a circle in the ΓS plane when plotted in a
Smith chart with radius and center given by:
√
1 − |S11 |2 1 − gS
rS = ,
1 − |S11 |2 (1 − gS )
(3.118)
gS S11∗
cS = .
1 − |S11 |2 (1 − gS )
If we again consider the BFU730F transistor and assume that this transistor is unilateral with
S12 = 0 and where the other scattering parameters are given by (3.103) at 400 MHz, we obtain the
constant-gain circles as illustrated in Fig. 3.43. Observe that the circle for MS = 0 dB passes the
centre of the Smith chart at ΓS = 0. It can be shown that this is always the case for any device.
The maximum value for this device is MS = 6.1 dB. In order to improve/simplify the matching
towards the source, we could compromise a few dB, for example by choosing the MS = 2 dB circle
in Fig. 3.43. In this case we would choose the point ΓS on the circle closest to the center of the
Smith chart as illustrated in Fig. 3.43. In a similar way, we can find the constant gain circles in
the load (ΓL ) plane. The radius and center in a Smith chart are given by:
√
1 − |S22 |2 1 − gL
rL = ,
1 − |S22 |2 (1 − gL ) (3.119)
gL S22∗
cL = .
1 − |S22 |2 (1 − gL )
where gL is the normalized value given by:
ML
(3.120)
gL = = ML 1 − |S22 |2 .
ML,max
The constant gain circles in the load plane for the BFU730F transistor at 400 MHz are shown in
Fig. 3.44.
3.14. LOW-NOISE AMPLIFIERS 69
MS,max= 6.1 dB
S
MS = 4 dB
MS = 2 dB
MS = 0 dB
Figure 3.43: Constant-gain circle of the BFU730F transistor at 400 MHz plotted in the source
(ΓS ) plane. For MS = 0 dB, we find cS = 0.495, 28◦ , rS = 0.495; for MS = 2 dB, we find
cS = 0.627, 28◦ , rS = 0.356; for MS = 4 dB, we find cS = 0.753, 28◦ , rS = 0.215. The transistor
is assumed to be unilateral.
3.14 Low-noise amplifiers

3.14.1 Noise in microwave circuits
In 1926 John Johnson at Bell Labs observed for the first time that a resistor generates noise in
the absence of external current biasing. This so-called thermal noise or Johnson noise is due to
the random motion of charges in the resistor material. Consider the resistor connected at the
input of a microwave amplifier, as illustrated in Fig. 3.45. The available noise power generated
by this resistor at the input is given by:
N = kb T0 B, (3.121)
where kb = 1.38 × 10−23 J/K is Boltmann’s constant and B [Hz] is the noise bandwidth of the
system. Note that by adding additional resistors in series or in parallel at the input would not
affect the available noise power, since the available noise power is independent of the resistor
value. As a result, the available noise power at the input of any receiver will always be equal to
-174 dBm/Hz at room temperature (T0 = 300 K). Note that dBm refers to decibel-milliwatts,
where 0 dBm corresponds to 1 mW of power. The noise figure F of a two-port microwave network
ML,max= 11 dB
ML = 8 dB
ML = 4 dB
ML = 0 dB
Figure 3.44: Constant-gain circle of the BFU730F transistor at 400 MHz plotted in the load (ΓL )
plane. For ML = 0 dB, we find cL = 0.5, 16◦ , rL = 0.5; for ML = 4 dB, we find cL = 0.73, 16◦ ,
rL = 0.27; for ML = 8 dB, we find cL = 0.89, 16◦ , rL = 0.1. The transistor is assumed to be
unilateral.
Ni=kbT0B No
Gav ,F
Figure 3.45: Thermal noise generated by a resistor at the input of a microwave amplifier.
is defined as the ratio between the input signal-to-noise ratio (SNR) and output SNR:
Si

N
i .
F = (3.122)
So
No
Since any microwave network will add noise to the available input noise power, we find that F > 1.
Usually the noise figure is expressed in [dB] according to:
NF = 10 log10 (F ) [dB]. (3.123)
Exercise
Determine the available noise power at the input of a Bluetooth receiver operating in the 2.4 GHz
band at room temperature. The channel bandwidth is 1 MHz. Answer: Ni = −114 dBm.
Let us go back to the amplifier of Fig. 3.45. The amplifier is characterized by the available
power gain Gav and noise figure F . By using relation (3.123) and (3.121), we can write the noise
figure in the following form:
Si Si

N kT B No
F = i = b 0 = . (3.124)
So Gav Si Gav kb T0 B
No No
The noise which is added by the amplifier ∆Niamp seen at the input of a noise-free amplifier with
available power gain Gav is given by:
No − Gav Ni (3.125)
∆Niamp = = (F − 1)kb T0 B.
Gav
Apart from the noise figure, it is often also convenient to express the noise performance of a
microwave network in terms of the equivalent noise temperature Te with
Te = (F − 1)T0 . (3.126)
In order to visualize the noise performance and dynamic range of microwave receivers, a
power level diagram can be a very useful tool. Fig. 3.46 illustrates an example of such a level
diagram. In this case a Bluetooth receiver is considered operating in the 2.4-2.48 GHz band with
an instantaneous channel bandwidth of 1 MHz. All power levels are determined at the input of
the receiver. The equivalent input noise power is kb T0 BF . The minimum required SNR required
to demodulate the modulated received signal is assumed to be So /No = 12 dB. From the level
diagram we can determine the dynamic range in which the receiver has to operate.
Microwave receivers usually consist of several cascaded stages, each stage being characterized
by an available power gain and noise figure. In Fig. 3.47 a two-stage network is shown, consisting
of a source and two amplifier stages. At the output of the first stage we find:
S1 = Gav1 Si ,
(3.127)
N1 = Gav1 kb T0 BF1 .
-10 dBm (Pmax)
Dynamic range of
receiver
-92 dBm (Pmin)

(So/No)min
-104 dBm (kbT0BF)
-114 dBm (kbT0B)
-174 dBm (kbT0)
Figure 3.46: Typical power level diagram of a Bluetooth receiver. All power levels are defined at the
input of the receiver at room temperature (T0 = 300 K). The channel bandwidth B = 1 MHz, the
minimum required SNR for demodulation is So /No = 12 dB, the receiver noise figure N F = 10 dB
and the maximum tolerated input power Pmax = −20 dBm.
As a consequence, the input signal power and noise power for the second stage are given by S1
and N1 and provide at the output of the second stage:
S2 = Gav2 S1 = Gav1 Gav2 Si ,

(3.128)
N2 = Gav2 N1 + Gav2 (F2 − 1)kb T0 B = Gav2 kb T0 B[Gav1 F1 + (F2 − 1)],
where relation (3.125) was used. The total noise figure of the cascaded two-stage network is then
given by
Si Si

N kb T0 B F −1
Ftot = i = = F1 + 2 . (3.129)
S2 Gav1 Gav2 Si Gav1
N2 Gav2 kb T0 B[Gav1 F1 + (F2 − 1)]
Equation (3.129) is also known as the Friis formula named according to the Danish electrical
engineer Harald Friis. As a consequence of this relation, the noise contribution of the second
stage is significantly reduced thanks to the available power gain of the first stage. In receiver
systems the first amplifier determines the noise performance and should have a low noise figure
and reasonable power gain. Often special (and more expensive) semiconductor technologies are
used to realize the first stage, for example by using III-V technologies such as Gallium Arsenide
(GaAs). Similar to (3.129) we can also determine the equivalent noise temperature of a cascaded
two-stage network, resulting in:
Te2 (3.130)
Tetot = Te1 + ,
Gav1
where Te1 and Te2 are the equivalent noise temperatures of the first and second stage, respectively,
which can be found using (3.126). Friis formula can be extended to a cascaded network with N
Gav1 ,F1 Gav2 ,F2
Ni=kbT0B N1 N2
RS
Si S1 S2
VS
-
Figure 3.47: Cascaded two-stage amplifier.
stages, providing:
F2 − 1 F3 − 1 FN − 1
Ftot = F1 + + + ··· + ,
Gav1 Gav1 Gav2 Gav1 Gav2 · · · GavN −1 (3.131)
Te2 Te3 TeN
Tetot = Te1 + + + ··· + .
Gav1 Gav1 Gav2 Gav1 Gav2 · · · GavN −1
Exercise
Determine the noise figure of a two-stage amplifier at room temperature. The first stage is a
high-performance GaAs low-noise amplifier with a 15 dB power gain and 1.6 dB noise figure. The
second stage is realized in low-cost Silicon with a power gain of 25 dB and 4 dB noise figure. Also
determine the overall equivalent noise temperature. Answer: N Ftot = 1.65 dB and Tetot = 138.1 K.
3.14.2 Noise Matching

Consider a low-noise amplifier which is connected to a source with complex source impedance
ZS = RS + XS . The noise figure of this amplifier can be determined from the following relation
[20], [23]:
Rn (3.132)
F = Fmin + |YS − Yopt |2 ,
GS
where the source admittance is YS = 1/ZS = GS + BS . The other three parameters in (3.132)
quantify the noise performance of a particular single-stage low-noise amplifier:
• Fmin , the minimum noise figure at particular DC-biasing of the transistor,
• Rn = 1/Gn , the equivalent noise resistance of the transistor,
• Yopt , the optimum source admittance that provides the minimum noise figure.
Relation (3.132) can also be rewritten in terms of impedance values:
Gn (3.133)
F = Fmin + |ZS − Zopt |2 ,
RS
Sometimes (e.g. in case of ultra low-noise amplifiers for radio astronomy) it is also convenient to
convert (3.132) to the equivalent noise temperature:
Rn (3.134)
Te = Tmin + |YS − Yopt |2 ,
GS
where the minimum noise temperature is Tmin = (Fmin − 1)T0 . It appears to be very convenient
for the design of a particular low-noise amplifier to use the Smith chart. In the Smith chart we can
indicate the region for the source impedance ZS for which the amplifier has a noise figure below
a certain value. In the same Smith chart other regions can be indicated for ZS in which other
relevant performance indicators, such as the available power gain Gav , is higher than a certain
minimum required value. If the regions overlap, a suitable choice for ZS can be obtained that
satisfies all requirements. In order to be able to use (3.132) in a Smith chart, we need to convert
the admittance values to reflection parameters. By using
1 − Γopt
Yopt = Y0 ,
1 + Γopt (3.135)
1 − ΓS
YS = Y0 ,
1 + ΓS
we can transform (3.132) in the following form:
4Rn |ΓS − Γopt |2 (3.136)

F = Fmin + ,
Z0 (1 − |ΓS |2 ) |1 + Γopt |2
where Γopt is the optimum source reflection coefficient that would result in a minimum noise figure
and ΓS is the realized source reflection coefficient. It can be shown that (3.136) represents a circle
for ΓS in the Smith Chart with center ON and radius RN given by [23]:
Γopt
ON = ,
1+N
r
1
(3.137)
RN = N 2 + N 1 − |Γopt |2 ,
1+N
F − Fmin
N = |1 + Γopt |2 ,
4Rn /Z0
where N is also known as the noise figure parameter. Let us now consider an example to illus-
trate how noise circles can be used. Consider the SiGe bipolar transistor BFU730F from NXP
Semiconductors. From the datasheet we find that at 400 MHz, with a DC collector current of
Ic = 10 mA and collector-emitter voltage VCE = 2 V, we get N Fmin = 0.57 dB, Rn = 6 Ω and
Zopt = 100 + 5.2 Ω. Suppose we want to design an amplifier using this transistor with a noise
figure below 1 dB. The noise circle that defines the area in which ΓS can be selected to obtain a
maximum noise figure of 1 dB is given by:
N = 0.443,
ON = 0.2315 + 0.016, (3.138)
RN = 0.53.
The resulting noise circle is shown in Fig. 3.48. We can observe that with a 50 Ω source connected
at the input of the transistor, a noise figure below 1 dB will be obtained.
Exercise
Determine expression (3.136) from (3.132) and (3.135).
Noise circle NF=1 dB
Minimum noise figure
Figure 3.48: Noise circle for N F = 1 dB corresponding to an amplifier based on the BFU730F
transistor at 400 MHz biased at Ic = 10 mA and VCE = 2 V. When the source impedance falls
within the noise circle, the resulting noise figure will be below or equal to 1 dB. The minimum
noise figure of 0.57 dB is obtained at Zopt = 100 + 5.2 Ω or Γopt = 0.334 + 0.023
Chapter 4
Antenna Theory
In this chapter we will introduce the general properties of radiated fields from antenna sources.
We will use Maxwell’s equations in the space-frequency domain as a starting point in which
the antenna is represented by a current or charge distribution. Next, we will determine the
electromagnetic field in the far field of an arbitrarily electric current or charge distribution. Based
on this, several elementary antenna structures will be analyzed, such as wire antennas and loop
antennas. Finally, we will introduce magnetic sources and the Lorentz-Larmor theorem in order
to extend the theoretical framework towards the analysis of aperture antennas, such as horns,
reflector antennas and microstrip antennas.
4.1 Maxwell’s equations

Maxwell’s equations describe the propagation of electromagnetic waves generated by a current or
charge distribution in any medium. In the time domain Maxwell’s equations are given by
~ r, t)
∂ B(~
~ r, t)
∇ × E(~ = − ,
∂t
~ r, t)
∂ D(~
~ r, t) =
∇ × H(~ + J~e (~r, t),
∂t (4.1)
~ r, t)
∇ · B(~ = 0,
~ r, t)
∇ · D(~ = ρe (~r, t),
where E~ is the electric field in Volt per meter [V/m], D

~ the displacement current in Coulomb
2 2 ~ the magnetic flux density in Weber per m2 [W/m2 ], H ~ the magnetic field in
per m [C/m ], B
Ampere per meter [A/m], ρe the electric charge distribution in Coulomb per m3 [C/m3 ] and J~e
the electric current distribution in Ampere per m2 [A/m2 ]. In addition, the continuity equation
provides a relation between the electric current and charge distribution:
∂ρe (~r, t) (4.2)

∇ · J~e (~r, t) + = 0.
∂t
77
78 CHAPTER 4. ANTENNA THEORY
Note that the continuity equation is not an independent equation, but can be derived from
Maxwell’s equations (4.1).
Since the propagation of electromagnetic waves from a transmit antenna to a receive antenna
usually takes place in free-space, we can state that:
~ r, t) = 0 E(~
D(~ ~ r, t),
(4.3)
~ r, t)
B(~ ~ r, t).
= µ0 H(~
Furthermore, we will again assume a time-harmonic dependence of all field components, currents
and charges, with angular frequency ω = 2πf , where f is the frequency expressed in Hertz [Hz].
In this way, the electric field is expressed as:
h i
~ r, t) = Re E(~
E(~ ~ r)eωt . (4.4)
A similar relation holds for the other components of (4.1). Maxwell’s equations now reduce to
the following set of equations:
~ r)
∇ × E(~ ~ r),
= −ωµ0 H(~
~ r) = ω0 E(~
∇ × H(~ ~ r) + J~e (~r),
(4.5)
~ r)
∇ · H(~ = 0,
~ r) ρe (~r)
∇ · E(~ = .
0
Next step is to determine the electric and magnetic fields, E(~ ~ r) and H(~~ r), due to the
~
source distributions, given by Je (~r) and ρe (~r). From these frequency-domain fields, we can then
determine the physical (time-domain) fields E(~ ~ r, t) and H(~
~ r, t) by using relation (4.4). Note that
in (4.5), we have used the linear properties of Maxwell’s equations, which allows to interchange
the vector operators ∇× and ∇· with the operator Re[]. In the rest of the derivation, we will
omit the use of (~r) in our notation. From ∇ · H ~ = 0 we can conclude that H ~ can be expressed in
terms of a vector potential A~ e according to
~ = 1 ∇×A
H ~ e. (4.6)
µ0
Note that A ~ e does not have a unique solution, since the vector A ~ 1e = A
~ e + ∇φe , with φe an
~
arbitrarily scalar potential, provides exactly the same solution of H. Substitution of (4.6) in the
first equation of (4.5) provides
~ = −ω∇ × A
∇×E ~ e, (4.7)
which leads to
~ = −ω A
E ~ e − ∇φe , (4.8)
4.1. MAXWELL’S EQUATIONS 79
where φe is an unknown scalar function of ~r. Substituting (4.8) in the second equation of (4.5)
gives
~ e = ω0 µ0 E
∇×∇×A ~ + µ0 J~e = k 2 A
~ ~
0 e − ω0 µ0 ∇φe + µ0 Je . (4.9)
Where k0 is the wavenumber in free space with k02 = ω 2 0 µ0 = ω 2 /c2 , with c the speed of light.
~ e = ∇∇ · A
By using the vector property ∇ × ∇ × A ~ e − ∇2 A~ e , (4.9) becomes
~ e + k2 A
∇2 A ~ ~ ~
0 e = ∇∇ · Ae + ω0 µ0 ∇φe − µ0 Je . (4.10)
Up to now, we did not specify the scalar potential φe . We will choose φe according to
~ e = −ω0 µ0 φe .
∇·A (4.11)
This particular choice is known as the Lorenz-gauge. Equation (4.10) can now be re-written in
the well-known Helmholtz equation
~ e + k02 A
∇2 A ~ e = −µ0 J~e . (4.12)
Note that we also need to satisfy the last equation of (4.5). This results in
~ = −ω0 ∇ · A
∇ · (0 E) ~ e − 0 ∇2 φe = ρe . (4.13)
Again, we apply the Lorenz-gauge, which provides us

ρe
∇2 φe + k02 φe = − . (4.14)
0
When considering antennas, the sources which are represented by the electric current density
~
Je and charge distribution ρe , will be located within a finite-size volume V0 . This is illustrated in
Fig. 4.1. As a first step, we will determine the solution of the Helmholtz equation for an electric
dipole (point source) oriented along the z-axis located at the origin of the coordinate system
(x, y, z) = (0, 0, 0) , where ~uz represents the unit vector along the z-direction. The configuration
is illustrated in Fig. 4.2. Now let I0 be the total current flowing through the dipole and l the length
of the dipole. Characteristic for the electric dipole is the fact that the length is much smaller than
the free-space wavelength, so l λ0 . The product I0 l = p is also called the ”dipole-moment”.
The current density along the electric dipole can now te expressed as:
J~e = I0 lδ(x)δ(y)δ(z)~uz = I0 lδ(~r)~uz . (4.15)
Here δ(r) represents the three-dimensional delta function. Since the current has only a component
in the z direction, the vector potential will also only have a component in the z-direction, i.e.
A~ e = Aez ~uz . The vector potential now satisfies the following equation
∇2 Aez + k02 Aez = −µ0 I0 lδ(~r). (4.16)
The scalar quantity Aez needs to represent a spherical behavior, since we have assumed a point
source. As a result Aez (~r) = Aez (r). Therefore, we can rewrite (4.16) in the following form
1 d dAez

r2 ~ ez = −µ0 I0 lδ(~r).
+ k02 A (4.17)
2
r dr dr
rd
Q r
ro
O
y
V0
Figure 4.1: The antenna current and charge distribution are located within the volume V0
Je
l
Figure 4.2: Electric dipole located in the origin and oriented along the z-axis with J~e = I0 l~uz .
4.1. MAXWELL’S EQUATIONS 81
Outside the origin, we now find two possible solutions of this differential equation:
C k0 r
A1ez = e ,
r (4.18)
C −k0 r
A2ez = e .
r
where C is a constant. The first solution A1ez represents a spherical wave propagating towards
the source, whereas the second solution A2ez represents a wave which propagates away from the
source. This is due to the fact that we have assumed a eωt time dependence of the fields. We will
use the second solution, where we still need to determine the constant C. We can determine C by
substituting k0 ↓ 0 in (4.17), which transforms into the Poisson equation known from electrostatics
with solution
µ0 I0 l (4.19)
Aez = .
4πr
µ0 I0 l ~ e of an electric
As a result, we find that C = . The general solution of the vector potential A
4π
dipole along the z-axis now takes the following form:
~ e = ~uz Aez = ~uz µ0 I0 l e−k0 r .

A (4.20)
4πr
The vector potential in (4.20) has the same direction as the dipole. This can be generalized
to all directions. Therefore, the vector potential of an arbitrarily oriented dipole located at
~r0 = ~ux x0 + ~uy y0 + ~uz z0 can be found by substituting r in expression (4.20) by |~r − ~r0 | =
p
(x − x0 )2 + (y − y0 )2 + (z − z0 )2 . We now obtain
−k0 |~
r−~
r0 |
~ e (~r) = I0~l(~r0 ) µ0 e
A
4π |~r − ~r0 | (4.21)
= I0~l(~r0 )G(~r, ~r0 ).
The function G(~r, ~r0 ) is known as the Greens function and represents the response of a point
source. The Greens function plays an important role in numerical electromagnetic methods (e.g.
Method-of-Moments) which are used to analyze complex antenna structures. More details can be
found in chapter 5.
The vector potential of a general electric current distribution J~e within the volume V0 can
be obtained by applying the superposition principle. The solutions of expression (4.12) and (4.14)
for an observation point P outside the source volume V0 can now be easily found by:
−k0 |~
r−~
r0 |
~ e (~r) = µ0 e
Z
A J~e (~r0 ) dV0 , (4.22)
4π V0 |~r − ~r0 |
1 e−k0 |~r−~r0 |
Z
φe (~r) = ρe (~r0 ) dV0 , (4.23)
4π0 V0 |~r − ~r0 |
where ~r0 indicates a source coordinate in the volume V0 and ~r indicates an observation point at
which the fields are being determined. It should be noted that the solutions of (4.12) and (4.14)
should meet the Lorenz-gauge. By using the notation
1 e−k0 |~r−~r0 |
ϕ(~r, ~r0 ) = , (4.24)
4π |~r − ~r0 |
we find that
∇r · J~e (~r0 )ϕ(~r, ~r0 ) = ϕ(~r, ~r0 )∇r · J~e (~r0 ) + (J~e (~r0 ) · ∇r ϕ(~r, ~r0 ))
= J~e (~r0 ) · ∇r ϕ(~r, ~r0 ) (4.25)
= −J~e (~r0 ) · ∇r0 ϕ(~r, ~r0 ).
Note that the operator ∇r is related to ~r and not to ~r0 . Furthermore, we have used the fact that
~ e.
∇r ϕ(~r, ~r0 ) = −∇r0 ϕ(~r, ~r0 ). As a next step, we will determine ∇r · A
Z
~ e (~r) = µ0
∇r · A ∇r · J~e (~r0 )ϕ(~r, ~r0 )dV0
V0 (4.26)
Z Z
= µ0 ϕ(~r, ~r0 )∇r0 · J~e (~r0 )dV0 − µ0 ∇r0 · J~e (~r0 )ϕ(~r, ~r0 )dV0
V0 V0
By applying Gauss theorem and by selecting a volume V0 much larger than the source region, the
second integral on the right-hand side of equation (4.26) will vanish. In the first integral we can
replace J~e (~r0 ) by means of the continuity equation (4.2) by ρe . This results in the Lorenz-gauge:
Z
~ e (~r) = −ωµ0
∇r · A ρ
~e (~r0 )ϕ(~r, ~r0 )dV0
V0 (4.27)
= −ω0 µ0 φe (~r).
~ and H
By using the Lorenz-gauge we are able to express the fields E ~ directly in terms of the
~
vector potential Ae .
~ = 1 ∇×A
H ~ e, (4.28)
µ0
~ = −ω A
E ~ e − ∇φe
∇∇ · A~e
~e +
= −ω A
ω0 µ0 (4.29)
1 h ~ e + ∇2 A
~ e + k02 A
~e
i
= ∇×∇×A
ω0 µ0
1 h i
~ e − µ0 J~e .
= ∇×∇×A
ω0 µ0
4.2. RADIATED FIELDS AND FAR-FIELD APPROXIMATION 83
Outside the source region, we now find that
~ = 1 ~ e,
H ∇×A
µ0 (4.30)
~ 1 ~ e.
E = ∇×∇×A
ω0 µ0
~ e only depends on J~e (and does not depend on the charge distribution
Since the vector potential A
ρe ), we only need to know the electric current distribution J~e on a particular antenna structure
in order to determine the radiated electromagnetic fields. This is a consequence of the applied
Lorenz-gauge.
4.2 Radiated fields and far-field approximation

In chapter 2 we have discussed the various field regions of antennas. Since most antennas are used
to transfer information over long distances, we are primarily interested in the fields far away from
the antenna. In this section we will determine a general analytic expression of the electromagnetic
field in the so-called far-field region of the antenna. The derivation starts with the expression of
the vector potential (4.22), which we derived in the previous section. Combining this with (4.30)
results in
~ = 1 ~e
H ∇×A
µ0 (4.31)
Z
= ∇r × J~e (~r0 )ϕ(~r, ~r0 )dV0 ,
V0
where ϕ has been defined in 4.24. Furthermore, operator ∇r × is performed on the field point ~r.
We now know that
∇r × J~e (~r0 )ϕ(~r, ~r0 ) = ϕ(~r, ~r0 )∇r × J~e (~r0 ) + ∇r ϕ(~r, ~r0 ) × J~e (~r0 ). (4.32)
Note that ∇r × J~e (~r0 ) = 0 and

1 −k0 |~r−~r0 | 1 1

∇r ϕ(~r, ~r0 ) = e ∇r + ∇r e−k0 |~r−~r0 | . (4.33)
4π |~r − ~r0 | 4π|~r − ~r0 |
By using the following relations
1 ~r − ~r0

∇r =− , (4.34)
|~r − ~r0 | |~r − ~r0 |3
and
(~r − ~r0 )
∇r e−k0 |~r−~r0 | = −k0 e−k0 |~r−~r0 | , (4.35)
|~r − ~r0 |
we find that
−k0 rd
1 1 e
Z
~ =
H −k0 − ~urd × J~e (~r0 )dV0 , (4.36)
4π V0 rd rd
where we have introduced the coordinate ~rd according to ~rd = ~r − ~r0 with rd = |~rd | and where
~rd
the unit-vector in the direction ~rd is given by ~urd = .
rd
We will now derive an approximation of the magnetic field H, ~ which is only valid at a large
distance from the antenna system. The distance rd between a field point P with coordinates
(x, y, z) and the source point with coordinates (x0 , y0 , z0 ) is given by
q
rd = ((x − x0 )2 + (y − y0 )2 + (z − z0 )2 ). (4.37)
This can also be written in the following form

q
r2 − 2(xx0 + yy0 + zz0 ) + (x20 + y02 + z02 ) ,

rd = (4.38)
with r2 = x2 + y 2 + z 2 . In the far-field region x0 , y0 and z0 will be very small as compared to r.

