Transformers, Kapp, 1908 PDF

A
LIST OF WORKS
OH, B.C.
Accur Voven Design,

Adam
Advic
Aeror Engineer's
net.
Alexa js. 6d. net.
tior !
Currents, $s.
2s. 6d. net.
Allso] is,
Alten Motors and
way Material,
4-r. Experiments,
Anab
andbook, 35.
s.
, is. Key,
Anat.
Andr
7s... 2s. net.
Apot, nent Making,

2S.
Arch: de, is. net.
net Magnetism,
Arith
is.
Arm?
net Dynamo, is.
Arno X.
Rays, 35-. 6d.
Analysis, los. bd. net.
Ashworth's Magnetism and Elec- Radium, is. net.
tricity, 2s. 6d. Wireless Telegraphy, 3^.
Heat, Light and Sound, 2s. net. Boyd's Coal Pits, 35. 6d.
Astronomy, Chambers, is. net. Petroleum, 2s.
Atkins' Sheet and Plate Metal Bridges, Moving Loads on Railway,
Work, 6s. net. Bamford, 4^. 6d.
Atlantic Ferry, Maginnis, 2s. 6d. British Locomotives, Cooke, Js. 6d.
Ballooning, May, 2s. 6d. Brodie's Dissections, i$s. net.
Moedebeck, los. 6d. net. Browne's (Jukes) Geology, 2s. 6d.
Bamford's Moving Loads on Rail- Builder's Quantities, Davis, $s. 6d.
way Under Bridges, ^s. 6d. net. net.
Barr's Electrical Engineering, ioj.net. Building Specifications, Farrow,
Barter's Drawing for Woodwork, 3 s. 6d. 35. 6d. net.
Woodwork, 6s. Burns & Kerr's Coal Mining, 10 Parts,
Batteries, Galvanic, Bottone, ,' 2s. each net.
(
2 )
Cabinet Making, Denning, 5$. Electric Ignition for Motor Vehicles,

Cables, Electric Light, Russell, ios.6d. Hibbert, is. 6d. net.
Carpentry and Joinery, Fletcher, 5^. Influence Machines, ?s.
net. Lamps, Maycock, 6s.
Central Station Electricity Supply, Light Cables, Russell, loj. 6d.
Gay and Yeaman, los. 6d. net. Light Fitting, Allsop, 5^.
Chambers' Astronomy, is. net. Light Installation Apparatus,
-
:
Chemical Analysis, Coppock, 2s. Salomons, 7-r. 6d.
-- Composition, Landolt, ^s. 6d.

Evolution, Rudorf, 4$-. 6d.
Manipulation, Hewitt, 45. 6d. Maycock, vol. i.
[net.
Lighting, Guide to, Bottone, is.
Lighting & Power Distribution,
6s. net. ; vol. ii.
Chemist's Compendium, Thompson, 7*. 6d.

2s. net. Lighting for Marine Engineers,
-
Chemistry, Physical, Reychler, 4*. 6d.
Chip
Harris, 3 vols. 45.
Carving Patterns, Andreen,
Walker, 5*.
Motors, Bottone, 3$.
Hobart, 12s. 6d. net.
7*. 6d. net. Maycock, 41. 6d. net.
Coal Mining, Burns & Kerr. 10 Punga, 45. 6d. net.
Parts, 2s. each net. Traction, Rider, los. 6d. net.
Coal Pits, Boyd, 3*. 6d. Transformers, Kapp, iOf.6d. net.
Coils, Induction,Bonney, 3*. Transmission of Energy, Kapp,
Cold Storage, Williams, IQS. 6d. ioj. 6d.
Colour in Woven Design, Beaumont, Transmission of Intelligence,
21S. Houston, 4.?. 6d. net.
Commutator Motors, Punga,4J-.6^/.net. Wiring, Fittings, Switches, and
-
Concrete-Steel, Twelvetrees, 6s. net.
Buildings, Twelvetrees, los. net.
Cooke's Locomotives, Js. 6d.
Lamps, Maycock, 6s.
Details Forms, 2s. 6d. net.
- Tables,
Maycock, y. 6d.
Coppock's Volumetric Analysis, 2s. Electrical Energy, Kapp, los. 6d.
Crapper's Electrical Measurements. Engineering, Barr, IQS. net.
Crellin's Bookkeeping, is. Key, Engineer's Pocket-Book, Edg-
is. 6d. net. cumbe, i>s. 6d. net.
Cullyer's Land MeasuringTables,2*.6^. Experiments, Bonney, 2s. 6d.
Davis' Quantities and Quantity Instrument Making, Bottone,
Taking, 3*. 6d. net. y. 6d.
Denning's Cabinet Making, 5^. Insulation, Turner and Hobart,
Designing and Drawing, Leland, 2s. i Of. 6d. net.
Dissections Illustrated, Brodie, i$s. Measurements, Crapper.
net. Measurements (Arithmetic),
Dod's Parliamentary Companion, Whittaker, is. net.
3-y. 6d. net. Measurements, Houston, 4-f. 6d.

Drainage, Nadieine, is. net.
Draughtsmen's Work, u. 6d. Electricity in Homes and Workshops,

--
Drawing and Designing, Leland, 2s.
for Marine Engineers,
Walker,
Supply,
5^. net.
Gay and Yeaman,
- Roberts, 6s.
Cards for Marine
Sothern, 3^. 6d. net.
Engineers,
los. 6d. net,
Gibbings, 6s.
and Magnetism, Ashworth,2j.6r/.
-- Woodwork, Barter, 6d. 35. Bottone, 2s. 6d. net.
Dynamo, Hawkins and Wallis, Houston, 2s. 6d. net.

-- Design, Hobart, i$s.
- How to
Js. 6d. net.
Manage, Bottone, is.
Maycock,
Electro-motors, Bottone,
2s. 6d.
3^.
Edgcumbe's Electrical Engineer's Electro-platers' Handbook,Bonney, 3^.
Pocket- Book, 35-. 6d. net. Electrolytic Analysis, Neumann, 6s.
Education, Practical, Leland, 6s. Elliott's Gas and Petroleum Engines,
Electric Batteries, Bottone, $s. 25. 6d.
- Salomons, 6s. net.

Bells, Bottone, 2s. net.
Industrial Electricity, 2s. 6d.
Engineer Draughtsmen's Work, I s. 6d.
3 )
Engineer Fitting, Homer, 51. Hawkins' and Wallis's Dynamo, 15^.

Locomotive, McDonnell, is. Heat, Light and Sound, Ashworth,
Engines, Model, Alexander, 55-. net. 2s. net.
English and American Lathes, Helical Gears, Horner, 5*.

Horner, 6s. Herbert's Telegraphy, 6s. 6d, net.
Experiments with Vacuum Tubes, Telephone System of the Post
Salomons, 2s. Office.
Electrical, Bonney, 2s. 6d. Hering, Electric Railways, qs.dd. net.
Explosives, Manufacture of, Guttmann, Hertzian Waves and Wireless Tele-
2 vols. 2/. 2s. graphy, Bottone, 3^.
Farley's Cellar Book, is. 6d. Hewitt's Organic Chemical Manipula-
Farman's Auto Cars, 55. tion, 4*. 6d.
Farrow's Specifications for Building Hibbert's Electric Ignition for Motor
Works, 3-y. 6d. net. Vehicles, is. 6d. net.
Stresses and Strains, 5*. net. Hills' Gas and Gas Fittings, 5$. net.
Field Work and Instruments, Hobart's Electric Motors, 12s. 6d. net.
Walmisley, 6s. net. Design of Continuous Current
Findlay's Railway Working, Js. 6d. Dynamos, 7^-' 6d. net.
Fitting, Engineer, Horner, 5^. Hobart and Ellis' Armature Con-

Electric, May cock, 6s. struction, 15-r. net.
Electric Light, Allsop, 5*. Hobart and Stevens' Steam Turbine
Fletcher's Architectural Hygiene, 55. Engineering, 2is. net.
net. Hobart and Turner's Insulation of
Carpentry and Joinery, 5*. net. Electric Machines, los. 6d. net.
Steam-jacket, is. 6d. Hornet's Helical Gears, 5-r.
Flying Machines, see Maxim. Ironfounding, $s. 6d.
Moedebeck. Lathes, 6s.
Foden's Mechanical Tables, is. 6d. Metal Turning, 35. 6d.
Friction and its Reduction, Wheeler, Pattern Making, $s. 6d.
3-r.
net. Principles of Fitting, $s.
Galvanic Batteries, Bottone, $s. Houston's Electrical Measurements,
Garratt's Modern Safety Bicycle, 3-r. 45. 6d. net.
Gas and Gas Fittings, Hills, 55. net. Electricity and Magnetism, 2 s. 6d.
Gas and Petroleum Engines, 2s. 6d. Transmission of Intelligence,
Gaseous Fuel, Thwaite, is. 6d. 4*. 6d. net.
Gay and Yeaman's C. S. Electricity Hutton's Mathematical Tables, I2s.
Supply, los. 6d. net. Hydraulic Motors and Turbines,
Gearing, Helical, Horner, 5*. Bodmer, i$s.
Geography, Bird, 2s. 6d. Hygiene, Architectural, Fletcher, $s.
Geology, Jukes-Browne, 2s. 6d. net. net.
Geometrical Optics, Blakesley, 2s. 6ct. Ibbotson & Arnold's Steel Works
German Grammar for Science Stu- Analysis, los. 6d. net.
dents, Osborne, 2s. 6d. net. Ignition, Electric, for Motor Vehicles,
Gibbings' Electricity Supply, 6s. Hibbett, is. 6d. net.
Gray's Electrical Influence Machines, Indicator Handbook, Pickworth, 2
vols. 31. each, net.
5*.
Greene's Saints and their Symbols, Induction Coils, Bonney, 3^.
3-T.
6d. Motors, Hobart, 12$. 6d. net.
Greenwell and Elsden, Road Con- Industrial Electricity, 2s. 6d.
struction, 5-r.
net. Insulation of Electric Machines,
Manures, 7*. 6d.
Griffiths' Turner and Hobart, los. 6d. net.
Guttmann's Manufacture of Explos- Iron and Steel Structures, Twelve-
ives, 2 vols. 2/. 2s.
, trees, 6s. net.
Percentage Tables for Analysis, Ironfounding, Horner, 3^. 6d.
net. Italian Dictionary, Baretti, 21.9.
3-r.
Harris' Practical Chemistry, 3 vols. $s. Italian Technical Dictionary, Webber,
Hatch's Mineralogy, 2s. 6d. 4-r. net.
Hatton's Mathematics, 2s. 6d. Jukes-Browne's Geology, 2s. 6d.
TRANSFORMERS
WHITTAKER'S PRACTICAL HANDBOOKS
ARNOLD & IBBOTSON Steel Works Analysis. IDS. 6d. net.
BAMFORD, H. . .
Moving Loads on Railway Underbridges.
45. 6d. net.
BARR, J. R. ... Direct Current Electrical Engineering. IDS. net.
BLAKESLEY, T. H. .
Alternating Currents of Electricity. 53.
BODMER, G. R. . .
Hydraulic Motors and Turbines. 155.
Inspection of Railway Material. 53.
GAY & YE AM AX . . Central Station Electricity Supply. IDS. 6d. net.
GRAY, J Electrical Influence Machines. 53.

HAWKINS & WALLIS The Dynamo : Its Theory, Design, and Manu-
facture. 153.
HIBBERT, W. . . Electric Ignition for Motor Vehicles, is. 6d. net.
HOBART, H. M. . . Continuous-Current Dynamo Design. 75. 6d.
net.
ElectricMotors Induction Motors and Con-
tinuous-Current Motors. I2s. 6d. net.
HOBART & ELLIS Armature Construction. 155 net.
KAPP, G Transformers for Single and Multiphase
Currents.
Electric Transmission of Energy. IDS. 6d.
LODGE, SIR O. . .
Lightning Conductors and Lightning Guards.
153.
MAYCOCK, W. P. . .
Alternating - Current Circuit and Motor.
43. 6d. net.
Electric Lighting and Power Distribution.
Vol. L, 6s. net. ;
Vol. II., 7s. 6d.
Electric Wiring, Fittings, Switches, and
Lamps. 6s.
MAZZOTTO, D. . . Wireless Telegraphy. 33. 6d. net.

MOEDEBECK, H. W. L. Pocket Book of Aeronautics. IDS. 6d. net.
PERKIN, F. M. The Metric and British Systems of Weights and
Measures, is. 6d. net.
PUNGA, F. . . .
Single Phase Commutator Motors. 43. 6d. net.
RIDER, J. H. Electric Traction. IDS. 6d. net.
RUSSELL, S. A. . . Electric Light Cables and the Distribution of

Electricity. IDS. 6d.
SALOMONS, SIR D. .
Management of Accumulators. 6s. net.
STEVENS & HOBART Steam Turbine Engineering. 2 is. net.
STILL, A ..... Alternating Currents and the Theory of
Transformers. 53.
Polyphase Currents. 6s. net.
TURNER HOBART . Insulation of Electric Machines. los. 6d. net.
WALKER, S. F. . .
Electricity in our Homes and Workshops.
53. net.
j> Electric Lighting forMarine Engineers. 53.

WHITTAKER'S . . . Electrical Engineer'sPocket Book. 33. 6d. net.
Mechanical Engineer's Pocket Book. 35. 6d. net.
,, . Arithmetic of Electrical Engineering, is. net.
WILLIAMS, H. . . Mechanical Refrigeration. IDS. 6d.
WHITTAKER & CO., 2 WHITE HART STREET, LONDON, E.G.

TRANSFORMERS
FOR
SINGLE AND MULTIPHASE CURRENTS
A TREATISE ON
THEIR THEORY, CONSTRUCTION, AND USE
BY
GISBERT KAPP
PROFESSOR OF ELECTRICAL ENGINEERING AT THE BIRMINGHAM UNIVERSITY
DH. ENG., MEMB. 1NST. C.E., MEMB. 1NST. E.E.
SECOND REVISED AND ENLARGED EDITION
WITH TWO HUNDRED AND NINETEEN ILLUSTRATIONS
WHITTAKER AND CO.

WHITE HART STREET, PATERNOSTER SQUARE, LONDON
AND 64 AND 66 FIFTH AVENUE, NEW YORK
1908
[All rights reserved]
V
RICHARD CLAY & SON?, LIMITED,

BREAD STREET HILL, E.G., AND
BUNGAY, SUFFOLK.
PREFACE
SINCE the first book appeared, now twelve
edition of this
years ago, enormous improvements have taken place in

the construction of transformers. Apart from the general
development of the science of electrical engineering,
which has affected all branches alike, these improvements
were mainly due and rendered possible by three causes :
the more general appreciation of oil as a cooling and
insulating medium, the advent of alloyed iron, and last,
but not least, the growing demand for large unit in
connection with power transmission. Whilst only a few

years ago a transformer of 200 or 300 kw. was considered
a very large unit, now-a-days units of iooo and more kw.
have become quite usual. At the same time the efficiency
has been raised, the regulation improved, and the price
reduced. It has been my aim in the present edition to
show how by a careful study of the thermic condition the
output of a transformer of given weight and cost may be

increased. have also considerably enlarged the part
I
dealing with the testing of transformers and iron samples.

Some readers might perhaps think that I have given too
much space to the treatment of ballistic and similar
methods, since these belong rather to a treatise on
physics than to an engineering handbook. Moreover,
there is now-a-days no lack of testing appliances which
even in unskilled hands may yield the desired information,
and for these reasons it may be argued that the practical
261199
V
RICHARD CLAY & SON?, LIMITED,

BUNGAY, SUFFOLK.
PREFACE
SINCE the first book appeared, now twelve
edition of this
years ago, enormous improvements have taken place in

the construction of transformers. Apart from the general
development of the science of electrical engineering,
which has affected all branches alike, these improvements
were mainly due and rendered possible by three causes :
the more general appreciation of oil as a cooling and
insulating medium, the advent of alloyed iron, and last,
but not least, the growing demand for large unit in
connection with power transmission. Whilst only a few

years ago a transformer of 200 or 300 kw. was considered
a very large unit, now-a-days units of iooo and more kw.
have become quite usual. At the same time the efficiency
has been raised, the regulation improved, and the price
reduced. It has been my aim in the present edition to
show how by a careful study of the thermic condition the
output of a transformer of given weight and cost may be

increased. have also considerably enlarged the part
I
dealing with the testing of transformers and iron samples.

Some readers might perhaps think that I have given too
much space to the treatment of ballistic and similar
methods, since these belong rather to a treatise on
physics than to an engineering handbook. Moreover,
there is now-a-days no lack of testing appliances which
even in unskilled hands may yield the desired information,
and for these reasons it may be argued that the practical
261199
vi PREFACE
engineer need not trouble about the scientific principles

of such apparatus, or use the more refined tests of the
physicist. admitting the commercial value

Whilst fully
of the modern and handy testing sets, I still think that a
thorough knowledge of their principles, and the

scientific
ability to do equal or better work by means of apparatus

built up for the occasion by using the appliances usually
found in the test-room of electrical
engineering works,
should be within the reach of the practical engineer, and
that is the reason why I have given a good deal of
space to a seemingly only theoretical subject.

Another addition is the chapter on the relations
between the transformer and its circuits. Although not
strictly aquestion of design, it is still a most important
matter, since it involves the safety of extended and, as a
rule, expensive networks.
Wherever have used graphic in preference
possible I
to analytical methods, and where the latter were unavoid-
able only the simplest analysis and the elementary

calculus have been employed.
In placing this book before the reader, I wish to
record sense of gratitude for the detailed information
my
which many Firms of the highest standing have kindly
permitted me to publish concerning their most recent
designs. These examples of modern transformer work
will be found in Chapter XIV.
GISBERT KAPP.
Birmingham,
CONTENTS
CHAPTER I
Principleof Action Magnetic Leakage Arrangement of Coils

Fundamental Equation . . . . i
CHAPTER II
Losses in Transformers Alloyed Plates Determination of the most

Advantageous Thickness of the Plates Influence of the E.M.F.
Curve on the Hysteretic Loss Influence of the Shape of Core
and Coil on the Losses Core and Shell Transformers . 16
CHAPTER III
Usual Types Construction of the Iron Part Proportions of the Iron

Part Heating of Transformers Results of Tests Circulation of
Cooling Medium Values of c Heat Conductivity Parallel and at
Right Angles to the Surface of Plates Theory of Heating and
Cooling Intermittent Load Influence of Linear Dimensions on
the Output Weight of Active Iron . . . . .
39
CHAPTER IV
The Use of Vectors Combination of Currents Combination of
Electromotive Forces Self- Induction and Capacity Influence
of Higher Harmonics Power of an Alternating Current . 76
CHAPTER V
The Magnetic Circuit Energy Stored in Magnetisation The
Hysteretic Loop No-Load Current of a Transformer Shape of
Exciting Current Choking Coil . . ,
'97
viii CONTENTS
CHAPTER VI
Design of a Core Transformer Best Distribution of Copper Losses
at Different Loads Time Constant for Heating Weight and
Cost of Active Material Best Distribution of Losses Trans-
formers for a Special Service Transformers for Power Trans-
formers for Lighting Annual Efficiency Economic Importance
of Small Losses Constructive Details 1 1 1
CHAPTER VII
Design of a Shell Transformer Fill Factor Winding Efficiency,

Weight and Cost Enlarging a Design . . .
-134
CHAPTER VIII
The Vector Diagram Application to a Transformer Working an
Open Circuit Working under Load Magnetic Leakage Work-
ing Diagram of a Transformer having Leakage Voltage Drop
Graphic Determination of Drop Drop Diagram Simplified
Correction for Eddy Current Losses .
.142 . .
CHAPTER IX
Calculation of Inductive Drop The Influence of Frequency on Drop
The Influence of Frequency on Output Equivalent Coils The
Self-induction of a Transformer Working Condition represented
by Vector Diagram Constant Current Transformer .
.176
CHAPTER X
The Dynamometer The Dynamometric Wattmeter Measurement of
the Power carried by Currents of Irregular Form The Induction
Wattmeter Measurement of Power in Three-Phase Circuit
Three- Voltmeter Method Three- Amperemeter Method .198 .
CHAPTER XI
Testing Transformers Testing Sheet-iron Special Implements by

Dolivo Dobrowolsky,Kapp, Epstein, Richter, Ewing The Ballistic
Method The Fluxometer Scott's Method Kapp's Method . 226
CHAPTER XII
Safety Appliances for Transformers Sub-station and House Trans-

formers Reducing Iron Losses Transformer for Three- Wire
System
Working
Balancing Transformers
Boosters Scott's
Auto-transformers
System ....
Series
262
CONTENTS ix
CHAPTER XIII
The Transformer in Relation to its Circuits Equivalent Coils in

Parallel and Series Connection Rise of Pressure through
Resonance Determination of the Dangerous Condition Rise of
Pressure on Loaded Lighting System is Small Influence of
Power Factor Breakdown of Cables in Large Networks .
291
CHAPTER XIV
Some Examples of Modern Transformers . .
305
APPENDIX
APPENDIX
I
II .... .
.
-355
35 6
APPENDIX
INDEX
III
......... -357
361
TRANSFORMERS
CHAPTER I
PRINCIPLE OF ACTION MAGNETIC LEAKAGE-

ARRANGEMENT OF COILS FUNDAMENTAL
EQUATION
Principle of action. If the magnetic flux
passing
<
through a coil changes, an E.M.F. is induced in the coil

which is
proportional to the time-rate of change /
^)
and the number of turns n. Conversely, if a current
be sent through the coil, it
produces a magnetic flux,
threading the coil, which is within certain limits
pro-
portional to the current. If this current changes, a
corresponding change takes place in the magnetic flux.

Let now two coils be so arranged that the flux produced
by the current in one passes wholly or partially through
the other, then any change in the current strength in
the former coil will produce an E.M.F. in the latter
coil. Such an arrangement is shown in Fig. i, where
a ring of iron is threaded through the two coils I, II.
A current passing through coil I produces a magnetic
field which passes partly through the iron ring and
partly through the air surrounding this coil. The flux
will therefore be strongest in the centre of the coil, at
a, and weakest at b, in the centre of coil II. The
iron ring acts as a vehicle for carrying the flux produced
by coil I through coil II, though as an imperfect vehicle,
since part of the flux is lost on the way. In a sense, the
iron ring may be regarded as a magnetic link between
the two coils. Even without iron the coils can be linked
V : l ^TRANSFORMERS
by ihe ^magnetic flux passing through

: - air.
together
Thus, in the position shown, the field produced by I
would in its
part pass through II, though strength
would be much diminished. The same holds good if
the two coils are laid upon each other, in which position
a somewhat stronger field would pass through II, though
not so strong as with an iron ring. If, however, whilst
are relatively so
stillomitting the iron ring, the coils
I lies in the plane of II, or vice
placed that the axis of
versa, then none of the lines produced by I can pass
through II, and a change in the current passing through

I cannot produce any E.M.F. in II. By suitably
the coils, an inductive effect of one upon the
placing
other can therefore be produced, even without the
use of an iron link, but the employment of such a link
has the advantage that not
only is the inductive action
increased, but it becomes to a
greater extent independent of
the mutual position of the
coils. An apparatus consist-
ing of two coils, interlinked
PIG. i. Magnetic and electric p
circuits interlinked. with an iron core common to
both, is called a transformer.
It has already been mentioned that the E.M.F. pro-
duced in II, which we may call the secondary coil, is
proportional to the time-rate of change of the current
in the primary coil I. Since the current in this coil
cannot alter indefinitely in the same sense without be-
coming infinite, it follows that periods of growing
current strength must alternate with periods of declining
current strength. If, then, with a growing current in
the primary coil, the E.M.F. induced in the secondary
coil acts in one way, it must act in the opposite way if
the current diminishes, and it is thus clear that changes
in the current strength in the primary coil, even if not
accompanied by changes in direction, must produce an
alternating E.M.F. in the secondary coil. This alter-

nating E.M.F. produces, in an external circuit connected
to the terminals of the secondary coil, an alternating
current. We are thus able to convert an unidirected
MAGNETIC LEAKAGE 3
pulsatory current into an alternating- current, but never

into a continuous current. Instead of using a pulsat-
ing current in the primary coil, we may with advantage
use an alternating current, and thus obtain from the
secondary coil another alternating current, the E.M.F.
of which is dependent on that of the primary current,
and on the ratio between the number of turns in the
two windings.
Magnetic leakage. Since the lines of force not only
pass through iron, but in a lesser degree also
through
air, it follows that only part of the magnetic flux at a
actually threads through the secondary coil at b, the rest

closing round the primary coil in air. The difference
between the flux in a and b will be the greater the
further the coils are from each other, and the
greater
the resistance which the iron offers to the passage of the
lines offeree. In consequence of this resistance (some-
times also called magnetic reluctance) lines of force are
caused to leak out laterally, and form thus a leakage
field which does not contribute in any way to the
production of E.M.F. in the secondary coil. The more
leakage there is, the smaller is consequently the E.M.F.
induced in the secondary coil through an alternating
current in the primary coil.
In order to understand the conditions which influence
leakage, we assume for the present that the primary coil
carries a continuous current, whilst through the secondary
coil there passes either no current at all or also a con-
tinuous current of such direction as will tend to weaken
the field produced by the primary current.
The coil I drives, then, a magnetic flux in a certain
direction through the iron ring. If no current flows
through coil II, then the lines of force have only to

overcome the magnetic resistance of the iron path, which
may be so small that comparatively few lines are crowded
out. If, however, the coil II also carries a current, it
will tend to produce a
magnetic flux in the opposite
direction, which, colliding with the original flux, must
cause a strong leakage field, thus weakening consider-
ably the flux actually passing through the secondary
coil.
4 TRANSFORMERS
This condition of things may easily be explained by
hydraulic analogy. Let, in Fig. 2, a ring-shaped tube of
material be filled with and immersed in water, and
porous
let the water in the tube, as shown by the arrow, be kept
in motion by a propelling fan, I. This fan produces a
difference of pressure between its inlet and outlet side,
which pressure is absorbed by the frictional resistance of
the tube. Since the pressure above the fan is greater
than that below, water will, as indicated by the dotted
lines, pass out through the pores of the tube in its upper
half, and enter the tube through the pores in its lower
half. The velocity of the water must consequently be
If the tube is wide and the pro-
greater at a than at b.
pelling power of the fan small,
little head overcome
will suffice to
the friction and the quantity of
;
water leaking out and in, as well

as the difference of velocity in a
and b, will be small. Let now a
second fan (II) be inserted at b y
which for the present we will

Jmagine to be frictionless then
;
FIG. 2.-iiydrauiic analogy this fan will be set in rotation by

of magnetic leakage. the stream of water, but it will not
increase the leakage or diminish
the velocity of the water. If, however, we retard the
motion of fan II by letting its spindle transmit mechanical

energy, the free flow of the water will be impeded, and the
difference of pressure between the upper and lower halves
of the tube will be increased. As a result, the leakage
will be augmented, and the quantity of water passing the
point a in unit time will be appreciably more than that

passing the point b. At the same time the speed of fan
II will be reduced; and this for two reasons. In the
first place because the load on the spindle of II must
retard its motion, and in the second place because the
velocity of the water is smaller than before. If we wish
to limit the loss of speed due to the latter cause, we can
do so by placing the fan I as near as possible to fan II.
Now let us substitute for the porous tube the iron
ring, and for the two fans the driving and the driven coil ;
MAGNETIC LEAKAGE
then we
see that the magnetic flux through the driven
coil (which corresponds to the velocity of the water at 6)
will be the smaller the stronger the current is in the
driven coil.
The arrangement of coils shown in Fig. i is bad, on
It does not
account of their great distance. give a strong
magnetic flux through the driven coil if a large current
is
permitted to flow through this coil. can improve We
the design by spreading the coils each over half the cir-
cumference of the ring, as shown in Fig. 3. In this case
the magnetic pressure tending to force the lines through
the air is no longer constant over each half of the ring,
but it attains its previous value only in the points c and
d. It diminishes on either side of the vertical diameter,
and becomes zero in a and b. The
leakage field is, therefore, not only
quantitatively smaller, but, owing
to its distribution and the distri-
bution the two windings, its
of
influence is also lessened,
qualitative
as compared with the arrangement
shown in Fig. i.
The distribution of the leakage FIG. Magnetic leakage. 3.

field may be approximately deter-
mined if we remember that the magnetic pressure,
which forces the lines to leave the ring at
any point,
is proportional to the ampere-turnscounted up to that
point. Imagine now the windings evenly distributed,
and the direction of the current thus, that the magnetic
pressure is from iron to air in the upper left quadrant,
and from air to iron in the lower left quadrant. Corre-
sponding pressures must of course exist in the right
quadrants. Let now the ring be cut at a and straightened
out, then the zigzag line in Fig. 4 gives a graphic repre-
sentation of the magnetic pressure producing leakage.
Positive ordinates represent a pressure from iron to air,
i. north polarity.
e. The leakage lines are shown dotted
in Pig. 3, but only inside the ring. There are, of course,
also leakage lines in the whole of the air space surround-
ing the ring. If we assume, as a very rough approxi-

mation, that the magnetic resistance along any path
TRANSFORMERS
through air is the same, then the number of lines passing
through unit surface in any point of the ring will be
proportional to the magnetic pressure at that point, and
the shaded areas in Fig. 4 may be taken to represent
roughly the leakage field. The assumption that all the
paths through air have the same magnetic resistance is
of course not strictly correct ;
as we are, however, at present

only concerned with a general
investigation
o of the leakageo
field, it is not necessary to
enter minutely into the ques-
tion of how the magnetic re-
FIG. 4. Leakage flux. sistance of any particular path
through the air varies, and we
may as a rough approximation assume that Fig. 4
represents the leakage field.
We have up to the present assumed that the two coils
carry continuous currents, but it is obvious that our
reasoning applies equally to the case of alternating-
currents, provided that the change in direction occurs
in both coils nearly simultaneously, a condition which is
always fulfilled in transformers when working under a

load.
Arrangement of coils. Wehave seen that in point

of leakage Fig. 3 is an improvement on Fig. i. We
may, however, carry the improve-
ment still further by subdividing
each unit into several parts. In
Fig. 5 we have six separate coils,
arranged to
uniformly cover the
ring, and connected alternately with
the primary and secondary circuit.
The greatest magnetic pressure is
also in this case at the junction of
two coils since, however, the num-
;
FIG. Subdivision of ber of turns in each coil is reduced

5.
coils. to one-third, this pressure is also
reduced to one-third of its previous
value. The surface through which lines can leak is
at the same time also reduced to one-third, so that the
ARRANGEMENT OF COILS 7
total leakage field now only amounts to \ x J = \ of

its previous value. If, instead of subdividing each coil
into three parts, we subdivide it into four, the leakage
field would be reduced to iV of its previous value, and so
It will thus be seen that, by
on. carrying the principle
of subdivision sufficiently far, we can reduce the leakage
field to any desired extent. It could even be reduced to
zero if we were to interweave the primary and secondary
coils. This would, however, lead to difficulties as regards
insulation, and is not necessary, since experience has
shown that it suffices for all practical purposes to sub-
divide the windings so far as to limit the effective ampere-
turns in each individual coil to a value which is roughly
proportional to the flux, and may amount to a few
hundred ampere-turns in small transformers to a thousand
or more ampere- turns in large
transformers.
Wehave hitherto assumed
that the magnetic link between
the two windings is a circular
ring, but it will be obvious that
its
geometric form is immaterial,
and that any shape of ring may magne
Rectangular
be used. Wemight, for instance,
employ a magnetic link in the shape of a rectangular
frame, and place the coils over the two longer sides of
the rectangle (Fig. 6). The arrangement shown on the
left corresponds to
Fig. 3. In this case there is only
one secondary and one primary coil, and the magnetic
leakage must therefore be very great. In the arrange-
ment shown on the right, the primary winding is sub-
divided into five coils, which alternate in position with
five coils of the secondary winding. The leakage is
thereby reduced to about A
part. Another, and with
regard to the reduction of leakage equally effective,
arrangement consists in placing the coils axially within
each other. This arrangement has the advantage of a
reduction in the number of separate coils to be wound and
handled, whilst at the same time the insulation between
primary and secondary coils is of simple shape (plain
cylinders), and can therefore easily be made perfect.
3 TRANSFORMERS
Fundamental equation. The E.M.F. induced in a
coil is, by a well-known law of electro-dynamics, propor-
tional to the number of turns of wire, and the time-rate
of change of the magnetic flux. In symbols
In order to be able to calculate the E.M.F. occurring

at any given moment of time, we must know the relation
between < and t. The magnetic flux < is produced by
the current passing through the primary coil, and if the
magnetisation of the iron core remains within such limits
that the permeability may be considered to remain con-
stant, then (/} may be considered to be proportional to the
primary current. We
assume for the present that the
secondary coil is open, so that no current can flow through
it which would mask or disturb the
magnetising effect of
the primary current. There are, as a matter of fact,
certain secondary actions such as leakage, ohmic resist-
ance and reluctance, which interfere with the strict
proportionality of primary current and magnetic flux, but
the consideration of these we must postpone. also We
assume that the primary current is obtained from an
alternator, the E.M.F. of which follows a true sine law.
This is not always, and indeed very seldom, the case ;
but it will be shown later on that the equations obtained

under these assumptions remain applicable in all cases
occurring in practice.
Imagine, then, a wire coil of one turn, including an
area of A sq. cm., traversed by a magnetic flux, which
varies according to a periodic sine function between the
limits + <f) and <. The maximum value of the induc-
tion is obviously B = A. <Call the time required for
:
the performance of a complete cycle from 4- </> to </>,
and back to + <, T, and the number of cycles occurring

per second v then
;
Since the E.M.F. is

dependent on the change in the
flux passing through the wire loop, but not on the angle
FUNDAMENTAL EQUATION
at which the lines of force thread through the loop, we
may replace the rectilinear and oscillating field by a

constant and homogeneous field, provided we revolve
the loop round an axis in its plane with a speed of v
revolutions per second. Let, in Fig. 7, the field be
represented by the vertical lines, and be the axis O
around which the wire loop is rotated. Let the rota-
tion take place in the direction shown by the arrow,
and, counting the time from the moment the loop is
horizontal, let, at time t, the coil occupy the angular
position a. Through the cutting of the lines of force
there will be induced in the upper half of the loop
an E.M.F. directed towards the
observer, and in the lower half
from the observer. In this and
all the
following diagrams we
mark these directions respect-
ively by a dot and a cross
inscribed into the little circle
representing the cross-section of
the wire these signs meaning
;
the point and the feathers of an

arrow which indicates the direc- FIG. 7. Rotating loop.
tion of E.M.F. or current.
Let co be the angular velocity of the loop, then a = a>/,
and co = 27rv ;
from which it follows that
a = 2irvt
The magnetic flux threading through the loop is
obviously <f>
cos a, and its rate of change
d(h cos a
-
. . da
sin a-
dt dt
Since =co = 27rv. we have for the instantaneous

dt
value of the E.M.F. the expression
E= 27TV(j)
sin a
The loop has only one turn of wire. If there are
n turns the same E.M.F. is induced in each, and the total
E.M.F., measured at the terminals of a coil of n turns, is

therefore in absolute measure
E = 27TV(t)n sin a
io TRANSFORMERS
To obtain it in volt we must multiply by io~ 8 If the .
coil is horizontal (a = o), the flux threading through it is a

maximum and the E.M.F. is zero. If the coil is vertical,
;
i. e.
parallel to the direction of the field, the flux passing
through the coil is zero, and the E.M.F. has its maximum
value
E = 27TV(f)nio~ s
(i)
The instantaneous value of the E.M.F. is therefore
E,= E sin a
and the instantaneous value of the magnetic flux is
cos a <j) t <j)
if by we denote its maximum value.

(f>
These equations show that between E.M.F. and

magnetic influx there is a difference in phase of a quarter
period.
Imagine now the terminals of the coil connected with
an incandescent lamp of resistance R. The current I
passing through the lamp must vary as the E.M.F. E.
Calling I the maximum value of the current, we should
Tj'
have I ,
and i I sin a. Although ordinarily the resist-
K.
ance of a lamp depends on and varies with the current,
we are justified in assuming the resistance in our case to
be constant, since the changes in current strength occur
so rapidly that the filament has no time to grow hotter or
cooler as the current grows stronger or weaker, but
assumes a mean temperature, and has consequently a
constant resistance. Let then the lamp be fed, first by
an alternating current derived from the coil, and secondly
by a continuous current, but let in both cases the filament
be raised to the same temperature, so as to emit the same
amount of light. The lamp will then in both cases re-
quire the same supply of electric power, and we may
consider the strength of the continuous current as a
measure of the effective strength of the alternating cur-
rent. Since i = I sin a, we may represent the instantane-
ous value of an alternating current by the projection of
a vector of length I, which revolves with an angular
speed of w
= 2?rv. In the same manner other quantities
FUNDA MENTA L. EQUA TION 1 1
of a periodic character may be represented, and the dia-

grams used for this purpose are called clock, or vector,
diagrams. Let then, in our case, the maximum value of
the current be graphically represented to a certain scale by
the length of a line I. Let the line revolve round one of
its ends, and take the projection of the other end at stated
times. The length of the projection, measured with the

same gives the instantaneous value of the current.
scale,
To energy which is supplied to the lamp
find the total
by the current during the time of one period, we should

divide the circle described by
the current vector into a suffi-
ciently large number of equal
parts, distant by the time interval
A^ from each other, project these
points to get i, and form the ex-
2
pression 2^ RA/. This would
be a laborious process, but it
can be simplified if we imagine
the additions made twice over
by counting together such posi- FlG 8 ._ C iock diagram.
.
tions of the vector I as are 90

apart. The members of the series we have to sum up
would then be of the form
R(Psin
2
a+P cos
2
a)A^ RPA/
as will easily be seen by reference to Fig. 8, in which
the vector is shown in two positions differing by 90.
The projections are OIj and OI 2 ,
and as the sum of
2
their squares is
obviously equal we find that each
I ,
member of our series has the same value, namely, RPA/.
Let m be the number of members, so that m= T ;

then
A^
we find the total work done by the current during one
period, by multiplying the value of one member of the
series by m
and dividing by 2m, the latter because we have,
by taking the vectors in pairs, counted them twice over.
The work done during one period, or the energy, is
c -RPT
12 TRANSFORMERS
and the power, that is, the rate at which work is
done, is
If the lamp is fed by a continuous current i the power y
is P= /
2
R. Let the power be the same in both cases,
then i may be
considered the effective value of the alter-
nating current, and will be given by the expression
X/2
This relation is of course only valid if the current is
a sine function of the time. If it follows any other law
the ratio between its maximum and effective value will
be given, not by \/2, but by some other coefficient.
If by I we denote the instantaneous, and by i the effec-
tive, value of the current, we have for any form of current
curve
rVT/ 1** 1
(3)
In words The effective current is the square

: root of the
mean squares of the instantaneous values. It is cus-
tomary to denote this by the abbreviation R.M.S., root

mean square.
The same reasoning applies, of course, to the pres-
coil, and in fact to any alter-
sure at the terminals of our
nating E.M.F. Since in all instruments intended for the
measurement of alternating pressure (hot wire, electro-

staticand electro- dynamic) the action is dependent on
the square of the E.M.F. applied to the instrument, the
quantity actually indicated is the square root of the mean
squares or in symbols
;
Any shape of current or pressure curve obtained from

FUNDAMENTAL EQUATION 13
an alternating current generator may be represented

analytically by a series of the form
A = A!, sin a +A 3 sin 301 +A 5 sin 5a+ . . .
which represents the resultant obtained by the super-

position of different sine curves or harmonics. Aj is the
greatest ordinate of the first harmonic, whose frequency
is v. A 3 is the greatest ordinate of the third harmonic,
a simple sine curve, whose frequency is 3^, and so on.
Let a be the effective value (either amperes or volts
accordingly as the crest values A lt A 2 A s of the ,
. . .
single harmonics represent amperes or volts), then we

have from (3) or (4)
2n
2
= - i- /r'(A! sin a +A 3 sin 306 +A 5 sin 5a+ . .
,)Wa
In squaring the series we get terms containing the

squares of sines of angles and other terms containing the
product of the sines of two angles, the angles being odd
multiples of a. By a well-known theorem of the integral
calculus the integral of these products taken within the
limits o and 2?r is zero, so that only the terms need be
considered which contain the square of the sines. We
thus find
(T = ~ /( X 2
sin
2
a+ A 3
2
sin V+A 5
2
sin V+
Let pi, pz, etc., be the proportion of the amplitude of
the third, fifth, etc., harmonic to the amplitude of the
first harmonic, then we can also write
Modern alternators give E.M.F. curves which do not

deviate greatly from a true sine curve. That is to say,
the values of p, especially for the higher harmonics, is
i
4 TRANSFORMERS
very small. We may therefore with sufficient approxi-

mation write
where E and
denote the crest values of the first har-
I
monic. The term in brackets may be considered as a cor-

recting factor to take into account the irregularity of the
curve. This factor is generally very small, as may be seen
from the following example, which represents a rather
unfavourable case. Let
A = 5% A = 3% A = 2% A = 3%
then the value of the correcting factor is 1-0024, that is
to say, the effective value has by the presence of the
upper harmonics only been increased by about per
cent. It will be obvious that for all practical purposes,
that is, in all cases where the current is supplied to the
transformer from a modern machine, we may neglect its
upper harmonics and assume that the relation between
E and e is that corresponding to a simple sine curve. If
then E changes according to a sine law, so that the
instantaneous value E, = E sin (27rvt), where E is the
maximum value of the E.M.F., then the effective value
is
e=^
X/2
....... (5)
that E = 27rv</2io~
8
It has been previously shown
represents the maximum value of the E.M.F. in volts
induced in a coil of n turns through which the magnetic
ftux^/passes, if the frequency is
v
complete cycles per
second. -
Combining equation (5) with this expression,
we obtain
X/2
= 4'44v<;zio~ 8
..... (6)
This is the fundamental equation for the determina-

tion of the effective E.M.F. in the coils of a transformer,
FUNDAMENTAL EQUATION 15
provided the E.M.F. curve is nearly sinusoidal. For

some special forms of curve varying greatly from a sine
curve the equation will be given in the next chapter.
Equation (6) applies, of course, to both the primary and
secondary coils. If we distinguish the corresponding
quantities by the indices i and 2, we have
These equations assume the same flux <f>

in both coils ;
they are therefore only applicable if there be no magnetic

leakage. If there be leakage, a correction must be
applied, as will be shown later on.

Since the current does work which is absorbed by
the primary coil, the direction of e 1 must, on the whole,
be opposed to the direction of the current. The
secondary coil gives energy to its external circuit, and
the secondary current must therefore, on the whole, be
of the same direction as e z The power supplied to or
.
given off by the transformer can, however, not generally

be considered to be correctly represented by the pro-
duct : effective current x effective pressure, since the
phases of these two quantities are, as a rule, not coin-
cident that is to say, the current attains its maximum
;
value at a different time from the E.M.F., and the times

at which they pass In
through zero are also different.
the primary coil the product of instantaneous current
and instantaneous E.M.F. is therefore not always nega-
tive, and in the secondary coil this product is not always
positive the energy impressed upon or given off by the
;
transformer during the time of a period is therefore

smaller than Tând Te 2 iz respectively. The deter-
mination of the true energy and true power will be given
in Chapter IV.
CHAPTER II
LOSSES IN TRANSFORMERS ALLOYED PLA TES

DETERMINA TION OF THE MOST AD VANTA GE-
O US THICKNESS OF THE PLA TES INFL UENCE
OF THE E.M.F. CURVE ON THE HYSTERETIC
LOSS INFLUENCE OF THE SHAPE OF CORE
AND COIL ON THE LOSSES CORE AND SHELL
TRANSFORMERS
Losses in transformers. The losses in transformers are
of various kinds. Firstly, those termed "current or
copper heat," which are due to the ohmic resistance of the
coils. These can easily be determined, and the method
of calculating them is so simple as to call for no particular
explanation. Next, loss may be caused by eddy currents
in the conductors or other metallic
parts external to the
iron. Their calculation is extremely difficult, and in some
instances impossible by an adequate design, however,
;
it is
easy so far to avoid the causes of these losses as
to render their effect negligible. Lastly, we have to
consider the losses in the iron core of the transformer,
which are due to two causes, firstly to hysteresis, and
"
secondly to eddy currents. These losses produce iron
heat."
If the induction an iron core passes through a
in
cycle from + B throughzero to - B, and back through

zero to + B, a certain amount of work is converted into
heat. This is the so-called hysteretic loss, and its amount
depends on the quality of the iron, and the corresponding
power depends on the periodicity, and it is directly
proportional to the weight of the iron and the periodicity,
no matter whether the curve which represents the
induction as a function of time is a sine curve or not. If
every half cycle has but one maximum of induction, then

16
LOSSES IN TRANSFORMERS i;
the hysteretic loss depends on this maximum alone, and

not on the manner in which it has been reached.
According to Steinmetz the hysteretic work per cycle and
per unit of weight is given by an expression of the
l
form
S = kW*
(7)
where h is a coefficient determined by the quality of iron

and the units of weight and energy chosen.
The surging of the flux in the iron generates
E.M.Fs. in its mass, and these lead to the formation of
eddy currents. Let us assume the iron core to be of
rectangular cross-section with a and & for dimensions, a
denoting the core width, & its The width a
thickness.
we will
imagine remain
to constant, the thickness
however to be varied. It is obvious that the E.M. F.
will have a maximum along the outer contour of the
parallelogram. This maximum is proportional to the

total number of lines of force, that is, a$B. For a given
value of B the E.M.F. close to the outer skin of the iron
core is proportional to a$, and the same obtains in regard
to the smaller values of the E.M.Fs. prevailing deeper
within the mass of metal. The currents generated
are inversely proportional to the resistance, i. e. the
greater S becomes, with the width a unchanged, the
more the resistance decreases and the currents increase.
An increase of & therefore causes a proportional rise
of the E.M.Fs. which generate the eddy currents,
whereas the latter themselves grow in quadratic ratio
with &. Consequently the loss by eddy currents is pro-
portional to the third power of 8. For circular cores the
E.M.Fs. are proportional to the square of the core
diameter, and the resistance of similar layers is indepen-
1
If the hysteretic loss is determined for that induction where the
permeability reaches the highest value, and next for inductions which
are about 20 per cent, greater and smaller, then the power of B can
be found from the results of the tests. As a rule the value thus found
will be smaller than r6, say 1*55 down to 1*5. In the following I
nevertheless retain the power 1-6 originally given by Steinmetz, simply
because it has become general practice, and also because after all it is
about as correct for values of the induction within wide limits as is the
more exact power 1-5 for values within narrow limits.
2
1 8 TRANSFORMERS
dent of it. Thus the current increases in quadratic
ratio to the core diameter, while its fourth power indicates
the growth of the power consumed. In order to keep
this loss as small as possible, the cores, instead of being
solid blocks, are made up of plates or wires. The loss
in each plate is proportional to the third power of the
thickness, that in each wire to the fourth power of the
diameter. The weight of plates being proportional to
their thickness, and that of circular cores to the square
of their diameter, the loss per unit of weight due to eddy
currents is proportional to the square of the thickness of
the plates or to the square of the diameter of the wire.
Wire cores are used but rarely. Where plates are
employed, the loss per unit of weight can be brought
down to one-quarter and one-ninth by a reduction
of the plate thickness to one-half and one-third
respectively. Thus the loss, if the plates used were
sufficiently thin, could be made disappearingly small.

A limit, however, is set in this direction by the waste
of space due to the insulation required between the
individual plates. The lamination of the iron is for
these reasons not carried further than necessary to render
the losses caused by eddy currents reasonably small. It
has been found that plates varying in thickness between

0*35 and 0*5 mm. (14 to 20 mils) can be used in practice.
The thinner plates are used for higher frequencies up to
about 100, whereas the thicker plates may be used for
lower frequencies up to about 50. Where the frequency
is very small and the magnetic saturation of the iron
fairly low, the plates can be even thicker than 0*5 mm.
An example may show the possible limit in this respect.
Let us assume that practical experience has proved
plates of O'5 mm. thickness quite suitable for a certain
quality of iron, v
= 50 and B = 4000. A new transformer
is now to be designed for v = 20 and B =
5000. What is
the upper limit of plate thickness, if the losses due to
eddy currents per kilogram of iron are to be the same as
before? In the first instance, the E.M.F. producing
the eddy currents was proportional to the product Bv =
200000. With plates of the same thickness it would in
the new transformer be proportional to the product
LOSSES IN TRANSFORMERS 19
Bv= 100000, or only one-half of the former, which means

that the losses in the new transformer, were the plates
equally thick in both, would amount to but one-fourth.
We may therefore increase the thickness of its plates
until the square of the proportion of old to new plate
thickness equals four. In other words, we can double
the thickness of the plates and make them i mm. (40
mils) thick.
The loss through eddy currents, as dependent on
periodicity, plate thickness and induction can be ex-
pressed by a simple formula. We have already seen
that this loss per unit weight is proportional to the
square of the plate thickness. The expression giving
the loss must consequently contain the square of the
plate thickness as a factor. Next, it is evident that
the E.M.Fs. which produce eddy currents at all depths
of the core must be proportional to the product vB
and the wasted energy to the square of that product.
Therefore (vB) 2 is a second factor in the expression. The
only further factor to be determined is a coefficient which
depends on the electric conductivity of the material.
The higher the latter, the greater are the eddy currents
corresponding to a given E.M.F., and the greater,
naturally, must be the losses. Knowing the conductivity,
it is
possible to calculate the loss by means of a formula
containing the factors above mentioned. The derivation
of this formula need not be given here, as it involves a
somewhat lengthy integration. The result is that the
factor depending on conductivity is 0*16 for iron having
an electric resistance about seven and a half times that of
pure copper. The loss per kilogram of iron is
expressed
in watts, the plate thickness in mm., the periodicity in
units of TOO, and the induction in units of 1000. It is,
however, advisable to assume a slightly larger coefficient,

for the following reason. In one and the same core the
length of lines of force differ, and therefore the flux does
not spread evenly over the area of the core, but is slightly
greater for the shorter, and somewhat smaller for the
longer lines of force, than the mean value B in the
formula. Hence, since the eddy current loss is a
quadratic function of B, a nonconformity of the latter
20 TRANSFORMERS
would give a greater total loss, and, in consideration of
this circumstance, it is advisable to assume a slightly
larger coefficiency than found by calculation ;
adding, say,
20 per cent, we get 0*19 instead of 0*16.
The formula is, therefore
(8)
X '
100
where P 7U is the power lost in watts due to eddy currents

in i
kg. of plates, A the plate thickness in mm., v the
periodicity and B the induction.
Bp Induction or number of lines ber aq cm

25,000
FIG. 9. Loss of power in ordinary transformer plates.
It is convenient to represent the formula (8) diagram-

matically by a curve, as shown in Fig. 9.
The lines P w
are drawn for a plate thickness of o'5 mm. (20 mils)
and for a periodicity of 100. The abscissae give the
induction and the ordinates the loss of power per kg.
of iron. In order to obtain the correct values for other
LOSSES IN TRANSFORMERS 21
thicknesses, A, of plates or other frequencies, v, the

ordinates must be multiplied by
Av\ 2
~
In Fig. 9 curves are also shown, which indicate the

hysteretic loss. The formula of Steinmetz is generally
quoted in C.G.S. units so as to give the loss of energy
in ergs per cycle and cub. cm. For practical purposes,
however, it is more convenient so to transcribe the
formula that it gives the loss of power for a certain
periodicity, for example, v=ioo, and a definite weight
of iron, say i kg. If h is the Steinmetz coefficient in
C.G.S. units, then the energy lost in one second, viz. the
energy in erg seconds, will be at v= 100
for i cub. cm. of iron. The specific weight of the plates

is 777, and, therefore, i
kg. of iron contains 1000:777
= 126*2 cub. cm. Hence the loss for each kg. of iron
in erg seconds will be
or = g
10'
By substituting B in units of 1000 we can write
10' \iooo
and (
i
ooo)
1
'
= 63 oo
1
16
hence
p
/ B \
'
(loooj
According to the quality of the plates the Steinmetz

h varies between 0*001 and 0*002. The lower
coefficient
value can be attained with alloyed iron (see next para-
graph), but for
ordinary transformer iron 0*0012 to
O'ooi 6 taken as mean values.
may be The curves in
Fig. 9 are drawn for h 0*001 5, where the loss of energy
22 TRANSFORMERS
per kg. of iron and 100 cycles expressed in watts, is
given
by the formula
For any other periodicity the loss will be

B
P A = o-i2 (-
100 \IOOO7j" (80)
Alloyed plates. Thanks to recent improvements, the

"
composition of the iron used for rolling cut into trans-
former sheets," a material may now be obtained which
has considerably less loss than the ordinary transformer
sheets hitherto used. This new material is known under
"
the name of " alloyed iron (or Stalloy, as one special
1
brand is
called), and although its permeability at high
induction is a little lower than that of ordinary transformer
iron, this is of no moment where, as in transformers, the
induction is moderate. In Figs, ga and gb the losses are
represented diagrammatically for alloyed plates. I have
to thank Professor Epstein for these curves, who has

obtained them with his apparatus, described in Chapter
XI. The ordinates give the sum of the hysteretic and
eddy current losses, and consequently that figure which
has a more immediate interest for the designer but the ;
curves may also serve to separate these losses. The

most convenient way to do this is by means of the lines
on Fig. gb, which represent the loss of energy per cycle
for 100 It will be seen that the
kg. of plates in joules.
total losses are represented by straight lines. These
lines would be horizontal if no eddy currents whatever
existed their slope is consequently a measure for the
;
loss due to eddy currents. This loss is given by the

difference of the ordinates at the start on the left and
that point on the right which corresponds to the chosen
periodicity. These curves show that the loss due to
eddy currents amounts to about one-half of the values
given by equation (8). This means that the eddy
1
Professor Turner has analysed this iron in the University of
Birmingham, and obtained the following composition Carbon, 0*03 ;
:
silicon, 3*40; sulphur, 0*04; phosphorus, o'oi ; manganese, 0*32;

iron, 96*2.
ALLOYED PLATES
current loss in alloyed plates may be calculated from
equation (8) if in the equation the factor o'l is substituted
in place of the factor 0*19. By means of this value it
will then be found from the curves in Fig. 90, that the
12000 (4000 6000 8000 10,000 12,000 14,000 16,000
FIG. ga. Loss of power in alloyed iron.
B= 14,000
10 20 30 40 50
Frequency
FIG. 9/>. Loss of energy per cycle in alloyed iron.
hysteretic loss is
given (in rough approximation) by the
formula (80), if cro8 and not 0*12 is taken as the factor.
Hence the eddy current loss of alloyed plates amounts

to about one-half, and the hysteretic loss to approximately
two-thirds of the corresponding losses in common trans-
former or dynamo plates. As regards the "ageing,"
24 TRANSFORMERS
i. e.the tendency of plates to show an increased loss
when heated for months, or for years, alloyed plates
seem to behave no better and no worse than the plates
of high quality hitherto used, in which this effect is not
noticeable to any very marked degree. The real cause
of the " ageing" has not yet been discovered; all that
can be said is that some plates show scarcely any signs
of ageing, while others degenerate to some extent in the
course of time, the additional loss, however, rarely
exceeding 15 per cent.
Determination of the most advantageous thickness of
the plates. The loss due to eddy currents being pro-
portional to the square of the thickness of the plates, this
loss can with a given induction be diminished by a corre-
sponding reduction of thickness. A lowest limit, how-

ever, beyond which the loss again increases, is prescribed
for such reduction by the condition that the induction
must be raised with the diminishing plate thickness
owing to the waste of space taken up by the insulation.
The thickness of the insulating layer is independent of
the plate thickness and measures about 0*05 mm. (2 mils),
so that the useful total cross-section is
For plates of ro mrn. (40 mils) . . .

95%
07 (28 )
. . .
93%
,, 0-5 (20 )
. . .
91%
;. o- 3 (12 )
. . .
86%
Where flux and total cross-section are given, the
induction, and with it the hysteretic loss, increases as the
plate thickness diminishes the loss due to eddy currents,
;
however, decreases. A certain plate thickness must

therefore exist, for which the sum of these two losses
attains a minimum. This most advantageous plate thick-
ness is naturally dependent on the magnetic qualities of
the iron, its electric conductivity, and the frequency. For
the standard plates on the market the most advantageous
thickness is
For a frequency of 50 . . .
0*25-0*4 mm.
25 0'35-07
The set of curves, Fig. 10, gives the total losses for
ADVANTAGEOUS THICKNESS OF PLATES 25
medium quality plates at periodicities of 50 and 25,

relatively to an ideal induction, which is obtained by a
Watts per kg.

ca
FIG. 10. Loss of power in ordinary transformer plates.
division of the total flux by the total cross-section (iron

plus insulation).
The curve, Fig. u, gives the total loss in o'35-mm.
(14 mils) plates of alloyed iron at 50 frequency. In
transformer it has become usual to
ordering plates specify
26 TRANSFORMERS
an upper limit for the loss at 50 frequency and 10,000
induction. This limit is called the figure of loss (in
012345 6 789
Induction in Units of 1000
10 11 12
FIG. II. Alloyed sheets 035 mm. thick of English make tested by the author.
German, Verlustziffer}. Thus the figure of loss in

o'35-mm. plates of alloyed iron is 1*96 per kg., and
in ordinary transformer plates about 3.
INFLUENCE OF THE E.M.F. CURVE 27
Influence of the E.M.F. curve on the hysteretic

loss. In the first chapter it has been shown that small
deviations from a true sine curve do not materially
influence the ratio between crest value and effective
value of E.M.F. but large deviations do not only
;
influence this ratio, but also the iron losses.