As a result, we can express rd in terms of a binomial distribution according to
" !
1 2 x2 + y02 + z02
rd = r 1+ − 2 (xx0 + yy0 + zz0 ) + 0
2 r r2
!2 
1 2 x2 + y02 + z02
− − 2 (xx0 + yy0 + zz0 ) + 0 + · · ··
8 r r2
" #
xx0 + yy0 + zz0 x20 + y02 + z02 (xx0 + yy0 + zz0 )2 (4.39)
= r 1− + − +···
r2 2r2 2r4
xx0 + yy0 + zz0
≈ r−
r
= r − ~ur · ~r0 ,
~r x2 + y02 + z02
with ~ur = . This approximation is only valid if the term 0 is small enough. The
r 2r
corresponding exponential factor will be close to 1. The criterion which is often used in literature
is
x20 + y02 + z02 λ0
< . (4.40)
2r 16
Consider an antenna with maximum dimension L and assume that the center of the antenna is
located in the origin of our coordinate system. Equation(4.40)now corresponds to the well-known
far-field criterion introduced in chapter 2
2L2
R> . (4.41)
λ0
e−k0 rd
We can now also replace the variable rd in the denominator of by r in expression (4.36)
rd
when the following conditions are fulfilled
max(xx0 + yy0 + zz0 ) r2 , (4.42)
and
max(x20 + y02 + z02 ) r2 . (4.43)

The second condition is fulfilled when r L. In addition, the first condition is then
also fulfilled
1

since x/r, y/r and z/r are both of order 1. If r λ0 , we can replace −k0 + by −k0 . By
rd
applying these approximations, we now obtain the following expression for the magnetic field H ~
−k0 e−k0 r
Z
~ =
H ~urd × J~e (~r0 )ek0 ~ur ·~r0 dV0 , (4.44)
4πr V0
If our observation point P is located far away from the antenna, the vector ~urd and vector ~ur will
be parallel to eachother. As a consequence, we can replace ~urd by ~ur . The unit vector ~ur does
not depend on the source point Q and can, therefore, by moved outside the integral. We then
finally obtain
−k0 e−k0 r
Z
~ =
H ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 , (4.45)
4πr V0
Expression (4.45) leads to several important conclusions:

~ has an
i. H 1
dependency.
r
ii. From e−k0 r and the time-harmonic behavior eωt , we can conclude that the waves propagate
away from the source.
~ · ~ur = 0. This implies that the radial component vanishes in the far-field region.
iii. H
iv. The magnetic field can now be expressed as

−k0 r
~ =e
H (~uθ Hθ (θ, φ) + ~uφ Hφ (θ, φ)). (4.46)
r
The components Hθ (θ, φ) and Hφ (θ, φ)) can be determined in the following way. We can
express the electric current density in the following form
J~e = Jer ~ur + Jeθ ~uθ + Jeφ ~uφ . (4.47)
We then obtain
~ur × J~e = Jeθ ~uφ − Jeφ ~uθ . (4.48)
~ can now also be written as

The field H
−k0 r
~ = −k0 e
Z
H (Jeθ ~uφ − Jeφ ~uθ )ek0 ~ur ·~r0 dV0 , (4.49)
4πr V0
This specific formulation of the far field was already used in (2.5) in order to determine the
radiation pattern of an antenna.
~ in the observation point P can be interpreted as the superposition of the contri-
v. The field H
butions from all source points. Note that the term ek0 ~ur ·~r0 corresponds to a phase difference
between source point Q (x0 , y0 , z0 ) and the origin (0, 0, 0)
The final step in the derivation of the radiated fields from a general antenna is the deter-
~ r) using expresssion (4.30)
mination of the electric field E(~
~ = 1 ~ e. (4.50)
E ∇×∇×A
ω0 µ0
~ e was already applied previously resulting in (4.36). The electric field
The vector operator ∇ × A
~ now takes the form
E
" −k0 rd #
1 1 e
Z
~ =
E ∇r × −k0 − ~rd × J~e (~r0 ) dV0 , (4.51)
4πω0 V0 rd rd2
Note that the operator ∇r is not only a function of the vector ~r0 but also of ~r due to the factor
1 e−k0 rd

~rd = ~r − ~r0 . With ψ(~r, ~r0 ) = −k0 − and ~a(~r, ~r0 ) = ~rd × J~e (~r0 ), we obtain
rd rd2
∇r × ψ~a = ∇r ψ × ~a + ψ∇r × ~a,

h i
∇r × ~a = ∇r × (~r − ~r0 ) × J~e (~r0 )
= −2J~e (~r0 ), (4.52)

!
1 e−k0 rd e−k0 rd 1

∇r ψ = −k0 − ∇r + ∇r −k0 −
rd rd2 rd2 rd
!
1 e−k0 rd e−k0 rd 1

= −k0 − ∇r + ~ur .
rd rd2 rd2 rd2 d
Furthermore,
!
e−k0 rd 1 1
∇r = −2e−k0 rd ~ur + (−k0 )e−k0 rd ~urd . (4.53)
rd2 rd3 d rd2
By substituting this expression in (4.52) we obtain
e−k0 rd 2 3 3

∇r ψ = k0 −1 − + ~urd . (4.54)
2
rd krd (k0 rd )2
~ we find that
When we apply the same kind of far-field approximation as used for H
−k02 e−k0 r
Z h i
~ =
E ~urd × ~urd × J~e (~r0 ) ek0 ~ur ·~r0 dV0 . (4.55)
ω0 4πr V0
~ in the far-field region of the

By substituting ~urd by ~ur , we finally arrive at the expression for E
antenna:
−k02 e−k0 r
Z
~ =
E ~ur × ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 . (4.56)
ω0 4πr V0
Note that the expressions of E ~ and H ~ given by (4.56) and (4.45) can also be obtained by
substituting the operator ∇× in equation (4.30) by (−k0 )~ur ×. Applying the vector identity
~ × [B
A ~ × C]
~ = (A~ · C)
~ B~ − (A
~ · B)
~ C~ to expression (4.56) we find
−k02 e−k0 r
Z h i
~ =
E (~ur · J~e (~r0 ))~ur − J~e (~r0 ) ek0 ~ur ·~r0 dV0
ω0 4πr V0 (4.57)
k02 e−k0 r
Z
= (~uθ Jeθ (~r0 )) + ~uφ Jeφ (~r0 )) ek0 ~ur ·~r0 dV0 .
ω0 4πr V0
From (4.57) we can conclude that E ~ · ~ur = 0, which implies that the radial component of E ~
~ is perpendicular to
vanishes in the far-field region of the antenna. As a result, the electric field E
~ur .
By combining expression (4.45) and (4.56) we can observe that
~
E ~ × ~ur ,
= Z0 H
(4.58)
~ = 1 ~
H ~ur × E,
Z0
q
µ0
where Z0 = 0 ≈ 377Ω is the free-space wave impedance. From (4.58) we find that the
electromagnetic waves in the far-field region behave locally as TEM waves. The Pointing vector
is now readily found from (4.58)
~p (~r) = 1 h
~∗
~ ×H
i
S Re E
2
1 h
~ ∗)
~ × (~ur × E
i
= Re E
2Z0 (4.59)
1 h
~ ∗ )~ur − (E
~ ·E ~∗
~ · ~ur )E
i
= Re (E
2Z0
1 ~ 2
= |E| ~ur ,
2Z0
~ · ~ur . As expected, the electromagnetic energy indeed flows in the radial direction.
since E
Summary
For an antenna described by an electric current distribution J~e (~r0 ) within a finite volume V0 , we
can express the corresponding electromagnetic fields in the far-field region by
−k0 e−k0 r
Z
~
H = ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 ,
4πr V0
−k02 e−k0 r
Z
~
E = ~ur × ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 ,
ω0 4πr V0 (4.60)
~
E ~ × ~ur ,
= Z0 H
µ0
r
Z0 = ≈ 377Ω.
0
4.3 Electric dipole

The most elementary antenna is the electric dipole. The electric dipole is an infinitely small
current element, which can be approximated in practice by a small linear antenna with a length
much smaller than the free-space wavelength λ0 . The electric dipole is located in the origin of our
coordinate system and directed along the z-axis, as indicated in Fig. 4.2. The current distribution
J~e along the electric dipole can now be expressed in terms of the delta function:
J~e = I0 lδ(x)δ(y)δ(z)~uz = I0 lδ(~r)~uz . (4.61)
Where I0 is the total current which we assume to be constant along the electric dipole. The length
of the antenna is l. The delta-function δ(z) belongs to the class of generalized functions with the
property:
Z∞
γ(z0 ) = δ(z − z0 )γ(z)dz. (4.62)
−∞
The delta-function can be visualized as follows


 1
− w2 < z < w




w 2
δ(z) = , (4.63)


w
 0 |z| >


2
where w needs to be very small. In order to determine whether (4.61) indeed represents an electric
dipole, we need to investigate the corresponding charge distribution ρe which can be found by
applying the continuity equation (4.2):
1 −I0 l
ρe (~r) = − ∇ · J~e (~r) = δ(x)δ(y)δ 0 (z). (4.64)
ω ω
By considering the derivative of (4.63), δ 0 (z), we can represent(4.64) by a set of two point charges
with opposite sign and located on the z-axis at z = w/2 and −w/2, respectively, since
1 w w

δ 0 (z) = δ z+ −δ z− . (4.65)
w 2 2
Let us now determine the electromagnetic fields in the far-field region of the electric dipole
by using the general expressions derived in chapter 4.2:
−k0 e−k0 r
Z
~ =
H ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 ,
4πr V0 (4.66)
~
E ~ × ~ur ,
= Z0 H
Substituting the current distribution of the electric dipole (4.61) in the above equation and using
the fact that
~ur × ~uz = −~uφ sin θ (4.67)

4.3. ELECTRIC DIPOLE 89
leads us to the magnetic field far away from the dipole

−k0 r
~ = Hφ ~uφ = k0 I0 le
H sin θ~uφ . (4.68)
4πr
The corresponding electric field in the far-field region is now given by
−k0 r
~ = Eθ ~uθ = k0 Z0 I0 le
E sin θ~uθ , (4.69)
4πr
where we have used the following relation
~uφ × ~ur = ~uθ . (4.70)
The time-average power flux density (Pointing vector) expressed in [W/m2 ] is found by using
equation (4.59), which provides
~p (~r) = 1 ~ 2
S |E| ~ur
2Z0 (4.71)
1 k02 Z0 (I0 l)2
= sin2 θ~ur .
2 (4πr)2
From (4.71) we can conclude that the electric dipole does not radiate uniformly in all directions.
In the direction along the z-axis (θ = 0) the dipole does not radiate al all. The power density
~p (~r) decreases as r−2 with increasing distance from the dipole. However, since the total power
S
radiates through a sphere with radius r and surface 4πr2 , the total radiated power is independent
of r. The total radiated power by the electric dipole is then found by
Z Z
Pt = ~p (~r) · ~ur r2 dΩ
S
Z2πZπ
1 k02 Z0 (I0 l)2
= sin3 θdθdφ (4.72)
2 (4π)2
0 0
k02 Z0 (I0 l)2
= ,
12π
where we have used the fact that
Zπ
4
sin3 θdθ = . (4.73)
3
0
As expected, the total power Pt does not depend on r.

Let us now explore some fundamental antenna parameters of the electric dipole. According
to (2.10), the normalized radiation pattern is given by
P (θ, φ) P (θ)
F (θ, φ) = = = sin2 θ. (4.74)
P (π/2, 0) P (π/2)
This pattern is plotted in Fig. 4.3. The 3-dB beam width of the electric dipole is 900 . The
directivity is the maximum of the directivity function, given in (2.11) by:
P (θ, φ) 3
D(θ, φ) = = sin2 θ, (4.75)
Pt /4π 2
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.3: Radiation pattern versus θ of an electric dipole of length l. Note that θ = 00 coincides
with the positive z-axis
3
which implies that the directivity is D = . Finally, let us determine the input impedance of the
2
electric dipole. The input impedance of the ideal electric dipole is purely resistive, so Za = Ra .
Ra , also known as the radiation resistance, can be found by using relation (2.16). Combining this
with (4.72) provides:
2
Pt k02 Z0 l2 l

Ra = 1 2
= 2 = 80π 2 . (4.76)
2 |I0 |
12π λ0
The length l of the dipole is small as compared to the wavelength. For example, let l = 0.01λ0 ,
which results in an input impedance of Ra = 0.08Ω. This imples that the electric dipole has a very
low radiation resistance which will makes it difficult to match to a transmission line since most
commonly used transmission lines have a much higher characteristic impedance (often Z0t = 50Ω).
This will result in a poorly matched antenna and a high corresponding reflection coefficient (2.19)
at the antenna port. Electric dipoles are only used in very specific applications in which power-
matching is not important. A nice example is low-frequency radio-astronomy [24].
The half-wavelength dipole has much better radiation and matching properties. This type
of antenna will be explored in the next section.
4.4 Thin-wire antennas

In the previous section we found that the radiation resistance of an electric dipole is very small,
which makes it difficult to match to a transmission line. The matching properties can be improved
by increasing the length of the dipole, resulting in a so-called linear wire antenna. This antenna
4.4. THIN-WIRE ANTENNAS 91
consists of a cylindrical metal wire with diameter d0 and length 2l, which has a small gap in the
center of the wire to accommodate a balanced feed using a so-called two-wire transmission line,
also known as twin lead or Lecher line. Furthermore, we will assume that the diameter d0 of the
wire very small compared to the wavelength d0 λ0 . Fig. 4.4 illustrates the linear wire antenna
connected to a transmission line.
Je
Generator
y
2l
Figure 4.4: Thin (lineair) wire antenna of length 2l connected to a two-wire transmission line.
The center of the wire is located in the origin of the coordinate system. The wire antenna is
directed along the z-axis.
The current distribution along the two-wire transmission line will be sinusoidal, when we
neglect radiation and other losses. The wire antenna can now be interpreted as the last part of an
open-ended two-wire transmission line. The radiation from this folded part of the transmission
line is much larger as compared to the spurious radiation from the transmission line itself. This
is, of course, our aim, since we want to construct an efficient antenna. Nevertheless, the current
distribution along the wire will remain the same (first-order approximation) as the un-folded open-
ended transmission line, so with the sinusoidal current distribution. The current distribution at
the end point of the antenna must be equal to zero. Furthermore, we will assume that the gap
at the center of the wire antenna is negligibly small. With all these assumptions, the current
distribution J~e along the wire takes the following form:
J~e = I0 δ(x)δ(y) sin [k0 (l − |z|)] ~uz , (4.77)
for −l < z < l. Lateron in this book we will use numerical methods to determine the exact
current distribution along the wire. It will be shown that (4.77) is an accurate approximation of
the current along the wire antenna when the diameter of the wire is smaller than λ0 /100. Some
examples of the current distribution for various lengths of the wire antenna are illustrated in
Fig. 4.5. The electric field in the far-field region of the wire antenna can be determined using
λ0/2 3λ0/4 λ0 3λ0/2 2λ0
Figure 4.5: Current distribution along a thin wire antenna for various lengths according to (4.77).
expression (4.60):
−k02 e−k0 r
Z
~ =
E ~ur × ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 , (4.78)
ω0 4πr V0
Since the current is directed along the z-axis we can rewrite ek0 ~ur ·~r0 into the expression ek0 z0 cos θ .
Furthermore,
~ur × ~ur × ~uz = ~uθ sin θ. (4.79)
As a result, we can conclude that the electric field E ~ only has one component, being Eθ . The
radiated field is linearly polarized and takes the following form
Zl
−k02 I0 e−k0 r
Eθ = sin θ sin [k0 (l − |z0 |)] ek0 z0 cos θ dz0 , (4.80)
ω0 4πr
−l
As a first step towards a closed-form solution of (4.80), we will determine the integral by
dividing it into two parts:

Zl
Iz0 = sin [k0 (l − |z0 |)] ek0 z0 cos θ dz0
−l
Zl Z0
= sin [k0 (l − z0 )] ek0 z0 cos θ dz0 + sin [k0 (l + z0 )] ek0 z0 cos θ dz0
0 −l
Zl Zl
= sin [k0 (l − z0 )] ek0 z0 cos θ dz0 + sin [k0 (l − z0 )] e−k0 z0 cos θ dz0
0 0 (4.81)
Zl
= 2 sin [k0 (l − z0 )] cos(k0 z0 cos θ)dz0
0
Zl Zl
= sin(k0 l − k0 z0 + k0 z0 cos θ)dz0 + sin(k0 l − k0 z0 − k0 z0 cos θ)dz0
0 0
2 [cos(k0 l cos θ) − cos(k0 l)]
= .
k0 sin2 θ
Substitution of (4.81) in (4.80) provides:
e−k0 r cos(k0 l cos θ) − cos(k0 l)

Eθ = Z0 I0 . (4.82)
2πr sin θ
Note that Eθ does not depend on the φ coordinate, which could have been expected as a result of
the symmetry. The corresponding magnetic field is now easily found, since in the far-field region
we have
~ = 1 ~
H ~ur × E
Z0
1 (4.83)
= ~ur × Eθ ~uθ
Z0
1
= ~uφ Eθ .
Z0
As a result, only the Hφ component is non zero and is given by
1
Hφ = Eθ
Z0 (4.84)
I0 e−k0 r cos(k0 l cos θ) − cos(k0 l)

= .
2πr sin θ
Note that also Hφ does not depend on the φ coordinate. The Pointing vector (power density in
[W/m2 ]) is found by using the expression (4.59):
~p (~r) = 1 ~ 2
S |E| ~ur
2Z0 (4.85)
2
Z0 |I0 |2 cos(k0 l cos θ) − cos(k0 l)

= ~ur .
8π 2 r2 sin θ
We will now determine the normalized radiation pattern for several wire antennas with
varying length. The radiation pattern is found by using expression (2.10).
i. λ0 /2 wire antenna.
When 2l = λ0 /2 we find that Eθ is equal to
e−k0 r cos((π/2) cos θ)

Eθ = Z0 I0 . (4.86)
2πr sin θ
The radiation pattern of the λ0 /2 dipole antenna is shown in Fig. 4.6. Note that the radiation
pattern is quite similar to the radiation pattern of the electric dipole, which is also shown
in the figure. The 3 dB beam width of the λ0 /2 dipole is 780 , while the beam width of an
electric dipole is 900 .
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.6: Radiation pattern versus θ of an λ0 /2 wire antenna (full line) and of an electric dipole
(dotted line).
ii. λ0 wire antenna.

Now 2l = λ0 and Eθ is given by
e−k0 r cos(π cos θ) + 1

Eθ = Z0 I0 . (4.87)
2πr sin θ
The corresponding radiation pattern of the λ0 antenna is shown in Fig. 4.7. The 3 dB beam
width is 470 .
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.7: Radiation pattern versus θ of an λ0 wire antenna (full line) and of an electric dipole
(dotted line).
iii. 3λ0 /2 wire antenna.

With 2l = 3λ0 /2 we find that Eθ takes the form:
e−k0 r cos((3π/2) cos θ)

Eθ = Z0 I0 . (4.88)
2πr sin θ
The corresponding radiation pattern of the 3λ0 /2 antenna is shown in Fig. 4.8. We observe
multiple beams in various directions. In addition, the antenna does not radiate energy in
several other directions (pattern nulls). Although the beam width is much smaller now, the
antenna generates multiple beams, which might not be very useful in most applications.
In order to determine the other antenna parameters of the wire antenna, such as the directivity
and radiation resistance, we first need to calculate the total radiated power Pt by integrating the
power density over a sphere according to (2.17)
Z Z
Pt = ~p (~r) · ~ur r2 dΩ
S
Zπ 2
Z0 |I0 |2 cos(k0 l cos θ) − cos(k0 l)
= 2π sin θdθ (4.89)
8π 2 sin θ
0
Z0 |I0|2
= Ip .
4π
A general analytic solution of the integral Ip is not easily found. However, in case of a λ0 /2-wire
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.8: Radiation pattern versus θ of an 3λ0 wire antenna (full line) and of an electric dipole
(dotted line).
antenna we can express the integral Ip in terms of the cosine integral. By applying the coordinate
transformation u = cos θ and dθ = −du/ sin θ we can write integral Ip in the following form
Zπ 2
cos(k0 l cos θ) − cos(k0 l)
Ip = sin θdθ
sin θ
0
Z+1" #
cos2 (πu/2)
= du
1 − u2
−1
Z+1" # Z+1" #
1 cos2 (πu/2) 1 cos2 (πu/2)
= du + du
2 1+u 2 1−u
−1 −1
Z2π" # Z2π" #
1 cos2 ((t − π)/2) 1 cos2 ((π − s)/2)
= dt + ds (4.90)
2 t 2 s
0 0
Z2π" #
cos2 ((t − π)/2)
= dt
t
0
Z2π
1 1 + cos(t − π)

= dt
2 t
0
Z2π
1 1 − cos t

= dt,
2 t
0
where we have used the substitutions 1+u = t/π and 1−u = s/π. Furthermore, we have used the
fact that cos2 (x/2) = (1 + cos x)/2. The integral can now be expressed in terms of the well-known
cosine integral ci(x)
Z∞
cos t
ci(x) = − dt. (4.91)
t
x
The cosine integral is well tabulated (for example in the famous book of I. S. Gradsteyn and I. M.
Ryzhik, “Table of integrals, series and products”, tabel (8.230), [26]) and is a standard function
in Matlab (function cosint). Expression (4.90) now takes the form
Z2π
1 1 − cos t

Ip = dt
2 t
0
1 (4.92)
= (C + ln(2π) − ci(2π))
2
2.437
= ,
2
where C = 0.577 is Euler’s constant. The total radiation power is now found from:
Z0 |I0 |2
Pt = Ip
4π (4.93)
1
= 73.1|I0 |2 .
2
By using definition (2.16) of the radiation resistance, we easily find that
Ra = 73.1Ω. (4.94)
This suggest that Ra is not frequency dependent. However, this is not true, since we have assumed
that the antenna length 2l = λ0 /2. Therefore, this antenna is a resonating structure, which implies
that the input impedance is strongly frequency dependent. The resonant wire antenna provides
a purely resistive load Ra seen at the center of the wire antenna (see Fig. 4.4). In general, the
load impedance at the termination of the two-wire transmission line will take a complex form
according to Za = Ra + Xa . Only in case of a resonating wire antennas with 2l = λ0 /2 and a
small diameter d0 λ0 , we find a purely resistive load impedance with Xa = 0 . The calculation
of the complex input impedance Za = Ra + Xa of a wire antenna outside resonance is more
complex and can only be done by using numerical techniques such as the Method of Moments. In
chapter 5 we will investigate this numerical method in more detail.
Finally, we can determine the directivity of the λ0 /2 wire antenna. By using the definition
of the directivity function (2.11), we obtain
Z0 |I0 |2
max(P (θ, φ)) 8π 2
D= = = 1.64. (4.95)
Pt /4π 73.1|I0 |2 /2
4π
The corresponding directivity expressed in Decibels (10log10 D) is DdB = 2.15 dB, which is slightly
higher than the directivity of the electric dipole D = 1.5 or DdB = 1.76 dB.
4.5 Wire antenna above a ground plane