To study this influence we may assume some extreme
cases of E.M.F. and flux curves. If we assume that the
flux has only one maximum value for each half-cycle,
then this value and the periodicity determine the hys-
teretic loss. In that case, as has been mentioned already,
the highest point only of the curve, which represents N
as a function of time, is of influence, and not its shape.
Now we are able to imagine a variety of flux curves
which may all rise to the same maximum and yet may
be built up with considerable divergence. All these
curves are equivalent as far as the hysteretic loss is con-
cerned, but they are not so in respect of the induced
E.M.F. Preference will have to be given to the curve
in which the effective E.M.F. attains the
highest value
for the same maximum flux N. As, however, the induc-
tion, and consequently its maximum value, depend on
the curve of the E.M.F., we can consider the problem to
be as follows : A
number of alternating o current machines
is
given, which all produce the same effective E.M.F.,
but differ in the shape of their E.M.F. curves. To
which curve corresponds a minimum flux, and consequently
the smallest total loss in the iron ?
In order to solve the problem, we must consider
various shapes of E.M.F. curve. We
may suitably start
the investigation with the sine curve and examine what
influence a modification of this curve, in one way or the
other, has on the relation of the effective to the greatest
value of the E.M.F., and also on the maximum value ot
the induction. The sine curve can be modified in two
ways we can either flatten it, or else make it steeper,
:
i.e. with a pronounced

peak. If we proceed with the
flattening of it to the utmost theoretical limit which in

practice would of course be impossible we obtain a
broken line composed of vertical and horizontal sections.
The vertical sections represent the sudden jump from
28 TRANSFORMERS
- E maximum to + E maximum, so that the length of each
horizontal section corre-
sponds to the time of one
half-cycle. Such a curve
could approximately be
obtained by commutation
of a continuous E.M.F.
With an alternating cur-
rent machine this shape
could not be very closely
approximated, so that the
E.M.F. curve of such
be accepted
Fig. 12. Graphic representation of S may
shape
v r .
E.M.F. and flux curves. as the extreme limit of

the flattened sine curve,
not quite attainable in practice. In this case
and E being constant, the fraction

-3- must also be con-
stant, which means that the curve of induction must form

a zigzag line, the points of which fall exactly into line
with the vertical sections of the E.M.F. curve (Fig. 12).
It will at once be seen from the diagram that -7. =
4?. We have for one turn E= f- and e = E, hence

1 at
e = 4.v<f>io~ 8 (9)
If the E.M.F. develops as sine curve, then, from
equation (6), we have
If now
the effective E.M.F. is to be the same in both
cases, then the flux for the shape of E.M.F. curve shown
on Fig. 12 must necessarily exceed the flux for a sine
curve in the ratio of 4*44 to 4. Consequently, if the
induction is the same, the volume of iron, and therefore
also the loss in the iron, must be about 1 1 per cent, greater.
But, as we already stated, the shape of curve drawn in
Fig. 1 2 forms an extreme case which cannot be attained
with an ordinary alternating current machine. In reality
INFLUENCE OF THE E.M.F. CURVE
the curve will deviate from the pronounced rectangular
shape and will approach the shape represented by the
dotted line. The loss in the iron will therefore not reach
the full theoretically possible i r per cent., but will have a
correspondingly smaller rise. At all events, the above
investigation shows that an E.M.F. curve of flattened
shape is unfavourable to transformers on account of the
greater loss in the iron.
Wewill next consider the alternative case, namely, a
pronouncedly peaked curve of the E.M.F. In this case

the zigzag or triangular shape cannot, from the outset,
be assumed to represent the limit. Machines exist
which produce an E.M.F. curve, showing a succession ot
triangles with hollow sides and therefore very steep peaks,
where consequently E is very great in proportion to e.
Now the mathematical investigation of such curves is
scarcely practicable, and

would be of no value for
the present purpose, since
the only decision to be
established is, whether, in
respect of the iron losses
in transformers, a peaked
shape of curve is preferable

to the sine shape. If the
is found to
triangular shape
be more suitable, then it is
,
i ,_
FIG. 13. Relation of E.M.F. and
obvious that its exaggera- flux curve.
tion must be better still.

Let us first determine the curve of induction for the
E.M.F. curve E in Fig. 13. Since for one turn ,

=
at
in absolute units, thiscurve must answer to the condition
that the trigonometrical tangent in any one point a on
the abscissa t is equal to the ordinate of the E.M.F.
curve at that point
=/ tan a = -
(
,J
where <j) {
= ft tan adt 4- const.
<f> f
= const. J/ 2 tan a.
30 TRANSFORMERS
We can determine the constant from the condition that
for t = o, $t
= fa,
we therefore write
fa
= (j)
- l^ 2
tan a
the equation of a parabola. Since E = T tan a we have
For t = T we have <j> t

=o ;
hence
4
TF i
or
8
= (f>
in absolute measure, and since î = v we
have
8
E-Sîo- volt
This shows that the maximum E.M.F. for a triangular

shape of curve is exactly twice as
great as for a rect-
angular shape. The point in question, however, is not
the maximum value of the E.M.F., but its effective value.
This latter is determined for a quarter-cycle as follows
,
t
=t tan a
74 -"
tanA? k tan
2
<^ i T 3
^
V -T- 364
nn
^ = tan a
4 x/3
^ = F= and E = e
By substituting this value in the above equation for

E, we have
e = 4'62V(/>icr 8
...... (10)
INFLUENCE OF THE E.M.F. CURVE 31
The
coefficient is slightly greater than with a sine-
shaped curve of the E.M.F., so that the field may be <$>
weaker at the same induction, and less iron is required to

produce the same effective E.M.F. The equations (6),
(9) and (10) collectively show that the effective value
of the E.M.F. induced in the coil of n turns can be
expressed generally by
where f
a coefficient which depends on the shape of
is
the E.M.F. curve. For this reason f

is called the form
factor it has the values

;
following
(1) The E.M.F. curve is composed of parallelograms
/= i -oo.
(2) The E.M.F. curve is a sine curve f= i *i i.
(3) The E.M.F. curve is composed of triangles

/= 1-16.
If then thesame transformer be successively connected
to three circuits, all of which carry the same effective
E.M.F., but the pressure curves of which have shapes
corresponding to the conditions given under i, 2 and 3,
then the induction B must obviously be different in the
three cases, being greatest in the first and smallest in the
third case. If, for the purpose of comparison, we assume
the induction for the sine-shaped E.M.F. curve as unity,

then it will be
For the E.M.F. curve of rectangular shape, rn.
For the E.M.F. curve of triangular shape, 0*96.
It consequently of a certain, although not of an
is
overwhelming, advantage, that the pressure curve of

machines used for transformers should not be a flattened
curve. power by hysteresis were the only con-
If loss of
sideration would be even advantageous to use E.M.F.
it
curves with very pronounced peaks, but then we would

have to face the difficulty that owing to the large crest
value the insulation would be unduly stressed. Taking
therefore all things into consideration, we find that it is
best to use machines, the E.M.F. curve of which deviates
as little as possible from a true sine curve.
Influence of the shape of core and coils on the
losses. Since the loss in the iron is proportional to its
32 TRANSFORMERS
weight, it must be the endeavour of the designer to keep
the latter as small as possible. The design of the iron
core, however, is somewhat restricted by the minimum of
cross-sectional core area required to convey the given
lines of force, and by the coils' which determine the core
length. On account of the copper heat the coils should
at the same time be of such shape as to have the shortest
possible length for the individual turns. These conditions
are to some extent contradictory, and therefore the best
design must be a compromise between these conditions,
and can therefore only be attained by a step to step
investigation.
The shape of the cross-section of the core is of
material influence on the length of the turns, and there-
fore on the resistance of the coils. A
rectangular cross-
section, for example, is less advantageous than one of
square shape, because, given a parallelogram and a
square of equal areas, more wire is required to enclose
the former than the latter. Generally speaking, a cir-
cular cross-section is preferable to the square. It may,
however, happen that the designer is prevented, by

certain considerations, from making the diameter of
the circle greater than the side of the square. If the
linear dimensions are the same, then the square core
contains times the volume of iron contained in the

7T
circular core, and is preferable to it. This is easily

proved by the following reflection. Let r be the radius
of the circle (?r the side of the square), % the thickness
of the insulation on the core, and d the depth of the
winding. The ratio of the E.M.Fs. for the same
induction is :rr
2
4^
:
2
. The mean length of one turn of
winding is for the circular core section w(2\r -f 8] + d\
and for the square core S(r + 8) + Trd. The E.M.F.
generated per unit length of wire is consequently pro-
2
portional in these cases to irr :
ir(2\r + 8] + d) and
4^ (8[r -f 8] 4- ird\ and the proportion of these values
:
is
> i
CORE AND SHELL TRANSFORMERS 33
i.e. the E.M.F. induced per metre of wire is greater for

the square than for the circular core, the difference in-
creasing with the depth of the winding. This apparent
contradiction is simply explained by the fact that the
square core has a larger volume of iron and the trans-
former of such design a greater output. The material
is always turned to better account in a
larger than in a
smaller apparatus.
It has already been stated that the iron cores of
transformers must be built up of plates or of wires, in
order that the losses due to eddy currents be sufficiently
reduced. Where wires (which need not be specially
insulated) are employed, from 78 to 80 per cent, of the total
space is actually taken
up by iron. Wire cores are used
but rarely now-a-days. Plates, on the other hand, must
be insulated from each other, the insulation consisting
either of a coating of shellac, a layer of oxide, some
special varnish, or else paper. The most reliable of
these is the last-named method of insulation. From 10
to 1 5 per cent, of the available space is lost in this manner,
so that on an average 87^- per cent, of the space actually
contains iron. Consequently the available space of the
coils is better used if plates are employed instead of wires,
and for this reason, and likewise on account of the better
mechanical design, plates are mostly used for the iron
cores of transformers.
Core and shell transformers. It has already been
stated in the opening chapter that the action of a trans-
former is due to the linking together of two independent
current circuits by a magnetic flux. This interlinking
of the electric and magnetic circuits can be carried out
in a great variety of ways. Fig. 6, page 7, represents
one of the simplest forms. The iron core is built up into
a rectangular frame, with its two longer limbs as cores
for the coils. Such an arrangement is called a core
transformer it
;
is characterised by the fact that the
greater portion of the iron is embedded in the coils and
that the external surface of the coils is exposed throughout.
Now we can also imagine an arrangement in which
the respective position of iron and the copper is reversed.
We can assume the rectangular frame in Fig. 6 to be
3
34 TRANSFORMERS
formed by the copper windings of the two coils, and iron
discs slipped over the longer limbs, surrounding them
like a shell. A transformer of such design is termed a
shell transformer, its characteristic feature being that the
coils are partly embedded in the iron.
The core type has a smaller weight of iron and a
short mean length of wire in each turn. The number
of turns, however, is comparatively large on account of
the smaller core section, and the weight of the copper
considerable, although the mean length of the turns is
small. The length of path of the lines of force is great,
and consequently the ampere-turns necessary for the
magnetisation are increased. On the other hand, the
open position and the accessibility of the coils are of
advantage.
Shell transformers have a short magnetic path, and
the magnetisation is therefore obtained with a smaller
number of ampere-turns the coils consist of fewer turns,
;
and, notwithstanding their greater length, require less

wire on the whole than the coils of core transformers.
On the other hand, the iron core is much heavier, the
ventilation of the embedded coils is not so good, and the
coils are only partly accessible.
In order first of all to obtain an approximate idea of
the influence of various arrangements, we will investigate
the problem by an example. For this purpose let us
consider all types calculated for the same output. The
product of current and E.M.F. must therefore be con-
stant. To simplify the matter we will also assume that
the current density in the wires, as well as the current
itself, shall remain constant. The same gauge of wire
must then be used for all transformer windings, and the
number of turns will be directly proportional to the wind-
ing space. The larger the winding space, the larger can
also be the number of turns on each coil, while the total
field N will decrease proportionately. If we further
assume the same induction for all cases (B constant),
then the cross-section of the core will be inversely pro-
portionate to the number of turns, i. e. to the winding
space. The weight of iron and the length of wire may
be taken as a basis to judge the design.
The transformer a, Fig. 14, has a cross-sectional core

area of 400 sq. cm. (inclusive of the space taken up by
the insulation) and a winding space of 60 sq. cm. The
weight of the iron is 200 kg., and the mean length of one
turn 1 19 cm., 100 turns on the primary coil will therefore
require 1 19 m. of wire for this coil. We

will now reduce
the cross-section of the iron to one-fourth by making the
L, 32
FIG. 14. Types of transformers.
core length 10 cm. instead of 40 cm. The number of

turns must consequently be multiplied by four to give the
same E.M.F. The cross-section of the winding space
= Thus we
will accordingly become 4x60 240 sq. cm.
arrive at type b. The mean length per turn now
measures only 78 cm., but since we require 400 turns,
the total length of wire has become greater. It is now
312 m., or nearly twice as much as before. This is

counterbalanced by a reduction in the weight of the
36 TRANSFORMERS
iron to 73 kg., or to nearly one-third. Type a will
therefore have preference where iron is high in quality
and low in price, and where copper is dear. If copper,
however, is cheap, and iron dear and of low quality,
then type b will be chosen. Both designs, however,
admit of considerable improvement. can, for We
instance, modify the design a in such a manner that
both sides of the coil become surrounded with iron, so
as to convert it into a shell transformer. By this means
we obtain the type c. Here the magnetic flux divides
and travels along both sides, so that the shell need only
have half the area of the core cross-section. As with
type a, here also the length of wire is 119 m., whereas
the weight of the iron has been reduced to 1 1 2 kg. The
type c, the iron of which is not excessive in weight,
requires but little copper, and represents the type of shell
transformer in general use.
By modifying the type b in similar manner, we arrive
at type d, which, strictly speaking, belongs likewise to
the shell transformers, without, however, possessing the
advantage of a low weight of copper. The length of
wire is again 312 m., but the weight of iron here amounts
to only 59 kg. This design may be considered an
extreme case of shell transformer. It is costlier in copper
than type c, but has the advantage of a larger cooling
surface. We may also alter the type b so that a saving
of copper results, and this can be attained by arranging
the coils on both the longer limbs, instead of on one only.
This leads to the type e, which is the type of a core trans-
former (compare Fig. 6) now in general use. The mean
length of one turn is now materially smaller than in b,
because the depth of winding has become smaller. The
length of wire is 236 m. and the weight of iron 73 kg.
For convenience of comparison we tabulate the
above results as follows
Type Weight of iron Length of wire

a . . . 210 kg. . . . 1
19 m.
b . . .
73 ... 312
C . . . H2 ,, ... 119 ,,
d . . .
59 ... 312
e . . .
73 ... 236
In these types the magnetic circuit is completely

all
closed, the useful lines of force travel through iron

i. e.
only. There is one more

type of transformer, in which
the path of the ftux lies only partly in iron and for the
rest in air. These are the so-called hedgehog trans-
formers, which were introduced by Swinburne with the
intention of diminishing the hysteretic loss. For this
purpose Swinburne uses a. bundle of iron wires as core
for his coils (Fig. 15, a). The
ends of the wires he opens
out semi-spherically, so-that
each transformer end bears
a semblance to the back of
a hedgehog. The lines of
force close through the air,
as indicated by the dotted Core type.
Hedgehog type.
lines. Thus the hysteretic FIG. 15.
loss is confined to the actual
core of the transformer, while in the shell of air no loss
occurs. This design has not proved successful in
practice. If we imagine two such transformers placed
side by side (Fig. 15, b] with their wire ends bent towards
each other, so that a close iron circuit results, we obtain
the ordinary core transformer. The hysteretic loss of
the transformer, obtained in this manner, can be but
insignificantly greater than the loss of two single hedge-
hog transformers (the additional loss being due to the
small extra length of the wires necessary to join the
cores), so that for this reason alone the hedgehog type
cannot possibly produce any material saving in the
hysteretic losses. On the other hand, this type must
have an increased loss, since, owing to the great
magnetic reluctance of the air shell, the induction must
be greater at the centre of the core than at its ends.
The E.M.F. is proportional to the mean induction,
whereas the hysteretic loss is
proportional to the Jy/of the
'
mean values of B 1
. From this it will at once be seen
that the loss must be greater the value of the induc-
if
tion in the core varies than if it remains constant. Apart
from this the hedgehog transformer possesses the further
disadvantage that it consumes an extraordinarily heavy
38 TRANSFORMERS
current on no load, whereas the types drawn in Fig.
14 have a no-load consumption amounting to only a few
per cent, of the primary current at full load, the hedge-
hog transformer (Fig. 15, a) consumes up to 60 per cent,
of the full-load current with the secondary circuit open.
On this account alone the hedgehog transformer would
be disqualified as distributor in connection with central
station supply. As choking coil, however, the hedgehog
transformer can be used with advantage in this respect
;
its peculiar capability of giving passage to heavy currents
at moderate E.M.F. is very valuable. For all other

purposes the types Fig. 14 c, d and e have in actual
practice proved the best.

CHAPTER III
USUAL TYPES CONSTRUCTION OF THE IRON

PART PROPORTIONS OF THE IRON PART-
HEATING OF TRANSFORMERS RESULTS OF
TESTS CIRCULATION OF COOLING MEDIUM
-VALUES OF c HEAT CONDUCTIVITY
PARALLEL AND AT RIGHT ANGLES TO
THE SURFACE OF PLATES THEORY O.F
HEA TING AND COOLING INTERMITTENT
LOAD INFLUENCE OF LINEAR DIMENSIONS
ON THE OUTPUT WEIGHT OF ACTIVE
IRON
Usual types. The designs commonly used belong all to
the two great groups of shell and core transformers.
The former are of the kind shown in Fig. 15^, where P
and S are the primary and secondary
coils somewhat
oblong in shape and
placed either within or upon each
other, whilst the iron part consists of
rectangular plates each with two
openings in which the winding is
embedded so that only the rounded
ends of the coils remain exposed. In
r , . f FIG. I5<r. -Typical shell
a variety of this design, introduced transformer.
by Mr. Berry, the coils are circular,

and the iron part is arranged in sections placed all round
the circle. This arrangement has the advantage of
exposing a large surface of iron to the cooling effect of
the surrounding medium.
The coils are first wound and the iron plates are
afterwards threaded over them in a manner to be pre-
sently described. The centre part of the plates K forms
the core proper which carries the flux through the coils,
whilst the external part M forms the shell.
39
TRANSFORMERS
In the core transformer the iron part has the shape of
a rectangular frame (Fig. 15^), the longer limbs forming
the cores K, and the shorter limbs the yokes J. The
primary and secondary coils (P arid S) are cylindrical,
and may be placed within each other as shown in the
figure, or as a pile of narrow discs upon each other. For
polyphase work the core type of transformer is almost
exclusively used, and Figs. 16 to 19 show arrangements
FIG. i$(L Kapp's core

transformer. FIGS. 1 6, 17. Three-phase core transformers.
FIGS. 18, 19. Three-phase core transformers.
of the iron part for this purpose. In Fig. 16 the yokes J

are formed of circular iron discs (armature discs may
conveniently be used), and the three cores are placed K
at an angular distance of 120 from each other and
pressed against the yoke discs by top and bottom plates
of cast-iron with slanting brackets engaging the tapered
ends of the core pieces. The coils are placed over the
cores as in Fig. 15^. In Fig. 17 the true symmetrical
arrangement is abandoned the three cores
; being in K
line and joined by top and bottom yokes J precisely as
in Fig. 15^. Each core is surrounded by the primary
CONSTRUCTION OF THE IRON PART 41
and secondary coil belonging to one phase. The advan-

tage of this design lies in the fact that the yoke and core
plates are parallel and not at right angles to each other,
which renders the construction mechanically easier and
magnetically more perfect. Fig. 18 shows also a design
in which the plates of core and yoke are parallel to each
other at the joint. The yokes are a little lighter than in
Fig. 1 6. In Fig. 19 yoke and core-plates cross each
other at right angles at the joints, and to avoid eddy
currents in those pla'ces it is necessary to insert thin
sheets of insulating material. In this respect Fig. 19 has
no advantage over Fig. 16 ;
but the mechanical con-
struction is a little more simple.
Construction of the iron part. In Figs. 15^, 15^ and
17 the plates forming the iron part are shown as complete
surfaces without any joint between that part which
forms the core and that which forms the yoke or shell.
The use of such plates is possible, and would have
the advantage that no interruption of any kind is
offered to the magnetic flux. On the other hand, there
would be the great disadvantage that the coils could
only be wound after the iron part is built up. The
wire would have to be threaded through the openings,
and could therefore not be wound on a lathe. Moreover,
if the wire is stout, it could
scarcely be wound properly
by hand, and to obtain a proper and reliable insulation
would be next to impossible, whilst a fault in the insula-
tion could only be detected after the transformer is com-
pletely finished. These are such serious drawbacks that
the use of closed plates is not commercially possible. It
is better to
arrange the iron part in such way that the
plates may be inserted after the coils are wound and
tested, and for this purpose some sort of joint in each
plate is necessary. It is true that the joint interrupts the
continuity of the magnetic path, and is in this respect an

imperfection, but the imperfection can with a suitable
arrangement of plates be so far reduced as to be without
appreciable influence on the magnetic conductivity of the
iron part taken as a whole. For this purpose the plates
are so arranged and laid upon each other that the joint
in one plate is covered by the solid part of the next.
TRANSFORMERS
The lines of force instead of leaping across the joint can
then in part pass through the solid metal of the neigh-
bouring plates, and since the surface through which they
pass from one plate to the other is immensely greater
than the sectional area of each plate, the magnetic resist-
ance of this by-pass becomes negligible, and we may
regard the iron part so built up as practically equivalent
to one having no joints whatever.
The principle here explained may be illustrated by
the way the iron parts of some of the usual types of
transformer are built up. In the smaller sizes of West-
inghouse transformer the plates are stampings of the form
shown in Fig. 20. Inclined slits are made to both sides
of the central bridge-piece forming the core, and the two
parts of the shell are bent up as shown. In this condi-
FIG. 20. Carcase of Westinghouse FIG. 21. Stamping for shell

transformer. transformer.
tion the plate is inserted into the coils and the bent-up
pieces are bent back so as to lie flat the next plate is
;
then inserted from the opposite side and its bent-up

pieces laid flat. In this manner the joints in each plate
are covered by the solid parts of its two neighbours, and
the continuity of the magnetic path is thereby secured.
Another way to avoid butt-joints in a small shell
transformer is to use E-shape stampings, in which the
middle and outer bars are of different lengths, and to
build up the iron part as shown in Fig. 21.
The carcase of large transformers of the shell type is
generally built up of rectangular plates as shown in Fig.
22. These plates are provided with holes for threading
over wooden pins to ensure exact alignment, and a space
aa is left in the central part to give the cooling medium
access to the inside of the core.
The core in Ferrantis original transformer (Fig. 23)

consists of a bundle of straight iron plates which, is
inserted into the coils. The plates are then one by one
doubled back to form the shell. The length of the plates
is such that at the
joint they overlap slightly, and thus
form an easy path for the magnetic flux.
FIG. 22. Method of building up carcase of FIG. 23. Carcase of Ferranti's

shell transformer. transformer.
In the core type of transformer (Fig. 15^) the plates

for the core and the yoke may be built up into slabs and
laid together with butt-joints if, however, it be desired
;
to avoid the magnetic resistance of these joints the
FIG. 24. Dovetail joint of Kapp core. FIG. 25. Crompton's core.
Author's method of dovetailing, as shown in Fig. 24,

may be employed: In order to show the construction
more clearly the thickness of the plates is much exagger-
ated in this figure. After the iron part is built up it is
held together at the corners by insulated bolts. Since
all the
plates are rectangular there is no waste of material
in
cutting out.
44 TRANSFORMERS
In Cromptons transformer of the older type
(Fig. 25)
the, plates are cut out in L form and inserted into the
coils alternately from one side and the other so as to
cover the joints in one layer by the solid parts of the
next. In cutting or punching the plates there is, on
account of the special shape, some loss of material.
In stamping the plates for the Westinghouse trans-
former there is also some loss of material, namely, that
punched out to form the two windows. To obviate this
loss Mordey has altered the
design in such a way as to
utilise the material punched out as the
bridge-piece or
core, Fig. 26. Each punching yields two pieces, the
shell, which is a rectangular frame, and the core, which is
laid across it. The shell plates are placed over and the
core plates through the coils. With the plates made in
FIG. 26. Mordey core. FIG. 27. Mordey's additional

stampings.
this waythe winding space is of course dependent on the

thickness d of the core, and cannot be chosen arbitrarily
as in Fig. 20. The height of each window is d, and its
width -, as will be seen by Fig. 26. The external
dimensions of the shell are $d and id. Contact between

core and shell plates only takes place over the shaded
areas ;
in other parts there is a clearance between
all
neighbouring plates equal to their thickness. In conse-

quence there is
only half the space within the coils
actually filled by iron, and to carry the same flux the
perimeter of the coils must be greater than in Fig. 20,
where the whole space is filled by iron. To remedy this
defect, Mordey uses additional stampings of the form
shown in Fig. 27, which can also be done without any
waste of material. The internal square of side d is used
to fill up the interstices in the core of Fig. 26, and the
two side pieces to fill up the interstices in the shell.

There are thus five pieces required for every two layers,
and all the material cut up is utilised. Although the
proportions resulting from this method of building up the
iron part are suitable, it is sometimes desirable to vary
slightly the dimensions of the winding space. This may
be done whilst still retaining the Mordey system of punch-
waste of material is then unavoidable.
ing, but a slight
the methods of building up illustrated in Figs.
In all
23 to 27, the principle of breaking joint is maintained,

and the iron part is magnetically equivalent with one
having no joints. It is, however, also feasible to abandon
this principle and build up the iron part with joints
between core and yoke or core and shell. In Figs. 16,
1 and 19 such joints are unavoidable; in transformers
8
of the type shown in Figs. 15^ and 17 they are sometimes
used to facilitate the fitting up. It is then possible to
completely finish each part of the apparatus and to piece

it
together with very little labour. In case of repair a
coil may be taken out and replaced without having to
take the plates out one by one, as is the case with trans-
formers constructed as described above. There are thus
from a practical point of view many advantages in the
use of joints there is, however, the disadvantage that the
;
no-load or idle current becomes greater, as will be

shown later on, and that unless the joints are very care-
fully made, there is loss of power through eddy currents
at the joint. To make the magnetic path as perfect as
possible, it is of course desirable to reduce the distance
between the edges of the plates at the joints as far as
practicable. It is, however, not practicable to reduce it to
zero, partly for obvious mechanical reasons, and partly

because direct contact between the edges would give rise
to eddy currents. With the design shown in Fig. 16 this
is at once obvious. The plates cross at right angles, and
if there were contact between them there would be
formed closed circuits through which considerable eddy
currents would flow and produce heat whilst wasting
power. To avoid this defect a sheet of insulating
material must be inserted at the joint, and the latter
can therefore not be magnetically perfect.
46 TRANSFORMERS
Even where the plates do not cross, but are parallel
to each other, the insertion of a sheet of insulating
material may become desirable if the joint were mecha-
nically perfect, as will be seen on inspection of Fig. 28,
which represents a joint, metal to metal. The thickness
of the plates is shown exaggerated, and the stout black
lines between the plates represent the insulation.
Although the thickness of the plates is supposed to be
the same, we cannot expect that at the joint insulation
and metal will register perfectly. There must be por-
tions on the surface where there is a slight displacement,
so that the insulation in part A
touches not the insula-
tion but the plate in part B, as shown in the figure.
It is obvious that in this case there is metallic connection
A ,, , , , ,
b
Lr\L
FIG. 28. FIG. 29.
between plates a and and that eddy currents as

6,
indicated by the wavy must flow. To interrupt

line
the path of such currents we must separate parts A
and B by inserting a sheet of insulation, as shown in
Fig. 29. Thus the magnetic resistance of the joint
must be increased in order to avoid heating and loss of
power. This refers only to mechanically perfect joints.
If the joint is not metal to metal all over the surface,
the insertion of an insulating separator is not necessary,
but the effect on the magnetising current is the same,
since the mechanically imperfect joint is also magneti-
cally imperfect, that is to say, has an extra amount of
magnetic reluctance.
Proportions of the iron part. The value of a
transformer design is largely dependent on the pro-
portions of the iron part. If, for instance, the cores in
Fig. i$d are made very short, the yoke pieces must be
made correspondingly long in order to obtain sufficient
winding space. At the same time, the depth of winding,
and consequently the weight of copper, is increased,
PROPORTIONS OF THE IRON PART 47
and we thus make the design worse by shortening the

cores. Onthe other hand, if to get a small perimeter
for the winding we make the cores very long, we get a
heavier carcase and a long magnetic path, and conse-
quently more loss in iron and a larger no-load current.
No hard and fast rules can be given for the best pro-
portion, and the only way to determine them is by a
method of trial and error in a series of tentative designs,
paying due regard to such questions as constancy of
secondary pressure, heating, efficiency at average and
maximum loads, the cost of iron and copper,
permissible weight, etc. As a
starting-point
for the design of core transformers, the
proportions shown in Fig. 30 may be taken.
The section of the cores and yokes is sup-
posed to be a square, and if the junctions
are made by dove-tailing, the edges of the
cores may be chamfered so as to reduce
the mean perimeter of the winding. It is
FIG. 30.
convenient to bring all dimensions into
relation with the thickness d of the core given in milli-
metres. We thus have as a first approximation to a
good design
a= 10 +
b= 100 + 2'6d
A= 10 +
B = 100 +
For three-phase transformers of the type Fig. 17 the
width of window should be r$d to i'6d and the length
4</to 6d. The other types (namely, Figs. 16, 18 and 19)
should be so proportioned that they will take the same
size of coils as Fig. 17.
For shell transformers *-- A -
--;
with short cores, the plates
of which are to be punched
without any loss of material,
we have (Fig. 31)
A= FIG. 31.
48 TRANSFORMERS
If however a slight waste of material is to be permitted, the
dimensions a and b may be somewhat increased so as to get
more room for winding and generally better proportions.
a = o'6d b = rid
A-3'2^ B = 2'id
or a = o'jd b= i'2d
A = 3'4<^ B = 2'2d
When using alloyed iron the size of the window may be
even larger still, since the greater permissible induction
means a thinner core. As extreme values may be taken
a =d b = i'd
The depth of the core and shell measured at right

angles to the plane of the plates may be d to /(d.
Heating of transformers. If a transformer is at

work, a part of the energy supplied to it is converted
into heat, which must be dissipated by radiation or con-
vection. Anecessary condition for this dissipation is an
increase of temperature of the transformer over its sur-
rounding medium, and this increase is naturally the
greater the smaller the surface as compared with the
power lost. The rise in temperature is thus a function
of the specific cooling surface <r = -S ,
if
by S we denote the
JL -u
total external surface of the heated body and by P^ the

power transformed into heat, that is lost. The character
of this function can only be determined experimentally,
and varies of course with the material tested. From
experiments made with coils of d.c.c. wire free in air
I have found that T= represents fairly well the tem-
<T*
perature rise up to about 70 C., the coefficient c vary-

ing slightly with the diameter of the wire. It is greater
for thin wire. This formula is, however, not simple
enough for practical use, and since c is not even constant,
it is better to abandon it for the
simpler expression
T=I
HEATING OF TRANSFORMERS 49
and to determine c experimentally for various materials

and cooling methods. If the arrangement is such that the
external air has free access to the iron part and the coils,
the cooling effect is greater, and therefore c smaller than
in transformers placed in a case. It is also obvious that
c will be smaller for the carcase than for the coils, be-
cause in the latter the heat has to travel alternately
through copper and cotton. If the transformer is placed
in a case, the rise of temperature will be greater if the
case contains air only than if it is filled with oil, which
conveys the heat more easily to the walls of the
casing.
The use of oil as a filling-in material not only
reduces the temperature rise, but has also the advantage
of maintaining the insulation when the transformer is
erected in the open air or in damp places. Since the
load is subject to fluctuations, the temperature of the
transformer as well as that of the air in its case must
vary. Thus the air in the case expands and contracts
slowly, and in this way damp air from the outside is
drawn in, and may in time injure the insulation. This
danger is avoided if the case be filled with oil, but since
the temperature coefficient of oil is great, provision
must be made for its expansion, either by not filling the
case completely or by providing a stand-pipe.
The heating of transformers is due to three causes :
hysteresis, current heat, and eddy currents. The hys-

teretic loss is dependent on the mass of iron, the induc-
tion and the frequency, but is independent of the load.
The heat produced by hysteresis is therefore constant if
pressure and frequency are constant, which is nearly
always the case in practice. This heat is developed in
the iron. The current heat is developed in the copper,
and is of course proportional to the square of the load.
The heat produced by eddy currents may be produced
not only in the iron, but also in the coils and in other
metallic parts, especially if the winding is so arranged
as to produce considerable magnetic leakage. It is
however always possible, and for other reasons generally
necessary, to so arrange the winding as to reduce mag-
netic leakage as much as possible. The eddy currents
4
50 TRANSFORMERS
are then only present in the iron and also independent
of the load.
We need therefore only consider the heat produced
in the iron, which is constant, and that due to the ohmic
losses in the windings, which is proportional to the
square of the current. The former may be determined
by measuring the power supplied to the primary circuit
when the secondary circuit is open the latter may be
;
calculated for any load by Ohm's law if the resistances

of the two circuits are known. We
thus find the total
loss. To determine the corresponding rise of tempera-
ture we must run the transformer under the correspond-
ing load and measure its temperature at intervals. It is
advisable to use a spirit and not a quicksilver ther-
mometer for this purpose,
since ifthere should be a
little magnetic
leakage the
metal in the bulb of the
thermometer would itself
become the seat of eddy
currents and get heated
on that account. It is,
-
however, better not to use
FIG. 32.-Shape of heating curve. a thermometer at all, but
to determine the tempera-
ture rise by resistance measurements. The thermometer
can only tell us the temperature on the outside of the
coil, and this imperfectly, whereas we require to know
the temperature inside the coil, which may be very
much higher. As regards the temperature of the carcase
there is not a very great difference between the inside
and the surface, and the use of a thermometer is not so
objectionable, but even here resistance measurements,
of, say, a wire coil wound and placed into the carcase in
such a way as not to be influenced by magnetic flux,
give more reliable results. If we plot temperature as
a function of time, we obtain a curve of the general
character shown by Fig. 32. The curve rises rapidly at
first, and then approaches a horizontal asymptote, the
orclinate of which is the final temperature. This final
is reached with small transformers
temperature slowly ;
HEATING OF TRANSFORMERS 51
only after some hours, and with large transformers of

high efficiency it may take several days. To carry out
the experiment in this way would not only require much
time, but the expenditure of considerable power over a
long period, and be therefore costly. To avoid this
difficulty we may use the expedient of working the
transformer light, but supplying to it that amount of
power which it would lose at full load. This may be
done in one of two ways we may work it light at
:
higher voltage so as to expend the whole of the power

in iron heat, or we may send continuous currents through
the windings and regulate them so as to expend the
whole of the power in current heat. In the first case we
heat only the iron and in the second only the copper, so
that neither method of heating corresponds exactly to
the conditions of regular work. By combining both
methods we obtain, however, a sufficiently close approach
to these conditions. It is convenient to do the heating by
alternating current during the day when the machinery
of the works is in motion, and that by continuous current
during the night, when a storage battery may be used.
If two equal transformers are to be tested, we can
arrange the connections in such way that although only

the lost power need be supplied from outside, each
transformer is working at full load. Details of this
method of testing are given in a subsequent chapter.
To shorten the time of the test the transformers may be
heated up beforehand in the drying stove, which is found
nowadays in all fully-equipped electrical engineering
works.
Resiilts of tests. I have in this manner made an
extended series of tests with various small sizes of trans-
formers and at various loads to find the relation between
rise of temperature and cooling surface per watt lost.
All the transformers tested were cased in, and the cases
sometimes filled with oil, but no air blast or artificial
means of cooling the oil was used. The results are
given in the two curves (Fig. 33). The transformers
stood on a cemented floor in a large covered space, so
that the air had access from all sides, and some of the
heat could also flow away through the floor. Had the
TRANSFORMERS
transformers been tested in the open, the rise of tempera-
ture would have been somewhat smaller if they had ;
been tested in a confined space it would have been

somewhat greater than shown by these curves. Where
the case was filled with oil, no mechanical means were
used to circulate the oil.
The experiments, of which Fig. 33 is a record,
were made with small or moderate-size transformers
(none exceeding 2okw.),
and no distinction was
made between the tem-
perature of iron and
copper. These did not
vary greatly from each
other, so that it appeared
permissible to take for
each load the mean tem-
perature of iron and
copper, and as specific
cooling surface the ratio
of the sum of exposed
surfaces of iron and
coils divided by the total
loss.
Circulation of cooling
medium. With larger
transformers, and es-
~6o pecially with trans-
FIG. 33. formers having artificial
Heating curves of small transformer,
cooling arrangements,
this simple way of calculation is no longer reliable, and
we must determine the temperature rise of iron and
coils separately. The cooling medium may be air or
oil. If air is used, we may either rely on the natural
movement of the air as it rises after coming into contact
with the heated body, and in this case it suffices to
give the air free access to the iron and coils, or we
may artificially increase the cooling effect by blowing
air through and round the transformer. The more
vigorous this fanning, the more efficient will be the
cooling effect, and we may thus distinguish between
CIRCULATION OF COOLING MEDIUM 53
cooling by moderate blast and strong blast, the terms

moderate and strong being defined by the volume of air
provided for each calorie to be carried off. When the
cooling medium is oil, it conveys the heat from the
transformer to the case, and the case must give it up to
the air. The efficiency of cooling will therefore depend
not only on the circulation of the oil, but also on the
greater or lesser ability of the case to give up the heat
to the surrounding air, that is to say, on the external
cooling surface of the case. Where this is insufficient,

it
may be supplemented by a coil of pipe placed within
the case and traversed by a stream of cold water.
Asimple way of moderate blast cooling is to use a
Level of Oil
FIG. 34. Oil-cooled transformer. FIG. 35. Oil-cooled transformer.

Natural circulator. Cooling and circulation assisted
by external channels.
case open at the bottom and placed over an outlet branch

of a ventilating duct which runs under the floor and may
be arranged to supply air to a number of transformers.
A fan is used to blow air into this duct. The air entering
the case of the transformer from below passes the carcase
and coils and escapes through the top of the case or the
upper part of its sides. For strong blast the case sur-
rounds the transformer completely, and the flow of air
through it is directed by suitable channels and slides.
Figs. 34 to 36 show some arrangements of oil cooling.
In Fig. 34 the case is provided with external ribs to
increase its surface (analogous to the ribs in radiators for
heating), and the of the oil is made more
circulation
efficient by the use of a diaphragm S, which separates
54 TRANSFORMERS
the upward stream of the heated oil from the downward
stream of the oil which has given up its heat to the case.
The same objectattained by the external circu-
is
lating pipes C in
Fig. 35. For this arrangement the
total quantity of oil is less than in Fig. 34, but the shape
of the case is more complicated, so that cooling by
external channels is only advantageous in large trans-
formers. For very large transformers a cooling worm, as
shown in Fig. 36, must be
provided. It placed in
is
the upper part of the case,

and consists of many turns
(for the sake of simplicity
of illustration only three
are shown in Fig. 36) of
iron or copper tubing tra-
versed by a stream of cold
water, the speed of flow
being from o'6 to i m. per
second. The oil heated by
the transformer rises in the
FIG. 36. Oil-cooled transformer. Cool-
middle, and after having
ing and circulation assisted by cold
water coil. given up some of its heat
to the worm, sinks down
again on the outside close to the case, where it gives
up the rest of its heat. To attain this circulation it is
essential that the worm should be placed above and not
below the transformer.
Values of c. If a body receives internally as much
heat per unit time as it gives up from its external surface,
then the temperature at any given' point inside or on the
surface remains constant, but from point to point there
is a
change of temperature. This condition may also be
expressed by the statement that there is a temperate
gradient from the inside to the surface. This gradient
may be graphically represented by the slope of a line,
the abscissae of which represent distance from the inner-
most or hottest point and the ordinates' temperature.
If in the direction of the flow of heat the body consists
of metal throughout, then the slope of the temperature
line will be very gentle if the continuity of the substance
;
VALUES OF c 55
is broken up by insulating and, therefore, badly heat-

conducting layers, the slope will be much steeper, and the
difference between the innermost and surface temperature
will be much greater than in the case of a purely con-
tinuous and metallic path. Both these conditions are met
with in transformers. In the iron carcase there is flow
of heat in a direction parallel to the plane of the plates
and also at right angles to it. Parallel to the plates we
have an unbroken metallic path, but at right angles the
heat has to traverse alternate layers of metal and insu-
lation. The heat generated inside the coils, when they
have more than one layer, has also to traverse alternately
copper and insulation, and if the coil is wound with round
wire, double cotton covered, the area of contact between
adjacent wires is smaller than corresponds to the diameter
of the wire, so that not only is the heat conductivity
diminished by reason of the interposition of layers of
cotton, but also because the cross-sectional area of these
layers is reduced. With coils immersed in oil the heat
conductivity from wire to wire is improved because the
oil fills up the interstices. It is also improved if the
wire consists of flat strip, because in this case the contact
surface between adjacent wires is greater than with round
wire. Also, the wire being stouter, the number of turns
is less than in coils wound with
many turns of fine wire,
and, therefore, the number of layers which offer an
obstruction to the flow of heat is reduced and the tem-
perature gradient is lessened.
In addition to the temperature gradient within the
mass of the body we have a sudden temperature drop
from the surface to the cooling medium, and this drop
depends not only on the nature of the cooling medium
(air or oil), but also on the manner in which it circulates
past the heated body.
These considerations make it obvious that the
coefficient c in the formula
must vary between very wide limits according to the

material of the heated body and the nature and move-
56 TRANSFORMERS
ment of the cooling medium. By T is meant the
difference in degrees centigrade between the hottest
point inside the body and the cooling medium passing
^ its surface.
along- In the case of the iron carcase the
surface to be considered in the determination of o- is the
surface of the edges of the plates, with certain correc-
tions for the flat parts which are given further on in this
chapter.
The temperature of the hottest part in a coil which
may still be allowed depends on the nature of the insu-
lating material. For cotton-covered wire 85 C. is the
limit, and with paper insulation 95 C. If thetempera-
ture is allowed to rise higher, there is
danger of the
insulation deteriorating in time. As regards the carcase
this danger is of no moment, but there is the other
danger that the iron may deteriorate magnetically

through "ageing," and for this reason it is also advis-
able to limit the temperature to about 70 C. In modern
transformers the temperature rise of any part over the
temperature of the outside air seldom exceeds 50 C.
In specifying temperature rise, account must, of course,
be taken of the locality in which the transformer is to
be erected. A rise of 50 C. may be quite admissible in
an English sub-station, and far too high for a trans-
former placed under the pavement of a tropical town.
For this reason the rise should always be specified with
due regard to the locality, so that the above limits of
temperature may not be exceeded.
If the transformer has a covering of perforated sheet
metal (intended primarily to prevent accidents by per-
sons touching a live part) we may consider that the
air has free access to iron and coils, and the tempera-
ture drop as defined above will be the difference of
temperature between the hottest part of carcase or coils
and the temperature of the room. If a blast is used,
the temperature drop must be reckoned not to the air of
the room but to the temperature of the air as it leaves
the transformer. The air in passing through the trans-
former gets heated itself, so that at the point of entry
the difference of temperature between body and air is
larger than at the point of exit. The cooling effect of
VALUES OF c 57
the blast will therefore not be the same all over the
surface of the body, but, as it would be difficult to
exactly discriminate between different parts, we may
assume that the blast has everywhere the exit tempera-
ture. In making this assumption we are slightly under-
estimating its effectiveness, but this is a fault on the
safe side. Let, then, / be the rise of temperature of
the blast, and, considering that the specific heat of air
is 0*24 and its
weight 1*28 per litre, we find that
gr.
each litre will
carry 0*307 t gr. cal., which is
off
equivalent to 1*28 t
joules. If we allow 20 per cent.
of the total drop to be lost in heating the blast, we get
about 10 C. as the rise of its own temperature,* and we
thus find that in round numbers 75 litres of air are
required for each kilo-joule of lost energy to be carried
away by the blast. If we double the quantity of air
blown through the transformer, its own rise of tempera-
ture will be halved, and about 150 litres will be required
for carrying off the heat corresponding to the loss of one
kilo-joule. Allowing 50 as the temperature rise of the

body in relation to the we would have, for
incoming air,
the determination of the cooling surface, the
specific
following temperature rises over the air at exit
With a moderate blast of 75 litres per second per kw.
T = 40
With a strong blast of 150 litres per second per kw.
The value of c in the formula T = - for air cooling
may be taken from the following table

58 TRANSFORMERS
When oil is used, the drop of temperature must, of
course, be reckoned not to the external air but to the
oil, and a further drop must be allowed for from the oil
through the case to the external air, or, if a cooling

worm be used, from the oil to the water in the worm.
Here, again, we reckon the temperature of the water at
the exit, not at an intermediate point. The value of o-
must be taken with reference to the external surface of
the case (including ribs) or the surface of the cooling
worm. The following values of c may be used for
cooling by oil-
Coils, d.c.c. round or flat wire . . ^ = 200
Coils, bare copper . . . .
=190
Iron edge surface . . . c 190
Oil to air through case . . c= 1500
Oil to water through pipe . .
= 230
The weight of oil required per kw. of output varies
from about 7 kg. in a ro-kw. transformer to about 3^ kg.
in a 5oo-kw. transformer.
Heat conductivity of carcase. Attention has already
been drawn to the fact that the flow of heat through the
packets of iron plate in a transformer is very different
according as the flow takes place in a direction parallel
or at right angles to the surface of the plates. To deter-
mine the conductivity in both directions, Mr. T. M.
Barlow 1 has made experiments with an electrically heated
carcase, and has found that the two conductivities are in
the ratio of about 90 or 100 to i. These figures agree
well with similar tests 2 made in Professor Linde's labora-
tory at Munich. Since the total heat which has to escape
at the surface is proportional to the distance travelled,
and the temperature gradient is also proportional to this
distance, it follows that if in one case all the heat were
to escape at the edge surface, and in another case at the
flat surface, of the plate, the temperature drop would in
"
1
"The Heat Conductivity of Iron Stamping, Journal of the Inst.
of Electr. Eng., 1908.
2
Mitteilungen ueber Forschungsarbeiten auf dem Gebiete des Ingenieur-
"
wesens, No. 35 and 36.
"
Dr. Ott :
Waermeleitvermoegen der lamme-
lierten Armatur.
VALUES OF c 59
the latter case be 100 times as great as in the former;

or if we wished to have the same temperature drop in
both cases the packet would have to be much thinner
when measured across the plates than along their sur-
face, the ratio being as vîoo to i, that is, TO to i. In
the heat travels in both directions simultaneously
reality
under the same temperature gradient, but the heat
flowing out through a unit of flat surface will be much
less than that flowing out through a unit of edge surface.
Let b represent the thickness of the packet and a its
width, then for unit axial length the flow of heat in the
direction of a will be proportional to ,

and in the
direction b it will be proportional to -.. Now let us
it were
suppose that possible to substitute for the flow
of heat across the packet of section a the flow of heat
along a packet of section a 19 the length of path being the
same as that of the heat emerging at b. In other words,
let us suppose that the flow of heat out of the flat surface
is stopped altogether, and to make up for it the surface b
is increased by the amount a^ We should then have

the equation
a _ a\
100^ a
IS
and from this we find a,=a 100

where r = -, the ratio of width to thickness of packet.
This result may be expressed as follows In calculating :
the total cooling surface of the carcase take the edge
surface in full and add -^ of the flat surface.

100
Theory of heating and cooling. Up to the present
we have considered the stable condition, namely, that
condition in which the heat liberated within the body in
unit time is equal to the heat dissipated from its surface
in unit time. We
must now consider the period pre-
ceding the stable condition, during which the tempera-
ture grows to its final value, and also the subsequent
60 TRANSFORMERS
period of cooling after the supply of heat has been cut

off or reduced. In order to simplify the mathematical
treatment we shall make no distinction between iron and
copper, but take the mean temperature rise as that of a
homogeneous body in which the sum of copper and iron
heat being the equivalent of the total loss of P watt is
liberated in unit time. If S is the sum of the cooling
surfaces for iron and copper, and T
the mean value of
the temperature rise, we have for the stable condition
C1 rr\
T=
r-r\
'
or T =
-s zs
where c or its reciprocal k may be taken as an average
from the values given in the tables on .pages 57 and 58.
During the heating time (the period preceding the stable
condition) part only of the power is dissipated, the
remainder going into and heating the body. During
the cooling time the store of heat in so far as it exceeds
the supply is dissipated.
Let Wc be the weight of copper and W^ the weight
of iron in kg., then the heat stored in consequence of a
temperature rise dy is, in kg. calories
f
(o ii W f + 0*093 W )dy c
o'n and 0*093 being the values of the specific heat of

iron and copper respectively. The corresponding energy
expressed in joules is
4160 (OTI W; + o O93

!
W c )^j/
= Cdy
ifby C we denote the number of joules required to raise
the temperature of the body by i C.
The energy dissipated in the time dx is
and that supplied is dx, which must obviously be equal

to the sum of energy stored and dissipated. We thus
arrive at the equation
dx Cdy Skydx = o
from which we find
P- S
THEORY OF HEATING AND COOLING 61
C -
C, / P
Since P = /ST we can also write
C \
_\
-/
T
the natural logarithm of -= becomes
unity, and we have corresponding to this a particular

value of x, which we may call t, and which depends on
the heat capacity of the body, its surface, and the effec-
tiveness of the cooling method. The time t is inde-
pendent of the power, and being a constant for a given
body and cooling method it is called the "time constant,"
and is given in seconds by
When using common logarithms we have the time in

hours to heat the body up to a temperature rise of
/C- T
x~ li/logf\T yl
\ .... (14)
3600
when T as the final temperature rise is
given by
p
T=
Since for y=T the log =,-- becomes infinite, it is
obvious that the heating curve approaches the final value

y = T asymptotically, that is to say, the final temperature
is
only reached after the lapse of an infinite time. If we
62 TRANSFORMERS
are satisfied with an approach to the final temperature to
within i per cent., the time this is reached is of course
finite, namely
#=-4- c^r hours
3600 Sk
In drawing the time-temperature curve we may con-
veniently take three points, namely-
Time t, to which corresponds
Time 2*3^, to which corresponds .
Time 4'6/, to which corresponds .
The slope of the curve at the origin is easily found

from the condition that for y = o the above fundamental
equation gives -4~

= ~. We also have T= -=-^ and t = ^-r,
dx C S/ SJk
so that
F-'.'C
Having thus found the slope at the origin and the

times corresponding to a rise up to 63, 90 and 99 per
cent, of the final value T, the heating curve is easily
drawn, and from this curve we may find the temperature
rise corresponding to any given time graphically. To
find it analytically we write
^T
-y,
T *
in the form ^ = e~~i
T-y
~
and from this we find y = Tî e~ t
)
having been at work a certain

If a transformer, after
time and acquired a temperature rise T, is switched off
on the primary and secondary, all further supply of heat
ceases and the temperature sinks gradually to that of
the cooling medium. If only the secondary is switched
off, the supply of heat is not completely cut off, but it
is reduced to the amount corresponding to the iron
losses. The temperature will also in this case be gradu-

THEORY OF HEATING AND COOLING 63
ally reduced, but only to the point which corresponds

to the iron losses. The same reasoning applies to the
case where only part of the load on the secondary is
switched off. Any reduction in load will be accom-
panied by a corresponding reduction in temperature.
The investigation of the cooling period is important
in cases where the load is intermittent, as in certain
circumstances it is possible to use a smaller transformer
than would be necessary with a continuous load. To keep
the reasoning general we shall assume that the supply
of heat has not been completely stopped, but only
reduced to a value corresponding to the lost power P .
The energy dissipated is now derived partly from the

supply P and partly from the store Gy in the heated
body. Wethus have the equation
the negative sign in C^y being used to indicate that

now we have to deal with a decrease of temperature.
The solution of this equation is the same as that given
above heating period, the
for the only difference being
in the limits. Now T
is the lower and y the upper limit.
We obtain for the cooling time in seconds to reach the
temperature y over that of the cooling medium the
expression
_ 41*. T TO
If the supply of heat is cut off altogether by switching

off on the primary and secondary side, we have T = o,
and
T
x=tln~
y
The temperature y is given as a function of the time by
The temperature constant has the same value as before,

namely
t-^.
f
-58
64 TRANSFORMERS
Fig. 37 shows the general character of heating and
cooling curves.
Intermittent load. Now let us suppose that a
transformer has been at work on an intermittent load
sufficiently long so that successive maxima of tem-
perature are equal, then the minima will also be equal.
Let T! be a maximum corresponding to the end of the
heating time of a seconds, when the lost power is P ;
and T 2 be a minimum corresponding to the cooling-

time of b seconds, when the lost power is P Further, .
let P! be that amount of lost power which continuously
applied would give the final temperature rise T lt It is

obvious that if a is small and b large as compared with
/, then P will be sensibly greater than P x The question .
FIG. 37. Heating and cooling curves.
is how much greater ? The criterion for the stable con-

dition is :
equality of temperature limits for heating and
cooling, or, mathematically expressed
T-Tx
Let e* =a and e* = |3
then we have the following two equations, from which
the two unknown quantites T and T 2 may be founda
T T = aT aT
Eliminating T 2 and solving for T, we find
a/3-
INTERMITTENT LOAD 65
Since temperature rise and lost power are proportional

we have also
P, 3-fl P.^-,
P a - I P a& - i
a+b
P,(i>
Pj is the lost power which on continuous load pro-

duces the temperature rise T\.
P is the lost power corresponding to the reduced
loadwhen working intermittently.
P is the lost power corresponding to the full load
when working intermittently, and which will produce the
same temperature as the lost power P x at continuous
working.
a is the period of full load, b is the period of reduced
load, and t is the time constant, all these values being of
course taken in the same time unit.
The application of the foregoing theory will best be
shown by an example. Let a 5o-kw. transformer have
800 watt iron loss and 300 watt copper loss in each
circuit at full load, so that the total losses amount to
2 '8 per cent, of the normal full output. Let the time
constant for heating be 10 hours and the final temperature
rise (which would be reached in about 50 hours) be
60 C. Let the transformer be worked intermittently,
the full load period a being 8 hours and the light load
period b 16 hours daily. Let in one case only the
secondary be switched off during the 16 hours, so that
P = 800 watt, and let in the other case the primary be
switched off so that P = o. What output may be taken
from the transformer during the 8 hours' full load daily,
if its maximum
temperature rise is to remain at 60 C. ?
We have
= 11 e =4*93
66 TRANSFORMERS
If the transformer remains magnetised during the
idle period we have
P = 1400(11-1) -8oo(4'93-i) =l78owatt
1 1
-4*93
and be switched
if it off on the primary side during the
idle period we have
P=
1 400
!
n-4'93
( 1 1
- i )
' = 2 100 watt
Since 800 watt is the constant iron loss in both

cases we retain for copper loss 980 watt
first in the
and 1500 watt in the second case, which permits of
an increase in output of i: V^800 and i: V~---

800
respect-
ively. For working intermittently we may therefore

load the transformer to
55 kw. if the carcase remains magnetised during the
idle period ;
68 kw. if the primary is switched off during the idle

period.
The by way of example can
calculation given here
be simplified the period of intermittence is very short
if
Then + and
a
as compared with the time constant.
become small fractions, and we may write
The expression for P becomes
P=
P=
as was to be expected, since

INFLUENCE OF LINEAR DIMENSIONS 67
is the mean power which continuously applied will heat

the transformer to the same maximum value P l as is
actually reached once in every period under the applica-
tion of a power, the magnitude of which alternates
between P and P.
of linear dimensions on the oiitput.

Inflttence Since
the output and other qualities of a transformer are influ-
enced by a number of sometimes contradictory conditions,
the best design must be a compromise arrived at by a
tentative method of trial and error. This is a laborious
process and
;
having once obtained a good design for a
particular type and size we may, when designing another
transformer of the same type but different size, shorten
the labour in its initial stages by the following general
considerations as regards the influence of a change of
linear dimensions on the output.
Let us assume that we have the design of a trans-
former which in actual work has proved satisfactory, and
that we wish to get out the design for a transformer of
larger size, the increase of linear dimensions being in
the ratio i m. : In determining the output of a large
transformer from that of the small or standard trans-
former we may proceed on various assumptions as
follows
I. The large transformer is in every respect an
enlarged copy of the standard transformer. The disposi-
tion of the cooling surfaces and the method of applying
the cooling medium remain the same.
II. The induction and current density in the large
transformer are the same as in the standard transformer.

The cooling arrangements are improved.
III. The induction and current density in the large
transformer are greater than in the standard transformer.
The cooling arrangements are very materially improved.
The condition for all three cases is equal temperature
rise.
Case I. Taking for the carcase hysteresis and eddy
currents loss together, we may assume this loss to be
proportional to a power of B a little higher than i *6, say
i
'63. Let symbols without index refer to the standard
transformer, and those with the index i to the larger trans-
68 TRANSFORMERS
former. Then we have for similar form and the same
cooling method
S =; 2
S k = k, =-,
1
P/ m
where P' and P/ is wasted power.
For the carcase P'= B 63 P/Es/ÎV 63
'
1
,
'
63
T T
- and B = 1
Since the number of turns remain unaltered the E.M.Fs.

are proportional to the fluxes B x i and B x x 2
which m ,
gives
D
= e =^ . f)
= e/;r -i .on
*
el w, l
or e l
The resistance of the windings is lower ;

the length of
wire is m times, its area - - times that of the standard

m 2
transformer, giving "a ratio of resistance of i : For

m
equal heating we have
o "o
rr = i
*
t r
-- = i
~2' 01 ~^2
-'^
/!>! m -
2
-^ m
i\
m
From which we find
This is under the supposition that in enlarging the wire

of the standard transformer we have enlarged its insula-
tion and the wasted space generally in the same ratio as
the diameter of the wire. This is not necessary. The
waste of space if the winding is for an E.M.F. not
very much greater than in the standard transformer
will be proportionately less, so that the cross-section of
the wire may be increased in a ratio a little greater than
i :m*. The result is that for equal heating the current
'
1 5
may be increased in a ratio a little greater than i :m ,
say i : m '
1 61
.
The output of the large transformer will then be

or proportional to the cube of the linear dimensions.