Up to now we have assumed that the antenna is located in free space. However, in practise the
antenna will often be placed close to objects, e.g. earth surface or in case of printed antennas
close to a ground plane of the printed-circuit board (PCB). As a result, the radiation properties
and input impedance will change. In this section we will investigate the effect of a perfectly
conducting ground plane on the radiation properties of wire antennas. We will assume that the
horizontally-oriented resonating wire antenna of length 2l = λ0 /2 is located above the conducting
ground plane at a height h. The configuration is shown in Fig. 4.9. In order to solve the antenna
Je wire antenna
h
Ground plane
x
h
Je,image image antenna
Figure 4.9: Thin horizontal wire antenna of length 2l = λ0 /2 above a perfectly conducting infinite
ground plane. The antenna is situated along the x-axis (y = 0).
problem of Fig. 4.9, we can use the so-called image method. This method is based on the fact
that the tangential component of the electric field E~ vanishes at the perfectly-conducting ground
plane. Exactly the same effect can be obtained by removing the conducting plate and placing a
second horizontal wire antenna (image antenna) at a distance 2h from the original wire antenna,
as illustrated in Fig. 4.9. The current distribution on the image antenna is 1800 out of phase with
respect to the current distribution on the original horizontal wire antenna. In this way, we have
defined an equivalent model of the original wire antenna above a ground plane. The equivalent
problem can be easily solved by applying the superposition principle and by using the expressions
found in section 4.4. Note that we have assumed that the current distribution of the resonant
wire antenna is still purely sinusoidal and not affected by the presence of the ground plane.
It is now interesting to investigate the radiation pattern of the horizontal wire antenna
4.6. FOLDED DIPOLE 99
above a ground plane in more detail in the φ = 900 -plane. Note that we had assumed that the
antenna is located along the x-axis. Based on the symmetry in the y − z plane, the radiation
pattern of a wire antenna in free-space would have been independent of the angle θ. The ground
plane makes the antenna θ-dependent. Now let E ~ o be the radiated electric field from a wire
antenna in free-space and E~ i the corresponding radiated electric field of the image antenna. By
applying the superposition principle we obtain the total field in the far-field region as:
~ tot = E
E ~o + E
~i
h i
~ f ree ek0 h cos θ − e−k0 h cos θ
= E (4.96)
~ f ree 2 sin(k0 h cos θ),

= E
where the vector E ~ f ree represents the electric field of a horizontal wire antenna in free space and
located at the origin of the coordinate system. The factor 2 sin(k0 h cos θ) describes the effect of
the ground plane. The corresponding radiation pattern, normalized to an horizontal wire antenna
in free-space (so ignoring the factor E ~ f ree in (4.96)) is now given for the φ = 900 plane by:
P (θ, φ = 900 )
Fr (θ) =
Pf ree (θ, φ = 900 ) (4.97)
2
= 4 sin (k0 h cos θ).
The antenna does not radiate energy in the horizontal plane with θ = 900 . Fig. 4.10 shows the
radiation pattern versus the height h above the ground plane. The directivity of a horizontal
wire antenna above a ground plane is larger as compared to a wire antenna in free space. This is
directly related to the fact that the antenna does not radiate energy in the half-space below the
ground plane. The directivity depends on the height h and can be up to a factor 4 (6 dB) higher
as compared to a free-space wire antenna. The beam width is also strongly dependent on the
height h (See Fig.4.10). In case h = 0.1λ0 , the beam width in the φ = 900 -plane is θHP = 940 .
4.6 Folded dipole

We have previously shown that a basic λ0 /2 wire antenna has an input impedance at resonance
of approx. 73Ω. A two-wire transmission line (Lecherline or twin-lead) has a somewhat higher
characteristic impedance, for example Z0t = 300Ω. As a result a matching circuit would be required
to match the antenna to the transmission line, resulting in additional losses and bandwidth
limitations. By modifying the basic λ0 /2 wire antenna to the structure of Fig. 4.11 we can realize
a much higher input impedance. This antenna structure is called folded dipole. It can be shown
that the input impedance of the λ0 /2 folded dipole is about 4 times higher as compared to the
normal λ0 /2 wire antenna, so Zin ≈ 4 · 73 = 292Ω. The radiation properties of the folded λ0 /2
dipole are quite similar to the characteristics of a normal λ0 /2 dipole.
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.10: Normalized radiation pattern versus θ in the φ = 900 plane of a horizontal wire
antenna above a perfectly-conducting infinite metal ground plane. A linear scale is used. With
h = 0.1λ0 full line, h = 0.5λ0 dotted line and h = λ0 line with dotts.
λ0/2
Two-wire transmission line
Figure 4.11: Folded λ0 /2 dipole.
4.7 Loop antennas

The loop antenna consists of a circular thin metal wire with radius a. In this section we will
assume that the current I0 along the loop is constant, which is only a valid approximation when
4.7. LOOP ANTENNAS 101
the radius is very small as compared to the wavelength, i.e a λ0 . For larger loop antennas,
this can only be realized by using additional phase-shifters in the loop. Figure 4.12 shows the
geometry of the loop antenna. The electric current distribution J~e can now be represented by:
z
P
I0
a y
φ
I0
Figure 4.12: Loop antenna consisting of a circular thin metal wire with radius a carrying a constant
current I0 . The antenna is situated in the x − y-plane.
q
J~e (~r0 ) = I0 δ x20 + y02 − a δ(z0 )~uφ0 . (4.98)
In chapter 4.2 we found that the far field can be determined with:
−k0 e−k0 r
Z
~ =
H ~ur × J~e (~r0 )ek0 ~ur ·~r0 dV0 ,
4πr V0 (4.99)
~
E ~ × ~ur .
= Z0 H
The integration for the loop antenna is carried out over a circle with radius r0 = a. We will use
polar coordinates to facilitate the integration. Furthermore
~ur · ~r0 = [cos φ sin θ~ux + sin φ sin θ~uy + cos θ~uz ] · [a cos φ0 ~ux + a sin φ0 ~uy ]
(4.100)
= a cos(φ − φ0 ) sin θ.
Substituting this in (4.99) provides
Z2π
~ = −k0 e−k0 r
H (I0 a)~ur × ~uφ0 ek0 a cos(φ−φ0 ) sin θ dφ0 . (4.101)
4πr
0
Due to the symmetry of the geometry, we can assume that the fields will not depend on the φ
coordinate. Therefore we can use φ = 0 in (4.101). In addition, ~uφ0 = −~ux sin φ0 + ~uy cos φ0 . In
the φ = 0 plane it is found that ~uφ = ~uy . We can now divide the integral in (4.101) into two parts
according to
Z2π
I~φ0 = ~uφ0 ek0 a cos φ0 sin θ dφ0
0
Z2π Z2π
k0 a cos φ0 sin θ
= −~ux sin φ0 e dφ0 + ~uy cos φ0 ek0 a cos φ0 sin θ dφ0 (4.102)
0 0
Z2π
= ~uy cos φ0 ek0 a cos φ0 sin θ dφ0 .
0
The first integral is equal to zero, since the integrand is an odd function with respect to φ0 . The
remaining integral can be analytically solved:
Z2π
I~φ0 = ~uy cos φ0 ek0 a cos φ0 sin θ dφ0
0
Zπ
= 2~uy cos φ0 ek0 a cos φ0 sin θ dφ0
0
π/2
Z
= −2~uy sin φ0 e−k0 a sin φ0 sin θ dφ0
−π/2
π/2
Z (4.103)
= 2~uy sin φ0 ek0 a sin φ0 sin θ − e−k0 a sin φ0 sin θ dφ0
0
π/2
Z
= 4~uy sin φ0 sin(k0 a sin φ0 sin θ)dφ0
0
= 2πJ1 (k0 a sin θ)~uy
= 2πJ1 (k0 a sin θ)~uφ
where J1 (z) is the Bessel function of the first kind with order 1 and argument z. Bessel functions
are well tabulated (e.g. [26]) and are available in Matlab and other mathematical tools. Finally,
we obtain the following expression for the magnetic field in the far-field region using ~ur ×~uφ = −~uθ :
−k0 r
~ = Hθ ~uθ = − k0 aI0 e
H J1 (k0 a sin θ). (4.104)
2r
4.7. LOOP ANTENNAS 103
Since ~uθ × ~ur = −~uφ , we can conclude that the corresponding electric field will only have a φ
component:
−k0 r
~ = Eφ ~uφ = k0 aZ0 I0 e
E J1 (k0 a sin θ). (4.105)
2r
4.7.1 Small loop antennas: magnetic dipole

A small loop antenna, also known as magnetic dipole, consists of a circular metal wire with radius
a λ0 and with a constant current I0 along the loop. We can use expression (4.104) and (4.105)
to determine the far fields, where we can approximate the Bessel function J1 (z) for small argument
z according to
z
J1 (z) ≈ . (4.106)
2
By substituting this in (4.104) and (4.105), we find that the far fields are given by:
k02 (πa2 I0 )e−k0 r

Hθ = − sin θ
4πr (4.107)
k02 Z0 (πa2 I0 )e−k0 r
Eφ = sin θ.
4πr
The factor πa2 I0 is also known as the ”transmitter-moment” and is the product of the current I0
and the surface of the loop. Usually it is denoted by m with
m = πa2 I0 . (4.108)
There is a strong analogy between the fields radiated by a magnetic dipole according to (4.107)
and the fields from the electric dipole, given by (4.68) and (4.69). The radiation patterns of the
magnetic and electric dipole are exactly the same. Now let us assume that both an electric as well
as an magnetic dipole are located in the origin of our coordinate system. Furthermore, assume
that the electric dipole is oriented along the z-axis and the magnetic dipole is situated in the
x-y-plane. When we choose the dimensions of both dipoles and current to fulfill the relation
p = −mk0 , (4.109)
the corresponding electromagnetic field generated by these two sources will fullfill the following
relation
~ = Z0 H.
E ~ (4.110)
The minus sign in (4.109) indicates that the currents in the electric and magnetic dipoles are 1800
out of phase. When the currents are in phase, i.e. p = mk0 , we find the following form of the
radiated electromagnetic field:
~ = −Z0 H.
E ~ (4.111)
Electromagnetic fields satisfying (4.110) or (4.111) have the unique property that the radiated far
fields are circularly polarized in all directions. Using (4.110) we find that
Eφ = Eθ , (4.112)
for all directions (θ, φ). Therefore, we satisfy the relation between Eθ and Eφ as described in
section 2.9 for a perfectly left-handed circularly-polarized field. Fields that satisfy (4.111) are
right-handed circularly polarized for all directions (θ, φ).
Exercise
Calculate the directivity of a magnetic dipole.
4.7.2 Large loop antennes

The electromagnetic fields in the far-field region of a loop antenna with uniform current distribu-
tion along the loop was already determined in (4.104) and (4.105). We can use these expressions
to determine the radiation pattern, resulting in
P (θ)
F (θ) =
max(P (θ))
(4.113)
|J1 (k0 a sin θ)|2
= .
max(|J1 (k0 a sin θ)|2 )
The shape of the pattern will be similar to the behavior of the Bessel function J1 (z). Fig. 4.13
shows the normalized radiation pattern of a loop antenna with radius a = λ0 /20, a = λ0 /2,
a = 2λ0 , respectively. Note that the direction of maximum radiation moves towards θ = 0 with
increasing loop radius (check Fig. 4.12) .
4.8 Magnetic sources

In order to be able to introduce aperture antennas, it is useful to extend Maxwell’s equations,
given by (4.1). We will do this by introducing magnetic currents and associated magnetic charge
distributions. Note that both magnetic currents as well as magnetic charges do not exist in the
real-world: both are non-physical quantities.. However, it will help us to introduce several new
electromagnetic concepts in an elegant way. For that purpose, we will introduce the magnetic
current distribution J~m (~r) and the magnetic charge density ρm (~r). When we assume that there
are no electric currents J~e (~r) = ~0 and electric charges ρe (~r) = 0, Maxwell’s equations take the
following form in the space-frequency domain:
~ r)
∇ × E(~ ~ r) − J~m (~r),
= −ωµ0 H(~
~ r) = ω0 E(~
∇ × H(~ ~ r),
(4.114)
~ r) ρm (~r)
∇ · H(~ = ,
µ0
~ r)
∇ · E(~ = 0.
4.8. MAGNETIC SOURCES 105
90
1
120 60
0.8
0.6
150 30
0.4
0.2
180 0
210 330
240 300
270
Figure 4.13: Normalized radiation pattern versus θ of a loop antenna with uniform current distri-
bution for various loop dimensions. With a = λ0 /20 full line, a = λ0 /2 dotted line and a = 2λ0
line with dots.
By combining the first and third equation from (4.114) we observe that the following continuity
equation yields:
∇ · J~m (~r) + ωρm (~r) = 0. (4.115)
The solution of these equations related to magnetic sources can be found by using a similar
approach as used in section 4.1 for the case of electric sources. As a first step, we will introduce a
~ m , using the fourth equation in (4.114).
vector potential, in this case the electric vector potential A
As a result, we find that
~ =−1∇×A
E ~ m. (4.116)
0
~ = −ω∇ × A
This gives ∇ × H ~ m . We can now write H
~ in the form:
~ = −ω A
H ~ m − ∇φm , (4.117)
where φm is (yet) an arbitrarily scalar function. By substituting this relation into the first equation
of (4.114) and using the Lorenz-gauge according to
~ m = −ω0 µ0 φm ,
∇·A (4.118)
~ m , given by
leads towards the well-known Helmholtz equation for the vector potential A
~ m + k02 A
∇2 A ~ m = −0 J~m . (4.119)
A similar expression can be found for the scalar potental φm directly from the Lorenz-gauge that
we used. In section 4.1 we solved Helmholtz equation for an arbitrarily electric current distribution
J~e by first looking for the solution for an elementary point source. The general solution was then
found by applying the superposition concept, since any electric current distribution J~e can be
written as a continuous superposition of point sources. By applying the same strategy here, we
find that the electric vector potential A ~ m due to an arbitrarily magnetic current distribution J~m
takes the following form:
−k0 |~
r−~
r0 |
~ m (~r) = 0 e
Z
A J~m (~r0 ) dV0 , (4.120)
4π V0 |~r − ~r0 |
where we have assumed that the observation point P , indicated by the vector ~r, is located outside
the source region V0 . The electromagnetic fields outside the source region V0 are now found by
~ 1 ~ m,
E = − ∇×A
0 (4.121)
~ = 1 ~ m.
H ∇×∇×A
ω0 µ0
Similar to section 4.2, we can determine the electromagnetic fields far away from the sources
(far-field region) by using several approximations. The fields in the far-field region than take the
following form:
k0 e−k0 r
Z
~
E = ~ur × J~m (~r0 )ek0 ~ur ·~r0 dV0 ,
4πr V0
−k02 e−k0 r
Z
~ =
H ~ur × ~ur × J~m (~r0 )ek0 ~ur ·~r0 dV0 , (4.122)
ωµ0 4πr V0
~
E ~ × ~ur ,
= Z0 H
We again can conclude that the fields in the far-field region locally behave as TEM-waves (E-
field perpendicular to H and perpendicular to the propagation direction ~ur ). The strong analogy
between magnetic and electric sources is known in literature as the duality concept. Always keep
in mind that magnetic currents and charges are mathematical quantities that cannot exist in
nature.
We have now introduced all tools to determine the radiated electromagnetic fields from a
general source, consisting of both electric and (fictive) magnetic currents and charges. We can do
this by using expressions (4.60) and (4.122) and by applying the superposition principle.
Exercise
In the origin of our coordinate system a magnetic current distribution exist with the following
properties:
J~m (~r0 ) = I0m lδ(x)δ(y)δ(z)~uz , (4.123)
where I0m is the total magnetic current. Show that the radiated electromagnetic fields due to this
source are identical to the fields due to the magnetic dipole (small loop antenna) as discussed in
section 4.7.1.
4.9. LORENTZ-LARMOR THEOREM 107
4.9 Lorentz-Larmor theorem

In order to analyze aperture antenna, it is convenient to introduce the Lorentz-Larmor theorem.
This concept defines the electromagnetic relation between two situations A and B. Later in this
section we will show that we can apply this theorem to replace a particular antenna problem,
represented by a source region V a with electric and magnetic (volume-) current distributions J~ea
and J~ma , by an equivalent and simpler antenna problem. In literature this concept is also known
as the Equivalence principle [27]. Figure 4.14 illustrates source region V a bounded by the surface
S a . The sources J~ea and J~m ~ a and H
a generate an electromagnetic field E ~ a outside source region V a .
It will be shown that the same fields outside V a can also be generated by an equivalent electric
surface-current distribution J~esa on S a and an equivalent magnetic surface-current distribution
J~ms
a on S a where
J~es
a ~ a,
= ~un × H
(4.124)
J~ms
a ~a
= E × ~un ,
where ~un is the normal vector on surface S a . The equivalent sources (4.124) produce within V a an
zero electromagnetic field and outside V a an electromagnetic field equal to the original problem.
un
S
V
 
J ea J ma
Va Sb
 
Sa J be J bm
Vb
Figure 4.14: Source region V a enclosed by the surface S a and source region V b enclosed by surface
S b . Both regions are located within volume V with corresponding surface S. Within source region
V a both electric and magnetic current distributions J~ea and J~m
a exist and the corresponding charge
a a
distributions ρe and ρm . Similar for source region V . b
Intermezzo: derivation of Lorentz-Larmor theorem

The derivation of the Lorentz-Larmor theorem provides interesting insight into the relation
and electromagnetic interaction between two source regions. As a first step, we will formulate
~ a and H
Maxwell’s equations in the space-frequency domain for situation A where the fields E ~a
are generated only by the sources in volume V a (see also Fig. 4.14):
~a
∇×E ~ a − J~m
= −ωµ0 H a
,
~ a = ω0 E
∇×H ~ a + J~ea ,
(4.125)
~ a ρam
∇·H = ,
µ0
~a ρae
∇·E = .
0
In situation B, a source distribution J~eb and J~m

b exist within the volume V b together with the
corresponding charge distributions ρbe and ρbm . The fields generated by these sources (so without
the situation A sources) comply to the following Maxwell equations :
~b
∇×E ~ b − J~m
= −ωµ0 H b
,
~ b = ω0 E
∇×H ~ b + J~b ,
e
(4.126)
~b ρbm
∇·H = ,
µ0
~b ρbe
∇·E = .
0
~a × H
The divergence of the vector function E ~b−E
~b × H
~ a is:
h i h i h i
~a × H
∇· E ~b−E
~b × H ~a × H
~a =∇· E ~b −∇· E
~b × H
~a . (4.127)
The first term of the right-hand side of equation (4.127) can be written as
h i
~a × H
∇· E ~b ~b·∇×E
= H ~a − E
~a · ∇ × H
~b

~ b · −ωµ0 H
= H ~ a − J~a − E
~ a · ω0 E
~ b + J~b (4.128)
m e
~b·H
= −ωµ0 H ~a−H
~ b · J~a − ω0 E
~a · E
~b − E
~ a · J~b ,
m e
~ × B)
where we have used the vector identity ∇ · (A ~ =B ~ ·∇×A ~ −A~ · ∇ × B.
~ In a similar way, we
can determine the second term in the right-hand side of equation (4.127) by exchange of a and b
in (4.128):
h i
~b × H
∇· E ~a ~a·H
= −ωµ0 H ~b−H
~ a · J~b − ω0 E
~b · E
~a − E
~ b · J~a . (4.129)
m e
Combining (4.128) and (4.129) provides the following result

h i h i h i
~a × H
∇· E ~b−E
~b × H
~a = E
~ b · J~ea − H
~ b · J~m
a ~ a · J~eb − H
− E ~ a · J~m
b
. (4.130)
Applying Gauss theorem according to

Z Z
~
∇ · AdV = ~ · ~un dS
A (4.131)
V S
on the volume V enclosed by surface S (see Fig. 4.14) provides

Z h i Z h i h i
~a × H
E ~b−E
~b × H
~ a · ~un dS = ~ b · J~ea − H
E ~ b · J~m
a ~ a · J~eb − H
− E ~ a · J~m
b
dV. (4.132)
S V
This relation is known as Lorentz’ theorem. In order to derive the Lorentz-Larmor theorem, we
will apply Lorentz’ theorem on several special cases.
i. Volume V encloses the entire free-space V∞ .

The corresponding surface S is now a sphere S∞ with infinite large radius. The sources
~ a, H
(E ~ a ) and (E
~ b, H
~ b ) are located within a finite region. At large distances from these
sources we have, according to (4.60) and (4.122), the following relations:
~a = 1 ~ a,
H ~ur × E
Z0 (4.133)
~b 1 ~ b.
H = ~ur × E
Z0
Furthermore, for a point with location vector ~r on the sphere S∞ we know that ~ur = ~un .
Therefore, the integrand of the left-hand side of expression (4.132) can be written in the
following form
h i h i h i
~a × H
E ~b−E
~b × H
~ a · ~un = ~a · H
~un × E ~ b − ~un × E
~b · H
~a
h i h i
= ~a · H
~ur × E ~ b − ~ur × E
~b · H
~a
(4.134)
~a·H
= Z0 H ~ b − Z0 H
~b·H
~a
= 0
The surface integral in (4.132) will vanish when S = S∞ . Therefore, expression (4.132) can
be reduced to
Z h i Z h i
~ b · J~ea − H
E ~ b · J~m
a
dV = ~ a · J~eb − H
E ~ a · J~m
b
dV. (4.135)
V∞ V∞
Although the integration in (4.135) is performed over the entire volume V∞ , it is evident that
the integration can be limited to the (finite) source volume.
S
V ub

J ea (r0 ) Q r - r0
P

J ma (r0 )
r0 r
Figure 4.15: Source distribution B is an electric dipole directed towards ~ub with dipole moment
p = 1.
ii. Source distribution B is an electric dipole.

Assume that our source distribution is an electric dipole with dipole moment p = 1 and
direction ~ub located at ~r outside the source regions. This is illustrated in Fig. 4.15. We now
obtain
J~m
b
= 0,
(4.136)
J~eb = ~ub δ(~r − ~r0 ),
where the vector ~r is fixed and ~r0 variable. When using this relation in (4.135) we find that
Z Z h i
~ a (~r0 )δ(~r − ~r0 )dV0 =
~ub · E ~ b (~r0 ) · J~ea (~r0 ) − H
E ~ b (~r0 ) · J~m
a
(~r0 ) dV0 , (4.137)
V∞ V a
or
Z h i
~ a (~r) =
~ub · E ~ b (~r0 ) · J~a (~r0 ) − H
E ~ b (~r0 ) · J~a (~r0 ) dV0 . (4.138)
e m
Va
In the special case that we only have electric sources within V a , expression (4.138) takes the
form:
Z
~ a (~r) =
~ub · E ~ b (~r0 ) · J~ea (~r0 )dV0 .
E (4.139)
V a
It appears that the electric field E ~ a (~r) due to source distribution A can be found by deter-
mining the electromagnetic fields of an electric dipole with p = 1 located at ~r with direction
~ub and substituting it in the volume integral. By choosing the direction of the dipole ~ub to be
~ux , ~uy and ~uz , respectively, we can determine the Cartesian components of E ~ a (~r). In section
4.1 we have determined the field generated by an electric dipole. Using relation (4.22) and
(4.30), we can rewrite (4.139) in the following form
1
Z
~ a (~r) =
~ub · E ~ e (~r, ~r0 )dV0 ,
J~ea (~r0 ) · ∇r0 × ∇r0 × A (4.140)
ω0 µ0
Va
with
−k0 |~
r−~r0 |
~ e (~r) = µ0 ~ub e
A . (4.141)
4π |~r − ~r0 |
Now introduce the variable ϕ according to
1 e−k0 |~r−~r0 |
ϕ(~r, ~r0 ) = . (4.142)
4π |~r − ~r0 |
We can rewrite (4.140) into:
1
Z
~ a (~r) =
~ub · E J~ea (~r0 ) · ∇r0 × ∇r0 × ~ub ϕdV0
ω0
Va
1
Z
= J~ea (~r0 ) · ∇r × ∇r × ~ub ϕdV0
ω0
Va (4.143)
1
Z
= ~ub · ∇r × ∇r × J~ea (~r0 )ϕdV0
ω0
Va
1
Z
= ~ub · ∇r × ∇ r × J~ea (~r0 )ϕdV0 ,
ω0
Va
where we have used the relation
∇r0 × ∇r0 × ~ub ϕ = ∇r × ∇r × ~ub ϕ,

(4.144)
~c · ∇ × ∇ × ~ub ϕ = ~ub · ∇ × ∇ × ~cϕ.
Note that expression (4.143) is identical to (4.22) in combination with (4.30). We can deter-
mine H~ a in a similar way by using an unit magnetic dipole directed along ~ub and located at
~r, while J~ea = ~0.
 
J ea J ma
 
J be J bm
S2
V=
2 V∞ − V
Figure 4.16: All sources are located within volume V .
iii. All sources in volume V .

Consider a volume in which the sources J~ea , J~ma, J~b and J~b and the corresponding charge
e m
distributions exist, see Fig. 4.16. Outside V , which is the volume V2 = V∞ − V , no sources
exist. We will again apply Lorentz’ theorem (4.132) on volume V2 = V∞ − V . This volume
is bounded by the surface S2 and by S∞ . Since there are no sources within V2 , the volume
integral in (4.132) is equal zero. Therefore, we find that
Z h i
~a × H
E ~b−E
~b × H
~ a · ~un dS = 0, (4.145)
S2 +S∞
in which ~un is the unit vector which is directed outwards from V2 . Based on the our findings
in case i), the contribution over S∞ will be zero. As a result (4.145) will take the form:
Z h i
~a × H
E ~b−E
~b × H
~ a · ~un dS = 0. (4.146)
S2
iv. Source distribution A in volume V and source distribution B in volume V1 .