The efficiency of the standard transformer is
P
which may, with very close approximation, be also

written in the form
<yi _
-P_
~F~
p' - T _ p'
P
For the large transformer we have similarly-
^= I
Pi'
-"rr
*i
m*P' =
i P'
~,W mP
As regards efficiency, there is a distinct gain by

enlarging the linear dimensions as regards weight of ;
active material per unit output there is no gain, because

weight and output both are proportional to m*, but as
regards cost there is a gain, since the labour on one large
transformer is less than that which has to be expended
m
on 3 small transformers. There is also a saving in oil
and housing of the one large transformer as compared
with a number of smaller transformers.
Case II. Since the magnetic and electric stresses are
equal, the losses rise proportionately to the weights. (In
reality, as far as copper heat is concerned, the rise is a
little smaller, since, as
already explained, the space wasted
is
proportionately less.) The cooling surface increases
for the same degree of subdivision as the square of the
linear dimensions. For the same heating we have
therefore
P' P'm*
k^ = mk
The effectiveness of the cooling agent must be in-
creased m fold;
or if this be impossible we must increase
70 TRANSFORMERS
also the cooling surface itself by more subdivisions, so
that
The
necessity to provide more cooling surface over
2
and above ;;z S means additional ducts, so that the
pro-
portional waste of space will be approximately the same.
The output will then be increased in the ratio of i m* as :
regards the E.M.F., and also in this ratio as regards the

current
P! = m*P
The improvement in efficiency is the same as before,
since
nPP' i P'
As regards weight of active material per unit of output,

there a decided gain.
is It is only the mth part of the
amount in the standard transformer.
Case ///. Let us assume that simultaneously with the
increase of linear dimensions in the ratio of i we : m
increase the magnetic
__
o and electric stresses in the ratio of
i :
fjm.
The loss in iron heat will then be
'
1 63
B)
It may be necessary to subdivide the carcase and

provide additional ducts, so that the loss will be in reality
greater than shown in the above expression. may, We
therefore, assume that the iron heat will be m* times that
of the Since ^ = ;;z 2 *Jm i and
standard transformer.
ri
= rm~ we find that the copper heat will also be m*
l
y
times that of the standard transformer, so that for equal

heating we have
S^j = m*Sk
Let S/ = /;2 2 S be that cooling surface which is obtained
merely by enlargement without greater subdivision and ;
let n be a coefficient to indicate the increase of this

natural cooling surface by the provision of further ducts,
then
The output increases in the ratio of w?\/^n on account of

the E.M.F., and in the same ratio on account of the
current, so that the large transformer has
5
w
times the
output of the standard transformer. The output per unit
weight is m* times that of the standard transformer. The
efficiency is given as before by
As an example for Case 1 1 1 we may assume a standard

transformer of 10 kw. with 220 watt loss in iron and
85 watt loss in each circuit. The carcase of this trans-
former, if of the shell type and if cooled
by air without a
blast, would weigh about 160 kg., and have a square core
of 14 cm. and a cooling surface of about 3000 sq. cm.
Let us now double the linear dimensions. The natural
cooling surface of the core is then 4 x 3000
= 1 2000 sq. cm.
As this would certainly be insufficient, let us double it
by providing additional ducts. We have now Sj = 24000.
has increased to 22ox 2 3 815 = 32OO watt, so that
'
The loss
<r becomes reduced from
13*6 to 7*5. We shall thus be
forced to either cool by air blast or put the transformer
into oil. Let us do the latter. The oil would then have
to transmit to the external cooling agent (water in a worm
or the air surrounding- the case) 3200 watt iron heat,
and 170 X2 4 = 2720 watt copper heat, or 5920, say in
round numbers 6000, watt in all. Let the temperature
drop from body to oil be 25 C., and that from oil to water
be 30 C, The worm will then require a surface of 46,000
sq. cm., which is that of 50 yards of i^-inch piping. The
cooling surface required for the iron is
S = 190 .
= 24300
or just about equal to what has above been estimated.

TRANSFORMERS
The cooling surface for the copper is
?72O =
S = 200 4 .
21700
25
whereas its natural cooling surface is
only 5000 x 2
2
=
n r C\C\
n = = roS*.
>~]
20000. In this case - The extra 8i

20000
per cent, of cooling surface may be obtained by splaying
out the heads of the single coils a little more.
The output of this transformer will be
P = 10 x 2
5
= 320 kw.
In this case the natural cooling surface had to be
augmented by
additional ducts,
such as a slotted
core, or the sub-
division of the
carcase into thin
packets. If we
had chosen m>2
so as to get more
than thirty-two
times the output,
then these means
would not have
been sufficient,
and we should
have had to pro-
vide further ducts
both within the
copper winding
and the carcase.
To provide the
former, the coils
may be separated
by distance pieces
forming a series
of flat channels
through which the
FIG. 38. Cooling channels in shell transformer. heated oil rises.
WEIGHT OF ACTIVE IRON 73
This, of course, increases the waste space so that the output

will be a little less than in the ratio of i:tn*. The further
ducts in the iron may be provided by splitting the carcase
into still thinner packets so as to utilise a greater number
of flat surfaces in addition to the edge surface. This
method has, however, the disadvantage of lengthening the
coils, andif the flat surfaces are horizontal some special
provision must be made to force the oil through, as its

greater buoyancy on being heated is in this case useless
for circulation. With blast cooling there is, of course, no
objection to horizontal ducts for the flat surfaces. These
are, however, in any case not very efficient, and where
the expenditure of a little extra iron is admissible it is
" ducts on the
better to do without flat," and substitute
cylindrical channels on the edge near the outer circumfer-

ence of the shell, as shown in Fig. 38. By this means the
natural cooling surface is increased by about 70 per cent.
It should be noticed that the additional iron in the shell
necessary for the channels is magnetically nearly inactive,

and does not sensibly alter the iron losses.
Weight of active iron. For shell transformers with
square cores the weight of the carcase is proportional to
cPc, if
by d we denote the width of the central core
measured parallel to the plates and by c its depth measured
at right angles to the plane of the plates. Taking these
dimensions in dm. the weight in kg. is G = $6cPc for a
type with windows dxo'jd to G = 460^ for a type with

windows i'$dx d.
For single-phase core transformers with square cores
the weight of the carcase is given approximately by the
formula
kilogrammes = 93^
In three-phase core transformers the weight of the carcase
varies from 150 to 2OO^3 kilogrammes, the width of the
core d being given in dm.
The output of small transformers working at the same
frequency and without special cooling arrangements is
very roughly proportional to the total core volume (yokes
or shell not counted). For v 50 and pressures not
exceeding 2000 volt we have, approximately
74 TRANSFORMERS
power in kw. =
where V is the volume of the core or cores in cub. dm.
and 7 is a coefficient which varies between 0*7 in small

(say 10 kw.) to i'2 in large (say up to 50 kw.) trans-
formers.
This rule can obviously only be considered as a rough
guide. A
more reliable, but also only an approximate,
formula for the w eight of the active iron in kilogrammes
7
is
.... (16)
IOO
Here P is the
output on a non-inductive load in kw., v
is the frequency, and c is a coefficient depending on the
type and cooling method. The formula is intended for
ordinary cases where the pressure is neither abnormally
high nor abnormally low. As coming within this limitation
we may consider any transformer in which the wire of
the high-pressure coil has a cross-section of not less than
6 sq. mm., and that of the low-pressure coil not more
than 200 sq. mm. The use of this formula should be
restricted to sizes up to about 100 kw. For larger
sizes the perfection of cooling arrangements and generally
the skill of the designer have so much influence that this
formula ceases to be applicable.
With a temperature of the external air of 15 to 20 C.,
and when using ordinary transformer iron, the value of c
in the above formula may be taken for a first approxima-
tion for determining the weight of the carcase when
getting out a new design for small and moderate-size
transformers as follows
Shell type, closed case, no oil . . 20

,, ,,
in oil, no worm . . .10
,, ,,
in perforated case, no blast .
15
,, ,, in closed case, with blast 7 to 10
Core
,,
type, closed case,
,,
,,
in oil,
in
no
no worm ...
oil
perforated case, no blast

.
10 .
15
8
in closed case, with blast 5 to 8

WEIGHT OF ACTIVE IRON 75
When using alloyed iron the values of this coefficient

may be reduced by from 30 to 40 per cent.
It must be borne in mind that the equation is not
intended to give the final and actual value for the weight
of the carcase, but merely a rough approximation, which
is to serve as a guide or starting-point when getting out
the design for a particular case.

CHAPTER IV
THE USE OF VECTORS COMBINATION OF CUR-

RENTSCOMBINATION OF ELECTROMOTIVE
FORCES SELF-INDUCTION AND CAPACITY-
INFLUENCE OF HIGHER HARMONICS POWER
OF AN ALTERNA TING CURRENT
The ^lse of vectors. The working condition of any
alternating current apparatus may be represented in
either of two ways. One is by means of formulae giving
the instantaneous values of current, pressure, power, etc.,
as a function of the time the other is by graphic repre-
;
sentation. The method is applicable whatever

analytical
the precise nature of the functions may be that connect
the current, E.M.F., flux, etc., with time the graphic ;
method, on the other hand, is restricted to those cases

where we may assume that the current or other magni-
tude is a simple harmonic of the time. With few excep-
tions this assumption is quite justified in practical cases,
and as the graphic method is simpler and more trans-
parent than the analytical, we shall in future employ it as
far as possible.
The method is based upon the conception of a vector,
that isa straight line representing by its length the

maximum value of the magnitude under consideration,
and by its projection on a fixed axis the instantaneous
value. The vector is supposed to rotate around a fixed
or a shifting point with constant angular velocity. A
displacement of the vector, parallel to itself, is therefore
permissible. The speed of rotation is such that the
vector sweeps through 360 in the time which elapses
between two positive maxima of the function. Let A
be the crest value of the magnitude under consideration,
and count the time from the moment when the vector A
76
THE USE OF VECTORS 77
isat right angles to the fixed axis, then for an angular

speed of 01, the instantaneous value after the time t will
be a = A sin >/.
/->
j-
The periodic time is T= (0

and the frequency is
(O
27T
A example of a periodic function is the
familiar
movement of the piston of a steam-engine if we assume
an infinitely long connecting rod. There the crank is the
revolving vector
(the centre of
rotation happens
in this case to be
fixed, so that
there is no dis-
placement of the
vector parallel to
itself),
and its
projection on an
axis parallel to
the piston rod
gives the excur-
sion of the piston
from its middle
position. It also
gives, if measured FlG. 39. Vector diagram.

with a suitable
scale, the acceleration, whilst the projection of the crank
on an axis at right angles to the piston rod can, by the
use of a suitable scale, be made to indicate the velocity
of the piston. In Fig. 39, OA
is the crank or vector,
and at the moment represented in the diagram Oa is the

excursion of the piston to the right of its middle position,
Oa being, of course, measured with the same scale to
which OAhas been drawn. The speed of the piston at
that moment is given by oa' but this distance must be
t
measured with a scale containing w divisions per unit of

length. Oa also represents acceleration, but it must be
measured with a scale containing 2 divisions per unit of
>
length.
78 TRANSFORMERS
Ifwe wish to get a graphic record of the movement
of the piston, we can do so by attaching a pencil to the
piston rod and letting it mark a line on a strip of paper
which is drawn with uniform velocity in a direction at
right angles to the piston rod. Since the crank revolves
with uniform velocity, the pencil will mark a sine curve.
To get the datum line for this curve representing the
mid position of the piston we need only stop the crank
when in the Y
axis and let the strip of paper be drawn
backward under the pencil which now is at rest. This
datum line will bisect the sine curve previously drawn,
and we can now distinguish between positive and nega-
tive ordinates, which represent excursions of the piston
to one side and the other of its middle position. Since
both motions are uniform, the abscissae of the sine curve
represent either time or angular displacement of the
crank. The ordinates represent not only linear displace-
ment of the piston, but to a suitable scale also the
acceleration or the force which has to be applied by the
crank pin to overcome the inertia of the moving mass
of the piston and rods. If we wish to get a curve of
velocity, we must rig up a parallel motion in the Y axis

and connect its pencil with the crank by a second con-
necting rod. Weshall thus get a second sine curve
connecting time and piston velocity, and if we lay the two

curves upon each other, taking care to make the times
correspond, we shall find that the crest value of the
velocity occurs simultaneously with the zero value of
the displacement or acceleration, and vice versa. The
crank and heavy piston is a perfect mechanical analogy
of the dynamics of an alternating current.
The linear displacement of the piston represents
quantity of electricity. The linear speed of the piston
represents current strength. The force of acceleration
represents electromotive force. The mass of the recip-
rocating parts represents electromagnetic inertia. The
angular speed expressed in revolutions per second repre-
sents frequency. The mechanical model represented by
Fig. 39 might be also used to represent the addition of
alternating currents having the same frequency. Imagine
the cylinder on one side of the piston filled with water
COMBINATION OF CURRENTS 79
and connection with a stand-pipe of the same diameter.

in
Mark the water level in the stand-pipe when the crank is
in the Y
axis. This corresponds to the datum line in
the sine curve representing displacement. As the crank
revolves, the water will rise and fall in the stand-pipe,
the extreme levels corresponding to the extreme positions
of the piston. The speed of water in the pipe joining
the cylinder with the stand-pipe will be proportional to
Qa! in Fig. 39. Now let us add another cylinder of the
same bore and another crank not necessarily of the same
length, and let the corresponding ends of the cylinders be
in freecommunication. Each piston will now push water
to and fro, and the level in the stand-pipe as well as the
speed in the communication pipe will be the resultant of

the two actions. Let OB represent the second crank set
at the angle /3 in relation to the first crank, then we find
the level in the stand-pipe by adding the projection of
OA to that of OB, that is to say, the resultant level is
Qa + Qb. In the same way we find the resultant speed
as the sum Qa! + O6'. A glance at the diagram shows
that the first sum is nothing else than Or, the projection
of OC on the X
axis, and the second is nothing else than
Or', the projection of OC on the Y axis, the point C
being obtained by shifting one of the vectors parallel to
itself to the end of the other vector. We can now replace
the two cylinders and two cranks by one cylinder and
one crank only, the length of the crank (for the original
diameter of cylinder) being the vectorial sum of the two
cranks, that is to say, a resultant vector which is obtained
by shifting one of the vectors parallel to itself so as to
join the other vector at its outer end.
Combination of ciirrents. The application of the
principle here explained by means of a mechanical
analogy to the combination of currents is obvious. Let
in Fig. 40 M and
1
M 2
be two alternators mechanically
coupled so that they have the same frequency and a
definite angular displacement between the currents pro-
duced. These currents flow jointly through the resistance
C. Each component current and the resultant current is
indicated on the instruments shown. Since all currents
follow the sine law, the readings on the amperemeters will
8o TRANSFORMERS
be proportional to the crest values of the currents, so
that the reasoning used in relation to
Fig. 39 is applic-
able also in this case. The problem is to predetermine
the current which will flow through C if the current
strength of the components and their angular displace-
ment are known. Let in Fig. 41 I' and I" represent the
maxima or crest values of the machine currents in their
correct phase relation.
At the moment to which the diagram refers, the
alternator M
a gives the current Oi' and the alternator 2 M
the current Oz", so that the total current
passing through
C is O*' + (V. If we draw the parallelogram OI'li"O
we see at a glance that the vertical distance between the
AT,
FIG. 40. Diagram of alternators coupled. FIG. 41. Clock diagram of coupled
alternators.
points I and I' is equal to the height of point I" over the
horizontal. In other words, the distance Oz is equal to
the sum of Oz" and Oi' so that Oi represents correctly
;
the strength of the resultant current at the moment for

which the clock diagram has been drawn. The length
Oi is
nothing else than the projection of OI upon the
vertical, and since the above reasoning holds good for
any position of the vectors, it is clear that the projection
of OI times give the instantaneous value of the
will at all
resultant current. We
may therefore imagine that the
conductor C is traversed by a single current, the maximum
value of which is graphically represented by the resultant
of the two current vectors I' and I", and the phase of
which lies between the phases of these two currents. If
we further imagine all lengths in the diagram reduced in
the ratio +/2 i there will be no change in the angles, nor
:
COMBINATION OF CURRENTS 81
in the ratios of vectors, but the resultant vector will then

represent not the maximum, but the effective value of the
resultant current. It will be clear that the above method
of combining currents can be
applied to more than two
currents. We
first find the resultant of two currents,
then combine this resultant with the third current, and
obtain a new resultant, and so on. It is not
necessary in
this operation to draw out the various
parallelograms
of currents all we
;
need do is to add the currents
graphically after the manner of the polygon of forces.
The last line which closes the polygon represents in
FIG. 42. Diagram of currents. FIG. 43. Two alternators mechanically

coupled.
magnitude and phase the resultant current. Let, in Fig.

42, the lines i to z'
4 represent four currents as regards
phasal position, direction, and magnitude, then by
drawing polygon of five sides, of which four
the
correspond to the current vectors, we obtain in the fifth
side the vector of the resultant current as regards phase,
direction, and magnitude.
Combination of electromotive forces. Electromotive
forces of the same frequency, but differing in phase and
magnitude, may be combined in the same manner.
Let, in Fig. 43, Mj and M
2 represent two
alternators
of the same type, and mechanically coupled together so
as to ensure equal frequency. Let three voltmeters, O, I
and II, be connected as shown then I will indicate the
6
82 TRANSFORMERS
terminal pressure of M 1} II that of M 2 and
,
O will show
the resultant pressure. The latter is not necessarily
equal to the algebraic sum of the two other readings, but
will as a rule be smaller. It depends on the magnitude
of the component pressures and their phasal difference.
After what has already been said with regard to the
combination of currents, we need not explain the
combination of electromotive forces at length. If the
three voltmeter readings are available, we can use them
to determine the difference of phase in the E.M.Fs. of
the two machines, as will easily be understood from Fig.
44. Let Oe 1 be the voltage indicated on I, and draw
round e l as centre a circle with radius equal to the voltage
indicated by II. The resultant
pressure must be a line joining O
with some as yet unknown point
on the circle. To find this point
we need only describe a circle round
O as centre with a radius equal to
the reading on the voltmeter O.
The two circles have two points of
intersection, either of which may be
the end of the vector of resultant
Fir, ^--Determination of E.M.F. If the phase of M, is in
phase. advance over that of M 2,
then the
vector of M 2 will lie behind and
therefore above that of M D the rotation of vectors being
clock-wise. The resultant E.M.F. will then have the
position shown in the diagram, and the angle of lag of e 2
is eOe 2 .
Self-induction and capacity. If a circuit be of such
shape that a current flowing through it produces a

magnetic field, the flux of which is interlinked with the
circuit, then any change in current strength will produce
a corresponding change in the flux, and the latter will
induce an E.M.F. in the circuit. This is called the
E.M.F. of self-induction. By Lenz's law the direction
of this self-induced E.M.F. must be such as to oppose
the change of current. Let a current i flowing through
a coil of n turns produce a flux which is interlinked
<
with all the n turns, then a change of flux by d$ in time

SELF-INDUCTION AND CAPACITY 83
dt produces a self-induced E.M.F. which in C.G.S. units is
given by
= n
dt
The negative sign is used to denote that a growing flux

and therefore a growing current produces an E.M.F.
opposed to the current. Let the magnetic reluctance R
be constant, so that flux and current are proportional,
then
or for an harmonic current with crest value I
,
_ n\sin<t
~R~
n*
e = col cos (of
K
Since R is the inverse of a length and n is a number
having no dimension, the fraction represents a

^-
length L measured in if e and
are given in absolute
cm. I
units. If we wish we have to multiply

to get e in volt
the right side of the equation with io" 8 and if we wish ,
to insert I in ampere we must divide it by io so that

the unit of L in the practical system becomes io 9 cm., or
10,000 km., or approximately an earth quadrant. This
unit is called the
Henry. have, therefore We
L= io"
e= Lft)I cos o>/
The E.M.F. is a sine function, and its

self-induced
negative crest value occurs when wt = o, or an even
multiple of TT, the positive crest values when >/ is an
odd multiple of TT.
Since the equation of the current is
i=l sin >/
we see that for ^ = o, when the current is

passing
84 TRANSFORMERS
through zero to become positive, the E.M.F. has its
greatest negative value, whilst for i I when a>t =
- the E.M.F. is
passing through zero to become positive.
The phase of this E.M.F. lags therefore a quarter
period behind the current, and since it must be com-
pensated by an equal and opposite E.M.F. impressed
on the coil, this must lead a quarter period in advance of
the current. We write, therefore
E = o,LI
where E and I are crest values, or if we use effective
values we have
e = t*>Lz (17)
e being the E.M.F. which must be impressed on the
circuit in order to balance and overcome the self-induced
E.M.F. Since L is a length and to an angular velocity,
ft) L is a linear velocity and has therefore the same
dimensions as an ohmic resistance. It is called reactance.
If an alternating E.M.F. be applied to a condenser, a
charging current will flow. Let C be the capacity in

farads, then for E volt continuous pressure applied
to the terminals, the charge in coulomb or ampere-
seconds is
Q = CE
A change de in applied pressure taking place in the
time dt produces a change dQ in the charge, and since
dQ = idt
we have Cde = idt and i = C de
T
dt
Since e is a harmonic function with crest value E so that

e =E sin cot and de = >E cos totdt we find
z = Co)E cos wt
e = E sin wt
Since e is proportional to the sine and i to the cosine
of the same angle, it follows that the vectors of
these quantities are in quadrature, the current vector
leading. The charging current attains its positive crest
SELF-INDUCTION AND CAPACITY
value for wt = o, or an even multiple of 2?r, and is
given
by-
if the capacity C is counted in microfarad and the

E.M.F. in volt. The effective value of the condenser
current is
* = o,C*io- (18)
If a circuit contains inductance L and resistance R, the

E.M.F. to be impressed in order to force the current
i through it must have two components one Rz in :
phase with the current and the other o>Le advance of in

the current by 90. The vectorial sum of these two is,
FIG. 45. Circuit containing resistance, inductance and capacity.
therefore, the hypotenuse of a right-angle triangle with

>Lz and R^ and the angle of lag L) 2 2

is called the impedance of the
circuit. If two or more such circuits are fed from the
86 TRANSFORMERS
same source of E.M.F. we can thus determine the
position and magnitude of the current vector for each
separately, and combine them, as already shown, to get
the position and magnitude of the resultant current
vector. Alternating current problems are best solved
graphically. As an example we may take the circuit
shown on the left side of Fig. 45. On the right side is
the vector diagram. L and R
being in series, we first
determine tg<p and draw the direction of the current
vector in relation to OE, the E.M.F. vector.
Wenext find the value of i in LR. Let this be OA.
The condenser current must be added vectorially to OA.
Since it is in quadrature with e and in advance by 90
we draw AB to the right to represent it. The current
through R0 is in phase with e. Its vector must, there-
fore, be drawn from B vertically upwards. Let BC be
the current through R0. We
finally have to add the
current through L0, which lags behind e by 90. Its
vector must, therefore, be drawn from C to the left.
We thus get ODas the vector of the total current to be
supplied from the terminals of the source of E.M.F. e.
If a self-induction is in series with a resistance and
a capacity the E.M.F. due to the condenser
no 6
e s=r
0)C
is
opposed to the E.M.F. of self-induction
e = o)Lz
and both are at right angles to R/, the E.M.F. due to
ohmic resistance. The equation for the current will,
therefore, be
i= -
.... (20)
V '
Suppose it were
possible to make R zero and let C,
L and w be so chosen that the term in brackets also
vanishes, then we should have a finite current when
the electromotive force impressed from outside on the
circuit is zero, or, in other words, a current once started
INFLUENCE OF HIGHER HARMONICS 8;
would oscillate through capacity and self-induction with-

out receiving any further external impulse, much in the
same way as a frictionless pendulum keeps oscillating
when once set in motion. The alternating current
flows thus with a periodicity peculiar to the circuit, and
determined by the condition that the term in brackets
is zero. From this condition we find
1000
"-
7CL
and since o> = 27rv we find the natural frequency of the
crcut
1 60
C being given in Microfarad and L in Henry.

Influence of higher harmonics. Up to the present
we have assumed that current and E.M.F. follow a
simple sine law. If, however, their curves contain
upper harmonics, the E.M.F. of self-induction, as well
as the charging current, will be somewhat altered.
one case
Since in
e=\^, and in the other
^'=C^ the
investigations can be carried on in the same way provided

we put for i or e the expression
a = AX sin (o>/) + A sin (3oi/) + A 3 5 sin (5>^)+ . . .
and determine the square root of mean squares. We then

find that besides the square of these terms, products of
two of them have to be integrated. The integral of
these products taken between the limits of = o and W
o>/ = zero throughout, so that only the integrals of
2?r is
the squares of the single terms remain, and we thus find
Ij being the crest value of the first harmonic, I3 that of

the third, and so on.
In the same way we find
88 TRANSFORMERS
Ei, E E
8, 5, etc., being the crest values of the different
harmonics.
Power of an alternating current. In order to
investigate the working condition of a transformer we
must be able to determine the power given to the
primary and taken from the secondary terminals. It is
therefore necessary that we should be able to find, either
by direct measurement or in some other way, the power
conveyed by an alternating current. We assume for
the present that current and E.M.F. follow a sine law.
This assumption is made for the sake of simplicity. It
is not always correct, but we shall see later on that the
methods of measuring power which are based on this

assumption are also
applicable in the
general case where
the current as well
as the E.M.F. fol-
low any law, pro-
vided the frequency
of both is the same.
Let, in Fig. 46,
the sine line I re-
FIG. 46. Curves of E.M.F. and current.
present the current
as a function of the
time, and the line E the E.M.F. impressed on any
two points of the circuit, say, for instance, the primary
terminals of a transformer. We
count the time in the
direction to the right. At the time o the current is
negative (the ordinate of the current curve I being
below the axis), and the E.M.F. is zero, At time t^
the current is zero and the E.M.F. has a certain positive
value. The maximum E.M.F. occurs at time 4 and
the maximum current a little later at time 4- At
time /4 the E.M.F. has decreased to zero, but the
current is still positive, though rapidly decreasing. It
reaches zero at time / 5 when the E.M.F. has already
,
a negative value. Since both curves follow the same

law the horizontal distances between their maximum and
zero values must all be the same, that is to say, the time
interval between any two pairs of corresponding points is
POWER OF AN ALTERNATING CURRENT 89
a constant. Thus 4 / 2 = 4 1 = t /6 etc. This time

,
difference between corresponding values of the E.M.F.

and current is the lag or lead of current or
called
E.M.F. respectively. In our example, where the E.M.F.
its zero and maximum values before the current
passes
passes through the corresponding values, we have a
lagging current as compared to the E.M.F., or a lead-
ing E.M.F. as compared to the current. The condition
under which this relation obtains is the existence, in
addition to the impressed E.M.F., of a second E.M.F.
which tends to oppose any and every change of current.
As already explained in this chapter, this is our E.M.F.
of self-induction, and is produced by the change in the
magnetic flux due to the current. If, however, instead
of this opposing E.M.F., there acts an E.M.F. in the
inverse sense, then every change in current strength is
thereby promoted, and the current attains its zero and
maximum values sooner than the impressed E.M.F., or
in other words, we have a current leading before the
impressed E.M.F. Such a second E.M.F. tending to

advance the current is produced by the insertion of a
condenser into the circuit. The condenser takes the
maximum positive charging current at the moment that
the impressed E.M.F. on its terminals passes through
zero in a positive sense. When the impressed E.M.F.
has attained its positive maximum the condenser is fully
charged, and the charging current is zero. When the
impressed E.M.F. now begins to decrease, it is still
positive, but the condenser begins already to discharge,
producing a negative current, which becomes a maximum
at the moment when the impressed E.M.F. passes through
zero, and so on. We
see thus that the condenser current
leads over the impressed E.M.F. by a quarter period.
In addition to the two cases here considered, a
third case is possible in which no second E.M.F. either
advancing or retarding the current is acting in this case
;
(glow lamps fed from a transformer) the current will

have the same phase as the impressed E.M.F., and its
strength will be simply determined by Ohm's law.
The periodic variation in current and E.M.F. may
be conveniently represented by a clock diagram. Let,
TRANSFORMERS
in Fig. 47, the outermost circle be used to mark the
time (somewhat in the fashion of a clock-dial), and let
O/ be the hand of a clock revolving with constant
angular speed. We
count the time from the moment in
which O^ stands horizontally to the left. Let in this
moment the E.M.F. be zero. Describe a circle the
radius of which represents to any convenient scale the
maximum or crest value of the E.M.F., then the pro-
jection of this radius on the vertical gives to the same
scale the instantaneous value of the E.M.F. at the time
to which the posi-
tion of O/ corre-
sponds. Thus at
the time t the
E. M. F. vector
occupies the posi-
tion OE, and the
instantaneous
value of the
E.M.F. is OE,.
We count the
E.M.F. as posi-
tive E, is above,
if
and negative
o if
E, is below the
axis.
FIG. 47. Clock diagram of current and E.M.F. The instantane-

ous value of the
current may be represented in a similar manner, but
the current vector must be drawn with an angular lag <p
behind the vector of E.M.F. Both vectors revolve in
the same sense, and with the same speed, thus always
preserving their relative angular position. Let the
vector OE occupy the position a at time then the if,
current vector will occupy the position a <p,

and the
instantaneous values of E.M.F. and current are E sin a
and I sin (a <p) respectively. The instantaneous value
of the power is
P = EI sin a sin (a <p)
If the vectors perform v revolutions per second so that

T=- is the periodic time, and if

by o> we denote the
angular speed, the following equations obtain

) = 27T
a = wl
da = wdt
da = 2
Since work is the product of power and time, we have
for the work performed by the current in the time dt the

expression
= Vdt
In Fig. 48 are drawn the curves E and I to
represent respect-
ively E.M.F. and
current. By multi-
plying their ordin-
ates we obtain the
ordinates of a third
curve marked P,
which represents
the instantaneous
value of the power,
whilst the area en-
closed between P FIG. 48. Curves of E.M.F., current and power.
and the horizontal
represents work. For ordinates above the horizontal,
the power is positive, or given to the circuit for those ;
below the horizontal it is negative, or taken from the

circuit. To obtain the work given to the circuit during
a complete cycle, we must measure the area of P between
the ordinates for / = o and / = T. counting O the small
shaded parts below the horizontal as negative. The
work corresponding to a complete cycle is
e =
The instantaneous power varies, as will be seen from

Fig. 48, between a small negative and a larger positive
maximum. Let us now suppose that we substitute for
this varying power the constant power of a continuous
92 TRANSFORMERS
current, so that the work taken over the time is the T
same in both cases, then the constant power (which in
future we will call effective power) is the quotient of
work and time, or in symbols
p=l
T
Substituting for P, dt, and T the values given above,

we obtain also
ml ft 6TT
P= / E I sin a sin (a <p) da

27T1JJ
or after integration
P= -I El cosp
2
The same result can be obtained graphically, as was

shown by Blakesley, by using the diagram Fig.
first 47.
Let E and I represent the instantaneous positions of the
vectors, then E sin al sin 6 represents the instantaneous
value of the power. To find the effective power we
would have to draw the vectors in a number of positions,
determine the sum of all the expressions E sin al sin ]3,
and divide by the number of positions. Instead of per-
forming this operation for neighbouring positions of the
vectors in their right sequence, we may do so for a series
of positions which are 90 apart. In this way we would
count the power twice over, and to get the true value of
the effective power we must divide the result by two
times the number of positions. The effective power will
therefore be given by the expression
P = -L 5 (E I sin a sin + EI cos a cos j3)

2m
m being the number of positions counted in their natural
sequence. The expression in brackets is El cos (a fi),
or El cos p, and is independent of the particular position
or value of a chosen. All members of the series
have therefore the same value, and the sum is

simply
;;/EI cos . We thus obtain
cos
p_ <p
2m
COS?
the same expression as before.

E and I are the maxima of E.M.F. and current
respectively. Between these and their effective values
= Er=.\ i= I
.
we have, as already shown, the relations e j=*

V2 V2
The effective power may therefore be expressed by
the equation
P = ei cos <p
..... (23)
In this expression e and i denote
the effective values of E.M.F. and
current respectively, and <p the angle
of lag between current and E.M.F.
For a given E.M.F. and current we
have maximum power, when p = o,
that is when E.M.F. and current
coincide in phase, whilst for p = 90
FIG. 49. Diagram of
power.
the power is zero.
Since i cos <p represents the projection of the current
vector on the E.M.F. vector, we may represent the
power graphically by the area of a rectangle, one side
of which is the E.M.F. and the other the projection of the
current upon the E.M.F. line. Or, we may project
the E.M.F. on the current line, and thus form the
rectangle. measuring the surface it is of course
In
necessary to bear in mind the scale to which E.M.F.
and current have been plotted. If, for instance, the
scale is i mm. per volt and i mm. per ampere, then
the number of square mm. contained in the rectangle is
the number of watt. If, however, we plot the current
to a scale of i mm. per ampere, and the E.M.F. to a
scale of i mm. per ico volt, then each square mm. of
surface represents 100 watt. In Fig. 49, Oi is the
effective current and Oe the effective E.M.F. which
94 TRANSFORMERS
leads by the angle p. The shaded rectangle gives the
effective power which is equivalent with that represented
by a current ih flowing under a potential difference but ,
having no lag. We may thus assume the real current i
replaced by two currents i^ and ih between which there ,
is a phase difference of 90. These can be regarded as

the components of the real current, the component ih con-
veying all the power, and being therefore called the watt-
component of the current, whilst i^ conveys no power,
and is called the wattless component, or briefly the idle
current.
If the E.M.F. and current curves contain higher
harmonics the expression for the power at any instant is
P= (Ii
sin (a <f)
+ I3 sin 3(0 f) + )(Ej sin # 4- E 3
sin 30 + )
where the crest values of the upper harmonics I 3, I5, I 7,
. .and E 3 E 5 E 7
.
may individually be
, , ,
. . .
positive,
or negative, or zero according to the character of the
curves.
The right side of the equation contains terms where
the product of the crest values of the same order, such as
l lt EX, I 3 E3 occur and also terms where the
, ,
. . .
product of two crest values of different order such as

I lt E 3 ;
I 3, E 7 ;
I 3, E 5 . . . occur. In the latter cases the
term also contains the product of two sines of different
order, such as sin(# <p)
sin 3*2 or sin 3(0 $) sin ya.
The differential of the energy is found by multiplying
these products with dt, or what comes to the same thing, by
. Now the integral taken between the limits o and
2?r of terms such as
sin m(oL $) sin nada.
where m and n are different odd numbers, is zero, so

that we need only consider the terms where m
and n are
equal, and we thus find the equation for the energy
27T
8==
~J [!I E I
sin ("-?) sin a
+ I3 E 3 sin 3(01 <p)

sin 301 + . .
.]dfa
We have generally
27T
/sin (ina. inq>)

sin
= /? . . .
\dmoi
J/(sin cos m0 sm cos ;;za sin ma)
m
The integral of the second term in the bracket is zero,
and that of the first term is
27T
cos m$ ma .
7=
/ TT cos m<p
m 9 -J
Since & = TP = 27T P we find the effective power

O)
P = -i-SIÊ IM cos
where ;;z has the values i, 3, 5, 7, etc., denoting the

different harmonics, or
P = J(I 1 E 1 cos f-f I8 ES cos 3<p-hI 5 E 6 cos 5<p-h. .

). (24)
The influence of the higher harmonics on the effective
power is not very great, as may be seen by an example
Let Ij= 100 ampere, E l
= 1000 volt
Is= O ,, E 3
= 300
Is= 20 ,, E 5
= 100 ,,
I7 = 10 E 7
= 80
The effective current 72*6 ampere and the effective

is
pressure 756 volt. Let the lag in reference to the first

harmonic be 30, which corresponds to a power factor
of cos<f
= o'8, then the true power as given by the above
formula is
P = 40880 watt
If we had only taken account of the first harmonics we
should have found 40,000 watt, or about 2 per cent, too
little. If on the other hand we had measured the effective
values of current and E.M.F. and calculated the power
on that basis we should have found 55,000 watt, that is
36 per cent, too much. The error is the greater the smaller
the power factor. For <f = o there is no error, but this
96 TRANSFORMERS
implies proportionality between current and E.M.F. crest
values of the same order. The practical conclusion to be
drawn from this investigation is that with very irregular
curves the power, even if lag be taken into consideration,
cannot be calculated from the indications of an ampere
and a voltmeter, but can only be accurately found by
means of a wattmeter.
CHAPTER V
THE MAGNETIC CIRCUIT ENERGY STORED IN
MAGNETISATION THE HYSTERETIC LOOP-
NO-LOAD CURRENT OF A TRANSFORMER-
SHAPE OF EXCITING CURRENT CHOKING
COIL
The magnetic circuit.
-
The relation between the
magnetising force In, given in ampere-turns, and the
resulting flux <, given in C.G.S. units, is represented by
the well-known relation
In = </>R
where R is the magnetic reluctance of the whole flux-

circuit, that is, the sum of the magnetic reluctances of
its individual parts.
R= '-si 1
O'47T S fJ.
O rv & I//!i
i
i
4
_
^ i
i_
\Si fa S 2 fa
In this expression /j means the length of the magnetic

path in cm. in that part of the magnetic circuit which
has a cross-section of B! sq. cm. and permeability fa,
and so on with the other members. In this formula
it is assumed that the same flux passes through the
different cross-sections S 1? S 2 etc., and this assumption
,
is justified in transformers which are generally so de-
signed as to reduce the leakage flux to a very small

fraction of the main flux.
By inserting the expression for into the formula R
for ampere-turns or exciting force X, and remembering
7 97
98 TRA NSFORMER S
that the flux <f> equals the product of induction B and
cross-section S, we also have
If the dimensions of a magnetic circuit be known, we

can find the values of B 1} B 2 etc., for any given value
,
of the total flux <. The corresponding values of the

permeability ja we can take from a magnetisation curve
of the material used, and the values of / can be
measured off on a drawing of the carcase. have We
thus all the data required to find the relation of X
and (f)
or I and <f>. In other words, we obtain as a <
function of I or as a function
of X. A
curve which repre-
sents this function is called
the char-acteristic curve of the
magnetic circuit. Its general
shape represented in Fig. 50.

is
In order to facilitate the

drawing of this curve, we may
use magnetisation curves,
which give the value
FIG. 50.
X o'8B
that the ampere-turns required for i cm. of path and
is,
various values of the inductions. then find We
If the magnetic path contains an air-gap of length S,

in which the induction is B, then this requires o'
ampere-turns, and the total exciting force is
Energy stored in a magnetic circuit* Let in Fig. 50

the current grow from zero to its final value Ij, and the
flux grow from zero to its final value fa. At a given
moment the current is i and the flux fa If the current
increases by di, the flux increases by tfy. If the increase
ENERGY STORED IN A MAGNETIC CIRCUIT 99
takes place in time dt, the E.M.F. generated in the

n turns of the magnetising coil is, in volt
C := I't
'
I O
The energy d& given to the circuit is eidt, or
But id$ is the area of the shaded rectangle, and it is

therefore obvious that the total energy stored in the
magnetic circuit carrying the flux ( b by reason of the
exciting force \^n ampere-turns, is the area enclosed
between the characteristic curve and the line, ^
multiplied with nio~ 8 . If the characteristic curve were
drawn with ampere-turns instead of ampere as abscissa?,
8
then the area multiplied with icr
would be the energy. It should be
noted that the energy is independent
of the number of turns in the exciting
coil. Since we used volt and ampere,
and e refers to the change of flux
per second, the energy is
given in
watt-seconds or joule.
The more exciting force is re-
FIG. 51.
quired to produce a given flux

the greater is the energy stored. If there were no
^
air space through which the flux has to pass, the
characteristic would for a moderate induction be very
steep, and the area enclosed between it and the axis
would be small. It follows from this that it is chiefly
that part of the magnetic circuit which lies in air which
forms the store of energy. For fairly low inductions,
that is, values of B which lie below the knee of the
magnetising curve and a long air space, the first term in
the equation for \n is enormously greater than the other
terms, so that we may neglect the latter and write
I=o*8B8
The characteristic then becomes a straight line, Fig.
51, and
ioo TRANSFORMERS
=tga. being the slope of the characteristic to the

.
rx
horizontal, and S the section of the air space. The

8
energy is the shaded area multiplied by io~ or ,
if we take as a unit for the flux the megaline, and as a

unit of exciting force 1000 ampere-turns, the energy is
in joule . . . 10 x shaded area

in km. . . .1*02 x shaded area
From X = o-8BS and = io~ 8 we find-
In this formula S is given in sq. cm. and in cm.; their

is the volume of magnetised air given in cub. cm.
product
Let this volume be given in cub. dm. or litres, and call
it V, then the energy in joules can be written
e = 4 v(-B_y .
(25)'
v
Viooo/
The energy contained in one litre of magnetised air
/By. i / B v
stores therefore I )
joule, or ) metre kilo-
^- ^-
gram, as shown in the following table
Energy stored in one litreof air traversed by a magnetic flux

with induction B.
= O'S I 3 8 12 IS 2O
1000
Joule =i 4 36 256 476 900 1600
Metre kg. = 1*02 4^08 3-67 26 48 92 163
The property of the magnetic circuit to act as a store
of energy is utilised in the construction of so-called choking
coils, as will be explained at the end of this chapter.
The hysteretic loop. It has already been stated that
the change of induction taking place continuously in the
carcase of a transformer involves a certain loss of energy
and a corresponding generation of heat. The energy lost
per cycle is the difference between that which has been
THE HYSTERETIC LOOP \>;&i
stored in one stage of the process and that which is

returned in another stage of the process. If the material
undergoing cyclic magnetisation is air, the whole of the
energy stored in magnetisation is again recovered in

demagnetisation but with iron this is not so.
;
If we
determine experimentally the magnetisation curve of any
sample of iron whilst this is being carried through a
complete cycle (methods for such tests are given in
Chapter VIII), we find that the I/z-B curve follows one
path for increasing values of I
and another for decreasing
values, the two curves forming
a loop, the so-called hysteretic
loop. This is
represented in
Fig. 52, where the magnetising

force is plotted horizontally and
the induction vertically. The
sense in which the cycle is
performed is shown by arrows.
Since the energy representing
the half cycle from - B to + B
is
proportional to the area en-
closed between the right side of
the loop and the B axis, and
since the energy in the return FlG 52
half cycle from +B to -B is
represented by the area between the left side of the loop

and the same axis, the energy lost in one complete cycle
is
proportional to the area of the loop, and is given by
the equation
6 = area x io~ 8 joule
Since we B and not

take ordinates to represent as <
before, the loss an element of the magnetic

refers to
circuit i sq. cm. in cross-section. Let / be the length of
this element, then the loss per cycle per cub. cm. will
be found by dividing the above expression by /, or in
mathematical language
+E
*X
102 TRANSFORMERS
Now, _
4^ = H, the magnetising force in C.G.S.
units, so that we may also write
-B
/"
= -. /H^B
H^Bio- 7
-^. joule
-B
.
(26)
-B
In this latter form the hysteretic loss per cycle for i

cub. cm. is
generally given in text-books.
No-load current of a transformer. Since the cyclic
magnetisation is accompanied with certain losses, the idle
or no-load current must have a component in phase with
the induced voltage, a so-called watt component. It
must also have a wattless component, that is, one lagging
90 behind the induced voltage, and consequently co-
phasal with the flux. The idle current is very small in
comparison to the full-load current, only a few per cent.
of it, and as the ohmic drop with full-load current is
only a few per cent, (sometimes less than i per cent.)
of the working or impressed voltage, we may, in cal-
culating the idle current, assume equality between
induced and impressed voltage, and determine the two
components of this current on the assumption that one is
in phase and the other in
quadrature with the impressed
E.M.F. We shall also assume that the impressed E.M.F.
and both components of the no-load current are sine
functions of their respective crest values. As will be
shown presently, this assumption is not strictly correct as
regards the wattless component, but we make it never-
theless in order to simplify the calculation. The error is
not important.
Let P be the power wasted in the iron in hysteresis
7j
and eddy currents, and e the effective value of the

NO-LOAD CURRENT OF A TRANSFORMER 103
E.M.F. ;
then the effective value of the watt component
for a single-phase transformer is
P*
The effective value of the wattless component is
where B is the induction in the air-gaps of the butt-joints

and & their combined length. The symbols .r and / under
the summation sign have the meaning already explained,
FIG. 53. Diagram of idle FIG. 54. Magnetic path

current and power. in shell transformer.
whilst n is the total number of turns traversed by the

idle current. The effective value of the latter is
The relation between these quantities is shown in

ig. 53, which also shows in the shaded area the power
wasted by the idle current.
In a core transformer having the same cross-section
in core and yoke / is the mean
length of the lines of force
taken round the rectangle of the carcase, and the sum-
mation in the formula for ^ has only one term. If the
yokes are of larger cross-section than the cores (they

would obviously not be made smaller) then the summation
has two terms, one for the two cores and the other for
the two yokes.
In a shell transformer / is taken round one window
as shown in Fig. 54. Since the flux divides in the
shell the cross-section of the latter may be half that of
the core. It may also be greater, but not smaller.
104 TRA NS FORMERS
To determine the idle current for a given impressed
voltage e we first calculate the flux < from
4'44 n
100
and then the various values of B = -
Ampere turns per centimeter

10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
2 3 6 6 7 ft ft 10 11 12 13 14 15 10 17 18
Ampere turns per centimeter
FIG. 55. Characteristic of Transformer Plates.
The length of air-gap at each butt-joint, if any, may

be taken as at most mm., or 8= 0*025, and the values
of :r corresponding to each value of B may be taken from
the curves, Fig. 55. It will be seen that up to a certain
value of the induction alloyed iron requires a little less
exciting force than ordinary iron only if the induction ;
SHAPE OF EXCITING CURRENT 105
is very large is this relation reversed. A very high

induction is, however, hardly admissible, since, even if it
be possible to keep the temperature down by some special
cooling arrangements, the large magnetising current is as
a general rule objectionable.
In three-phase core transformers each limb serves
one phase, and since one phase is magnetically and
electrically always in series with one or two others, the
value of / and & is only half what it would be in a
single-phase transformer. Thus the wattless current of
one phase must be calculated on the assumption that
it has to
propel the flux through its own core, two butt-
joints, and about half the yokes.
Shape of exciting current. A knowledge of the
hysteretic loop, sheered over so as to include the influence
of butt-joints if any, is necessary for the determination
of the shape of the exciting or magnetising current and
the resulting idle current. We assume, as before, a sine
wave of E.M.F., and consequently also a sine or rather
cosine wave of induction. To each value of B corre-
spond two values of z'
M ,
the one given by the ascending,
the other by the descending branch of the loop. The
graphic construction of the z^-time curve is then as shown
in Fig. 56. The hysteretic loop on the left gives B as
a function of i^ the sine curve on the right gives B as a
function of the time. By combining the two it is easy
to find the curve giving ^ as a function of the time.
Thus a particular value of B, to which correspond the
currents zi and z 2 occurs at the times ^ and t^.
, Then
plot o^ over / lf and oi2 over t# giving z\ and i2 as two
points of the curve required. The whole curve ^ may
be drawn in this way. The curve ih is simply a sine
curve in quadrature with B. The idle current curve is
z'
the resultant of these two components.

Choking coil. A coil having large inductance and
small resistance, so that it will cause a lag of current of
nearly 90 behind the E.M.F. impressed on its terminals,
is called a
choking coil. Such a coil might be obtained
by altering a transformer so that it will take a large no-
load current. The alteration would consist in enlarging
the butt-joints and suppressing the secondary winding as
io6 TRANSFORMERS
superfluous. Since now the current is large, the ohmic
loss no longer negligible and the watt component of
is
the current will be increased. If R is the resistance of
the winding and the loss in iron is as before P //}

we
have
CHOKING COIL 107
for calculating the wattless component we may neglect

and simply write
n \/
The lag being intended to be nearly 90 we must so
design the coil that ih will be very small in comparison
with i^ and then i will be very nearly equal to i^. We
can therefore write
o-SBS
t = or i =
n\/2
TOO
if by P we denote the apparent power of the coil

expressed not in watt but in volt-ampere. The true
power taken by the coil is
P =P A +R; 2
and its power factor is
cos = -Pj ^
ei
This is be a minimum.
to It is obvious that maxi-
mum lag and therefore minimum power factor will be
reached with one particular value of i, which we find
from the condition that
d
this gives P A = Rt* ..... (27)

which means that greatest choking effect combined with
a minimum waste of power will be obtained if the coil is
so designed as to make the loss in iron equal to the
ohmic loss. For purely constructive or commercial
reasons it may sometimes be necessary to slightly depart
from this Thus, it might be desirable to use
rule.
existing stampings for the carcase, or a particular gauge

of wire in stock, etc. If, by accommodating the design
to commercial requirements, the difference between the
two losses does not exceed 10 or 15 per cent, of their
1 08 TRANSFORMERS
sum, a departure to this extent from the rule of best
proportions is admissible.
The equation which connects current and induction
in the air-gap may be written in the form
if
by I we denote the crest value of the current. Let
similarly E be the crest value of the E.M.F., then we have
the relation
E=
where > = ITTV. Since current and E.M.F. are in quadra-
ture, their simultaneous values at any instant are
i=\ sin a
e = E cos a
The instantaneous value of the power is-
ei= El sin a cos a
It is positive for all values of a between the limits o and
,
or arand ;
it is
negative for all values of a between
the limits and TT, or TT and 2?r. We see thus that during
one complete period the choking coil takes in energy
twice and gives off energy twice. To find the amount
of energy stored in a quarter period we determine
= /E I sin a cos adt
7T
/~2
da.
S= J/El sin a cos a-
CO
El-
e=-CO
i
2
=
CO
i
>-27TV
ei
.... /
(28)
Q x
then frequency, choking volt and ampere to be

If
passed are known, we can determine the energy to be

CHOKING COIL 109
stored in the air-gap. Inserting the values for E and I
we obtain
1"\ CV
I
~
2(0 C
=
o4r-
or if we take for the volume the litre as unit
B \2
*-4V(-5L\
\ ooo/ i
the same expression which has already been found on

page 100. Each litre of magnetised air is the carrier of
220
As an example for the applica-
tion of these formulae take the
case of a choking coil which is
required to let pass 10 ampere at
a choking pressure of 100 volt
when the frequency is 50. We
have then = 314 and ei=* 1000.
This gives
FIG. 57. Choking coil.
1000
e = = 3-2 joule
.
If we assume an induction in the air-gap of 5000,

one litre of air will store 200 joule, so that we require a
*"?
*
O
total volume of air-gaps of - cub. dm., or 32 cub. cm.,
100
which may conveniently be subdivided into four gaps, as
shown in Fig. 57. The dimensions are figured in mm.
We make the central core 5 cm. square, and the shell
of double the cross-section of the core. The latter is
2 i 7 sq. cm. The total weight of iron is 1 2 kg., and with
an induction of 11,600 in the core the total loss of
power in the carcase 14 watt, alloyed iron being
is
used. The flux is 0*25 megalines, and this requires 180

turns of wire for 100 volts choking E.M.F. The winding
space allows of the use of 3-mm. wire, to which
corresponds a resistance of 0*14 ohm and an ohmic loss
110 TRANSFORMERS
of 14 watt. The excitation required for the air-gaps
is o'8 x 0*64 x 5000 = 2550
ampere-turns, or with 180
turns 1 4' i ampere crest value, corresponding to TO
ampere effective value. The power factor is
14+ 14 =
0-028
1000
to which corre-
sponds an angle of
lag
<
= 88 24'
The coil shown

in Fig. 57 being
of small power, no
special provision
need be made for
cooling. For large

power coils it is,
however, advisable
to provide air chan-
nels. Fig. 58 shows
a design which I
have found to take
up a large apparent
power per unit of
weight, and yet
keep fairly cool
when in continuous
use.
It is advisable to
FIG. 58. Ventilated choking coil. put a hardwood
lining into the gaps
so as to preserve the value of 8 correctly.
CHAPTER VI
DESIGN OF A CORE TRANSFORMERBEST DISTRI-
BUTION OF COPPER LOSSES AT DIFFERENT
LOADS TIME CONSTANT FOR HEATING-
WEIGHT AND COST OF ACTIVE MATERIAL-
BEST DISTRIBUTION OF LOSSES TRANS-
FORMERS FOR A SPECIAL SERVICE TRANS-
FORMERS FOR POWER TRANSFORMERS FOR
LIGHTING ANNUAL EFFICIENCY ECONOMIC
IMPORTANCE OF SMALL LOSSES CONSTRUC-
TIVE DETAILS
Design of a core transformer. As an example of the

practical of the formulae developed in the
application
preceding chapters we will now get out the design of a
2O-kw. transformer of the air-cooled core type, for a
secondary pressure of 160 volt on open circuit at the
usual frequency of 50. The primary pressure is 3120
volt, giving a transforming ratio of 19*5 to i. trans- A
former of this pressure rnay be used to supply lighting
current to 5o-volt metallic filament lamps which are
arranged in three circuits and balanced by an autotrans-
former, as will be explained in Chapter XII. The lamps
being three in series, require 150 volt, so that 10 volt
remain for covering ohmic losses in the transformer and
lamp circuits.
The coefficient c in formula (16) for the weight of
iron is about 8, if we use alloyed iron. This gives
170 kg. To get the side, d, of the core we may use
the formula
G = 6o(^+ o*2) 3= 70
1
This gives d= 1*22 dm., or if we chamfer the corners

d= 1 2 '5 cm. We can now design
&>
the carcase and
in
I 12 TRANSFORMERS
determine its exact weight. The shape will be as shown
in Fig. 59.
To find approximately the cooling surface we assume
for the present that the outside diameter of the coils will
be 26 cm. and the inside diameter 15*6 cm. Of the
inside surface only about one-half can be considered as
available for cooling, whilst the whole of the outside
surface is of course effective. The total cooling surface
of the coils is 8800 sq. cm. The edgeways cooling
surface of the carcase is 2720, and to this has to be
added about 80 sq. cm. for the
flat surfaces, so that the total
410 cooling surface is 2800 sq. cm.

The area of the core, allowing
for the chamfered edges, is 1 30,
that of the yoke, which is
perfectly square, is 136. Allow-
ing an induction of 7900 in the
core (7550 in the yoke) we find
8
with alloyed iron a loss of 190
watt. This is permissible,
since it gives <r = 147, to which
corresponds T = 44 C.
The total weight of iron is
178 kg., that is, only slightly

more than estimated above.
To reduce the perimeter of
the coils, we chamfer the edges
FIG. 59. of the core as shown in Fig. 60.
This is done by stepping the
width of the outer plates. After all the plates are
assembled the core is bound up by strong tape to keep
the plates from bulging. The thickness of this tape
serving is about 2 mm. The primary and secondary
winding we arrange as concentric cylinders, the secondary
being inside, as shown in Fig. 60. In order to enable
this coil to be easily slipped on, we require a clearance of
2 mm. over the corners, so that the inner diameter of the
supporting cylinder becomes 160 mm. Its thickness

need not exceed 5 mm., bringing the inner diameter of
the coil itself to 170 mm. The depth of winding and
DESIGN OF A CORE TRANSFORMER 113
the mean perimeter can at present only be fixed approxi-
mately from the following consideration. The dimen-

sions of the iron part being given, we find the distance
of centres of cores to be 125 + 160 = 285 mm. The
external diameter of the primary winding can therefore
not exceed 285 mm., and must in reality be somewhat
smaller, since a certain clearance is necessary not only to
avoid touching of the two coils, but also to make up for
possible irregularities in the manufacture, and leave a

margin in case the same size of transformer should have
at any future time to be made for a higher primary
pressure, when finer wire requiring more space would

have to be used. A certain clearance is also necessary
to allow for the circulation
of the cooling medium. If
we fix this clearance at
20 mm., we obtain for the
external diameter of each
primary coil 265 mm., so
that between the inner
surface of the secondary
and the outer surface of
the primary winding, there
remains a distance of
1(265 7 170)
= 47 '5 mm.
This distance is made up FIG. 60. Design for core.
of the depth of winding

a,
in thesecondary coil b, the clearance between the latter
;
and the inner side of the primary cylinder c, the thick- ;
ness of the latter and d, the depth of the primary

;
winding.
For clearance we may reckon 4 mm., and for the
thickness of primary cylinder 5 mm., Ieaving47'5 9 38*5
mm. to be apportioned between the two windings. It
is obvious that the
depth of primary winding will be more
than half this amount, because not only is more space
required for the insulation of the thinner wire, but the
perimeter being larger, the gauge of the wire must be
correspondingly increased in order to avoid getting too
much resistance in the primary. As a first attempt in
apportioning the space we may assume that 60 per cent*
Ii 4 TRANSFORMERS
of it will be filled by the primary, and 40 per cent, by the
secondary winding.
Best distribution of copper. The correct distribution
of the available space between the two windings is in so far
important as the total loss by ohmic resistance becomes a
minimum equally divided between the two windings.
if it is
This may be shown as follows. Let the total winding

space available for both coils be b (in our case
b = 38-5
mm.), and the depth of the secondary winding a, then b a

is the depth of the primary winding. Let be the inner D
diameter of the secondary coil. Its resistance is propor-
tional to the number of turns n.2 and to the mean perimeter
of one turn ?r(D -M) it is
inversely proportional to the
;
cross-section of wire or to put it in another way, the

;
resistance with a given length of wire inversely propor-

is
tional to the depth of winding, a. By comprising all

constants into one coefficient K, we have for the loss in
the secondary coil
P =
a
For the primary coil we obtain a similar expression,
except that the coefficient K must be multiplied by the
ratio of windings n lt n 2 and ,
we thus obtain
P
r __* D + a+. -i
.
2
! rv% -
LI
b a n.2
2 2
Since n*i* ;* 2 /2 ,
we can also write
The total loss by ohmic resistance is therefore
b-a
.
a
In order to make this loss a minimum, a must be so
chosen that the term in brackets becomes a minimum.
By differentiating and equating to zero, we obtain after
some reduction
BEST DISTRIBUTION OF COPPER 115
Since a negative depth of winding is physically

impossible, we take the upper sign
4 2
By equalling the two terms in brackets we also get
=o
2
and we see from this that there will be minimum loss by

ohmic resistance if the loss is equally divided between
the two circuits.
If, instead of using concentric cylinders, we arrange
the two circuits in a group of flat coils alternating primary
and secondary, as shown in Fig. 6, p. 7, we should also
have to consider the question how the available winding
space must be distributed between the two circuits in order
to obtain minimum total loss. In its general form the
problem is as follows Given
: a volume of copper v, the
lengths /! and /2 of the two circuits and the currents ^
and z*2 determine the cross-sections q l and q^ of the re-
;
spective conductors, that is to say, the most advantageous

division of v into
so that the total loss

P*=^ 8
+ ^ 8
shall become a minimum. In this expression ^ and

are the resistances which can be found from
r,
= K-
1
fi
The condition of minimum loss of power is
d> d-
P v = o,
r>
or -,
r>
P v
=o
agi dqi
Selecting the first, we have
ii6 TRANSFORMERS
Since = -- we have
q^ j-
4
,
also
d /K/i .g
-
K/
-- 2
-y- I- --
I? + 7 /
î \
f i (^-A
from which we find
the .condition for minimum total copper heat. It is equal

current density in both circuits. But since the mean
perimeter is the same in both circuits, and the number
of turns is inversely as the current, the volume of copper
is the same in both circuits. The loss being proportional
to current density and volume, we find that also in this
case,, the same as in the case of concentric winding, the
total loss is a minimum for equal copper heat in primary
and secondary.
In the example under consideration, we have chosen
cylindrical coils, and have assumed as a first approxima-
tion that the available winding space will be divided in
the proportion of 40 per cent, for the inner (secondary)
cylinder and 60 per cent, for the outer (primary) cylinder.
We must now investigate whether this division is in
conformity with the law just passed. The flux is
1
30 x 7900 = o6 i *O2 7 . i .
To get 1 60 volt on open circuit on the secondary

we require 70 turns
;^ 2 = 70
%=i365
If for these figures and the original assumption of
40 and 60 per cent, winding space we determine the size

of wire and the losses, we shall find that the loss in the
primary is greater than that in the secondary. The

40 60 division is, therefore, not correct. By a method
:
of trial and error, which need not be repeated here, we

find ultimately that the division in the ratio of 37:63
gives equality of losses. thus find We
Best radial depth of inner cylinder .
14 mm.
outer ,, .
24 ,,
BEST DISTRIBUTION OF COPPER iij
We may now draw the coils, and determine from

the drawing the exact mean perimeter of each. This
gives
^2 = 0*575 m. 77-1-0755 m.
The cross-section of wire may now be determined.
For fixing the length of the coils we have to consider
the height of the window in the iron frame (in our case
45 cm.), and leave sufficient space for clearance "and 'the
end flanges of the cylinders. The total space required
for these purposes is about 3^ cm., leaving 41-5 cm. net
length of coil. Each secondary coil must contain 35

turns of wire. If these were arranged in a single layer,
the wire would have to be wound on edge. Although
this presents no difficulty with naked wire which is after-
wards insulated by paper insertion, it is not so easy with
cotton-covered wire, and in this case it would be better
to wind the wire on the flat and make two layers, one
with 1 8 and the other with 17 turns. Since the space of
one turn is lost in crossing over from the lower to the
upper layer, we must arrange the width of the wire to be
not xVth, but TVth of the net winding space. This gives
415/19 = 21*8 mm. The thickness of the wire is already
determined by the depth of winding, which we found
must be 14 mm. Allowing 0*5 mm. for the thickness
of covering (or i mm. in all), we find that the section of
the wire will be 6 x 20*8 mm. Since it is, however,
scarcely possible to lay on succeeding turns with mathe-
matical accuracy, it will be advisable to take the width a
little less, say 20 mm., so that the actual cross-section of
the wire becomes 6 x 20 = 120 sq. mm. The length
of winding is 70 x 0*575 40*5 m., and if we allow 0*5
m. for connections, we can take 4T m. as the basis oil
which to calculate the resistance of the secondary winding.
The formula for the resistance, taking rise of temperature
into account, is
D
R 0*02/2
--^-
/,being the length in metre and q the cross-section in
square millimetre. We
thus obtain
R = 0*00682
2
1 1 8 TRA NSFORMER S
A similar calculation made for the primary winding
shows that we have to use round wire of 3*1 mm.
diameter (covered to 3*67 mm.) in six layers of 122
turns, and one layer of ten turns on one and eleven turns
on the other limb. The length of wire is
/!
= 1030 m.
and its resistance warm is
R! = 2-8
Losses at different loads. have now all the data We

required for calculating the losses at different loads.
They are given in the following table
Output in kw. .10 12 15

. . 20 25
Secondary current, ampere 63 76 96 128 161
Primary current, ampere 3*4 3*9 5 . 6*6 8*4
Total loss in copper, watt 59 Si 224
133 374
Loss in iron, watt . .
190 190 190
190 190
Total losses, watt . .
249 271 323 414 564
Efficiency, per cent. . .
97*6 97*8 977 98 97*8
.1
s surface of
Specific cooling )
\ 149 IOQ
^
66 r
36 23
coils, sq. cm. J
Temperature rise of coils)

15 24 45 /o
if air cooled, deg. C. }
The specific cooling surface of the carcase is 147, to

which corresponds a temperature of 44 C.
Time constant for heating. The time constant is
found from-
CT
P
where P is the lost
power, in our case for a load of
20 kw., 190+224 = 414 watt, and C the number of
watt-seconds required to raise the temperature of iron
and copper by i C. if radiation be neglected. We have
178 kg. of iron and 112 kg. of copper; hence
C 4200 (178 x o'n + 112 x 0*093) = J 26000
126000 _._45
/= = T 3600 seconds
414
t = hours
WEIGHT AND COST OF ACTIVE MATERIAL 119
For the copper only the time constant is 2*4 hours.