Consider the volume V in which the sources J~ea , J~m a and corresponding charge distributions
exist. Outside V , in V1 , we find the sources J~eb and J~m

b with corresponding charge distributions
(see Fig. 4.17). Outside V + V1 no sources exist. By again applying Lorentz’ theorem (4.132)
on volume V1 provides:
Z h i Z h i
− ~a × H
E ~b−E
~b × H
~ a · ~un dS = − ~ a · J~b − H
E ~ a · J~b dV, (4.147)
e m
S V1
where we have used relation (4.146), while the minus-sign in front of the surface integral
accounts for the fact that ~un is directed inwards V1 instead of outwards V1 . Assume that
un
V
 
V1 J ea J ma
 
J be J bm
S2
Figure 4.17: Source distribution A is located in volume V and source distribution B is located in
volume V1 .
source distribution B consists of an electric dipole with direction ~ub and dipole moment
p = 1. Furthermore assume that J~mb = 0. This is the same situation as used in case ii). From
(4.147) we arrive at
Z h i h i
~ a (~r) =
~ub · E ~a × H
E ~ b · ~un − E
~b × H
~ a · ~un dS. (4.148)
S
~ × B)
Using vector relation (A ~ ·C
~ = −(A
~ × C)
~ ·B
~ finally provides
Z h i h i
~ a (~r) =
~ub · E ~a ·E
~un × H ~b − E
~ a × ~un · H
~ b dS. (4.149)
S
With these four special cases, we can make our final last step in the derivation of the Lorentz-
Larmor theorem. For that purpose, we should compare expression (4.149) and (4.138). Both
expressions are derived for the same situation as shown in Fig. 4.15. Therefore, we can formulate
the following important conclusion:
The sources J~ea and J~m

a and the corresponding charge distributions within V generate outside
V an electromagnetic field E ~ a, H
~ a . The volume V is enclosed by surface S. The fields E~ a, H
~a
outside V can also be assumed to be generated by an electric surface current distribution J~es on
a
S and an magnetic surface current density J~ms a on S, where
J~es
a ~ a,
= ~un × H
(4.150)
J~ms
a ~ a × ~un .
= E
These equivalent sources generate inside volume V an electromagnetic field equal to zero.
End intermezzo: derivation of Lorentz-Larmor theorem
We can further derive the expression for the electric field at location ~r outside the source
~ b and H
region V (4.149) by applying the derived expressions for the dipole fields E ~ b . From (4.30)
we find that
~ b (~r0 ) 1
E = ∇r × ∇r0 × ~ub ϕ,
ω0 0 (4.151)
~ b (~r0 ) = ∇r × ~ub ϕ,
H 0
where ϕ is defined in (4.142). Substitution of(4.151) in (4.149) provides

1
Z h i Z h i
~ a (~r) =
~ub · E ~ a (r~0 ) · ∇r × ∇r × ~ub ϕdS −
~un × H ~ a (r~0 ) × ~un · ∇r × ~ub ϕdS.(4.152)
E
0 0 0
ω0
S S
By using the vector relations provided in (4.144) and relation ∇r0 × ~ub ϕ = −∇r ϕ × ~ub we obtain
1
Z h i Z h i
~ a (~r) = ∇r ×
E ~ a (r~0 ) ϕdS +
~un × E ∇r × ∇ r × ~ a (r~0 ) ϕdS.
~un × H (4.153)
ω0
S S
with
1 e−k0 |~r−~r0 |
ϕ(~r, ~r0 ) = . (4.154)
4π |~r − ~r0 |
The corresponding magnetic field H ~ a (~r) outside source region V can be found by placing a mag-
netic dipole of unit strength and direction ~ub at the location indicated by the position vector ~r.
From (4.147) we than obtain
Z h i h i
~ a (~r) = −
~ub · H ~a × H
E ~ b · ~un − E
~b × H
~ a · ~un dS
S (4.155)
Z h i h i
= − ~a
~un × E ~ + ~un × H
·H b ~ a ~b
·E dS.
S
~ b and H
The (magnetic) dipole fields E ~ b have already been derived in section (4.8) and are given
by (4.156). This results in
~ b (~r0 )
E = −∇r0 × ~ub ϕ,
(4.156)
~ b (~r0 ) = 1
H ∇r × ∇r0 × ~ub ϕ.
ωµ0 0
Substitution of (4.156) in (4.155) results in

1
Z h i Z h i
~ a (~r) =
~ub · H ~ a (r~0 ) · ∇r ×~ub ϕdS −
~un × H ~ a (r~0 ) · ∇r × ∇r ×~ub ϕdS.(4.157)
~un × E
0 0 0
ωµ0
S S
By again applying vector relations we can rewrite (4.157) in the following form
1
Z h i Z h i
~ a (~r) = ∇r ×
H ~ a (r~0 ) ϕdS −
~un × H ∇r × ∇ r × ~ a (r~0 ) ϕdS.
~un × E (4.158)
ωµ0
S S
Summary
Assume that the sources J~e , J~m and corresponding charge distributions exist within volume V ,
enclosed by the surface S, The generated electromagnetic field outside V is then found by:
1
Z h i Z h i
~ r)
E(~ = ∇r × ~ r~0 ) ϕdS +
~un × E( ∇r × ∇ r × ~ r~0 ) ϕdS,
~un × H(
ω0
S S (4.159)
~ r~0 ) ϕdS − 1 ∇r × ∇r ×
Z h i Z h i
~ r ) = ∇r ×
H(~ ~un × H( ~ r~0 ) ϕdS,
~un × E(
ωµ0
S S
where J~es = ~un × H

~ and J~ms = E ~ × ~un are the equivalent electric and magnetic surface current
distribution on the surface S, respectively. When ~r is within volume V we find that
~ r)
E(~ = ~0,
(4.160)
~ r) = ~0.
H(~
This is the Lorentz-Larmor theorem also referred to as equivalence principle
It is now fairly straightforward to derive expressions for the electromagnetic fields far away
from the sources, in the so-called far-field region. The integrand in (4.159) consists of products
of vector functions and scalar functions, where the vector function does not depend on ~r, while
the operator ∇r × operates only on ~r. Formulas with such a structure have been discussed before
in section 4.2. We concluded that for far-field points ~r the following relations can be used:
∇r × → −k0 ~ur ×,
(4.161)
∇r × ∇r × → (−k0 )2 ~ur × ~ur × .
Furthermore, in the far-field region the following approximation can be used:
e−k0 |~r−~r0 | e−k0 r k0 ~ur ·~r0

≈ e . (4.162)
|~r − ~r0 | r
Substitution of (4.161) and (4.162) in (4.159) provides the following result
−k0 e−k0 r
Z h i h i
~ r)
E(~ = ~ur × ~un × E( ~ r~0 )
~ r~0 ) − ~ur × ~un × Z0 H( ek0 ~ur ·~r0 dS,
4πr
S (4.163)
−k0 e−k0 r
Z h i h i
~ r) =
Z0 H(~ ~ur × ~ r~0 ) + ~ur × ~un × E(
~un × Z0 H( ~ r~0 ) ek0 ~ur ·~r0 dS.
4πr
S
This can also be written in a more compact form:
~ r) −k0 e−k0 r ~
E(~ = Fa (~ur ),
4πr (4.164)
~ r) = −k0 e−k0 r
Z0 H(~ ~ur × F~a (~ur ).
4πr
with
Z h i h i
F~a (~ur ) = ~ur × ~ r~0 ) − ~ur × ~un × Z0 H(
~un × E( ~ r~0 ) ek0 ~ur ·~r0 dS. (4.165)
S
The subscript a is used to indicate that the function Fa is found by integrating the fields on the
antenna aperture (= surface S). Note that the following relation also holds in the far-field:
~ = Z0 H
E ~ × ~ur . (4.166)
The formulas (4.164) can now be used to determine the radiation properties of arbitrarily-shaped
aperture antennas. A proper choice of the surface S is very important. In the next sections we will
show how S can be chosen in a convenient way for some well-known aperture antenna concept.
In this way, we are able to find an analytic expression of the radiated fields.
4.10 Horn antennas

A well known example of an aperture antenna is the horn antenna as illustrated in Fig. 4.18. Horn
antennas are available with rectangular, square and circular apertures and are usually fed from
a waveguide. Fig. 4.18 also shows how we could select the closed surface S with the equivalent
sources in order to apply the Lorentz-Larmor concept as discussed in the previous section. S
consists of two part, namely Sa and Sc . Sa is the actual radiating aperture of the horn antenna,
while Sc contains the outer (metallic) surface of the horn antenna and generator. Therefore, all
sources are enclosed by the surface S. We can now make the following assumptions:
• The outer surface of the antenna and the generator are perfectly electric conductors (PEC).
~ = ~0 on Sc .
This implies that the tangential electric field along this surface is zero: ~un × E
• The (secondary) currents along the outer surface of the antenna and generator are very
~ = ~0 on Sc .
small and can be ignored in our analysis, i.e. ~un × H
4.10. HORN ANTENNAS 117
generator waveguide horn

Sc Sa
Figure 4.18: Example of horn antennas and choice of the enclosed surface S as Sa + Sc .
Figure 4.19: Rectangular uniform aperture with E ~ a = − E0 ~ux for − a < x <
~ a = E0 ~uy and H a
Z0 2 2
b b
and − 2 < y < 2 .
As a result, the integration in (4.163) is limited to Sa . As a first-order approximation we will

assume that the aperture fields are uniformly distributed in the x-y plane, as indicated in Fig.
4.19. Note that the derivation for the case of a more realistic aperture distribution is quite similar,
although the mathematics are more complex or even require a numerical integration. The fields
of the uniformly illuminated aperture can be described as:
~a
E = E0 ~uy ,
(4.167)
~a E0
H = − ~ux ,
Z0
valid for the rectangular region − a2 < x < a
2 and − 2b < y < 2b . Furthermore, E0 is a constant and
we have assumed that the characteristic impedance is equal to Z0 . The equivalent electric and
magnetic surface currents on Sa are now given by
J~ms = E
~ a × ~un = E0 ~uy × ~uz = E0 ~ux ,
(4.168)
J~es ~ a = − E0 ~uz × ~ux = − E0 ~uy .
= ~un × H
Z0 Z0
We can calculate the electric field in the far-field region by applying expression (4.163) on aperture
S = Sa . We now find
Za/2 Zb/2
−k0 e−k0 r
~ r) =
E(~ ~ur × −J~ms − Z0 ~ur × J~es ek0 ~ur ·~r0 dx0 dy0
4πr
−a/2 −b/2 (4.169)
k0 e−k0 r
~m + L
~e ,

= L
4πr
~ m and L
where the integrals L ~ e are given by
Za/2 Zb/2
~m =
L ~ur × J~ms (~r0 )ek0 ~ur ·~r0 dx0 dy0 ,
−a/2 −b/2
(4.170)
Za/2 Zb/2
~e
L = Z0 ~ur × ~ur × J~es (~r0 )ek0 ~ur ·~r0 dx0 dy0 .
−a/2 −b/2
We can determine the integrals easily by writing the in- and outer-products in (4.170) in terms
of spherical coordinates (r, θ, φ):
~ur × ~ux = ~ur × (~uθ cos θ cos φ − ~uφ sin φ) = ~uφ cos θ cos φ + ~uθ sin φ,
~ur × ~ur × ~uy = −~uθ cos θ sin φ − ~uφ cos φ,

(4.171)
~ur · ~r0 = (~ux x0 + ~uy y0 ) · (~ux sin θ cos φ + ~uy sin θ sin φ + ~uz cos θ)
= x0 sin θ cos φ + y0 sin θ sin φ = x0 u + y0 v,
with u = sin θ cos φ and v = sin θ sin φ. Substitution of (4.171) and (4.168) in (4.170) provides for
~m
the first integral L
Za/2 Zb/2
~ m = E0 (~uφ cos θ cos φ + ~uθ sin φ)
L ek0 (x0 u+y0 v) dx0 dy0 ,
−a/2 −b/2 (4.172)
sin(k0 au/2) sin(k0 bv/2)

= abE0 (~uφ cos θ cos φ + ~uθ sin φ) ,
k0 au/2 k0 bv/2
4.11. REFLECTOR ANTENNAS 119
where we have used the well-known integral

Za/2
sin(βa/2)

βx
e dx = a . (4.173)
βa/2
−a/2
~ e as
Similarly, we can determine the integral L
~ e = abE0 (~uθ cos θ sin φ + ~uφ cos φ) sin(k0 au/2) sin(k0 bv/2)

L . (4.174)
k0 au/2 k0 bv/2
Finally, by substituting (4.174) and (4.172) in (4.169) we obtain the following expression for the
electric field in the far-field region of the horn antenna
~ r) = k0 e−k0 r ~ ~e

E(~ Lm + L
4πr
k0 abE0 e−k0 r sin(k0 au/2) sin(k0 bv/2)

= [~uφ cos φ(cos θ + 1) + ~uθ sin φ(cos θ + 1)] (4.175)
4πr k0 au/2 k0 bv/2
= Eθ ~uθ + Eφ ~uφ .
We can observe that E ~ has two components, Eθ ~uθ and Eφ ~uφ . In the principle planes of the
antenna, i.e. the φ = 0 plane (H-plane) and φ = π/2 plane (E-plane), one of both components is
equal to zero. The magnetic field H ~ in the far-field region is easily found by using the far-field
~ = Z0 H
relation E ~ ×~ur . The radiation pattern is determined according to the approach in chapter
2 by using relation (2.8) up to (2.10). The resulting normalized radiation pattern then takes the
following form:
|Eθ (θ, φ)|2 + |Eφ (θ, φ)|2
F (θ, φ) = . (4.176)
|Eθ (0, 0)|2 + |Eφ (0, 0)|2
The corresponding half-power beam width is given by θHP = 2 arcsin(0.433λ0 /a) in the H-plane
and θHP = 2 arcsin(0.433λ0 /b) in the E-plane, respectively. When the dimension of the horn
increases, the beam width decreases, so the antenna will become more directive. The directivity
is found from D = 4π λab0 . Fig. 4.20 shows the three-dimensional radiation pattern in [dB] for a
uniform aperture with dimensions a = 4λ0 and b = 2λ0 . The corresponding cross section in the
φ = 0 plane (v = 0) is illustrated in Fig. 4.21. The first sidelobe has a peak at −13.26 dB w.r.t.
the main lobe. The 3-dB beam width in the φ = 0 plane of this aperture antenna is θHP = 12.40
and the directivity D = 20 dBi.
4.11 Reflector antennas

Reflector antennas can take many shapes, e.g. parabolic, spherical or cylindrical. The parabolic
reflector antennas is the most commonly used type. Reflector antennas provide a large antenna
gain and associated effective antenna aperture, which makes them very suited for long-range
wireless applications, such as satellite communication, point-to-point (PtP) backhauling and long-
range radar. The parabolic reflector antenna is fed by a small antenna element(feed), such as
Figure 4.20: Three-dimensional normalized radiation pattern versus (θ, φ) of a uniform aperture
with dimensions a = 4λ0 and b = 2λ0 . All values are in dB.
Uniform rectangular aperture φ=0 plane, a=4λ0 and b=2λ0

0
−5
−10
Normalised radiation pattern [dB]
−15
−20
−25
−30
−35
−40
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 4.21: Cross section in the φ = 0 plane (H-plane) of the normalized radiation pattern versus
θ of a uniform aperture with dimensions a = 4λ0 and b = 2λ0 . All values are in dB. The 3 dB
beam width in this plane is θHP = 2 arcsin(0.433λ0 /a) = 12.40 and directivity D = 20 dBi.
a dipole antenna or horn antenna. The feed is placed in the focal point of the reflector and
illuminates the reflector. As a result of this, electric currents are generated on the metal reflector
surface. These currents generate the scattered field of the reflector. Fig. 4.22 shows an illustration
of the parabolic reflector and feed antenna. The electromagnetic field in the far-field region can
be determined using equation (4.163). In this case, we will choose S = Sa + Sc as indicated in
Fig. 4.22a. The sources of the scattered field are now all located within the surface S. We will
assume that the reflector is made of a perfectly-conducting metallic material. Furthermore, we
Sc Sc
Sa Sa
F
F F
Feed antenna Feed antenna
Parabolic reflector Parabolic reflector
Figure 4.22: Parabolic reflector (cross section) with feed placed in the focal point. (a) The choice
of the surface Sa according to the aperture method, (b) surface Sa is selected according to the
current-integration method..
will ignore the currents flowing on the back side of the reflector. As a result
~ = ~0
~un × E on Sc ,
(4.177)
~ = ~0 on Sc .
~un × H
Therefore, we can limit the integration in (4.163) to the aperture Sa . The radiation pattern of
the feed antenna located in the focal point F determines the field distribution over the aperture
plane Sa . This method is known as the aperture method. However, it is also possible to select the
~ = ~0 on Sa , and as a result
surface S as indicated in Fig. 4.22b. In this case we find that ~un × E
(4.163) reduces to
k0 e−k0 r
Z h i
~ r) =
E(~ ~ur × ~ur × ~ a (r~0 ) ek0 ~ur ·~r0 dS,
~un × Z0 H (4.178)
4πr
S
The magnetic field in the aperture H ~ a on Sa can be easily determined from the radiation charac-
teristics of the feed antenna located in the focal point F . This method is known as the current
integration method.
In this section we will use the apeture method to determine the radiation characteristics
of a parabolic reflector antenna. Now consider the configuration of Fig. 4.23. The reflector has
A B
E
D
y
Feed antenna
ZoH
Sa Equi-phase plane
Aperture
Parabolic reflector
Figure 4.23: Circularly-symmetric parabolic reflector with feed located in the focal point. The
radius of the circularly symmetric aperture is a = D/2.
a diameter D = 2a. A feed antenna is located in the focal point F . Since the feed antenna is
relative small (typically 0.5λ0 − 4λ0 ), a point A on the reflector will be located in the far-field
region of the feed antenna, that is the distance between the feed and the reflector is larger than
2L2
, with L the largest dimension of the feed antenna. In the far-field region of the feed antenna,
λ0
the equi-phase phase front is spherical. Perpendicular to these planes, we can define so-called
rays directed along the Pointing vector. Now assume that a ray that originates from the feed hits
the reflector at point A on the reflector surface, as indicated in Fig. 4.23. We will assume that
the reflector at point A is locally flat. Furthermore, the incident wave at point A is locally flat as
well, represented as a plane wave. As a result, we can use the well-known reflection rules which
yield for a plane wave incident on a large and perfectly-conduction flat plate. This provides us
the field in point B, see Fig. 4.23. In this way, we can determine the fields in the aperture Sa .
The radiation characteristics of the feed antenna have a very large impact on the final radiation
characteristics of the reflector antenna. In many applications the feed is designed in such a way
that the edges of the reflector are less illuminated as compared to the center region. This is called
amplitude tapering.
For the sake of simplicity, we will first assume that the feed antenna provides a uniform
illumination of the aperture Sa . Now let the incident electric field E~ i coincide with the x-direction
and the magnetic field H~ i with the y-direction, as illustrated in Fig. 4.23. In a circular-symmetric
aperture, we then have the following field distribution:
~a
E = E0 ~ux ,
(4.179)
~ a = E0 ~uy ,
Z0 H
where E0 is a constant. The electric field in a point P far away from the aperture, can be
calculated by using expresion (4.163)
−k0 e−k0 r
Z h i h i
~ r) =
E(~ ~ur × ~ a (r~0 ) − ~ur × ~un × Z0 H
~un × E ~ a (r~0 ) ek0 ~ur ·~r0 dS. (4.180)
4πr
Sa
We have already shown in section 4.2 that the electric field ~

h E in the far-field region only has itwo
components: Eθ and Eφ . Therefore, we can express ~ur × ~un × E ~ a (r~0 ) − ~ur × ~un × Z0 H
~ a (r~0 ) in
the following form:
h i
~
M ~ a − ~ur × ~un × Z0 H
= ~ur × ~un × E ~a
= E0 ~ur × [~uz × ~ux − ~ur × ~uz × ~uy ] (4.181)
= E0 (1 + cos θ) (~uφ sin φ − ~uθ cos φ) .
Furthermore, the inner product (~r0 · ~ur ) can be written in polar coordinates according to
~ur · ~r0 = r0 sin θ cos(φ − φ0 ). (4.182)
The electric field in the far-field region then takes the following form
Z2πZa
~ r) = −k0 e−k0 r
E(~ (1 + cos θ) (~uφ sin φ − ~uθ cos φ) E0 ek0 r0 sin θ cos(φ−φ0 ) r0 dr0 dφ0 . (4.183)
4πr
0 0
Since we have assumed a uniform aperture illumination, E0 is constant and can be put outside
the integral. The φ0 integral can be expressed in terms of a Bessel function according to
Z2π
ek0 r0 sin θ cos(φ−φ0 ) dφ0 = 2πJ0 (k0 r0 sin θ), (4.184)
0
where J0 is the Bessel function of the first kind with order zero. The remaining integration w.r.t.
r0 can now be expressen in closed-form as:
Za Z1
2
J0 (k0 r0 sin θ)r0 dr0 = a J0 (k0 aρ sin θ)ρdρ
0 0 (4.185)
J1 (k0 a sin θ)
= a2 ,
k0 a sin θ
where J1 is the Bessel function of the first kind with order one. Inserting this in (4.183) provides
a2 k0 E0 e−k0 r J1 (k0 a sin θ)

Eθ = (1 + cos θ) cos φ ,
2r k0 a sin θ (4.186)
−a2 k0 E0 e−k0 r J1 (k0 a sin θ)
Eφ = (1 + cos θ) sin φ .
2r k0 a sin θ
The radiation pattern is found by using relations (2.8)- (2.10):
|Eθ (θ, φ)|2 + |Eφ (θ, φ)|2

F (θ) =
|Eθ (0, 0)|2 + |Eφ (0, 0)|2 (4.187)
2
J1 (k0 a sin θ)

= (1 + cos θ)2 .
k0 a sin θ
J1 (z) 1
Note that we used the relation lim = . The radiation pattern (in [dB]) is shown in Fig.
z
x→0 2
4.24 in case of a uniformly-illuminated aperture with a diameter D = 2a = 4λ0 . The cross section
in the φ = 0-plane is shown in Fig. 4.25. The first sidelobe is found at a level of −17.6 dB w.r.t.
the main beam and the 3-dB beam width is θHP = 1.02λ0 /(2a) = 14.60 . The beam width scales
directly with the size of the reflector: the larger D, the smaller the beam width. Note that the
first sidelobe of a circular aperture is lower than the first sidelobe of a rectangular aperture (-17.6
dB versus -13.3 dB).
Figure 4.24: Three-dimensional radiation pattern versus (θ, φ) of a parabolic reflector with uniform
aperture illumination with diameter D = 2a = 4λ0 . All values are in dB.
Up to now, we have assumed a uniform illumination of the aperture: we have assumed

E0 to be constant over the aperture surface Sa . The drawback of a uniform distribution is the
relative high first sidelobe level of -17.6 dB. We will now extend our analysis to a more general
Uniform circular aperture, D=2a=4λ0

0
−5
−10
−15
−20
−25
−30
−35
−40
−45
−50
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 4.25: Radiation pattern in the φ = 0-plane θ of a parabolic reflector with uniform aperture
illumination with diameter D = 2a = 4λ0 . All values are in dB. The 3 dB beam width in this
plane is θHP = 1.02λ0 /(2a) = 14.60
aperture distribution. For that purpose, we will assume that the aperture distribution on Sa can
be expressed in the following form:
~ a (r0 )
E = W (r0 )~ux ,
(4.188)
~ a (r0 ) = W (r0 )~uy ,
Z0 H
with aperture distribution (or weighting) function given by

" 2 # p
r0

W (r0 ) = E0 1 − . (4.189)
a
The specific case p = 0, which provides a uniform illumination, is already discussed in this
section. For other valus of p we have to return to expression (4.183) and replace E0 by the
aperture distribution function W (r0 ). This provides us with the following integral
Za Z2π" 2 #p
r0
Z
k0 (r0 ·~
ur )
W (r0 )e dS = E0 1− ek0 r0 sin θ cos(φ−φ0 ) dφ0 r0 dr0 . (4.190)
a
Sa 0 0
The φ0 integration is done in a similar way as in (4.184). As a result we obtain the following
integral along r0 :
Za " 2 # p
r0

2πE0 1− J0 (k0 r0 sin θ)r0 dr0 . (4.191)
a
0
Substituting ρ = r0 /a and ua = k0 a sin θ results in

Z1 h ip
2
2πa E0 1 − ρ2 J0 (ua ρ)ρdρ. (4.192)
0
This integral can be expressed in terms of a Bessel function by applying the integral of Sonine
(see [28], p.373)
π/2
z ν+1
Z
Jµ+ν+1 (z) = ν Jµ (z sin θ) sinµ+1 θ cos2ν+1 θdθ, (4.193)
2 Γf (ν + 1)
0
Here Γf is the Gamma function [26] for which Γ(p + 1) = p!. Now let µ = 0, ν = p, z = ua and
sin θ = ρ. We then find
Z1
Jp+1 (ua ) 1 h ip
= p J0 (ua ρ) 1 − ρ2 ρdρ. (4.194)
up+1
a 2 Γf (p + 1)
0
As a result, the integral takes the form