If this transformer be required to work only four hours
every evening, the loss in the copper may be increased,
as shown on page 65, to
_ 2-4
P =224-5-
e - i
P = 224 X I '21
The output may therefore be 20^1*21 = 22 kw. with-

out exceeding the temperature rise of 45 C.
Weight and cost of active material, In designing
this transformer we have used stampings of alloyed iron.
Had we used ordinary transformer sheets we should, in
order to remain within the same temperature limit, have
been obliged to reduce the induction to B = 5600, thus
reducing the secondary terminal pressure to about 1 10
volt, and the output to 14 kw. Alloyed iron costs about
twice as much as ordinary transformer sheet, or, say,
i6d. against &d. per kg. Copper may be taken on an
average at is. 9^. per kg. The weight of active material
and its cost is as under
Quality of iron.
Ordinary. Alloyed.
Shillings. Shillings.
Weight of carcase .
178 kg. cost . 120 240
Weight of coils . 112 ,, ,.,
.
196 196
290 316 436
Output, kw. . . . .
14 20
Cost per kw. output . . . .
227 217
It will be seen from this table that it
pays to use the
more expensive iron.
Best distribution of losses. In designing this trans-
former we paid no heed to the question whether the
losses between iron and copper are correctly distributed.
We designed mainly with a view to moderate and equal
temperature rise in both parts, and it so happens that
the loss in iron is not very different from that in copper.
Theoretical conditions demand that they should be equal,
or nearly equal. The latter is the case in the transformer
120 TRANSFORMERS
under consideration. The proof of the rule for minimum
lossis as follows. Let a given transformer working on a
constant pressure circuit be loaded to different degrees
corresponding to various values of the primary current i.
Since primary and secondary current are at all loads
very nearly proportional, the total loss in the windings
2
will be correctly represented by Ri whilst the loss in ,
the iron is a constant quantity C. The efficiency is

ei cos
73
= -.

= o. or
di
(ei cos <f

2
+ Rz + C) e cos <f
= ei cos <p (e cos <f
+ 2 Re 2
)
ei cos <p + Rr + C = ei cos <p

-f 2 Rr
C = R* 2
. ... . .
(29)
The condition of maximum efficiency is therefore equality

of copper and iron losses. If, then, it is desired to
obtain maximum efficiency at any particular load, the

transformer must be so designed that at that particular
load the losses in the iron and copper shall be equal.
Absolute equality is the theoretical requirement. The
practical requirement is that there shall not be too great
a difference between the two losses. glance at the A
table on page 1 8 will show this. The least proportionate
1
loss, namely 2*07 per cent., occurs at the load of 20 kw.,

when the copper and iron losses are very nearly equal.
At 15 kw. and at 25 kw. there is certainly not equality
between the two losses, and yet the proportionate loss is
not so very much increased in either case, being 2*16 per
cent, of the output at 15 kw., and 2*26 per cent, of the
output at 25 kw. At the lower output the iron loss is

nearly two-thirds of the total loss, instead of one-half, as
theory requires, and yet the efficiency is only diminished
by about one-third per cent. and even at half-load, when
;
the iron loss is as much as three-quarters of the total

loss, the efficiency is only diminished by 0*4 per cent.
It is necessary in this place to refer to the rule for
BEST DISTRIBUTION OF LOSSES 121
5
1
the best distribution of losses given by Arnold, because
this being different from the rule deduced above,
may cause the reader to doubt the correctness of the
latter. Arnold argues as follows For a given carcase :
there must be a most efficient winding, so that at a

specified output the losses, shall be a minimum. This
will certainly give the best efficiency for that specified
output, but not the best efficiency the transformer is

capable of giving, if the output is so chosen as to make
the losses equal. The most efficient winding as defined
by Arnold may be determined from the following con-
sideration. For a given pressure the number of turns is
inversely proportional to the induction B. The cross-
section is inversely proportional to the number of turns,
and the resistance is directly proportional to the number
of turns and inversely proportional to the cross-section.
The resistance is therefore proportional to the square of
the number of turns, or, which comes to the same thing,
the resistance is inversely proportional to the square of
the induction, and the copper losses with a given current
can be represented by
where C is a constant. The iron losses due to hysteresis

and eddy currents for a given pressure, to which the
induction B corresponds, are represented by
QB^ + CaB"
where C^ and C 2 are constants pertaining to the given
carcase. Since the output is given the efficiency will be
a maximum for minimum loss, the condition for which is
r6C B + 2 C B = o
1
0<6
2
This equation means that the copper losses must be

equal to the sum of 80 per cent, of the hysteretic loss,
and the full eddy current loss in the carcase.
1
Arnold, Die Wechselstrointechnik, Vol. II., p. 80.
122 TRANSFORMERS
According to Arnold's law the copper losses should be
smaller than the iron losses according to the law deduced
;
on page 120 they should be equal. Both laws have been

mathematically deduced. They are different, and therefore
one must have been deduced from wrong premisses, or its
interpretation must be wrong. The fact that in practical
work absolute equality of losses is not essential does not
eliminate the difficulty, since we are dealing with a
principle and not with the magnitude of an error when
departing from the principle. To show where the fallacy
lies let us assume that we have wound a
given carcase
according to Arnold's law, so that at a given output
(which must obviously be smaller than the possible output)
the losses shall be a minimum. Let P be the output, C
W
L
the iron loss, which constant, and

is 1 the copper loss,
which is proportional to the square of the current, and

therefore also proportional to the square of the out-
put, or
2
w^yp,
By Arnold's law W, < C
and
will be a maximum for the particular value of the output

PX for which the winding has been designed. Now let
the transformer be worked at another output, which we
call P, and determine its value so that TJ
shall be an
absolute maximum. We form
and find
P + 2C
Since there can be only one maximum, and since y is
an absolute maximum, it follows that
The discrepancy is now explained. The law of equal

TRANSFORMERS FOR A SPECIAL SERVICE 123
losses gives the maximum efficiency of a transformer

which is the right size for the load Arnold's law gives
;
it for a transformer which is slightly too large for the

load.
Transformers for a special service. In getting out
the design of the 2okw. transformer we have only
considered heating and efficiency, and have aimed at a
type which is satisfactory in both respects. It may be
called a good design for general purposes. The question
now arises whether the design or method of working may
be modified with advantage if the service is of a special
character, such as exclusively for power or exclusively
for lighting. In the first case we have a long hour
service at a fairly steady load, and in the second case
we have a load fluctuating between wide limits with a
maximum lasting only a short time.
Poiver transformers. Wehave seen that a fairly
wide departure from the law of equal losses does not
very materially lower the efficiency. We
may therefore
increase the output of a given transformer by working
it at higher pressure, that is, at a
higher induction and
with an increased iron loss, without seriously impairing
its efficiency. The increased iron loss may force us to
use specially effective cooling appliances, but where, as in
power plants, the units are large, the extra cost on this
head is not a very serious item in comparison with the
increased power obtainable from the given active material.
There is, however, the objection that the heavier iron loss
will materially reduce the efficiency during times of light
load, and will also be a serious charge on the plant during

the time of no load. Whatever the weight of this objec-
tion in a power plant may be, it must be greater in a
lighting plant, because with the latter the no-load and
light-load time is longer. It is also conceivable that in a
power plant this objection has no weight at all, as for

instance in the case that the water power is of a nature
to preclude storage or that the transformer can be
switched off during the idle time. In such a case only
the efficiency at full load is of importance, and it is
obviously permissible to make the iron losses large.
This course has the advantage of reducing the capital
1
24 TRA NSFORMERS
charge per kw. Let us then see how our transformer

might be worked if on a power plant where reduction of
efficiency at light loads is of no moment. If we increase
the pressure so far that the induction becomes 50 per
cent, greater we shall have increased the output to 30
kw.i and only doubled the magnetising current, as will be
seen by reference to Fig. 55. The iron loss will have in-
; crease.d to 380 watt, making the total loss 604 watt, or
only about 2 per cent, of the output. The efficiency has
therefore not been diminished by working the iron at a
higher induction, and if it were possible to work the
copper at a higher current density without overstepping
a safe temperature limit, the efficiency would be even
improved. Thus at 40 kw., or a secondary current of 170
ampere, the copper loss would be (j) 224 = 4oo watt,
2
total losses 780 watt, or 1*95 per cent., and the efficiency
a little over 98 per cent. It will be seen that we have
nearly doubled the total losses (41410780 watt), and this

will require very perfect cooling appliances. It may not
pay to provide them for the small transformer under con-
sideration, but
it will
certainly pay to do so in a plant
where the transformer unit is several hundred kw. The
cost of active material per kw. has therefore been brought
down to something like us. even in this small size, and
in large units it would of course be less still, It may be
asked, why not go further in this direction and bring the

.cost lower still? The precise limits for induction and
current density may possibly not have been reached in
;our example, which is given merely to show the tendency
of design, but the limits exist, and are fairly well defined.
As
regards the induction the limit set by the exciting
force, will probably be reached sooner than that set by
heating. By providing additional cooling channels, vigor-
ous circulation of oil and large cooling worms we can
dissipate a very large amount of heat, but the high induc-
tion requires also a large magnetising current, which,
being in quadrature with the pressure, reduces the power
factor, and thus reacts unfavourably on the generating
plant. As regards the copper heat, this is always more
difficult to dissipate than iron heat, and as the heat is
proportional to the square of the current the point at

TRANSFORMERS FOR LIGHTING 125
which a further increase of current becomes inadmissible

on account of temperature is defined within fairly narrow
limits.
Transformers for lighting. Since a lighting trans-

former must be under pressure continuously, and since
fullload only lasts a short time, a reduction of iron losses
will save more energy than the same reduction of copper
losses at full load. We have seen that a moderate
departure from the law of equal losses has very little
influence on the efficiency, and we may therefore make
such a departure in whichever direction is suitable to the
special service. In a transformer for power we found
that it pays to make the iron losses greater than the
copper losses, since we increase thereby the output
obtainable from a given weight of active materials, and
thus get a cheaper transformer without sacrificing
efficiency. Where power is cheap, this can also be done
with lighting transformers, but when the commercial
value of the electric power is that usually charged for
lighting current, say 3^. to $d. per kw.-hour, then the
iron loss becomes important. We have long hours of
reduced load or no load. We must therefore reduce the
losses, which are going on all the time (that is, the iron
losses), even if we increase" the copper losses.
Let us now see how the transformer to which the
table on page 1 1 8 refers could be modified to make it
suitable for lighting. If we wish to reduce the iron
losses by one-half we shall have to lower the induction
to B = 5000. With the same winding this would give on
the secondary terminals 100 volt, a suitable pressure for
glow lamps. We may allow a maximum ohmic drop of
1*4 per cent, at full load, giving 270 watt copper loss.
The full output would then be 14 kw., and the total losses
365 watt, or 2 -6 per cent, of the output. The cost of the
active material is 31^. per kw. At half load the copper
loss drops to 68 watt, and the total losses are 163 watt
or 2*33 per cent, of the output. It might be asked why
limit the output to 14 kw. ? In the transformer for power
we have assumed that the copper heat due to a second-
ary current of 1 70 ampere, which is 400 watt, could
be dealt with satisfactorily. If copper heat were the only
126 TRANSFORMERS
consideration we could work the transformer at 17 kw. ;
but we have also to consider the question of voltage

drop, which in this case would be from ohmic resistance
about 1 7 per cent. As will be shown later, there is also
a drop due to magnetic leakage, and this is proportional
to the load. The total drop must be kept within a
narrow limit, and it is this consideration which really
restricts the output. It will thus be seen that for equal
maximum output a lighting transformer must be more
expensive than a power transformer.
Yearly efficiency. When considering efficiency we
have up to the present used the term in its generally
understood sense, namely, as the ratio between the power
given out to the power put in. There is, however,
another kind of efficiency also to be considered, namely,
the ratio to the energy given out to the energy put in,
both being taken over a period sufficiently long to
embrace all the fluctuations in the demand, say over a
year. From the financial point of view this is the more
important efficiency. To illustrate the effect of design
on the yearly efficiency we must take some special case,
both for power and light. We
shall assume that the
two transformers are not switched off during the time
of no load, and that the demand taken over the whole
year is as follows
For power 100 per cent, of full load for 500 hrs.
5 M 1000
25 ,, ,, 1500
idle for . . .
5760 ,,
For light 100 per cent, of full load for 200 hrs.
50 i
8o
10 ,, ,, 3000
,, idle for . . . .
4760
This refers to a transformer at a sub-station where

the demands from customers of various types overlap
and thus lengthen the time of service. The service
hours of a transformer for a single house would be con-
siderably shorter. We
can now determine the losses
YEARLY EFFICIENCY 127
at different loads for the three types of transformer,

namely
G the design for general purposes
MM
. .
P . .
power
L . .
,, ,, ,, lighting
To make the figures directly comparable we assume

that three types have the same full load output of 40 kw.
all
In increasing the G
and L transformers so as to give the
same output as the P transformer, we must of course
take account of the advantage derived from the increase
of linear dimensions as explained in Chapter III. The
output increases with the third, the losses only with the
second power of the enlargement factor m*
It is now possible to apprise commercially the relative

value of the three designs. We
have found that the
cost of active material for P is 4365. The cost of active
material of the other two types will be greater in the
ratio of the third power of the enlargement factor.
The P type being the smallest, labour, case, oil, terminals
and other accessories will add a larger proportion to the
cost of materials than in the other two types. Taking
128 TRANSFORMERS
these things into consideration, we may take it that for
the same (moderate) voltage and a full load output of
40 kw. the cost for the three types will be as follows
For G .
P .
55
L 125
The question which
should be used can be
type
answered if we know
the price of the kw.-h. to be charged
to the energy lost in the transformer. With power
current under' id. there can be no doubt that the P type
is
commercially the best. With lighting current at $d.
or \d. the L type is more advantageous. Reckoning
10 per cent, of the capital outlay as an annual fixed
charge, the working cost of transformation per annum
comes out as given in the following table
It will be seen that for cheap power current the

choice lies between G and P, whilst for the more expen-
sive lighting current L is the best type. The differences
in annual working charges are not very great. This is
due to the use of alloyed iron had we used ordinary ;
transformer sheet in these designs the annual charges

and their differences would have come out larger.
Economic importance ofsmall losses. Station engineers
generally reckon the cost of lost energy not at the selling-
price of the current, as we have done here, but at the so-
called engine-room cost, which of course, much lower.
is,
In this connection it is
interesting to notice that even if
lostenergy be only reckoned at engine-room figures (say
about id. per kw.-h.), a transformer with large iron loss
ECONOMIC IMPORTANCE OF SMALL LOSSES 129
FIG. 61. 20 kw. transformer. Scale I :6.

1
30 TRANSFORMERS
is a heavy charge on the working expenses. Most of
the transformers now in use for lighting and general
purposes have been made some years ago, and may be
assumed to have from 2 to 3 per cent, iron loss. This
means that each kw. of transformer installed uses up
250 kw.-h. in iron heat annually, whereas a modern
transformer made with alloyed iron would only use up
about 80 kw.-h. By replacing the old with modern
transformers a saving of 170 kw.-h., or 145-. worth of
current per kw. installed, would annually be made.
Against this must be set the annual charge of, say, 10
per cent, on the capital outlay, which on an average
may be taken at 2 per kw. Of the 14^. saved 4^.
must therefore be set aside for the annual charge,
leaving a clear saving of icxr. per annum, so that after
four years the cost of the new transformers could be
completely written off.
Constructive details. After this digression into what
may be termed the financial side of transformer design,
we return to the particular 2O-kw. transformer for general
purposes dealt with in the beginning of this chapter.
The core with the coils in place is shown in Fig. 61.
This transformer is put into a perforated sheet-metal
case, and is therefore air cooled by natural draught. It
can with such a covering only be used indoors, and then
only in dry places. For use in damp places and out of
doors a perforated covering is of course inadmissible.
We must put the transformer into a tight case of cast-
iron. As this would greatly diminish the effectiveness of
the enclosed air as a cooling medium, we use oil as the
internal cooling medium, and provide the case with ribs
to increase the cooling effect of the external air. See
Figs. 62, 63 and 64.
In building up the carcase, the plates for the core and
yoke are cut to size and punched for the bolt-holes, then
laid together with an insertion of very thin paper. Some
makers use varnish instead of paper, but this is not so
reliable an insulation. Plates are now on the market
which have one side covered with a very thin insulating
film. These may be used without paper insertion. In
building up, the lower yoke and the two cores are first
CONSTRUCTIVE DETAILS
132 TRANSFORMERS
made up, the coils are then inserted, and lastly the plates
of the top yoke are put in. The coils are wound on
paper cylinders, which at their lower ends are provided
with flanges to prevent the coils slipping. In winding
the coils it is advisable to wrap each layer with a sheet of
thin paraffined calico, which is doubled back at the ends
so as to give additional insulation between adjacent layers.
The thickness of the cotton covering on the wire
depends on its diameter
(or equivalent diameter if rect-
angular wire be used), the voltage, and the quality of the
FIG. 64. Plan of 20 kw. transformer.
cotton and number of coverings. There must at least

be two coverings, though treble covering with very fine
cotton is still better. For very stout wires braiding is
advisable. The thickness of the covering in millimetres
should not be less than
S = o'i3+o'o6^ (30)
when d is the diameter (or equivalent diameter) of the
naked wire in millimetres. The diameter of the covered
wire is then
CONSTRUCTIVE DETAILS 133
Wire of large rectangular section may also be wound

naked, suitable strips of fibre or other insulating material
being wound in, or afterwards inserted.
The resistance of the coil must be calculated with
reference to its temperature ;
as a first approximation,
based on a temperature of 75 C, the following formula
may be used
R = CTO2/_ ohm
r> ..
i
where / is the length of wire in metres and q its area in

square millimetres.
To promote dissipation of heat, the casing may, as
already mentioned, be provided with external ribs or gills.
Small internal ribs are also provided to hold the trans-
former securely. The main cover is fitted with a small
auxiliary cover to give access to the terminals without
the necessity of breaking the joint of the main cover.
The leading-in wires may be taken through stuffing-
boxes, as shown in Fig. 63, or they may be passed
through bushed holes which are afterwards cast out
with insulating compound. The latter arrangement is
generally adopted in large transformers. When the
pressure is
very high the bushes take the form of long
glass or porcelain tubes.
CHAPTER VII
DESIGN OF A SHELL TRANSFORMER FILL

FACTOR WINDING EFFICIENCY, WEIGHT
AND COST ENLARGING A DESIGN
Design of a shell transformer. As an example of how
to design a small air-cooled shell transformer we take
a 7 kw. transformer to give at v = 50 with 2000 volt on
the primary terminals 64 ampere at no volt on the
secondary terminals, or allowing for a moderate drop, 113
to 114 volt on open circuit. If we make the height of
the windows equal to the thickness d of the core and
their width 70 per cent, of the height, the total weight of
iron in kg. is for a depth of core c
G=
d and c being given in dm. Using alloyed iron, the
formula giving the approximate weight is
G=io- -?- .... (16)
A/
V T r^r*
100
This gives for P = 7, G-8okg. and (Pc = 2'22. For a
core 15 cm. deep, the thickness would be 12*1 cm., or say
in round numbers 12
cm. Its area is 156
sq. cm.
Fig. 65 is a sketch
of the carcase, the
dimensions being
mm. The next point
to be determined is
__i the induction permis-
.
65.
sible with regard to
FILL FACTOR 135
temperature rise. In a small transformer the windows

have to be closely packed with wire and cannot contribute
to the cooling surface. The flat surface of the plates is
nearly worthless for cooling, so that we can only reckon
on the outer edge surface. This is 1670 sq. cm. Allowing
a temperature rise of 55 C. we get a specific cooling
surface of <r = 12 '8, to which corresponds an iron loss ot
130 watt or 1*62 watt per kg. The permissible induction
is 9000, and the corresponding number of secondary
turns 36'6. Since we cannot have a fraction of a turn,
we make
^ = 36
and to this corresponds
ih
= 640
The induction will then be 9250 and the iron loss
136 watt, giving a temperature rise of 58 C., which is
still
permissible.
Fill factor. The design of the coils is a somewhat
tedious process, as it can only be done by a method of
trial and error. We
have to assume certain sizes of
wire and try whether the winding can be housed in the
available space. As a rough guide to shorten this labour
we make use of the conception of the Jill factor, that is
the ratio of the space actually filled by copper to the total
available winding space. The fill factor must naturally
vary with the gauge of wire and the size of the transformer,
and is therefore
largely a matter of experience obtained
with previous designs. In small transformers for com-
paratively high pressure, that is to say with a large
number of primary turns, the average fill factor for
primary and secondary taken together is a little under
50 per cent. Taken for the primary alone it is con-
siderably less than 50 per cent, and taken for the
secondary alone it is more than 50
per cent. We
shall
therefore have to allot the whole window area of about
10,000 sq. mm. in two unequal portions to the primary and
secondary, giving the former the larger share.
Before proceeding with the details it is, however,
advisable to check in a rough way the possibility of
housing the wires. Take 48 per cent, as the average
136 TRANSFORMERS
fill factor, then 4800
sq. mm. are available for twice 36
turns of secondary. This gives q^ = 66 sq. mm. The
perimeter of one turn we take from the sketch, Fig. 66,
and find 0*85 m. The resistance of the two windings in
series is 0*0185 ohm, and the total copper loss is 76
watt. This is permissible. For reasons which are
stated in the next chapter it is necessary to subdivide
each circuit into several coils, and we therefore group the
winding into three secondary and two primary coils.
Winding. The detailed calculation as to the exact
gauge of wire for each winding can now be made. As
already mentioned, it is
a somewhat laborious
process, but presents no
difficulties, and need
therefore not be set
down here. The result
is that the primary is
wound with 2 mm. d.c.c.

round wire, and the
secondary with naked
strips (two in parallel
for easier handling) of
together 80 sq. mm.
FIG. 66. cross section. There
are three secondary coils
each containing 12 turns, and the three coils are coupled
in series,two inside ends, namely b and c, Fig. 66, being
joined, whilst the inside end a of the first coil and the
outside end of the third coil are connected to the ter-
minals. This method of connecting up is permissible,
because the pressure is low it cannot be used for the
;
primary, where, on account of the high pressure, inside

connections must be avoided. For this reason each
primary coil is wound in two halves, which are separated
by an insulating partition, and both the beginning and
the end of the winding are on the outside. The coils
are flanked by thin fibre discs and served with
insulating
tape. To increase the cooling surface the heads of the
end coils are splayed out as shown in Fig. 67. The
particulars of the winding are
EFFICIENCY, WEIGHT AND COST 137
Primary. Secondary.
Number of turns .
640 36
Cross section sq. mm. 3'M 80
Length in m. 545 31
Resistance ohms .
3'5 0*0078
Loss in watt .
45 32
Weight of copper, kg. 22*2
The total surface, counting that between the

cooling
splayecl-out heads as of half value, is 2640 sq. cm., and
the total copper loss is 77 watt, making
(7 =
34-5, to which corresponds a tem-
perature rise of 47C.
It be noticed that primary
will
and secondary copper losses are not
equal. Equality of losses would be 240-
theoretically preferable, but practic-

ally it is not always possible to
arrange for equality because the ratio i

of depth to width of each half of a
primary coil can only be varied
within narrow limits, and it is there-
fore necessary to take the nearest
gauge of wire that will fit into the FIG. 67.
space.
For the
rest it should be noted that a moderate de-
parture from the law of equal losses has only a very small
influence on the efficiency.
Efficiency, weight and cost. The total losses are 213
watt or 3 'i per cent, of the full load output. The
efficiency at various loads is as follows :
Load in
per cent. 100 50 25 10 5
Efficiency per cent. 96-9 957 92-5 837 72
The weight of active material is 1177 kg. or 167 kg.

per kw. In calculating the cost of active material it
should be remembered that the primary is wound with
fine, and therefore more expensive wire also, that by ;
reason of using E-shaped stampings for the carcase the

cost of sheets is increased by the unavoidable waste. To
take these circumstances into account we shall reckon
138 TRANSFORMERS
the iron at is. $d. and the copper at 2s. 6d. the kg. The
cost of active material is then 193^., or at the rate of
about 28^. per kw. for the 7 kw. air-cooled transformer.
Enlarging a design. The influence of an enlarge-
ment of linear dimensions on the output of a transformer
was discussed in Chapter III. We
may now show the
practical application of the various relations then found
to the problem of deducing from the design of the 7 kw.,
"
which we may call the " small transformer, a design
for a large transformer of, say, 200 kw. One way, but
certainly not a good way, of getting out the new design
is to simply enlarge the linear dimensions in the ratio of
It is not a good way, because the large transformer

would have the same weight per kw. as the small, and
would therefore be far too costly. In order to get an
economic design we must increase the magnetic and
electric stresses and improve the cooling.
As regards the induction, it has already been pointed
out in the previous chapter (under the side - heading
Power Transformers] that this is limited not so much
by heating as by the magnetising current, since too large
a magnetising current lowers the power factor of the
whole plant. We decide, therefore, on an induction
which should be well below the knee of the curve given
in Fig. 55 for alloyed iron, say
B = 12000
Since with an enlargement factor m the core area
increases by m z
,
and since the induction of the small
transformer was 9200, we find that for the same number
of turns the E.M.F. e of the large transformer will be
e = m* I2OOO-
2
eQ
9200
eQ being the E.M.F. of the small transformer
e = m* 1-3*0.
In the large transformer the windows will not be

completely filled by wire owing to the better fill factor
;
ENLARGING A DESIGN 139
with stouter wire, we shall be able to spare a duct of

about 2 cm. thickness and width d on the outside of the
coils, so that their whole outer perimeter will be cooling
surface. Weshall also splay out all the coil heads.
Had these two improvements been possible in the small
transformer, the cooling surface for copper would have
been
S = 3820 sq. cm.
The cooling surface for copper in the large transformer

will therefore be
S = ;^ 2 382O
As cooling medium we must use oil, and this in turn
will be cooled partly by contact with a corrugated case
and partly by a cold-water worm. We allow a tempera-
ture drop of 20 from copper to oil, 10 in the oil itself
for producing circulation, 25 from oil through the case
to air or to the water in the worm. The water at the
exit will then have the same temperature as the air.
Let P be the power lost in the coils, then with 20
drop to oil we have <r
= 10*5 and
In the small transformer the power lost is P =77

and the current z" Since for the same number of turns
.
the resistances vary inversely as m, we have for a current

i in the large transformer the relation
P - 1 ^
PO m z'
2
^365 _ i
2
77 z'o
2OO
7x 2-83
From which we find the enlargement factor
m i
'94
140 TRANSFORMERS
The weight of iron will be 580 kg., and the loss
1800 watt. The cooling surface of the carcase will be
15000 sq. cm., giving <r = 8'3, to which corresponds a
temperature drop to oil of 23 C.
The
cooling surface of the coils will be 14500 and
the copper loss 1380 watt, giving <r= 10*5 as stipulated
above. temperature drop to oil will be 20. The
The
total loss being 3180 watt or i'6% of the output, the
efficiency is 98*4%.
We have assumed the same number of turns as in
the small transformer. The secondary pressure then
was no volt. Now it will be
e nfr^ x 1 10 = 540 volt
The primary pressure be 9800 volt. The primary

will
current will be 20 ampere. The only question yet to be
decided is whether the magnetising current is sufficiently
small. The mean path of the lines of force in the small
transformer was 60 cm. now it is m times as great,
;
or 117 cm. We require for B = 1 2000 an exciting

force of 4 ampere-turns per cm., or 468 ampere-turns
crest value, that is 332 ampere-turns effective value.
Since n = 640 we find
?;
= 0'52
or 2*6% of full load current. This is permissible.
The case for this transformer would be 100 cm. x 60
cm. and about 140 cm. high. Allowing for corrugation,
the surface wetted by oil would be about iioooo sq. cm.
Since the drop from oil through case to air is to be 25,
the specific cooling surface is
25
that is to say, the case can dissipate
noooo =
1 840 watt
60
leaving 1320 watt to be taken up by the cooling water.