Jp+1 (ua )
Z
W (r0 )ek0 (r0 ·~ur ) dS = 2πa2 E0 2p p! . (4.195)
up+1
a
Sa
The electric field in the far-field region is now easily found by inserting (4.195) in (4.183). Some
calculated results are summarized in table 4.1. From this table we can conclude that the uniform
illumination (p = 0) provides the highest directivity and antenna gain. However, the first sidelobe
level is relative high. By using a weighted aperture distribution (tapering), we can reduce the
sidelobe level at the expense of a reduced directivity and corresponding efficiency, which is as low
as 44% for p = 2. The requirements for a particular application will determine the optimal shape
of the aperture distribution function W (r0 ).
Antenna parameter p=0 p=1 p=2
λ0 λ0 λ0
3-dB beam width in degrees 29.2 36.4 42.1
a a a
λ0 λ0 λ0
Beam width between zeros in degrees 69.9 93.4 116.3
a a a
First sidelobe level [dB] -17.6 -24.6 -30.6

2 2 2
2πa 2πa 2πa

Directivity 0.75 0.56
λ0 λ0 λ0
Table 4.1: Overview of the most relevant antenna parameters as a function of the order p of the
p
aperture distribution function W = E0 1 − (r0 /a)2 in case of a parabolic reflector antenna with

diameter D = 2a.
4.12. MICROSTRIP ANTENNAS 127
4.12 Microstrip antennas

Microstrip antennas, also known as patch antennas or printed antennas are resonant antennas
which can be realized on printed-circuit boards (PCBs). In order to determine the radiation
properties of microstrip antennas, we can use the same method as used for aperture antennas
in the previous sections. Microstrip antennas can take many forms and shapes. We will limit
ourselves to some basic structures. Fig. 4.26 shows the side- and top-view of a rectangular and
circular microstrip antenna fed by a coaxial cable. The inner conductor of the coaxial feed is
connected to the metal patch and the outer conductor is connected to the ground plane. The
dielectric substrate has a thickness h and dielectric constant r . The dielectric losses are indicated
by the loss tangent tan δ and will be neglected here. Microstrip antennas can be manufactured
with standard PCB etching techniques to realize printed copper structures with accuracies well
below 100µm. The patch can also be fed by a microstrip line or by using a resonant slot in the
ground plane.
z
P
patch
θ
εr h
r
ground plane coax feed
patch
a h
φ ρ
εr
ground plane
x
a) Side view and top view of rectangular microstrip antenna b) Circular microstrip antenna
Figure 4.26: Rectangular and circular microstrip antenna fed by a coaxial cable. The metallic
patch is printed on a substrate with permittivity r .
In literature, various models have been introduced to determine the radiation properties of
microstrip antennas, including analytical models such as the cavity model and transmission line
model, and numerical models such as the method of moments (MoM)[3]. In this section we will
explore the cavity model in more detail for the case of a circular microstrip antenna as illustrated
in Fig. 4.26. We will assume that all metal surfaces (patch and ground plane) are perfect electric
conductors (PEC). Furthermore, we will assume that the thickness of the substrate h is very thin
as compared to the wavelength, i.e. h λ0 . We will use cylindrical coordinates with (ρ, φ, z)
to describe the antenna configuration. With these assumptions, we can introduce the following
simplifications:
i. The electric field will be concentrated between the circular patch and the ground plane within
a cylinder with radius a and height h. This cylinder is called cavity. The cavity is filled with
a homogeneous diectric material with permittivity r .
ii. Due to the small thickness of the substrate, the fields within the cavity will not depend on
the z-coordinate.
~ = Ez ~uz .
iii. The electric field has only a component in the z direction, so E
iv. The current distribution J~e along the edge ρ = a of the circular patch will be directed parallel
to the edge of the patch. This implies that the magnetic field given by H ~ = ~uz × J~e , will be
~ ~
perpendicular to the side walls of the cavity, so ~un × H = 0. Note that ~un is the normal on
the side walls of the cavity.
With these assumptions, our problem is reduced to a cavity structure with the following boundary
conditions:
~ = ~0
~un × H on the sidewalls
(4.196)
~ = ~0
~un × E on the top and bottom side of the cavity,
where ~un is the normal vector on the surface under consideration. Since E~ will only have a z-
~ = ~0 when ρ = a in terms of a condition for
component, we can express the condition that ~un × H
Ez , since
~ = −1 ∇ × E.
H ~ (4.197)
ωµ0
As a result, we can express the components of the magnetic field in the following form:
−1 1 ∂Ez
Hρ = ,
ωµ0 ρ ∂φ (4.198)
1 ∂Ez
Hφ = .
ωµ0 ∂ρ
Apparantly, on the side walls the following condition holds
∂Ez (4.199)
= 0, when ρ = a.
∂ρ
~ = Ez ~uz by using
Inside the cavity, we can derive the Helmholtz equation for the electric field E
Maxwell’s equations for a source-free region (4.5). We now obtain
~ + k2 E
∇2 E ~ = ~0, (4.200)
√
where k = k0 r represents the wave number in the substrate. The Helmholtz equation in
∂Ez
cylindrical coordinates then takes the following form (with = 0)
∂z
∂ 2 Ez 1 ∂Ez 1 ∂ 2 Ez
+ + + k 2 Ez = 0. (4.201)
∂ρ2 ρ ∂ρ ρ2 ∂φ2
The solution of this second-order differential equation is
Ez = E0 Jn (kρ) cos(nφ), (4.202)
where Jn (kρ) is the Bessel function of the first kind with order n and E0 is a constant which is
determined by the applied voltage at the feed point. The other field components are given by
n
Hρ = E0 Jn (kρ) sin(nφ),
ωµ0 ρ (4.203)
k ∂Jn (kρ)
Hφ = E0 cos(nφ).
ωµ0 ∂ρ
All other field components are zero, so Eρ = Eφ = Hz = 0. The side wall of the cavity acts as a
~ = ~0 or Hφ = 0. This provides us the resonance condition:
magnetic wall with ~un × H
∂Jn (ka)
= 0. (4.204)
∂ρ
∂Jn (ka)
Since has multiple zeros for a particular values of n, we will find multiple resonance
∂ρ
∂Jn (ka)
frequencies for a given radius a and n. Now let Knm be the m-th zero of . Table
∂ρ
4.2 summarizes a number of zeros with corresponding values of ka. The resonance frequency
corresponding to a particular mode can be determined from the following equation
2πf √
ka = r a = Knm , (4.205)
c
resulting in
Knm c
fnm = √ . (4.206)
2πa r
The mode with (n, m) = (1, 1) provides the lowest resonance frequency for a particular radius a
of the circular patch. Note that the cavity model does not account for fringing fields at the edges
of the cavity. These fringing fields will slightly change the resonance frequency. This effect can
be included in formula (4.206) by introducing an effective radius ae . Closed-form expressions for
ae can be found in literature [29]-[30]:
s
2h πa

ae = a 1 + ln + 1.7726 . (4.207)
πar 2h
Near resonance it is quite easy to match the input impedance of a microstrip antennas to
a transmission line, such as a 50Ω coax cable. The bandwidth of microstrip antennas (defined
as the frequency band over which the antenna can be matched to a transmission line with a
Mode (n, m) zero ka
0, 1 0
1, 1 1.841
2, 1 3.054
0, 2 3.832
3, 1 4.201
4, 1 5.317
1, 2 5.331
5, 1 6.416
∂Jn (ka)
Table 4.2: Zeros of .
∂ρ
characteristic impedance Z0 ) strongly depends on the electrical thickness of the substrate. An

accurate estimation of the input impedance of microstrip antennas can be best obtained using
numerical methods, such as the Method-of-Moments (MoM), see also chapter 5. Fig. 4.27 shows
the realized bandwidth of a coax-fed microstrip antenna versus electrical thickness, according to
MoM simulation from [3]. A larger bandwidth can be obtained by using stacked patches [3] or by
using a resonant slot feed in the ground plane.
The radiation properties of microstrip antennas can be easily determined by using the
aperture method according to the Lorentz-Larmor theorem. We will select the surface S as
Sp + Sa + Sg , where Sp is the patch, Sa the side wall of the cavity and Sg the ground plane. We
can also apply the image principle (see section 4.5) in order to account for the ground plane. At
the side wall of the cavity we then obtain the following magnetic current distribution
J~ms = 2E
~ × ~un = 2Ez ~uφ when ρ = a and 0 < z < h. (4.208)
As a result, we find that the field components Eθ and Eφ in the far-field region are given by
n hak0 E0 Jn (ka)e−k0 r
Eθ = cos nφ [Jn+1 (k0 a sin θ) − Jn−1 (k0 a sin θ)] ,
2r (4.209)
n hak0 E0 Jn (ka)e−k0 r
Eφ = cos θ sin nφ [Jn+1 (k0 a sin θ) + Jn−1 (k0 a sin θ)] ,
2r
Figure 4.27: Bandwidth of a single-layer microstrip antennas versus electrical height of the dielec-
tric substrate.
where we have assumed that the patch is fed by a transmission line in the φ = 0 plane. For the
fundamental mode with (n, m) = (1, 1) we obtain:
hak0 E0 J1 (ka)e−k0 r
Eθ = cos φ [J2 (k0 a sin θ) − J0 (k0 a sin θ)] ,
2r (4.210)
hak0 E0 J1 (ka)e−k0 r
Eφ = cos θ sin φ [J2 (k0 a sin θ) + J0 (k0 a sin θ)] .
2r
Fig. 4.28 shows the radiation pattern in the φ = 00 plane and the φ = 900 plane of a circular
microstrip antenna operating in the (n, m) = (1, 1)-mode and resonating at f = 12 GHz. The
radius of the patch is a = 4.6mm and substrate permittivity r = 2.56. The 3-dB beam width is
1040 in the φ = 00 plane and 800 in the φ = 900 plane. Due to its omni-directional character,
microstrip antennas are very usefull in phased-array antennas that require wide-scan electronic
beamforming.
Circular Microstrip Antenna, a=4.6mm, f=12GHz, εr=2.56

0
−2
−4
−6
−8
−10
−12
−14
φ=00
−16 φ=900
−18
−20
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 4.28: Radiation pattern of a circular microstrip antenna printed on an electrically thin
substrate in the φ = 0 plane and φ = 900 -plane. The antenna operates in the fundamental
(n, m) = (1, 1)-mode at f = 12 GHz. The radius of the patch a = 4.6 mm and r = 2.56. All
values are plotted in dB.
Chapter 5
Numerical electromagnetic analysis
In chapter 4 we have determined the radiation characteristics of several antenna types with a given
electric and/or magnetic current distribution. In section 4.4 we have investigated wire antennas,
where we assumed a sinusoidal current distribution along the wire. When the antenna is operated
close to its resonance, this will be a very good approximation of the real current distribution.
However, at other frequencies the current distribution will be different. In general, it is not easy
(or not possible) to derive an accurate analytic description of the currents on antenna structures,
even not for a fairly simple λ0 /2 dipole (wire) antenna. For that purpose, antenna designers
use numerical methods to determine the current distribution on the antenna structure. Several
numerical approaches are described in literature, including the Finite-Element-Method (FEM),
Finite-Difference-Time-Domain (FDTD) and the Method-of-Moments (MoM), see for example in
[31]. One of the more common methods for antenna analysis is MoM, which is implemented in
several commercial microwave and antenna design tools, for example in Momentum which is part
of the EDA tool Advanced Design System (ADS) [32]. In this section we will show in a step-by-
step approach how MoM can be used to determine the current distribution on a cylindrical wire
antenna.
5.1 Method of moments (MoM)

The method of moments is a general numerical procedure for solving integral equations. One
of the first books which describes MoM for electromagnetic problems is the book of Harrington
[33]. In this section we will illustrate how MoM can be used to determine the current distribution
along a simple cylindrical wire antenna. Note that MoM can also be used to determine equivalent
magnetic currents. The following step-wise approach is used:
• determine the integral equation of the antenna problem at hand,
• discretize the antenna structure in small segments (also known as sub-domains),
• derive the matrix equation for the (still) unknown current distribution along the antenna,
• calculate the elements of the matrix and solve the matrix equation.
133
134 CHAPTER 5. NUMERICAL ELECTROMAGNETIC ANALYSIS
Now consider the cylindrical wire antenna in free space as illustrated in Fig. 5.1. The antenna
is symmetrically oriented along the z-axis with length 2l and diameter 2a. The antenna is fed
in the center at z = 0. This is exactly the same structure as discussed in section 4.4. We will
2l J
2a
Figure 5.1: Electrically-thin cylindrical wire antenna of length 2l and diameter 2a. The center of
the wire is located at the origin of the coordinate system and is oriented along the z-axis.
assume that the diameter of the wire is much smaller than the wavelength, so 2a λ0 . Therefore,
we can ignore electric currents at the end surfaces of the cylinder, and as a result, the electric
current distribution will only have a component along the z-axis, i.e. J~ = Jz ~uz . Note that we
have removed the subscript e in J~ in order to simplify the notation in the rest of this section.
The antenna configuration of Fig. 5.1 imposes two boundary conditions:
i The tangential field at the perfectly-conducting cylinder with diameter 2a has to be equal to
zero.
ii The current at both ends of the cylinder is zero.
In our MoM approach we will only apply the first boundary condition directly. This boundary
condition can be describes as
~un × E~tot (~r) = ~un × (E~ex (~r) + E~s (~r)) = ~0, ~rS0 , (5.1)
where the surface S0 is the outer surface of the cylinder and where E~ex (~r) and E~s (~r) are the
excitation field and the scattered field, respectively. Furthermore, ~un is the normal vector on
the metal cylinder. The scattered field is the field due to the induced currents on the wire
antenna. The excitation field is the field induced by the voltage source which is connected to the
differential port at the center of the antenna. The scattered field can be expressed in terms of
the (yet) unknown electric current distribution J~ and the free-space Greens’ function G(~r, ~r0 ) by
5.1. METHOD OF MOMENTS (MOM) 135
using expressions (4.30), (4.21) and (4.22) from chapter 4. In case the antenna is not located in
free-space, i.e. in case of a microstrip antenna, we can still use the proposed approach. However,
the Greens function will be more complicated, see for example in [3] for the case of multi-layer
microstrip antennas. For our free-space configuration of Fig. 5.1 we can express equation (5.1) in
the following form:
ZZ
~ ex (~r) = ~un × ωµ0
~un × E ~ r0 )dS0
G(~r, ~r0 )J(~
(5.2)
ωµ0
ZZ
+~un × ∇ ∇· ~ r0 )dS0 , ~rS0 .
G(~r, ~r0 )J(~
k02
Integral equation (5.2) can be solved numerically by using the method of moments. The first
step is the expansion of the yet unknown current distribution J~ on the wire antenna in expansion
functions according to
~ y, z) = In J~n (x, y, z),

X
J(x, (5.3)
n
where In are the mode coefficients which we need to determine. The expansion functions J~n (x, y, z)
are also called basis functies and can take many forms. When we want to have an exact solution
~ we need to use an infinite summation in (5.3). In addition,
of the surface current distribution J,
the set of basis functions need to be complete in the sense that all physical effects are included.
In practice we will limit the summation in (5.3) to a maximum number of basis functions of
n = Nmax . In this way, we can approximate the exact solution with a limited computational
effort. There are many types of basis functions. We can subdivide them into two main categories,
that is in global basis functions and local basis functions. Global basis functions are non-zero over
the entire domain, in our example over the entire cylindrical wire, whereas local basis functions are
only non-zero over a very small part of the entire domain. For that reason, local basis functions
are also known as sub-domain basis functies. Two examples of sub-domain basis functions are
shown in Fig. 5.2, where piece-wise constant (PWC) and overlapping piece-wise linear (PWL)
basis functions are shown that approximate a particular function.
When using Nmax basis functions to describe the yet unknown current distribution on the
antenna, we obtain
NX
max
~ y, z) =
J(x, In J~n (x, y, z). (5.4)
n
~ s (x, y, z) can be written in terms of the current distribution J(x,

The scattered electric field E ~ y, z)
~ s (x, y, z) = L{J(x,
E ~ y, z)}, (5.5)
where L is a linear operator. By combining (5.5) and (5.4) we obtain

NX
max NX
max
~ s (x, y, z) =
E In L{J~n (x, y, z)} = ~ ns (x, y, z).
In E (5.6)
n=1 n=1
Piece-wise constant (PWC) basis functions Piece-wise linear (PWL) basis functions
z0 z1 z2 z3 z4 z5 z0 z1 z2 z3 z4 z5
Figure 5.2: Piece-wise constant (PWC) and Piece wise linear (PWL) approximation of a function.
Substitution of the expansion (5.6) in equation (5.1) provides

NX
!
max
~un × ~ ns (x, y, z)
In E ~ ex
+ E (x, y, z) = ~0 , (5.7)
n=1
~ according to
on the outer surface S0 of the wire antenna. Let us now introduce the residue R
NX
!
max
~
R(x, y, z) = ~un × ~ ns (x, y, z)
In E ~ ex
+ E (x, y, z) . (5.8)
n=1
This residue has to be zero over the entire outer surface of the wire antenna. We will relax this
requirement by weighting the residue to zero with respect to some weighting functions J~m (x, y, z),
such that
ZZ
~ ; J~m i =
hR ~
R(x, y, z) · J~m (x, y, z) dS = 0 , (5.9)
Sm
for m = 1, 2..., Nmax , where Sm represents the surface on which the weighting function J~m is
nonzero. The set of weighting functions J~m , is also known as a set of test functions. Often the
set of test functions is the same as the set of expansion functions in (5.3). This particular choice
is called the Galerkin’s method [27]. Inserting (5.8) in (5.9) provides the following set of Nmax
linear equations
NX
max ZZ ZZ
In ~ s (x, y, z) · J~m (x, y, z) dS +
E ~ ex (x, y, z) · J~m (x, y, z) dS = 0,
E (5.10)
n
n=1 Sm Sm
for m = 1, 2..., Nmax . We can write this set of linear equations in a more compact form:
NX
max
In Zmn + Vmex = 0, (5.11)
n=1
for m = 1, 2..., Nmax . In matrix notation we now obtain:
[Z][I] + [V ex ] = [0], (5.12)
where the elements of the matrix [Z] and the elements of the excitation vector [V ex ] are given by
ZZ
Zmn = ~ ns (x, y, z) · J~m (x, y, z) dS,
E
Sm (5.13)
ZZ
Vmex = ~ ex (x, y, z) · J~m (x, y, z) dS,
E
Sm
The matrix [Z] includes Nmax × Nmax elements, [I] is a vector with Nmax unknown mode coeffi-
cients and [V ex ] is the excitation vector with Nmax elements. The matrix equation (5.12) can be
solved rather easily by using standard numerical linear-algebra routines such as those available in
MATLAB. In practise it will turn out that the largest computational effort is claimed for the cal-
culation of the elements of the matrix [Z]. After matrix equation (5.12) has been solved, we obtain
a solution for the mode coefficients and we can find an approximation of the current distribution
J~ on the antenna surface by inserting the mode coefficients in expression (5.4). Once the current
distribution is known, we can calculated the far-field pattern and other antenna characteristics
by using the concepts from chapters 2 and 4.
The elements of the matrix [Z] can be found by using the expression for E ~ s as applied in
(5.2), which results in:
ZZ ZZ
Zmn = −ωµ0 G(~r, ~r0 )J~n (x0 , y0 , z0 )dS0
Sm Sn
(5.14)
ωµ0
ZZ
−∇ ∇· G(~r, ~r0 )J~n (x0 , y0 , z0 )dS0 · J~m (x, y, z) dS,
k02 Sn
In general it will not be possible to find an analytic expression for the elements of [Z]. We will
need to use numerical methods. Note that, depending on the particular antenna structure under
investigation, several numerical problems (e.g. singularities) might occur while evaluating (5.14).
More on this can be found in [31] and other literature.
We will now return to our example, the cylindrical wire antenna as illustrated in Fig. 5.1.
Since we have assumed that the diameter 2a λ0 , the electric current distribution will only have
a component along the z-axis, i.e. J~ = Jz ~uz . Therefore, the expansion and test functions will
only have a component in the z direction, so J~m = Jzm ~uz . We can now write an element of the
matrix [Z] in the following form
!Z
1 ∂2
Z
Zmn = −ωµ0 1+ 2 2 G(z, z0 )Jzn (z0 )dz0 Jzm (z) dz
k0 ∂z
m n (5.15)
!
1 ∂2
Z Z
= −ωµ0 1+ 2 2 G(z, z0 )Jzn (z0 )dz0 Jzm (z) dz
k0 ∂z
m n
where the Green’s function G(z, z0 ) is given by

e−k0 Ra
G(z, z0 ) = , (5.16)
4πRa
with
q
Ra = a2 + (z − z0 )2 . (5.17)
Note that Ra is nonzero for z = z0 . This is the result of our choice (in fact approximation) to
define the expansion function on the outer cylindrical surface (ρ = a) of the wire antenna, whereas
the test functions are defined along the axis of the wire antenna (ρ = 0). In this way, we avoid a
singularity of the Green’s function G(z, z0 ) at z = z0 . The integral equation for the wire antenna
that leads towards expression (5.15) for Zmn is known as the Pocklington integral equation. Next
to this equation, we can also derive an alternative integral equation, known as the Hallén integral
equation, see also [33],[31].
In our example, we will choose piece-wise constant (PWC) expansion- and test functions.
The m-th basis function of this set is defined as



 1, zm−1 < z < zm+1 ,


Jzm (z) = (5.18)


 0, elsewhere,


where the index m = 1, 2, .., Nmax . For the wire antenna of Fig. 5.1, we get z0 = −l and
zNmax +1 = l. Note that the PWC basis functions do not explicitly satisfy the physical requirement
that the current at both ends of the wire should be equal to zero. This requirement can be satisfied
when using PWL basis functions (see Fig. 5.2). For that reason, PWL basis functions will provide
a more accurate result for a given maximum of sub-domains. By substituting (5.18) for PWC
functions in the expression for Zmn (5.15), we obtain
zZ
m+1 zZ
n+1
e−k0 Ra
!
1 ∂2
Zmn = −ωµ0 1+ 2 2 dz0 dz (5.19)
k0 ∂z 4πRa
zm−1 zn−1
The double derivative w.r.t. z in (5.19) can easily be obtained analytically, see [34], [30]. The
resulting double integral then needs to be determined numerically, for example by using standard
integration routines available in MATLAB. The last step in the calculation of the mode coefficients
with (5.12) is the calculation of the excitation vector [V ex ]. The excitation field in expression
(5.13) can be an incident plane wave or a locally generated field. In this example, we will assume
that a voltage generator is connected to the input port of the wire antenna, located at the center
of the wire. The voltage generator can be approximated by a so-called delta-gap generator, which
generates locally a very strong electric field directed along the z-axis. This excitation field can be
expressed as follows:
~ ex = Vg δ(z − zg )~uz ,
E (5.20)
where Vg represent the generator voltage and where zg is the position of the delta-gap generator.
When we connect the delta-gap generator at the center of mode m = i, with zg = zi , the excitation
vector [V ex ] takes the following form:

   

 V1ex   0
 


   
   
   

 V2ex   0
 


   
   
   
.......   .......
   
 
ex
[V ] =  = . (5.21)
   
   
Viex   Vg
   
 
   
   
   
   

 .......   .......
 


   
   
   
VNexmax 0
As a result, only element i of the vector [V ex ] is nonzero. We have now determined all elements of
[Z] and [V ex ] in matrix equation (5.12). By solving this equation (numerically), we can determine
the unknown mode coefficients and related current distribution along the wire antenna. The far-
field pattern is then determined by using the expressions introduced in section 4.2. The input
impedance is easily found as the relation between the input voltage and input current:
Vg
Zin = , (5.22)
Ii
where Ii is the mode coefficient of mode i.
Chapter 6
Phased Arrays and Smart Antennas
Phased-array antennas with electronic beamsteering capabilities have been developed in the period
between 1970-1990 mainly for military radar applications to replace mechanically rotating antenna
systems. Reflector antennas have a high antenna gain, but have the disadvantage that the main
lobe of the antenna has to be steered in the desired direction by means of a highly accurate
mechanical steering mechanism. This means that simultaneous communication with several points
in space is not possible. Small antennas, like a dipole or microstrip antenna have a close to omni-
directional radiation pattern, but have a very low antenna gain. Therefore, small antennas cannot
be used for large-distance communication or sensing. By combining a large number of small
antennas, we can create an array of antennas. When each of these individual antenna elements
is provided with an adjustable amplitude and phase, we obtain a phased-array antenna. In this
way, a phased-array antenna can communicate with several targets which may be anywhere in
space, simultaneously and continuously, because the main beam of the antenna can be directed
electronically into a certain direction. Another advantage of phased-array antennas is the fact
that they are usually relatively flat. Examples of modern phased array systems are shown in Fig.
6.1 and Fig. 6.2 used in radar and radio astronomy, respectively. Phased arrays are also very
suited for future millimeter-wave 5G and beyond 5G wireless communication systems, where a
high antenna gain is required in combination with electronic beamsteering.
In this chapter, we will first investigate the general properties of phased arrays, starting with
linear arrays of isotropic radiators. Then, we will extend the theory to two-dimensional arrays.
As a next step, we will introduce real antenna elements, like dipoles and microstrip antennas [25].
We will show that mutual coupling between the elements can limit the performance of the array.
Finally, we will investigate the effect of errors, noise and present a method to calibrate the array
to compensate for errors. In this chapter we will again assume to have time-harmonic signals and
waves.
6.1 Linear phased arrays of isotropic antennas

6.1.1 Array factor
We will investigate the basic principle of operation of linear arrays by considering the array in
receive mode, as illustrated in Fig. 6.3. Due to reciprocity, the behaviour of the array antenna
141
142 CHAPTER 6. PHASED ARRAYS AND SMART ANTENNAS
Figure 6.1: Example of a phased-array radar operating at X-band frequencies, courtesy: Thales
Netherlands.
Figure 6.2: Example of a large phased-array antenna used in radio astronomy, [24].
in transmit mode is exactly the same. However, in practical active phased-array systems, the
functionality of the transmit electronics may differ from the receive electronics. The linear array
of Fig. 6.3 consists of K identical antenna elements with k = 1....K, separated by an element
spacing of dx . Element k = 1 is located in the origin of the coordinate system at (x, y, z) = (0, 0, 0).
In this section, we will assume that all antenna elements are isotropic (omni-directional) radiators.
Although isotropic radiators cannot exist in reality, they allow us to investigate the basic array
properties. Each antenna element k includes a circuit which can perform a complex weighting
of the received signal sk , by multiplication with the complex coefficient ak = |ak |e−ψk . This
complex coefficient consists of an amplitude |ak | (taper) and a phase shift −ψk . The set of
complex coefficients is also often called the array illumination. In general the complex weighting
6.1. LINEAR PHASED ARRAYS OF ISOTROPIC ANTENNAS 143
will be realized by means of an electronic circuit and can be done both in the analog and digital
domain. Finally, all signals are summed in the summing network.
z
θ
Antenna
element
1 2 K-1 K
dx dx
s1 s2 sK-1 sK
|a1|exp(-jψ1) |a2|exp(-jψ2) |aK-1|exp(-jψK-1) |aK|exp(-jψK)
SUMMING NETWORK
S
Figure 6.3: Linear phased-array antenna consisting of K identical antenna elements, separated by
a distance dx . The signals received by each of the antenna elements are multiplied by a complex
coefficient with amplitude |ak | and phase −ψk .
Now let us investigate the receive properties of a linear array of isotropic antennas in more
detail. We will assume that a time-harmonic electromagnetic plane wave is incident on the array
from an angle θ0 with respect to the z-axis. This plane wave can be assumed to be generated by
another (transmit) antenna located far away from the array. The received signal of array element
k can now be written as:
sk = ek0 (k−1)dx sin θ0 , (6.1)

2π
where we have normalized the received signal s1 of element 1 to be equal to 1. In addition, k0 =
λ0
is the free-space wavenumber. The received signals at each array element can be physically
interpreted as the voltage along the output port of each antenna element or can represent the
(complex) amplitude of the guided wave along the transmission line which is connected to the
antenna element, for example in case of a waveguide or a microstrip transmission line. Since the
incident plane wave illuminates the array with an angle θ0 , a phase difference will occur between
the signals received by each of the array elements. This phase difference could also be interpreted
as a time delay, since element k will receive the signal a fraction τ earlier as compared to array
element (k − 1), where τ is given by:
dx sin θ0
τ= , (6.2)
c
where c = 3.108 is the speed of light in free-space. By using the superposition concept of electro-
magnetic waves we can write the total received signal at the output of the array of Fig. 6.3 in
the following form:
K
|ak |e[k0 (k−1)dx sin θ0 −ψk ] .
X
S(θ0 ) = (6.3)
k=1
The function S(θ0 ) is called the array factor. The array factor is the response of the array of
Fig. 6.3 on an incident plane wave that hits the array surface under the angle θ0 , expressed in
the coefficients ak = |ak |e−ψk with k = 1...K − 1. The array factor is a periodic function of sin θ0
with a period of λ0 /dx .
A similar analysis could be done for an array in transmit mode. Besides some constants
and the factor e−k0 r /r, we would obtain a similar expression as (6.3) for the electromagnetic field
far away from the array (far field).
The maximum of the array factor occurs when the applied phase shift −ψk by the electronics
connected to each of the receiving antenna elements exactly compensates the phase of each of the
received signals. This occurs when:
ψk = k0 (k − 1)dx sin θ0 . (6.4)
By changing the set of applied phases ψk , we can manipulate the direction θ0 for which the array
is maximized. In other words, by changing the set of phases ψk , we can electronically scan the
beam of the array. Because of this, this type of antenna is called a phased array.
Let us now assume that we use a set of excitation coefficients ak of which the phase ψk is
given by (6.4). The resulting array will have a maximum sensitivity when incident plane waves
hit the array with an angle of θ = θ0 w.r.t. to the z-axis. Let us now introduce the variables u
and u0 according to:
u = sin θ
(6.5)
u0 = sin θ0
If we now assume that a plane wave is incident on the array under an angle θ, we can re-write
the array factor (6.3) by means of the following expression
K
X
S(u) = |ak |e[k0 (k−1)dx (sin θ−sin θ0 )]
k=1 (6.6)
XK
= |ak |e[k0 (k−1)dx (u−u0 )] .
k=1
6.1.2 Uniform amplitude tapering

In case of a uniform amplitude weighting (tapering), where |ak | = 1 for k = 1...K, we can express
the summation of (6.6) in terms of an analytical expression. By introducing β = k0 dx (u − u0 ),
we can re-write the array factor:
K
X
S(β) = e(k−1)β . (6.7)
k=1
By multiplying both left- and right-side of (6.7) with eβ we obtain:

K
X K
X
β β (k−1)β
S(β)e =e e = ekβ . (6.8)
k=1 k=1
By subtracting the relation (6.7) from (6.8), we get:

S(β) eβ − 1 = eKβ − 1. (6.9)
In other words:
" #
eKβ − 1
S(β) =
eβ − 1
" #
(K−1)β/2 eKβ/2 − e−Kβ/2 (6.10)
= e
eβ/2 − e−β/2
sin(Kβ/2)

(K−1)β/2
= e .
sin(β/2)
Expressed in u coordinates, we then finally find the following expression for the array factor of
an array with uniform tapering:
sin(Kk0 dx (u − u0 )/2)

S(u) = e (K−1)k0 dx (u−u0 )/2
. (6.11)
sin(k0 dx (u − u0 )/2)
The antenna radiation pattern can by found by normalizing the received power w.r.t. its maximum
value, according to the definition introduced in chapter 2. When we apply this to our array factor,
we obtain the following expression for the normalized (power) radiation pattern of the linear array:
2
|S(u)|2 sin(Kk0 dx (u − u0 )/2)

= . (6.12)
|S(u)|2max K sin(k0 dx (u − u0 )/2)
For small values of the term k0 dx (u − u0 )/2, this reduces to
2
|S(u)|2 sin(Kk0 dx (u − u0 )/2)

≈ . (6.13)
|S(u)|2max Kk0 dx (u − u0 )/2
Expression (6.13) is exactly equal to the radiation pattern of a uniformly illuminated continuous
line source of length L = Kdx . This can easily be verified by comparing (6.13) with expression
(4.175), which resembles the radiation pattern of a uniform aperture.
As an example, let us investigate a linear array with uniform tapering consisting of K = 16
isotropic antenna elements, separated by a distance dx = λ0 /2. The normalized radiation pattern
can be found by using equation (6.12). Fig. 6.4 shows the radiation pattern in case the main
beam of the array is directed towards the angle θ0 = 00 . Fig. 6.5 shows the radiation pattern
when the main beam is scanned towards θ0 = 400 . The first sidelobe of the radiation pattern
is located at a level of -13.2 dB below the main lobe level. The height of the first sidelobe is
directly related to our choice of a uniform tapering. It can be shown that the 3 dB beam width
0.8858λ0
of a linear array with uniform tapering for large values of K equals: θHP ≈ [35]. With
Kdx cosθ0
K = 16, dx = λ0 /2 and θ0 = 00 , we obtain θHP ≈ 6.30 . The exact 3 dB beam width in Fig. 6.4
is θHP = 6.20 .
Array pattern of 16−element uniform array with λ0/2 spacing

0
−5
−10
Normalised Array Pattern [dB]
−15
−20
−25
−30
−35
−40
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 6.4: Normalized radiation pattern according to (6.12) of a uniformly illuminated linear
array of 16 isotropic antennas. The main beam is located at broadside (θ = 00 ). The element
spacing is dx = λ0 /2.
Array pattern of 16−element uniform array with λ0/2 spacing

0
−5
−10
Normalised Array Pattern [dB]
−15
−20
−25
−30
−35
−40
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 6.5: Normalized radiation pattern according to (6.12) of a uniformly illuminated linear
array of 16 isotropic antennas. The main beam is located at an angle of θ = 400 . The element
spacing is dx = λ0 /2.
6.1.3 DFT/FFT relation

When we investigate expression (6.6) more carefully, we can recognize a Discrete Fourier Trans-
formatie (DFT). The DFT can be written in the following general form:
K
X 2π(k−1)(n−1)
F (n) = f (k)e K 1 ≤ n ≤ K. (6.14)
k=1
We can rewrite the array factor (6.6) in a similar form as (6.14) by choosing the sample points in
the u space according to:
λ0 (n − 1)
u − u0 = . (6.15)
dx K
For the specific case that the number of array elements K is a power of 2, the array factor simplifies
in to a Fast Fourier Transform (FFT). By using FFT processing, we can efficiently compute the
radiation pattern of very large phased arrays. FFT processing is also often used in the digital
back-end of phased array antennas with digital beamforming.
6.1.4 Grating lobes

Previously, we have already shown that the main beam of a linear array can be scanned to any
angle θ0 by applying a linear phase distribution along the array according to (6.4). The array
factor as defined in (6.6) is a periodical function of the variable u−u0 . In other words: besides the
main beam, the array factor can have more maxima in the visible space, the so-called grating lobes.
The existence and exact location of the grating lobes w.r.t. the main beam depends on the element
spacing dx . In most practical systems, dx will be chosen in such as way as to avoid the existence
of grating lobes in the visible space, where the visible space is defined by −1 ≤ u = sin θ ≤ 1.
A grating lobe will appear in the direction um when the argument of the exponent in expression
(6.6) can be written as an integer number of 2π
k0 (k − 1)dx (um − u0 ) = 2πm, (6.16)
2π
where m is an integer number and where k0 = . For a given scan angle of the main beam
λ0
u0 , a grating lobe will appear in the visible space when |um = 1| or θm = ±900 . The relation
between the element spacing and the maximum scan angle θ0max of the main beam of the linear
array without the appearance of a grating lobe is given by:
dx 1
≤ . (6.17)
λ0 1 + | sin θ0max |
A lot of applications require scanning over the entire visible space. By substituting θ0max = 900
in expression(6.17), we can observe that grating lobes are avoided when the element spacing
dx ≤ λ0 /2. This corresponds to the well-known Nyquist sampling criterium used in analog-to-
digital convertors (ADC).
6.1.5 Amplitude tapering and pattern synthesis

In section 6.1.2 we have shown that a linear array with uniform amplitude tapering, i.e. |ak | = 1
for all k, generates quite high sidelobes in the radiation pattern, corresponding to the behavior of
the sinc function. This results in a first sidelobe level of -13.2 dB w.r.t. the main beam power level.
We can reduce the sidelobe level of phased arrays considerably by using an amplitude weighting
along the array elements. The weighting function or amplitude taper needs to be chosen in such
a way that contributions of the edge elements to the total array factor are reduced. In other
words, the illumination of the edges is reduced. One of the well-known taper functions is the
cosine-power taper given by:
πxk Kdx

m
|ak | = h + (1 − h) cos , |xk | ≤ , (6.18)
(K − 1)dx 2
in case of an array along the x-axis, where xk represents the x-coordinate of array element k. Note
that we have assumed in this case that the center of the array is located at x = 0. The parameter
h allows us to use an offset (also known as pedestal). When h = 1, we again obtain a uniform
taper. Commonly used taper functions are the cosine taper (m = 1) and the cosinus-square taper
(m = 2). The corresponding normalized radiation patterns are shown in Fig. 6.6 and Fig. 6.7 in
case of an array with K = 16 elements and element spacing dx = λ0 /2. The cosine taper results
Array pattern of 16−element array with cosine (m=1) taper and λ0/2 spacing
0
Cosine taper
Uniform taper
−5
−10
−15
Normalised array pattern [dB]
−20
−25
−30
−35
−40
−45
−50
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 6.6: Normalized radiation pattern of a linear array with a cosine taper (m = 1) with h = 0
and K = 16 array elements. The highest sidelobe is found at a level of -23 dB. The main beam is
directed towards broadside θ = 00 . The element distance dx = λ0 /2. The corresponding radiation
pattern of an array with uniform tapering is also shown.
in a peak sidelobe level of approx. -23 dB and the cosine-square taper results in a peak sidelobe
level of -32 dB. This is a significant improvement as compared to a uniform taper. Other well-
known taper functions are the Taylor taper and the Dolph-Tchebycheff taper [35]. The Taylor
tapering results in a monotonic decrease of the sidelobe level away from the main beam, whereas
the Dolph-Tchebycheff taper creates a constant sidelobe level. An example of the Taylor taper is
Array pattern of 16−element array with cosine−squared (m=2) taper and λ0/2 spacing
0
Cosine taper
Uniform taper
−10
−20
Normalised array pattern [dB]

−30
−40
−50
−60
−70
−80
−80 −60 −40 −20 0 20 40 60 80
θ [deg]
Figure 6.7: Normalized radiation pattern of a linear array with a cosine-square taper (m = 2) with
h = 0 and K = 16 array elements. The highest sidelobe is found at a level of -32 dB. The main
beam is directed towards broadside θ = 00 . The element distance dx = λ0 /2. The corresponding
radiation pattern of an array with uniform tapering is also shown.
shown in Fig. 6.8 for a 16-element linear array with a spacing of dx = λ0 /2. The Taylor taper
provides low sidelobes with a peak sidelobe level of 30 dB in this case. As a side effect, the beam
width of the main beam increases, and as a result, the directivity will be lower as compared to
an array with uniform tapering, see also next section.
Figure 6.8: Normalized radiation pattern of a 16-element linear array with a Taylor taper with 30
dB peak sidelobe level. The main beam is directed towards broadside θ = 00 . The element distance
dx = λ0 /2. The corresponding radiation pattern of an array with uniform tapering is also shown.
One of the main benefits of phased arrays is that the pattern (main beam and sidelobes) can
be controlled by using specific sets for the complex weight of each of the array elements. Several
pattern synthesis methods are described in literature that control the main beam and/or sidelobe
level. An overview of these methods can be found in [36]. The synthesis method introduced by
Schelkunov [37] provides some physical background in sidelobe control of pencil-beam patterns.
The array factor of a linear array (6.6) can also be written as a polynomial of a new variable z,
resulting in the so-called Schelkunov polynomial:
S(z) = a1 + a2 z + a3 z 2 + .... + aK aK−1 , (6.19)
where z = ek0 dx sin θ and ak = |ak |e−k0 dx sin θ0 . Note that |z| = 1, so located on a unit-circle in
the complex z plane. The Schelkunov polynomial (6.19) can also be written in terms of a product
of K − 1 factors:
S(z) = aK (z − z1 )(z − z2 ).....(z − zK−1 ), (6.20)
where z1 , z2 , .... zK−1 represent the zeros of the polynomial. A zero in the polynomial corresponds
with a null-location in the radiation pattern of the linear array. As an example, we will consider
a linear array with K = 8 elements with spacing of d = λ0 /4 and main beam directed towards
broadside (θ0 = 0). Let us first consider a uniform amplitude tapering. Fig. 6.9 shows the
locations of the zeros in the Schelkunov polynomial plotted in the complex z-plane, together with
the radiation pattern. The null at θ = −300 in Fig. 6.9 corresponds with the following location
in the complex z-plane:
2π λ0
λ sin θ
z = ek0 dx sin θ = e 0 4
(6.21)
 π2 sin θ  π4
= e =e
The separation between the nulls of a uniformly illuminated array in Fig. 6.9 is equidistant. If
we want to create low sidelobes, we should try to put the nulls closer together and move the first
nulls away from the main-lobe. This is done when using a Taylor of Dolph-Chebychev tapering.
This is illustrated in Fig. 6.10, where the null location of a Taylor taper with -20 dB peak sidelobe
is compared to the uniformly illuminated array. The corresponding radiation pattern and taper
function along the array is provided in Fig. 6.11. By comparing the location of the nulls with Fig.
6.11, it can be observed that the nulls have indeed shifted away from the main beam and are now
located closer together. In addition, the beam width of the main beam has broadened, which will
result in a reduction of the directivity as we will see in the next section. Low sidelobes can also
be obtained by choosing a specific (often randomized) array grid. A special case is discussed in
[38], where a sequential rotation technique is used to realize circularly-polarized arrays with low
sidelobes.
6.1.6 Directivity and taper efficiency

The directivity of an array directly determines the range of a wireless link or radar system as
stated by the radio- and radar equation (see chapter 2). The directivity of a linear array of
Receive patterns at f=12 GHz phi=0 deg 16-Apr-2016

Im(z) 0
-5
zeros
-10
-15
Array pattern [dB]

Main beam -20
-25
-30
Re(z)
-35
-40
-45
-50
-80 -60 -40 -20 0 20 40 60 80
Theta [deg]
Figure 6.9: Zeros of the Schelkunov polynomial plotted in the complex z-plane and relation to the
nulls in the radiation pattern. Linear array with uniform illumination (|ak | = 1) and with K = 8
elements spaced d = λ0 /4. The main beam is directed towards broadside (θ0 = 0).
Uniform taper Taylor taper
Im(z) Im(z)
zeros zeros
Main beam Main beam
Re(z) Re(z)
Figure 6.10: Zeros of the Schelkunov polynomial plotted in the complex z-plane for a uniform and
Taylor taper. Linear array with K = 8 elements with spacing of d = λ0 /4 and main beam directed
towards broadside (θ0 = 0).
isotropic radiators can be determined by substituting the array factor (6.6) into equation (2.11).
Figure 6.11: Radiation pattern (left) and amplitude taper along the array (right) of a linear array
with K = 8 elements with spacing of d = λ0 /4 and main beam directed towards broadside (θ0 = 0).
In this way, we can obtain the directivity in the direction θ0 :
P (θ0 )
D(θ0 ) =
Pt /4π
4π|S(θ0 )|2 (6.22)
= Z ,
|S(θ)|2 dΩ
V∞
where the integration is done over the entire visible space. The integral can be solved by using a
new coordinate ξ, where ξ is defined as the opening angle with the positive y-axis in the y − z-
plane. The differential solid angle is now given by dΩ = cos θdθdξ. In case of a linear array
located along the x-axis, the array factor will not depend on the ξ coordinate. Let us assume
that the main beam of the array is directed towards broadside, so θ0 = 00 . By using u = sin θ
and du = cos θdθ, we can rewrite (6.22) in the following form:

K
!2
X
2 |ak |
k=1
D =
Z1
|S(u)|2 du
−1
K
!2
X
2 |ak |
k=1
=
Z1 X
K X
K
|ak ||al |ek0 dx u(k−l) du
−1 k=1 l=1 (6.23)
K
!2
X
2 |ak |
k=1
=
K X
X K Z1
|ak ||al | ek0 dx u(k−l) du
k=1 l=1 −1
K
!2
X
|ak |
k=1
= K X
K
.
X
|ak ||al |sinc[k0 dx (k − l)]
k=1 l=1
When k0 dx = mπ (which implies that dx = mλ0 /2), we find that sinc[k0 dx (k − l)] can only be
non zero when k = l. Therefore, the directivity can be written as:
K
!2
X
|ak |
k=1
D = . (6.24)
XK
|ak |2
k=1
As a result, we can observe that the directivity of a uniformly-illuminated linear array with
element spacing dx = mλ0 /2 is equal to D = K. Consider for example, an array with K = 1000
(isotropic) elements. The corresponding directivity is in this case D = 10 log10 (1000) = 30 dB.
When we apply an amplitude tapering with |ak | ≤ 1, the directivity will be reduced as
compare to the uniform case according to:
D = Kηtap , (6.25)
where the taper efficiency ηtap is given by

K
!2
X
|ak |
k=1
ηtap = . (6.26)
XK
K |ak |2
k=1
An example is shown in Fig. 6.12 for a linear array with K = 64 elements and a spacing of
dx = λ0 /2. The figure shows the radiation pattern in case of a Taylor taper with 40 dB peak
sidelobe as compared to the uniform case. The Taylor taper provides low sidelobes, but also a
directivity reduction of 1.2 dB (ηapp = 0.76). In other words, the effective antenna aperture has
reduced with 24% as compared to the uniformly-illuminated array.
Figure 6.12: Effect of Taylor amplitude tapering on the radiation pattern of a linear array with
K = 64 elements with element spacing of d = λ0 /4 and main beam directed towards broadside
(θ0 = 0). The taper efficiency ηapp = 0.76.
6.1.7 Beam broadening due to beam scanning

The beam width of an antenna has been defined in chapter 2. In this section, we will again
consider a uniformly illuminated linear array of isotropic radiators. The array factor is given by
expression (6.12). From this expression, we can observe that the array factor depends on the
coordinate u − u0 , where u0 is a fixed value corresponding to the direction of the main beam (scan
angle). Therefore, we can conclude that the array factor does not depend on the absolute value of
u. As a result, the 3 dB beam width ((i.e. the u value for which the amplitude of the array factor
√
is equal to 1/ 2 w.r.t. the maximum value) will also not depend on u. To conclude, the beam
width will be constant in the u-domain. However, when we express the array factor in terms of
θ, the situation is different since the relation between u and θ is given by:
u = sin θ, (6.27)
which implies that

du
dθ = . (6.28)
cos θ
As a result we can express a small difference ∆θ around θ0 in the following way:
∆u
∆θ ≈ . (6.29)
cos θ0
Assume that ∆u indicates the difference in u corresponding to the 3 dB beam width in the
expression of the array factor (6.12). The corresponding beam width expressed in θ will then
have a cos1θ0 dependence. As a result, we can write the 3 dB beam width in the following form:
θ0
θHP = . (6.30)
cos θ0
As expected, the beam of the array becomes broader when the main beam is scanned towards the
horizon (θ0 = ±900 ).
6.1.8 Phase shifters (PHS) versus true time delay (TTD)

Traditionally, phased-arrays use phase shifters (PHS) to steer the beam of the antenna. As an
alternative true-time delays (TTD) can be used. In order to understand the differences and
whether PHS or TTD should be used in an application, let us investigate the array factor in more
detail. The general form of the array factor is given by (6.6):
K K 2πf0
|ak |e (k−1)dx u−ψk
X X
S(u) = |ak |e[k0 (k−1)dx (u−u0 )] = c , (6.31)
k=1 k=1
where f0 is the frequency of operation, c is the speed of light and ψk is the applied phase shift
to array element k at the frequency f0 . In a PHS unit, the realized phase shift is constant
versus frequency. In a TTD unit, the phase shift is realized with transmission lines of a specific
length. As a consequence, the realized phase will depend linearly on frequency. An example of a
4-element beamformer using 2-bit TDU units is shown in Fig. 6.13. When we look carefully at
Figure 6.13: 4-element beamformer using TDU units including 4 dipole-like antenna elements.
The switches in the TDU units are realized with MEMS technology [39]
equation (6.31), we can observe that the required phase shift ψk should have a linear dependency
with frequency, since the term 2πf
c (k − 1)dx u also shows a linear behavior versus frequency. As
0
a consequence, PHS units can only be applied in phased-array systems with a relative small
operational frequency bandwidth. At least when a fixed scan angle is required. The effect of
so-called beam squinting is illustrated in Fig. 6.14, where the radiation pattern of a 64-element
linear array is shown at 10, 12 and 14 GHz. The center frequency of the array is f0 = 12 GHz at
which the optimal phase shift settings were chosen to scan the array towards 400 . The main beam
is clearly squinting to another direction when the frequency of operation is changed according to:
f u0
= ,
f0 u (6.32)
f0
u = u0 ,
f
where we assumed that f0 is the design frequency of the array (in this case 12 GHz) and f is the
frequency of interest. Note that beam squinting can also be useful: it is a way to realize beam
steering without the need of adjustable phase-shifters.
Figure 6.14: Beam squinting in a linear array of 64 elements. Phase-shifter settings are optimized
at the center frequency f0 = 12 GHz. By using TDU units, the main beam would remain fixed at
θ0 = 400 over the entire frequency range.
6.2 Planar phased arrays of isotropic radiators

The theory of linear arrays can be easily extended to planar arrays. Fig. 6.15 shows the geometry
of a two-dimensional planar array, in which the elements are placed on a rectangular grid. Let us
6.2. PLANAR PHASED ARRAYS OF ISOTROPIC RADIATORS 157
again first assume that all array elements are isotropic radiators. The element distance along the
x-axis and y-axis are given by dx and dy , respectively. j denotes the element index according to
j = (l − 1)K + k. The position of the j-th element in the array is denoted by ~rj according to:
~rj = xj ~ux + yj ~uy

(6.33)
= (k − 1)dx ~ux + (l − 1)dy ~uy .
We will again investigate the array in receive-mode first. Now let us assume that a plane wave
illuminates the array from the direction (θ, φ), denoted by the unit vector ~ur . The array factor
can now be obtained by summing all received contributions from the array elements according to:
K X
L
|aj |e[k0 ~ur ·~rj −ψj ] ,
X
S(θ, φ) = (6.34)
k=1 l=1
where
aj = |aj |e−ψj , j = (l − 1)K + k, (6.35)
represents the excitation coefficient of element j with phase ψj and amplitude |aj |. The inner
product (~rj · ~ur ) can be found by
~ur · ~rj = [(k − 1)dx ~ux + (l − 1)dy ~uy ] · [sin θ cos φ~ux + sin θ sin φ~uy + cos θ~uz ]
(6.36)
= (k − 1)dx u + (l − 1)dy v.
with u and v given by
u = sin θ cos φ,
(6.37)
v = sin θ sin φ.
As a result, we can rewrite the array factor (6.34) in terms of (u, v)-coordinates according to
K X
X L
S(u, v) = |aj |e[k0 (k−1)dx u+k0 (l−1)dy v−ψj ] . (6.38)
k=1 l=1
When we want to direct the main beam of the array towards (θ0 , φ0 ), the phase shift ψj at each
array element should be chosen as:
ψj = k0 (k − 1)dx u0 + k0 (l − 1)dy v0 , (6.39)
with u0 = sin θ0 cos φ0 and v = sin θ0 sin φ0 . Substituting (6.39) in (6.38), results in the expression
for the array factor of a planar array with the main beam directed towards (u0 , v0 ):
K X
X L
S(u, v) = |aj |ek0 [(k−1)dx (u−u0 )+(l−1)dy (v−v0 )] . (6.40)
k=1 l=1
dy
dx − jψ j
aj e
Figure 6.15: Planar phased-array antenna consisting of K × L elements separated by a distance

dx along the x-axis and dy along the y-axis. The element index is j = (l − 1)K + k. The signals
received by each antenna element are fed into an electronic receiver that takes care of the amplitude
weighting |aj | and phase shifting −ψj of the received signal.
Now consider a uniformly illuminated array with |aj | = 1, for all j. The array factor can now
be expressed in terms of a product of two array factors corresponding to two linear arrays with
uniform illumination:
K L
! !
X X
k0 (k−1)dx (u−u0 ) k0 (l−1)dy (v−v0 )
S(u, v) = e e
k=1 l=1
= e(K−1)k0 dx (u−u0 )/2 e(L−1)k0 dy (v−v0 )/2 (6.41)

" #
sin(Kk0 dx (u − u0 )/2) sin(Lk0 dy (v − v0 )/2)

× .
sin(k0 dx (u − u0 )/2) sin(k0 dy (v − v0 )/2)
Fig. 6.16 shows the normalized radiation pattern of a planar array consisting of 64 isotropic
antennas (K = L = 8). The element spacing is equal to λ0 /2 in both directions. The main
beam is scanned towards (θ0 = 400 , φ0 = 00 ). We can clearly observe the beam broadening of
the main beam and of the first sidelobes. The peak value of the first sidelobe is again -13.2 dB,
corresponding to the sinc function in expression (6.41). Lower sidelobe levels can be obtained by
applying amplitude tapering, in a similar way as done for linear arrays.
In our analysis of linear arrays, we have already observed that multiple maxima (grating
lobes) can occur in the array factor. This happens when the distance between the array elements
is too large. Grating lobes in a planar array occur when the argument in the exponent of (6.40)
6.3. ARRAYS OF NON-IDEAL ANTENNA ELEMENTS 159
Figure 6.16: Normalized radiation pattern of a uniformly illuminated planar array of isotropic
antennas with 8 × 8 = 64 elements. The main beam of the array is directed towards (θ0 =
400 , φ0 = 00 ). The element spacing is dx = dy = λ0 /2.
is a multiple of 2π. As a result, grating lobes will appear in the directions indicated by (um , vm )
when the following condition is fulfilled:
k0 [(k − 1)dx (um − u0 ) + (l − 1)dy (vm − v0 )] = 2πm, (6.42)
for integer values of m, where m = 0 corresponds to the main beam. In case of a rectangular
array grid, the grating lobes will occur along a rectangular grid in the (u, v)-plane given by:
i
u − u0 = i = 0, ±1, ±2, ....
dx /λ0
(6.43)
j
v − v0 = j = 0, ±1, ±2, ....
dy /λ0
The rectangular grid of grating lobes in the (u, v)-plane is illustrated in Fig. 6.17 in case of an
dx dx
array with = 0.5 and = 1 with the main beam directed towards broadside u0 = v0 = 0. We
λ0 λ0
can clearly see that all grating lobes appear outside the unit circle. As a result, no grating lobes
will appear in the visible space. In case of scanning, the whole pattern will shift according to the
λ0
arrows as indicated in Fig. 6.17. In case 2, the element spacing is larger than . As a result,
2
grating lobes will appear in the visible space. In most applications grating lobes will deteriorate
the performance of the antenna system.
6.3 Arrays of non-ideal antenna elements

Up to now, we have assumed that the individual array elements are ideal isotropic radiators with
omni-directional radiation characteristics. As already discussed before, isotropic radiators can
physically not exist. Therefore, we need to introduce physical antenna structures in our array
theory. However, most of the array features that we have already discussed in this chapter will
Figure 6.17: Grating lobe pattern in the (u, v)-plane of a planar array placed on a rectangular grid
dx dy dx dy
with: case 1 = = 0.5 and case 2 = = 1. The blue dot shows the main beam scanned
λ0 λ0 λ0 λ0
towards broadside (u0 = v0 = 0). The entire pattern will shift according to the direction of the
arrows in case of scanning.
not change significantly by using real antenna elements. Several types of antennas can be used as
a building block in a phased array system, such as dipole antennas (usually with a length of λ0 /2)
and microstrip antennas. These individual antennas should have a broad beam width, else the scan
range of the phased array will be very limited. More details about the radiation characteristics
of various types of antenna can be found in other chapters. An example of a realized array for
radio-astronomy is shown in Fig 6.18, where a 64-element planar array of bow-tie antennas is
shown [24].
Now let us assume that the far-field pattern of an isolated array element is given by f~(u, v),
which can physically be interpreted as the electric field generated by an isolated element. The
radiation pattern of an individual array element f~(u, v) is known as the element factor or element
pattern. Let us for now assume that the radiation characteristics of all array elements are identical.
We can then simply apply the superposition principle in order to find the far-field of a planar
array of real antenna elements:
K X
L
~ v) = |aj |f~j (u, v)ek0 [(k−1)dx (u−u0 )+(l−1)dy (v−v0 )] .
X
S(u, (6.44)
k=1 l=1
Since we have assumed here that all array elements are identical with f~j = f~, we can rewrite
(6.44) in the following form:
K X
L
~ v) = f~(u, v)
X
S(u, |aj |ek0 [(k−1)dx (u−u0 )+(l−1)dy (v−v0 )] . (6.45)
k=1 l=1
6.4. MUTUAL COUPLING 161
Figure 6.18: The one-square metre array (OSMA), a pathfinder for the square kilometer array
(SKA) radio telescope [8], [24]. OSMA consists of 64 bow-tie elements with a bandwidth of approx.
an octave from 1-2 GHz.
The normalized radiation pattern can be determined from 6.45 by:

~ v)|2
|S(u,
F (u, v) = . (6.46)
~ 0 , v0 )|2
|S(u
6.4 Mutual coupling

In the previous sections we have assumed that mutual electromagnetic interaction between the
individual array elements can be neglected. However, in practise the array elements will be placed
on an array grid with a typical element spacing of only λ0 /2. Since most antenne elements only
operate at resonance, the physical size of the antenna elements will not differ that much from the
element spacing. As a result, the individual antenna element will interact with each other, causing
so-called mutual coupling. Mutual coupling will affect the element pattern of each array element
and will have an influence on the input impedance. Since each antenna element will observe
a (slightly) different electromagnetic environment, the element pattern and input impedance of
each element will be different. A rigorous electromagnectic analysis of mutual coupling in phased-
arrays is outside the scope of this book. Numerical electromagnetic methods (e.g. MoM, see
chapter 5) can be used to determine the mutual coupling. In addition, in [3] a method is provided
to analyze mutual coupling in finite arrays of microstrip antennas.
In this book we will restrict ourselves to the introduction of the key performance parameters
that describe the effect of mutual coupling. We will use the concept of microwave networks (see
chapter 3) in which we will use propagating waves along transmission lines instead of currents
and voltages. Let us consider the microwave network of Fig. 6.19. This black-box representation
is an equivalent microwave network of a phased-array antenna consisting of K × L elements (see
also Fig. 6.15). A forward wave will propagate with complex amplitude aj along the transmission
line connected to the input port of each antenna element j. In addition, a reflected wave with
complex amplitude bj will propagate along the same transmission line in the opposite direction.
The complex amplitudes of both the forward and reflected waves can be directly measured with
a vector network analyzer (VNA). When we consider the array in transmit mode, the reflected
signal, represented by bj , is unwanted since it reduces the total signal that is radiated by array
element j. As an example to further explore the concept, we will consider a very small array with
Microwave network
a1p b1p a jp b jp aKp ×L bKp×L
Figure 6.19: Microwave network representation of a phased-array antenna consisting of K × L

antenna elements.
only two antenna elements, so K = 2 and L = 1. The relation between the forward and reflected
waves at the input of the network can be written in the following form:
    
 b1   S11 S12   a1 
    
 =  , (6.47)
    
    
b2 S21 S22 a2
where [S] is the de scattering matrix. Note that S11 represents the reflection coefficient at port 1
when port 2 is terminated with a matched load impedance, also known as the reference impedance.
In practise the reference impedance will often by equal to 50Ω. Similarly, S22 is the reflection
coefficient at port 2 when port 1 is terminated with a matched load impedance. The coefficient S21
is the coupling coefficient between antenna element 1 and 2. It is a result of the electromagnetic
interaction between both antennas. Since array antennas are passive, reciprocity applies. As a
result S21 = S12 . The coupling coefficient can be obtained by using
b2
S21 = |a =0 . (6.48)
a1 2
In case the array is used in transmit mode, the reflected signal at port 1 will include two compo-
nents according to
S11 a1 + S12 a2
R1act = , (6.49)
a1
6.4. MUTUAL COUPLING 163
where a1 and a2 represent the excitation coefficients which have already been introduced in (6.35).
Since a1 and a2 both depend on the scan angle of the main beam of the array (θ0 , φ0 ), the resulting
reflection coefficient R1act will also depend on the scan angle. Therefore, R1act is referred to as the
active reflection coefficient of array element 1. Similarly, R2act is the active reflection coefficient of
array element 2. From (6.49), we can observe that a large coupling coefficient S12 can significantly
deteriorate R1act and R2act Note that in practise, the active reflection coefficients not only depend
on the scan angle, but also on frequency.
For an array consisting of K × L elements, the active reflection coefficient of array element
j is given by
1 K×L
X
Rjact (u0 , v0 ) = Sij ai (u0 , v0 ), (6.50)
aj i=1
where the excitation coefficients ai with i = 1, 2, ... are given by (6.35). Since part of the incident
power will be reflected, a power loss equal to Lj = 1 − |Rjact |2 will occur. Lj is the scan loss and
can be interpreted as a reduction of the antenna gain of array element j. The element factor in
6.45 should then be replaced by:
r
2
f~j (u, v) = f~(u, v) 1 − Rjact (u0 , v0 ) . (6.51)
We can also define the active input impedance of the j-th array element as:
K×L
Vj Ii
Zjin,act (u0 , v0 ) =
X
= Zji , (6.52)
Ij i=1
Ij
where Vj and Ij are the voltage and current at the input port of the j-th element and where Zji
is the mutual impedance between array element j and i. The relation between the active input
impedance and the active reflection coefficient is now given by:
Zjin,act (u0 , v0 ) − Z0
Rjact (u0 , v0 ) = , (6.53)
Zjin,act (u0 , v0 ) + Z0
where Z0 is the port impedance, usually Z0 = 50Ω.

The effect of mutual coupling on the performance of phased arrays can be best illustrated
with some examples. Let us first consider a large X-band phased-array radar with 4096 open-ended
waveguide radiators as presented in [40]. The measured coupling coefficients of a small sub-array
of 127 elements near the center of the array and the associated active reflection coefficient are
provided in Fig. 6.20. It can be observed that the active reflection coefficient depends on the scan
angle, but remains quite low up to a scan angle of +/ − 600 in both planes. The measured element
patterns in the H-plane (φ = 00 ) are provided in Fig. 6.21. The element patterns fluctuate around
a cos θ pattern, resulting in about 3 dB of gain reduction at a scan angle of θ0 = 550 . Another
example of the effect of mutual coupling is shown in Fig. 6.22. Due to a periodic effect in this
array of 1024 folded-dipoles, a so-called blind scan angle occurs at a scan angle of θ0 = 600 . In
this case, all power is reflected and as a result, the directivity will be very low. More details can
be found in [41].
H-plane (φ0=00)
y
l
x
E-plane (φ0=900)
k
Sij [dB] ∠Sij [deg]
Figure 6.20: Measured coupling coefficients (left) of the center part of 127 array elements and
the corresponding active reflection coefficient (right plots) in the H-plane (φ0 = 00 ) and E-plane
(φ0 = 900 ). The coupling coefficients between the center element (k = 4, l = 9) and the other
elements is shown. More details on the measured data can be found in [40].
Center-element element pattern Measured element patterns of 64 elements, H-plane, f=f0
Figure 6.21: Measured and simulated element pattern of the center element (left) and measured
element patterns (right) of the center part of an array with 127 array elements in the H-plane
(φ0 = 00 ). More details can be found in [40].
6.5. THE EFFECT OF PHASE AND AMPLITUDE ERRORS 165
Figure 6.22: Photo and the measured element pattern of the center element of a 1024 phased-array
of folded dipoles operating at L-band frequencies. More details can be found in [41].
6.5 The effect of phase and amplitude errors

In this section we will investigate the effect of phase and amplitude errors on the radiation
characteristics of phased arrays. Errors in phase and amplitude at the output of each array element
will occur due to non-perfect electronics, mechanical tolerances and environmental changes, like
a change in temperature. Let us first consider a linear array of isotropic radiators with uniform
amplitude distribution (|ak | = 1). In addition, we will assume that the main beam of the array is
directed towards broadside (θ0 = u0 = 0). This configuration is shown in Fig. 6.3. Without any
errors the array factor is now given by (see also (6.6))
K
X
S(u) = ek0 (k−1)dx u . (6.54)
k=1
Let us now first determine the effect of phase errors. Lateron, we will generalize our analysis to
include amplitude errors as well. Phase errors will affect the main beam (directivity, beam width)
and the sidelobes.
As a first step we will investigate the main beam. Assume that the applied phase of each
element has an error, say δk according to a Gaussian or uniform probability density function with
mean value 0 (since we are considering the non-scanning case here first). The array factor at
u = 0 can now be written as:
K K K
e−δk =
X X X
Smδ = ek0 (k−1)dx u−δk |u=0 = [cos δk −  sin δk ] . (6.55)
k=1 k=1 k=1
Assume that δk is small, so the sum over the sin δk terms can be neglected, resulting in:
K K K
1 2 1X
X X
Smδ = cos δk ≈ 1 − δk = K − δ2, (6.56)
k=1 k=1
2 2 k=1 k
where we have applied a Taylor expansion of cos δk . The main beam average complex power is
given by:
K
!2
2 1X
|Smδ | = K− δ2
2 k=1 k
(6.57)
KK
1X X
= K2 − K δk2 + δk2 δl2 ,
k=1
4 k,l
Assume that the variance of δk of all elements is equal δk2 = δ 2 . This implies that for large K and
small δ 2 :
δk2 δl2 = δ 4 4Kδ 2 ,

K
|Smδ |2 = K 2 − K
X
δk2 (6.58)
k=1

= K 2 1 − δ2 ,
since the last term in (6.57) can be neglected.

As a second step, we will determine the effect of phase errors on the sidelobes. Let us
assume that we are at a null of the error-free radiation pattern. For the non-scanning case the
array factor is now written as:
K K
ek0 (k−1)dx u e−δk =
X X
Ssδ = eβk [cos δk −  sin δk ] , (6.59)
k=1 k=1
with βk = k0 (k − 1)dx u. Since we are at a null of the error-free pattern, the cos δk term vanishes:
K
X K
X
Ssδ = = − eβk sin δk ≈ − eβk δk . (6.60)
k=1 k=1
We have now transformed the phase-error in an amplitude error occuring in the error-free array
pattern. The average power at a null of a sidelobe now becomes:
"K #" K #
2
X X
|Ssδ | = eβk δk e−βl δl
k=1 l=1
K
(6.61)
X X
= δk2 + δk δl e(βk −βl )
k=1 k6=l
= Kδ 2 ,
with again δk2 = δ 2 and we have assumed that δk and δl are uncorrelated.
The average normalized sidelobe power due to phase errors at a null of the error-free array
pattern is now given by:
|Ssδ |2 Kδ 2 δ2
Rp2 = = ≈ . (6.62)
|Smδ |2 K 2 1 − δ2 K
6.5. THE EFFECT OF PHASE AND AMPLITUDE ERRORS 167
From (6.62) we can conclude that large arrays can tolerate much larger phase errors for a given
required sidelobe level. The is due to the fact that we have assumed the errors to be uncorrelated.
The directivity and associated antenna gain will be reduced in case of phase errors. Starting
from the definition of the directivity (6.22), we can determine the directivity in case of errors De
of a linear array with uniform illumination with u0 = θ0 = 0 as:
P |Smδ |2
De = =
Pt /4π |Sideal (u)|2 + |Ssδ (u)|2

K 2 1 − δ2 1 − δ2 (6.63)
= =K
K + Kδ 2 1 + δ2
1
≈ K .
1 + δ2
In (6.63), the average radiated power |Sideal (u)|2 of an error-free linear array with uniform tapering
(|ak | = 1) is equal to K. The normalized directivity loss then becomes:
De 1 (6.64)
= ,
D 1 + δ2
since the directivity of an ideal linear array with uniform tapering is D = K.
Electronic phase shifting is commonly realized using discrete settings. Let us consider a
phase shifter with P bits. The total number of settings is 2P . The incremental phase step
[radians] is:
2π (6.65)
∆ψk = .
2P
Assuming an uniformly-distributed random phase error in the range [−∆ψk /2, ∆ψk /2], we find
that the variance is given by:
2
2π
2P π2 (6.66)
δ2 = = .
12 3(22P )
Example: 4-bit phase shifter.

Let us consider a 4-bit phase shifter used in a linear array of K = 64 elements with uniform
tapering. The phase step is 22.50 and the corresponding maximum phase error is now ±11.250 .
We now find that the variance is given by:
π2 (6.67)
δ2 = = = 0.0129 [rad2 ],
3(22P )
which corresponds to an average phase error of 6.50 (rms). The average sidelobe level at a null of
the error-free pattern and the directivity loss and can be found by:
Kδ 2
Rp2 = = −36.9 [dB],
K 2 1 − δ2
(6.68)
De 1
= = −0.0557 [dB].
D 1 + δ2
The corresponding radiation pattern is shown in Fig. 6.23. The ideal pattern has low sidelobes
which is achieved by using a Taylor tapering with 30 dB peak sidelobe level.
Figure 6.23: Effect of phase errors using 4-bit phase shifters on the radiation pattern of a linear
array with uniform taper with K = 64. The ideal pattern is designed to have a peak sidelobe level
(SLL) of 30 dB using a Taylor taper with n = 8.
As a final step we can generalize to the case of a planar array with both amplitude and
phase errors and where the main beam is scanned towards any direction. Let us consider a planar
array with K × L elements with phase and amplitude error of element j given by δj and ∆j ,
respectively. We again will assume that the variance of the phase and amplitude errors of all
elements are identical and given by δ 2 and ∆2 , respectively. It can be shown that the average
sidelobe level at a null of the error-free pattern and the directivity loss are given by:
δ 2 + ∆2
Rp2 = ,
D0 (6.69)
De 1
= ,
D0 1 + δ 2 + ∆2
where D0 is the directivity of the array. Fig. 6.24 provides design guidelines to determine the
required amplitude and phase accuracy of the electronic circuits used in a phased array. Three
array sizes are compared with K = 64, 1024, 4096 elements.
6.6 Noise Figure and Noise Temperature

Let us first do a quick review of noise in microwave receivers by considering the configuration
shown in Fig. 6.25, see also section 3.14. A resistor is connected to the input of an amplifier. The
6.6. NOISE FIGURE AND NOISE TEMPERATURE 169
Figure 6.24: Design guidelines to determine the required accuracy of the phase shifter and variable
gain amplifier used in a phased array with K × L antenna elements. Random errors are assumed.
maximum noise power N delivered by a resistor R operating at the environmental temperature T0

is N = kb T0 B, where kb = 1.38 × 10−23 J/K is Boltmann’s constant and B [Hz] is the frequency
bandwidth of the system. The amplifier is characterized by its available power gain Gav and noise
Ni=kbT0B No
Gav ,F
Figure 6.25: Noise in microwave receivers.
figure F , which is defined as the ratio of the signal to noise ratio at the input versus the signal to
noise ratio at the output:
Si Si

N
i = b 0
kT B
F =
So Gav Si
No No (6.70)
No
= .
Gav kb T0 B
Therefore, the additional noise ∆Niamp added by the amplifier seen at the input of a noise-free
amplifier with available power gain Gav is now:
No − Gav Ni (6.71)
∆Niamp = = (F − 1)kb T0 B.
Gav
Now let us connect an antenna at the input of the receiver. We will assume that the noise
generated by the antenna can be considered to be white noise, so it would not depend on the
frequency within the bandwidth of interest. In this case, we can model the antenna noise as a
thermal noise source, equivalent to the case of a resistor at the input of the receiver. The noise
power Na from a single antenna element with antenna noise temperature Ta is now equal to
Na = kb Ta B. The antenna noise temperature can be determined by integrating the antenna gain
function of the antenna over the sky-noise distribution:
1
Z Z
Ta = T (Ω)G(Ω)dΩ, (6.72)
4π
where Ω is the solid angle with dΩ = sin θdθdφ. Note that Ta is often much lower than the
environmental temperature since we weigh the sky-noise temperature distribution T (Ω) with the
gain G(Ω) of the antenna. For example in satellite TV operating in the Ku-band (11-14 GHz), a
typical value is Ta = 100 K.
Now consider a phased array with K elements with uniform amplitude taper, as illustrated
in Fig 6.26. Each antenna element is connected to a low-noise amplifier with available power gain
Gav and noise figure F . The noise power received by a single antenna element k is Nka = kb Tka B.
Let us assume that Tka = Ta . The K : 1 power combiner coherently combines the antenna noise
z
θ
Antenna
element
1 2 K-1 K
d d
s1 s2 sK-1 sK
Gav ,F Gav ,F Gav ,F Gav ,F
Ideal noise-free K:1 power combiner

S
Figure 6.26: Array with K elements. Each antenna element is connected to a low-noise amplifier
with available power gain Gav and noise figure F .
power contributions from all elements, since (worst-case) the wanted signal and the antenna noise
could be co-located at the same spherical coordinate (u0 , v0 ). The total output noise power due
6.7. SPARSE ARRAYS 171
to the antenna noise is given by:
NaT = KGav Nka = kb Ta BKGav . (6.73)
The noise from the electronics at the input of the power combiner is given by:
Ne = kb T0 B(F − 1)Gav . (6.74)
The transmitted noise due to the electronics at the output of the power combiner now becomes:
kb T0 B(F − 1)Gav (6.75)

Neo = .
K
And as a result, the total noise at the output of the combiner will become:
NeT o = KNe0 = kb T0 B(F − 1)Gav . (6.76)
The signal power at the output of the combiner is:
So = KGav Sa , (6.77)
where Sa is the signal power received by a single array element. The signal-to-noise ratio at the
output of the combiner is therefore given by:
S So KGav Sa

= =
N out NaT T
+ Ne o kb Ta BKGav + kb T0 B(F − 1)Gav
(6.78)
Sa
= .
kb Ta B + kb T0 B(F − 1)/K
From (6.78) we can conclude that the noise performance of a phased array is identical to the
case of a single equivalent large antenna (e.g. a reflector antenna) with antenna gain equal to the
antenna gain of the phased array. The equivalent antenna is connected to a low-noise amplifier
with an available power gain Gav and noise figure F as illustrated in Fig. 6.27.
6.7 Sparse arrays

Most application of phased-arrays do not tolerate grating lobes. In radar applications, grating
lobes would create confusion about the position of the target and it would spread the radiated
power into multiple directions. In telecommunication applications, grating lobes could allow
strong interfering signals to saturate the receiver electronics. However, some applications could
allow grating lobes to appear. An example is radio astronomy. Sparse arrays are arrays of which
the inter-element distance (dx , dy ) is much larger than 0.5λ0 . Fig. 6.28 illustrates the basic
concept of sparse arrays. Sparse array can have a regular grid or irregular grid. Fig. 6.29 shows
the radiation pattern of a linear array with K = 32 elements placed on a dense versus sparse
regular grid. The sparse array clearly has a much narrower main beam. The price to pay is the
z
θ
Antenna Gain = K
Antenna
element
1 2 K-1 K
d d
s1 s2 sK-1 sK Gav ,F
Gav ,F Gav ,F Gav ,F Gav ,F
Ideal noise-free K:1 power combiner

S
Figure 6.27: Noise in a phased-array can be modelled by considering an equivalent single antenna
with the same antenna gain and connected to a single low-noise amplifier with available power
gain Gav and noise figure F . In case of an ideal array with uniform taper the antenna gain is
equal to K.
Figure 6.28: Dense regular versus sparse irregular arrays
appearance of several grating lobes. A way to remove the unwanted grating lobes in sparse arrays
is to use an irregular array grid. In this case the radiated power in the grating lobes is spread
out over the entire visible space. As a result, the average sidelobe level will be higher and will
depend on the amount of array elements. An example of a very efficient irregular sparse array
configuration is the sun-flower topology, as shown in Fig. 6.30. The average element distance
in both the regular and sun-flower configuration in Fig. 6.30 is about 2λ0 . Fig. 6.31 shows
the corresponding radiation patterns of both arrays. Clearly, the sunflower array suppresses the
grating lobes at the expense of a higher average sidelobe level. When the element distribution
in a sparse array is random or quasi-random, it can be shown that the average sidelobe level for
6.7. SPARSE ARRAYS 173
Figure 6.29: Radiation pattern of a dense regular array and sparse regular array. A linear array
of 32 elements with uniform tapering is assumed.
Figure 6.30: Sparse regular array (left) and sparse irregular array (sun-flower arrangement). Both
arrays have 289 elements and have an average element spacing of about 2λ0 .
large arrays with K × L elements, is approximated by:
1 (6.79)
SLL = .
KL
So in case of a sparse-random array with 1000 elements, the average SLL will be -30 dB below
the main beam gain.
Figure 6.31: Radiation pattern in the (u, v)-plane of the sparse regular and sun-flower array. Main
beam is scanned towards (θ0 , φ0 ) = (300 , 00 ). A uniform amplitude tapering is used. Courtesy
Yijing Zhang 2015.
6.8 Calibration of phased-arrays

Amplitude and phase errors will occur in phased-array systems. These errors can have a systematic
time-invariant nature, but can also be caused by changes in the environmental conditions, e.g.
temperature changes. Calibration must be used to correct for these errors. A schematic diagram
of a calibration concept which is often used in modern phased arrays [8] is shown in Fig. 6.32.
Two types of signals can be injected into each array element: an external sigal generated by a
Figure 6.32: Calibration principle used in phased arrays. Only element k is shown.
far-field source HkF F , or a local calibration signal Hkcal injected via a special calibration network
within the array or by means of injection via mutual coupling with a special calibration element,
as illustrated in Fig. 6.32. The first step in the calibration procedure is to relate the received
signals at the output of each array element from a test signal of one of the calibration elements
to the received signals due to one or more incident plane waves from a far-field source. This part
6.9. FOCAL-PLANE ARRAYS 175
of the calibration procedure is called off-line calibration, since it needs to be done only once in a
high-quality anechoic facility, e.g. by using a near-field scanner or compact antenna range. The
element calibration coefficients are calculated and stored in a look-up table. With ak the received
signal from the far-field source and bk the received signal from the calibration element, we obtain
the following calibration coefficients in our look-up table:
ak H F F Hkcom HF F (6.80)
ck = = kcal com = kcal .
bk Hk Hk Hk
The second step is to measure the amplitude and phase at the output of each array element due
to a signal from one of the calibration elements. This part is called on-line calibration, since it
can be done while the system is operation in the application environment. The measured signal
from the calibration element is now:
b̃k = Hkcal H̃kcom . (6.81)
By combining the results from both calibration steps, we can re-construct the signal transfer from
a far-field source to array element k:
!
HkF F
ãk = HkF F H̃kcom = Hkcal H̃kcom = b̃k ck . (6.82)
Hkcal
Phased arrays often consist of many elements, e.g. common values are 1024 and 4096 elements.
Therefore, the on-line calibration could take a lot of time. Several techniques have been pro-
posed in literature to overcome this problem. An example is the multi-element phase togging
(MEP) technique [8], in which the FFT principle is used to calibrate groups of array elements
simultaneously.
6.9 Focal-plane arrays

A major draw-back of phased-arrays is the large amount of individual antenna elements and
associated electronic circuits that is required, leading to high cost and high power consumption.
However, several applications do require a high antenna gain, but do not require a very wide
scan-range, instead these applications only use a limited field-of-view (FoV). In this case, the
hybrid concept of focal-plane arrays (FPAs) is very interesting. FPAs combine the benefits of
phased-arrays and traditional reflector-based solutions by offering a high antenna gain, relative
low costs and electronic beam scanning over a limited FoV. Therefore, FPAs have already become
an interesting alternative for conventional horn-fed reflector antennas in a number of applications,
e.g. in radio astronomy and in satellite communication. Moreover, emerging applications such as
point-to-point wireless communications, 5G new-radio millimeter-wave (mm-wave) wireless and
low-cost Ka-band multi-function radars could be areas in which FPAs can play a major role.
Fig. 6.33 shows the concept of a basic FPA system. A phased-array feed (PAF) is placed in the
focal-plane in front of a parabolic reflector antenna with diameter D and focal length F . When
an incident plane wave illuminates the reflector, the location of the beam focus moves along the
phased-array feed depending on the scan-angle. Therefore, for each scan angle a limited number of
array elements will be active at the same time. The antenna gain of a FPA is mainly determined
by the size of the reflector and not by the number of array elements in the PAF. In this way, a
high antenna gain can be realized with electronic beam-scanning capabilities. For large F/D ratio
(a) (b)
Figure 6.33: Focal-plane array (FPA) concept in which a phased-array feed is placed in the focal-
plane in front of a parabolic reflector antenna, from [42]. When an incident plane wave illuminates
the reflector, the location of the beam focus moves along the phased-array feed with varying angle
of incidence.
and small scan angles the electric field distribution in the focal plane of a prime-focus parabolic
reflector antenna can be approximated by:
2J1 (k0 r sin Ψ0 ) (6.83)

E(r) = ,
k0 r sin Ψ0
where J1 is the Bessel function of the first kind with order 1, Ψ0 = π/4, the subtended angle of
the reflector, k0 is the free-space wavenumber and r is the distance from the center of the focal
plane (aperture radius), see Fig. 6.33. The field distribution in the focal plane is illustrated in
the example of Fig. 6.34 in case of broadside incidence of a plane wave. The rings in the field
distribution are also known as Airy rings. We can clearly see that only a limited number of array
elements will be active simultaneously for a particular angle of incidence (scan angle). In case of
scanning, the pattern will shift along the focal plane. The aperture efficiency η as a function of
the radius of the PAF feed rp of the FPA can be determined by integrating the electric field in
the focal plane and normalize it to the total power Ptot obtained by the reflector:
Z2πZrp
1
η(rp ) = |E(r)|2 drdφ, (6.84)
Ptot
0 0
More details on the scanning capabilities and limitations can be found in [42], [43].
6.9. FOCAL-PLANE ARRAYS 177
(a) (b)
Figure 6.34: (a) Focal-plane array illuminated by a plane wave (broadside incidence), (b) Field
distribution in the focal plane according to (6.83) for a classical prime focus parabolic reflector
with F/D = 0.6 and f = 30 GHz. Dimensions in focal plane in [m]. Figure from [42].
Appendix A
Vector calculus and coordinate

systems
A.1 Vector algebra
~a · ~b = ~b · ~a,
~a · (~b × ~c) = ~b · (~c × ~a) = ~c · (~a × ~b),
~a × ~b = −(~b × ~a),
~a × (~b × ~c) = (~a · ~c)~b − (~a · ~b)~c.
A.2 Vector operations in rectangular (Cartesian) coordinates
~ = ~ux Ax + ~uy Ay + ~uz Az ,

A
∂Ψ ∂Ψ ∂Ψ
∇Ψ = gradΨ = ~ux + ~uy + ~uz ,
∂x ∂y ∂z
~ = divA
∇·A ~ = ∂Ax + ∂Ay + ∂Az ,
∂x ∂y ∂z
~ux ~uy ~uz

∂Az ∂Ay ∂Ax ∂Az
= ~ux ( − ) + ~uy ( − )
~ = curlA
∇×A ~= ∂ ∂ ∂ ∂y ∂z ∂z ∂x
∂x ∂y ∂z ∂Ay ∂Ax
+~uz ( − ),
∂x ∂y
Ax Ay Az
∂2 ∂2 ∂2
∇2 Ψ = + + Ψ,
∂x2 ∂y 2 ∂z 2
~ = ~ux ∇2 Ax + ~uy ∇2 Ay + ~uz ∇2 Az .
∇2 A
179
180 APPENDIX A. VECTOR CALCULUS AND COORDINATE SYSTEMS

uz

uy
 P
ux

r
y
Figure A.1: Rectangular coordinate system with ~r = x~ux + y~uy + z~uz .
A.3 Vector operations in cylindrical coordinates

uz

uφ
P

uρ
y
φ
ρ
Figure A.2: Cylindrical coordinate system with x = ρ cos φ, y = ρ sin φ, z = z.
∂Ψ 1 ∂Ψ ∂Ψ
∇Ψ = ~uρ + ~uφ + ~uz ,
∂ρ ρ ∂φ ∂z
~ = 1 ∂ (rAρ ) + 1 ∂Aφ + ∂Az ,
∇·A
ρ ∂ρ ρ ∂φ ∂z
A.4. VECTOR OPERATIONS IN SPHERICAL COORDINATES 181
1 ∂Az ∂Aφ ∂Aρ ∂Az 1 ∂ ∂Aρ

~ = ~uρ
∇×A − + ~uφ − + ~uz (ρAφ ) − ,
ρ ∂φ ∂z ∂z ∂ρ ρ ∂ρ ∂φ
1 ∂

∂Ψ
2
1 ∂ Ψ ∂ Ψ 2
∇2 Ψ = ρ + 2 + .
ρ ∂ρ ∂ρ ρ ∂φ2 ∂z 2
A.4 Vector operations in spherical coordinates

ur

uφ
P

uθ
θ r
y
φ
Figure A.3: Spherical coordinate system with x = r sin θ cos φ, y = r sin θ sin φ, z = r cos θ and
dS = r2 sin θdθdφ = r2 dΩ, where dΩ = sin θdθdφ and Ω is the solid angle.
∂Ψ 1 ∂Ψ 1 ∂Ψ
∇Ψ = ~ur + ~uθ + ~uφ ,
∂r r ∂θ r sin θ ∂φ
~ = 1 ∂ (r2 Ar ) + 1 ∂ 1 ∂Aφ
∇·A (sin θAθ ) + ,
r2 ∂r r sin θ ∂θ r sin θ ∂φ
1 ∂ ∂Aθ

~
∇ × A = ~ur (sin θAφ ) − +
r sin θ ∂θ ∂φ
1 ∂Ar 1 ∂ 1 ∂ ∂Ar

~uθ − (rAφ ) + ~uφ (rAθ ) − ,
r sin θ ∂φ r ∂r r ∂r ∂θ
1 ∂ ∂Ψ 1 ∂ ∂Ψ 1 ∂2Ψ

∇2 Ψ = 2 r2 + 2 sin θ + 2 2 .
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2
A.5 Commonly used vector products
~ur × ~uθ = ~uφ ,

182 APPENDIX A. VECTOR CALCULUS AND COORDINATE SYSTEMS
~ur × ~uφ = −~uθ ,

~uθ × ~uφ = ~ur .
~ur × ~ux = ~uφ cos θ cos φ + ~uθ sin φ,

~ur × ~uy = ~uφ cos θ sin φ − ~uθ cos φ,
~ur × ~uz = −~uφ sin θ.
~ur × ~ur × ~ux = ~uφ sin φ − ~uθ cos θ cos φ,

~ur × ~ur × ~uy = −~uφ cos φ − ~uθ cos θ sin φ,
~ur × ~ur × ~uz = ~uθ sin θ.
~ur · ~r0 = x0 sin θ cos φ + y0 sin θ sin φ = x0 u + y0 v, when ~r0 = (x0 , y0 , 0)

u = sin θ cos φ, v = sin θ sin φ.
~ux = ~ur sin θ cos φ + ~uθ cos θ cos φ − ~uφ sin φ,

~uy = ~ur sin θ sin φ + ~uθ cos θ sin φ + ~uφ cos φ,
~uz = ~ur cos θ − ~uθ sin θ,
~ur = ~ux sin θ cos φ + ~uy sin θ sin φ + ~uz cos θ,
~uθ = ~ux cos θ cos φ + ~uy cos θ sin φ − ~uz sin θ,
~uφ = −~ux sin φ + ~uy cos φ.
A.6 Vector identities
∇(Φ + Ψ) = ∇Φ + ∇Ψ,
~ + B)
∇ · (A ~ = ∇·A
~ + ∇ · B,
~
∇ × (A~ + B)
~ = ∇×A~ + ∇ × B.~
∇(ΦΨ) = Φ ∇Ψ + Ψ∇Φ,
~ = Φ∇ · A
∇ · (ΦA) ~+A~ · ∇Φ,
~ = Φ∇ × A
∇ × (ΦA) ~−A~ × ∇Φ.
∇(A~ · B)
~ = (B
~ · ∇)A
~ + (A
~ · ∇)B~ +B~ × (∇ × A)
~ +A ~ × (∇ × B),
~
~ × B)
∇ · (A ~ = B
~ · (∇ × A)
~ −A ~ · (∇ × B),
~
∇ × (A~ × B)
~ = (B
~ · ∇)A
~ − (A
~ · ∇)B~ −B~ (∇ · A)
~ +A~ (∇ · B).
~
~ = ∇∇ · A
∇2 A ~ − ∇ × ∇ × A,
~
∇ · ∇Ψ = ∇2 Ψ,
∇·∇×A ~ = 0,
∇ × ∇Φ = ~0.
Bibliography
[1] M. E. J. Jeuken,
“Elektromagnetische antennes I: de grondslagen,”
Lecture notes (TUE Dictaat), second edition, in Dutch, 1988.
[2] R. E. Collin en F. J. Zucker,

“Antenna theory: part 1 and part 2,”
McGraw-Hill, 1969
[3] A. B. Smolders,
“Microstrip phased-array antennas: a finite array approach ,”
Ph.D. Thesis, Eindhoven University of Technology, 1994.
[4] H. Hertz
“Gesammelte werke von Heinrich Hertz,”
Leipzig: J.A. Barth, 1895.
[5] D. Marconi,
“My father, Marconi,”
London: McGraw-Hill, 1962.
[6] G. A. Deschamps,
“Microstrip microwave antennas,”
Third U.S.A.F. Symposium on antennas, USA, 1953.
[7] B. B. Adela, P. T. M. van Zeijl, U. Johannsen and A. B. Smolders

“On-chip antenna integration for millimeter-wave single-chip FMCW radar, providing high
efficiency and isolation,”
IEEE Transaction on Antennas and Propagation, no. 8, pp. 3281-3291, Aug. 2016.
[8] A. B. Smolders and G. A. Hampson,

“A fast and accurate scheme for calibration of active phased-array antennas,”
Proceedings IEEE Antennas and Propagation Symposium, Orlando, USA, pp. 1040-1043, 1999.
[9] G. A. Hampson, A. B. Smolders and A. Joseph,

“One square of a million,”
Proceedings 29th European Microwave Conference 1999, pp. 111-114, 1999.
183
184 BIBLIOGRAPHY
[10] A. B. Smolders and G. W. Kant,

“Thousand element array (THEA),”
Proceedings IEEE Antennas and Propagation Symposium 2000, Davos, Swiss, pp. 162-165,
2000.
[11] A. J. van den Biggelaar, A. J., Johannsen, U., Geurts, M. and A. B. Smolders,
“Assessment on the frequency dependent performance of active phased arrays for 5G,”
Proceedings of the 13th European Conference on Antennas and Propagation, EuCAP 2019.
Piscataway: Institute of Electrical and Electronics Engineers, 4 p. 8739544, 2019.
[12] S. Bronckers, A. Roc’h and B. Smolders,

“Wireless receiver architectures towards 5G: where are we?,”
IEEE Circuits and Systems Magazine, 17 (3), pp. 6-16, 2017.
[13] J. R. Hampton,
“Introduction to MIMO communications,”
Cambridge University Press, 1st Edition, 2014.
[14] G. E. Evans,
“Antenna measurement techniques,”
Norwood, MA:Artech House, 1990.
[15] A. C. F. Reniers, Q. Liu, M. H. A. J. Herben, H. J. Visser, A. B. Smolders,

“Statistical analysis of an extended method to measure circularly-polarized millimeter-wave
antennas,”
IEEE Antennas and Propagation Magazine, Feb. 2019.
[16] M. I. Skolnik,
“Introduction to radar systems,”
McGraw-Hill, International Edition, 1981.
[17] D. M. Pozar,
“Microwave engineering,”
Addison-Wesley Publishing, 1993.
[18] R. E. Collin,
“Foundations for microwave engineering,”
McGraw-Hill Series in Electrical Engineering, 2nd edition, 1992.
[19] G. L. Matthaei,
“Design of wide-band (and narrow-band) band-pass microwave filters on the insertion loss
basis,”
IEEE Transactions on Microwave Theory and Techniques, p. 580-593, November, 1960.
[20] G. Gonzalez,
“Microwave transistor amplifiers,”
Prentice Hall, 2nd edition, 1996.
BIBLIOGRAPHY 185
[21] J. M. Rollett,
“Stability and power gain invariants of linear twoports,”
IEEE Transactions on Circuit Theory, p. 29-32, March, 1962.
[22] M. L. Edwards, J. H. Sinsky,

“A new criterion for linear 2-port sta-bility using a single geometrically derived parameter,”
IEEE Transactions on Circuit Theory, p. 2303-2311, December, 1992.
[23] R. Ludwig, P. Bretchko,

“RF circuit design, theory and applications,”
Prentice Hall, 2000.
[24] Astron,
“The square kilometer array (SKA),”
www.astron.nl.
[25] B. B. Adela, M. C. van Beurden, P. T. van Zeijl and A. B. Smolders,

“High-isolation array antenna integration for single-chip millimeter-wave FMCW radar,”
IEEE Transactions on Antennas and Propagation, Vol. 66, nr. 10, p. 5214-5223, 2018.
[26] I. S. Gradsteyn en I. M. Ryzhik,

“Table of integrals, series and products,”
New York: Academic Press, 1965.
[27] R. F. Harrington,
“Time harmonic fields,”
McGraw-Hill, 1961.
[28] G. N. Watson,
“Theory of Bessel functions,”
Cambridge University Press, 1966.
[29] I. J. Bahl en P. Bhartia,

“Microstrip antennas,”
Massachusetts: Artech House, 1980.
[30] C. Balanis,
“Antenna theory,”
John Wiley & Sons, 1982.
[31] M. N. O. Sadiku,
“Numerical techniques in electromagnetics,”
Boca Raton: CRC Press, 1992.
[32] Advanced Design System ,”

www.keysight.com.
186 BIBLIOGRAPHY
[33] R. F. Harrington,
“Field computation by moment methods,”
New-York IEEE Press, 1993.
[34] J. D. Kraus,
“Antennas,”
McGraw-Hill, 2nd Edition, 1988.
[35] R. C. Hansen,
“Phased array antennas,”
New-York: John Wiley and Sons, 1998.
[36] R. J. Mailloux,
“Phased array antenna handbook,”
Boston: Artech House, 2005.
[37] S. A. Schelkunov,
“A mathematical theory of linear arrays,”
Bell-labs Tech. Journal, pp. 80-107, 1943.
[38] A. B. Smolders, H. J. Visser,

“Low side-lobe circularly-polarized phased arrays using a random sequential rotation tech-
nique,”
IEEE Transactions on Antennas and Propagation, 62 (12), pp. 6476-6481, 2014.
[39] R. M. C. Mestrom, T. J. Coenen and A. B. Smolders,

“Adaptive downtilt for cellular base stations,”
Smart Antennas (WSA), 2012 International ITG Workshop on, pp.162-167, 7-8 March 2012.
[40] A. B. Smolders,
“Design and construction of a broadband wide-scan angle phased-array antenna with 4096
radiating elements,”
Proceedings of International Symposium on Phased Array Systems and Technology, Boston,
MA, pp. 87-92, 1998.
[41] M. C. van Beurden, A. B. Smolders, M. E. J. Jeuken, G. J. Werkhoven and E. W. Kolk,

“Analysis of wideband infinite phased arrays of printed folded dipoles embedded in metallic
boxes ,”
IEEE Transaction on Antennas and Propagation, pp. 1266-1273, 2002.
[42] A. Dubok, A. Al-Rawi, N. Tessema, E. Tangdiongga, M. H. A. J. Herben, G. Gerini, and

A. B. Smolders.
“Double reflector configuration for optimal exposure of wideband focal plane arrays with
optical beamforming,”
IEEE Transaction on Antennas and Propagation, 65(8), 2017.
[43] A. B. Smolders, A. Dubok, N. M. Tessema, Z. Chen, A. Al-Rawi, U. Johannsen, T. Bressner,

D. Milosevic, H. Gao, E. Tangdiongga, G. Gerini, P. G. M. Baltus, M. Geurts, and A. M. J. T.
BIBLIOGRAPHY 187
Koonen.
“Building 5G millimeter-wave wireless infrastructure: wide-scan focal plane arrays with broad-
band optical beamforming,”
IEEE Antennas and Propagation Magazine, 67(4), p. 2305-2319, 2019.

Modern Antenna and Microwave Circuits

Uploaded by

Copyright:

Available Formats

Modern Antenna and Microwave Circuits

Uploaded by

Document Information

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

Modern Antenna and Microwave Circuits

Uploaded by

Copyright:

Available Formats

arXiv:1911.08484v3 [eess.

SP] 15 Sep 2020

Version September 2020

A catalogue record is available from the Eindhoven University of Technology Library

Subject headings: Antennas, microwave engineering, phased-arrays, wireless communications.

c 2020 by A.B. Smolders, H.J. Visser, U. Johannsen, Eindhoven, The Netherlands.

2 System-level antenna parameters 11

3 Transmission line theory and microwave circuits 23

3.13.3 Amplifier design using constant gain circles . . . . . . . . . . . . . . . . . . 67

5 Numerical electromagnetic analysis 133

6 Phased Arrays and Smart Antennas 141

A Vector calculus and coordinate systems 179

1.1 Purpose of this textbook

1.2 Antenna functionality

1.3 History of antennas

Reflector antenna (Westerbork Synthesis Radio Telescope) Wire antenna

Horn antenna Microstrip antenna

Figure 1.2: Some well-known antenna concepts.

antennas can be realized as two-dimensional structures on PCBs or as three-dimensional struc-

current status of U.S. allocations.

Office of Spectrum Management

ALLOCATION USAGE DESIGNATION

1st Capital with lower case letters

U.S. DEPARTMENT OF COMMERCE

Facsimile: (202) 512-2250 Mail: Stop SSOP, Washington, DC 20402-0001

National Telecommunications and Information Administration

AND TIME SIGNAL

THE RADIO SPECTRUM

TIME SIGNAL SATELLITE

STANDARD FREQUENCY AND

** EXCEPT AERONAUTICAL MOBILE

* EXCEPT AERONAUTICAL MOBILE (R)

THIS CHART WAS CREATED

ISM - 40.68 ± .02 MHz

FIXED FIXED FIXED FIXED

MOBILE MOBILE MOBILE

EXPLORATION- EXPLORATION- EXPLORATION- EXPLORATION-

59.0 5.47 698.0

ISM – 5.8 ± .075 GHz

ISM - 61.25± 0.25 GHz

ISM - 6.78 ± .015 MHz

ISM - 915.0± .13 MHz

mation can be found at: www.ntia.doc.gov and www.itu.int

FIXED Space research (deep space)(space-to-Earth)

RADIO- RADIO RADIONAVIGATION-SATELLITE AERONAUTICAL

ISM - 122.5± 0.500 GHz

ISM - 13.560 ± .007 MHz

RADIO RADIODETERMINATION- AERONAUTICAL MOBILE SATELLITE

and time signal

141.0 14.8 1668.4 AMATEUR

RADIO ASTRONOMY SATELLITE 15.35 1670.0 149.9

RADIO ASTRONOMY SPACE(passive) SATELLITE (passive) 1675.0 RADIONAVIGATION- MOBILE-SATELLITE MARITIME

Radiolocation RADIOLOCATION 17.97

1780.0 156.2475 AERONAUTICAL MOBILE (OR)

164.0 17.1 MARITIME MOBILE RADIONAVIGATION

182.0 (passive) (space-to-Earth) SATELLITE (passive) 19.99

flight testing of manned

telemetry, but excluding

SATELLITE 21.4 Amateur FIXED Land mobile 22.855

RADIO SPACE RESEARCH EARTH EXPLORATION - AMATEUR