The worm will be a pipe of i" diameter, and have a
total length of 12 yards. With a 5 rise in the tempera-
ture of the water (10 at inlet and 15 at outlet) we
ENLARGING A DESIGN 141
require 240 litres, or 54 gallons, of cooling water per

hour.
The weight and cost of active material will be as
follows :
Iron . .
580 kg. . .
5805-.
Copper .
300 ,, 5255-.
880 uo5<r.
Per kw. . .
4*4 . .
5$. 6d.
Thecost of the complete transformer will be about

^120, or 125. per kw.
It goes without saying that over all dimensions,
weight and cost would not be altered if the winding were

modified for slightly different pressures. The efficiency
will be
Load in % . 100 50 25 10
Efficiency % .
98*4 97-9 96-3 91
The improvement as regards weight and cost brought

about by enlargement of linear dimensions and more
vigorous cooling is enormous. It would have been still
greater with further enlargement. Thus in a transformer

r 10000
tor volt for 500 kw. at v = 50 the active material
\J
would weigh about 3*6 kg. and cost about 45. per kw.
It need hardly be mentioned that the method here used
is not intended to make a careful design with attention
to all details superfluous it is
only intended as a first
;
step to obtain some leading data which may be taken as

a basis for working out a detailed design.
CHAPTER VIII
THE VECTOR DIAGRAM APPLICATION TO A

TRANSFORMER WORKING ON OPEN CIRCUIT
WORKING UNDER LOAD MAGNETIC LEAK-
AGEWORKING DIAGRAM OF A TRANS-
FORMER HA VING LEAKAGE VOLTAGE DROP
-GRAPHIC DETERMINA TION OF DROP DROP
DIAGRAM SIMPLIFIED CORRECTION FOR
EDDY CURRENT LOSSES
The vector diagram. The working condition of a trans-
former or any alternating current apparatus may be
represented in two ways, either analytically by certain
more or less complicated formulae, or graphically and
more simply by the use of a vector or, as it is also
called, a clock diagram. The latter method is only
applicable if all quantities follow a simple sine function
of the time. It has already been mentioned that this is
nearly true for the current and pressure given by modern

machines, and it is only in some special cases, such as
the magnetising current of a transformer, that the influence
of the hysteretic loop produces a departure from the
sine curve. Even in such cases the error caused by
substituting for the actual curve an equivalent sine curve
is not very serious, and as the use of vectors very
materially simplifies the investigation, we shall in study-
ing the working condition of a transformer assume that all
quantities follow a simple harmonic law. Some use has
already been made of vectors in previous chapters, but
in order to represent the matter in a comprehensive way
I shall not refer to the
previous examples, but assume
that this is the first introduction of the reader to the use
of vectors.
A vector is a line drawn to scale, which by its length
142
THE VECTOR DIAGRAM 143
gives the magnitude of the quantity represented, and by

its angular position in reference to some base line the
phase of that quantity. The vector rotates with an

angular velocity o>
corresponding to the frequency, so
that
ft) = 2TTV
From the definition it follows that a vector may be

displaced parallel to itself without ceasing to represent
the quantity correctly. Strictly considered, the length
of the vector should represent the crest value of the
magnitude its
projection on the base line will then
;
represent the instantaneous value, and this will be

positive or negative accordingly as the projection lies to
one side or the other of that point on the base line which
corresponds to zero value. Sometimes it is convenient
to let the vector represent not the crest value, but the
effective value of the quantity.
This is permissible since the o
ratio between crest and effective 2
value is a constant for all quan- FlG> 6 8. -Conception of vectors.
namely ^/2 an d passing

tities, >
from one to the other only means an alteration of the

scale in this ratio.
Vectors which represent different magnitudes having
the same phase must be drawn parallel to each other,
and if drawn from the same origin will lie upon each
other. Thus in Fig. 68 OI, ON, OX
represent respectively
current, and exciting force. They are co-phasal,
flux,
and for this reason drawn as parallel lines. They may
be drawn from the same origin so that all lie along OA.
The length of each will depend on the magnitude of the
quantity and the scale chosen to represent it. If we
select, for instance, the scale so that i mm. shall
represent
i
ampere, or i
megaline or i
ampere-turn, then the number
of turns n will be represented
OX and the
by the ratio -=-=->
R OX
magnetic resistance by the ratio -r- By altering
the scales for current and flux in these ratios it is obvious

that O A OX
~ may be made to represent not only ampere-
144 TRANSFORMERS
and flux. In order that
turns, but also current may OA
represent current we shall use a scale, the divisions of
which are not i mm. but n mm. apart, and for the flux
we must use a scale, the divisions of which are R mm.
apart. The value of R
is found from the following
consideration. The well-known law of the magnetic
circuit is
TT _
where H is the magnetic force, and / the length of path.

Let A be the area, B the induction, and p the perme
ability, then we have for the flux in megalines
" -X
N= o-8/i o ~R
!
p.
A
or for a magnetic circuit composed of different materials,
p.
A
In a transformer the magnetic circuit consists of iron,
and if there are butt joints air. For
air /A is constant, namely i, but for iron
it varies with the induction. As, how-
ever, transformers are mostly worked at
a fixed pressure, and therefore at a
constant induction, R will also be a
constant, and we are therefore justified
in using the same vector for flux and
exciting& force, provided we alter the

PIG. 69. E.M.F. 1-1 ir v u
and current vector,
,
scale accordingly. If X
be given not
as a crest value, but as an effective
value, we have
Let, in Fig. 69, OE

represent the crest value of the
E.M.F. impressed on a circuit, for instance on the primary
terminals of a transformer, and OI the crest value of the
APPLICATION TO A TRANSFORMER 145
current. The phase difference is

$, and if the vectors
rotate as shown by the arrow, the current lags and is <p
Let the base line (not shown in the

positive. diagram)
be vertical, then the E.M.F. has at the moment its positive
maximum value, and the current is approaching its
positive maximum value. We may also mark off on the
same vectors the effective values, thus, that
OE
The and pressure have not been

scales for current
altered but
;
we enlarge them in the ratio of i ^2,
if :
OE will represent the effective E.M.F. and OI the

effective current. We can now even drop the concep-
tion of rotating vectors, provided we take the arrow
to mean that OE leads and OI lags.
The power will be given by the product of effective
E.M.F. and the projection of effective current on OE,
each measured with its proper scale. If, for instance,
1 mm. 10 effective and i mm.
represents volt, represents
2 effective ampere, then the power in watt will be 20
times the area
OE x OI x cos <f
given in sq. mm.

Application to a transformer. As by definition and
according to the original conception of vectors, any
vector may be displaced parallel to itself, it is clearly
permissible to combine vectors representing like quantities
in the same manner as forces, the closing line of
the^
polygon being the resultant vector. Thus in a trans-
former the exciting forces exerted by the two windings
individually may be combined to a resultant exciting
force. In order to be able to use the same current scale
for both windings convenient to assume for both cir-
it is
cuits the same number of turns ofwire, which makes the

transforming ratio equal to unity. Such an assumption
is permissible, as will be seen from the consideration that
without altering the gauge of wire on the primary, we

10
146 TRANSFORMERS
can by simply grouping the turns differently bring a
number of wires into parallel connection to
sufficient
produce the effect of a winding of fewer turns, but of

stouter wires. If, for instance, the transforming ratio is
in reality 2000 and the high pressure coil has 800

to 100,
turns, we may assume these 800 turns to be grouped in
20 parallels of 40 turns each. The current will now be
20 times as large as before, and the E.M.F. will be
reduced to the 2Oth part that is to say, it will not be
;
2000 volt but only 100 volt. In assuming this alter-

ation in the connections we have not altered in any way
the heating, the percentage of idle
current, the efficiency, etc., but we
have obtained the advantage that
the E.M.F.s in both circuits are
now of the same order of magni-
tude, and can therefore be con-
veniently represented in our clock
diagram to the same scale. It is
obvious that the current must be
increased in the same ratio as the
E.M.F. is reduced, and that the
resistance is reduced in the square
of this ratio.
Working an open circidt. As
an introduction to the use of clock
FIG. 70. No-load current. diagrams we may investigate the
simplest case, namely, a trans-
former working on open circuit. Let in Fig. 70 OI
represent the idle current to any convenient scale, and
OI and OI M its two components, which can be found
/t
by calculation as shown in Chapter IV. The component

I
M magnetises the
iron of the transformer, the total
magnetic flux being represented to any convenient
scale by the vector ON. The current and flux have
of course the same phase in the diagram. At the
moment to which the diagram refers the projection of
the flux on the vertical is zero and the E.M.F. has
its maximum value,
8
namely 27rv<;no" volt. Since the
E.M.F. in the primary must try to prevent the growth
of the current (and flux) its direction in the diagram must
WORKING AN OPEN CIRCUIT 147
be vertically downwards. Let this be represented by

the line OE
plotted to any convenient volt scale. It is
obvious that the impressed E.M.F. must be equal and
opposite vector will therefore occupy the position
;
its
OEj. We
neglect in the diagram the influence of the
resistance of the primary coil, since the error thereby
committed is insignificant. The diagram then gives OE X
as the E.M.F. impressed on the primary terminals, OE

as the E.M.F. obtained on the secondary terminals, and
OI as the primary current, the secondary current being
zero.
T1 COS 0Ei
I he power supplied
I' 1 ^ ,1
to the primary is
I/,Ei
-
- - = -In .
or if, instead of using maximum values, we use effective

values
z = cos
The apparent power supplied to the primary is iê l and

the ratio between effective and apparent power, that is
cos <p, is called the power factor.
It is
interesting to note that the flux which is ON
produced by the current z does neither in position nor
magnitude correspond with the flux that would be pro-
duced by a constant current of the strength Yet the z' .
equivalent alternating current passes actually through

z'
the primary coils, and we should expect that it must

produce the corresponding magnetisation. This is,
however, not the case. The magnetisation is less than
corresponds to the ampere turns of exciting power
carried by the coil and it lags behind the current by
;
the angle <p. /\2T ^f)

This apparent contradiction can, however, be easily
explained. The loss on open circuit is due to hysteresis
and eddy currents. If we could obtain a magnetic-
ally perfect iron, and if it were possible to so design

the transformer as to be absolutely free from eddy
currents, then ih zero and ^ = 7 The power factor .
will also be zero. then that it were

Assuming possible
to obtain such a perfect transformer, we could by the
addition of a third winding supplying current to an
inductionless resistance so alter it that its
working
148 TRANSFORMERS
diagram would be the same as that of the transformer
practically possible. The only condition to get the two
transformers into 'thesame state is that the power which
is transformed into heat in the third coil and resistance
of the perfect transformer equals the power wasted in

hysteresis and eddy currents in the imperfect transformer.
It obvious that the current generated in the third coil
is
must be in phase with the impressed E.M.F., and there-

fore opposed to it. It must to that extent weaken the
magnetisation and this circumstance

;
explains why only one component of

the idle current (namely, that which
stands at right angles to the com-
ponent wasting power) is effective in
producing magnetic flux.
Working under load. We may
now consider the working diagram of
a loaded transformer. For the sake
of simplicity we make at first the
two following assumptions first, the
transformer has no magnetic leakage ;
secondly, the load is an absolutely

non-inductive resistance. Under these
circumstances the secondary current
must be proportional to the E.M.F.
induced in the secondary coil, and
must coincide with it in phase. If
i^
in Fig. 7 1 represent the magnetis-
FIG. 71.
ing current, then e\ the E.M.F. in
the secondary lags by 90 behind ^
and e\ the E.M.F. in the primary is in advance of z'M by
90. In this, as in the following diagrams, we assume
clockwise rotation of the vectors.
The pressure at the secondary terminals is repre-
sented by the line Oe 2 thus the distance eê\ represents
;
the voltage drop due to the resistance of the secondary

coil. To find the pressure at the primary terminals as
regards magnitude and position in the diagram, we
proceed as follows. We plot to an arbitrary scale
the
ampere turns in the secondary on the vector of the
secondary E.M.F. Let this be the distance Oi n
2 2.
To
MAGNETIC LEAKAGE 149
the same scale we plot Oa, which represents the ampere

turns of the idle current in the primary. It is obvious
that Oa must be the resulting ampere turns of the two

coils, and since the ampere turns of the secondary coil
are known, we find those of the primary coil by drawing
the parallelogram as shown in the figure. thus We
obtain i^n^ and therefore also the primary current ilt
This current is being driven through the apparatus by
an E.M.F. which must obviously have two components :
first, the component Ob required to overcome the ohmic

resistance of the primary coil, and, secondly, the com-
ponent Oe\ to balance the induced E.M.F. The result-
ant, also obtained by constructing a parallelogram, is Oe^
which line therefore represents the E.M.F. that has to
be supplied at the primary terminals in order that the
current i.2 may be drawn from the secondary terminals
under an E.M.F. 2 A glance at the diagram shows
.
that j> the difference being in the present case very

2>
marked, because for the sake of greater clearness we

have exaggerated all losses and assumed too large an
exciting force. If the vectors represent effective values,
the followingo relations obtain :
Power supplied equals . . .

e^ cos <p
Power given off equals . . . e 2 i%
Efficiency 73 equals .... cos
In good transformers the idle current is only to ^ ^

of the primary current at full load. The distance Oa is
therefore, as compared to the distance Oi 2 n 2 exceedingly ,
small when the transformer is working under full load.

It is obvious that the line Oi-^n^ that is, the vector of
the primary current, becomes nearly vertical, making the
angle $ so small that its cosine can, without appreciable
error, be considered equal to unity. The efficiency
therefore becomes TJ
= -
2 2
. .
.
Magnetic leakage. Up to the present we have

assumed that the transformer is free of magnetic leakage,
so that exactly the same magnetic flux passes through
TRANSFORMERS
the primary and secondary This condition is only
coil.
approximately realisable practice, because the two

in
circuits must be separately wound and insulated from
each other, whereby intervening spaces are produced
which admit of the passage of leakage lines. The result
of this leakage flux is to increase the counter E.M.F. in
the driving (primary) coil, and to reduce the E.M.F. in
the driven (secondary) coil, an effect which will obviously
be the greater the heavier the transformer is loaded. If
the transformer is not loaded at all, that is to say, if it is
worked on open the effect vanishes almost com-
circuit,
pletely, ratio between the primary and secondary
and the
electro-motive force is equal to the ratio between the
primary and secondary
1
number of turns. At
load, however, the trans-
forming ratio rises, by
reason of the leakage
effect just mentioned,
and the secondary volt-
age drops below its
value on open circuit.
Let in Fig. 72 P
FIG. 72.-Magnetic leakage. and S represent the
primary and secondary
coils of a shell transformer traversed at any particular
moment by the currents indicated by the dots and
crosses. The driving coil P produces a magnetic field
N! the lines of which flow in the sense indicated by
the arrows shown in full lines. At the same moment
the driven coil has the tendency to produce a field the
lines of which have the direction shown by the dotted
arrows. This field cannot in reality be produced, because
the driving power of the primary coil overpowers that of
the secondary, so that on the whole the direction of the
lines in the core within the coils is vertically downwards,
A small change in this ratio may occur owing to hysteretic and

1
eddy current losses. It has been shown that such losses may be repre-
sented by the heat developed in a third circuit of appropriate resistance.
The current in this circuit opposes the primary current, and may cause
a little leakage.
MAGNETIC LEAKAGE
but the effect of the opposition offered by S is to crowd
lines offeree laterally outward through the sides AB, so
that the flux N l passing through the primary coil is larger
than the flux N
2 passing through the secondary coil, the
difference N^ being squeezed laterally through the spaces

ab. The E.M.F. induced in the secondary coil is there-
fore not proportional to N x but to N x N f When work- .
ing on open circuit, the secondary coil carries no current,

and, apart from the effect mentioned in the footnote on
the previous page, which is generally negligible, there is
consequently no tendency to force part of the flux later-
ally outward through A and B. In this case N^ o and
NO = N!. The E.M.F. in the secondary is now propor-
tional to N l5
and there-
fore greater than be-
fore, when we assumed
the transformer to have
been loaded. It follows
that with a constant
primary pressure, the
secondary pressure is
a maximum on open
circuit and decreases
as the load on the FIG. 73. Grouping of coils.
secondary increases.
"
This so-called " drop of a transformer must in most cases
be considered as an imperfection, and the question arises
as to what means can be used to minimise this imperfec-
tion. Since the drop is due to the flux N f our aim must ,
be to reduce the latter, and this can be done in either or

both of the two following ways. We
may reduce the
magnitude of the exciting force of each coil, and we may
increase the magnetic resistance of the leakage path.
Thus the leakage path in Fig. 72 is shorter and wider
than in Fig. 73, and we shall in the latter arrangement
obtain a smaller drop than in the former. We can carry
this improvement a step further by subdividing the
the
primary coil into two parts, which are separated by
secondary coil as in Fig. 74. In doing so we have
halved the ampere turns which produce the leakage flux,
whilst at the same time still further reducing the width
152 TRANSFORMERS
of their path. The same object is attained in the arrange-
ment of Fig. 75. Here we have subdivided each coil
into a number of parts, each carrying only a corresponding
fraction of the total exciting power. The primary coils
are sandwiched between the secondary coils, the leakage
flux is produced by a
much smaller exciting
power, and has a path
of much higher resist-
ance than in Fig. 72.
This arrangement is
also convenient because
the coils need not be
wound upon each other,
but may be wound,
FIG. 74. Subdivided primary. insulated, and
tested
independently of each
other. the arrangement always employed in shell
It is
transformers, whilst for core transformers the concentric

as well as the sandwiched arrangement is used.
From what has been said above, it will be clear that
the effect of leakage can be represented by the addition
of two magnetic fields
(one interlinked with
a coil in the primary
rp i np~~i and the other with a
i gi coil in the secondary
circuit) to the main

HP i field common to both
sn
i
coils. We may thus

imagine the practically
possible transformer,
;
FiG. 75. which has leakage, re-
Primary and secondary subdivided.
placed by a perfect
transformer T, Fig. 76, having no leakage, to which
however are added two choking coils I and II, the
terminals i i, 2 2 being outside the latter. At full
load the choking coil I produces the field x and the N
choking coil II the field 2 N
For the sake of simplicity,
.
we assume the dimensions of the choking coils such that

each contains the same number of turns at the corre-
DIAGRAM OF A TRANSFORMER 153
spending transformer coil. The field common to both

transformer coils is N. Then the E.M.F. induced in
the primary transformer coil
is
AT
*1
= 8
and that induced in the chok-

ing coil I is /
x 8
FIG. 76.
Similarly we have for the secondary circuit
In plotting these values in a clock diagram we must

not forget to place e s2 at right
angles to z'2 and e sl at right
angles to ilt
Working diagram of a
transformer having leakage.
We are now in a position to
draw the clock diagram of a
transformer having leakage.
Let for the present its load
be non-inductive (glow-lamps
for instance), and assume that
we have, as mentioned in the
beginning of this chapter,
arranged the primary winding
so as to have the same number
of turns as on the secondary
coil.
Let O/2 in Fig. 77 repre-
sent the secondary current,
O^. 2 the pressure at the
FlG. 77. Vector diagram of
secondary terminals, ^.2 / 2 the transformer.
loss of pressure clue to the
ohmic resistance of the secondary, then (V 2 must be
the resultant of the E.M.F. induced by the main field N
and the E.M.F. of self-induction clue to the leakage
154 TRANSFORMERS
field N 2. The of the latter must lie in such
vector
position that e s2 tries to prevent the decrease of 2 that z'
;
is to say, it must be drawn horizontally to the right as

shown at es2 in the diagram, which is plotted to the same
volt- scale as 2 O^ By constructing the parallelogram
.
we obtain Oe 2 the E.M.F. which must be induced in

,
the secondary coil in order that the current z 2 may flow

under the terminal pressure e k 2 . .
The magnetising current i^ must be at right angles to

Oe2 and in advance of it, whilst the current representing
loss of power ih must be in line with e^ and of the same
sense as e^. The position and magnitude of the idle or
no-load current is thus completely defined. Weobtain
for it in the diagram the vector z The vector of the
.
primary current we find by combining z'2 and This

z' .
gives z'j. The E.M.F. of self-induction (due to leakage)

in the primary must stand at right angles to and must
z'j
follow it. In our diagram, e sl must therefore be drawn

to the left. The E.M.F. supplied to the primary
terminals must obviously contain three components.
One component must be equal and opposite Oe>2 .
This is given by the vector O^.

One component must be equal and opposite O^, and
one component must be provided to overcome the ohmic
resistance. Let the vector of the latter be Oa.
By adding these three components graphically we
obtain the point ekl Oekl is the vector of the E M.F.
.
supplied to the primary terminals. A

glance at the dia-
gram shows that ekl is greater than ek ^ the difference
being the more marked the greater are the ohmic resist-
ances and the E.M.F.s of self-induction in the two
windings. In both respects the diagram Fig. 78 has
been exaggerated, so that the influence of each part may
be more clearly seen.
It is interesting to investigate the case of a transformer
the secondary terminals of which are short-circuited by
a stout copper wire and amperemeter, thereby making
^ 2
= o. Weassume the primary E.M.F. to be so
adjusted that this amperemeter shows the normal second-
ary current corresponding to full load under normal
working conditions. The diagram then assumes the
DIAGRAM OF A TRANSFORMER 155
form shown in Fig. 78. The lettering is the same as in

FiV.
o 77. *
be seen from this diagram
It will o that although
o
no pressure is obtained at the secondary terminals, a
pressure equal to ekl must be supplied to the primary
terminals in order that the current 2 may flow through z*
the short circuit.

If, as is in modern transformers of
always the case
good design, the resistance of the windings is very small,
and the no-load current t is only a very small fraction of
/!, then z'
2 very nearly in a straight line, and
and i will lie
e l and 2 very nearly in a straight line.
will lie
With a symmetrical arrangement between the two
windings (and the assump-
tion that the number of
turns is the same in both)
we have e^ e^ and
The E.M.F. necessary

to overcome self-induction
and ohmic resistance can
thus be found by a very
simple experiment. We
short-circuit the
secondary
terminals by means of an
amperemeter (having itself
FIG. 78. Vector diagram for short-
as induction as pos-
little
circuited secondary.
sible), and supply the
primary terminals with current of normal frequency and
such E.M.F, that the normal secondary current is
indicated on the amperemeter. One-half the E.M.F.
supplied to the primary equals the E.M.F. required for
the primary winding. The E.M.F. required for the
secondary winding is equal to this value divided by the
transforming ratio. Take as an example the case of a 10
kw. transformer wound for a ratio of 2000 volt to 100 volt.
In testing this transformer, as above explained, it is
found that 100 volt must be supplied on the primary at
v=5o in order that 100 ampere may be driven through
the short-circuit. We
have then e l = 50 and e^ = 2*5.
The E.M.F. has two components. One is a wattless
156 TRANSFORMERS
component at right angles to the current, and is due to
self-induction, and the other is in phase with the current,
and is due to resistance. There may be another watt
component due to eddy current losses in the copper or
other metal parts including the carcase. Such eddy
currents may be produced by the leakage field passing
laterally through the wires, plates, or other metal parts.
Not to complicate the investigation, we neglect the (in
any case small) influence of such eddies for the present.
We also assume for the present that on short-circuit
the currents are inversely proportional to the respective
numbers of turns. The watt component of impressed
primary E.M.F. is then simply the product of current
and resistance taken for both windings.
As we measure the pressure in the primary circuit
and the current in the secondary, R 2 2 has to be reduced
z'
to the primary. We
have therefore the watt component
of the impressed E.M.F.
n n
!
- l2 ,
R
+ -D 2
-l
n, n2
Since the wattless component must be at right angles
to e r and since both together give the total impressed
,
E.M.F. e we find the wattless component e^

,
that is, the
E.M.F. of self-induction
e^ may be divided into two parts inversely proportional

to the turns, so that we get the E.M.F. of
numbers of
self-induction separately for each winding.
Voltage drop. If by making the experiment above
described we have found how much E.M.F. is produced
by magnetic leakage in each coil, we can use this informa-
tion to determine the voltage drop at various loads. In
this determination it is convenient and permissible to
assume exact opposition in the phases of primary and
secondary current. Modern transformers with closed
magnetic circuit require so little magnetising current,
that even at load this assumption is very nearly true.
Let, in Fig. 79, A
represent the pressure at the secondary
terminals, AB
the ohmic loss of pressure, BC = , 2 the
VOLTAGE DROP 157
E.M.F. due to self-induction; and therefore OC = 2 the

E.M.F. induced in the secondary. Let the transforming
ratio be reduced to unity, then OC = e is also the E.M.F.
l
induced in the primary, and with symmetrical windings

CD = BC the E.M.F. of self-induction in the primary,
so that esl = es2 The ohmic voltage loss in the primary
.
is DE = AB if the losses are

equally divided between
the two windings as required by a good design. The
line joining A, C and E is therefore a straight line, and
its inclination to the vector of
secondary terminal pressure
is the same for all loads. At a smaller load, for instance,
producing the ohmic loss A'B the terminal pressure
would be OA' in the secondary and OE' in the primary.
FIG. 80.
FIG. 79.
The ratio of the length of the lines AE and A'E' is

the same as that of the lines AB
and A'B, and the
length of the line AE is directly proportional to the
current.
Let us now assume that we are able to vary the
primary E.M.F. in any way which may be required to
keep the pressure at the secondary terminals constant
for all loads. draw the line AE (Fig. 80) for full
We
current, and make an ampere scale which corresponds
with this length, then we can, by using this scale, mark
off on the line AE the points E', E", etc., corresponding
to other currents, and thus find the primary E.M.F.
vector OE', OE", etc., corresponding to these currents.
It is thus possible to determine the primary E.M.F. as
i
58 TRANSFORMERS
a function of the load, if the secondary terminal pressure
is to be a constant.
Thishowever not the case generally met with

is
in practice. As a rule the E.M.F. in the primary or

supply-circuit is constant, and it is required to find the
secondary terminal pressure at various currents. This
problem can also be solved graphically in a very simple
manner.
Graphic determination of drop. It has already been
shown that in all OAE, OA'E', etc., the
the triangles
obtuse angle at A, A', the same.
etc., is The longest
side of the triangle represents the
E.M.F. impressed on the primary,
and the shortest side the current
in the secondary. We
may now
imagine all the triangles in Fig. 80
so enlarged or reduced that all
the points E lie on a circle de-
scribed round O
as centre, with
a radius equal to the impressed
E.M.F. Let OE
in Fig. 81
represent this E.M.F. at full load

(current represented to a suitable
scale by the lengths AE) and
E', E" the positions of E for
smaller loads, then the length
FIG. 81.
OA, OA', OA", etc., gives the
corresponding pressures at the
secondary terminals. As a matter of convenience we
may also plot the secondary current on a horizontal o\
to a suitable scale, and find the points E by projection
from the points I, as shown by dotted lines.
Let us now apply this method to our previous
example of a 10 kw. transformer. We
have assumed
that 100 volt must be impressed on the primary in order
to produce 100 ampere in the short-circuited secondary ;
that is, 5 per cent, of the normal primary voltage. Let

us further assume that the watt component as calculated
from resistance measurements has been found to be
2 per cent. The wattless component is therefore
2
2
= 4'58 percent.
VOLTAGE DROP 159
Theslope of the line AE in Fig. 80 is 2 in 4*58,

and its
length is 5, whilst the
length OAis 100. The
angle at Ois therefore
very acute, and the difference
between OA
and OE, that is, the drop at full current and
a non-inductive load, is 2 per cent. At half-load it would
be i per cent., and so on. If the transformer had con-
siderably more leakage, say 20 per cent, instead of 4*58

per cent, then the drop, even at non-inductive load, would
be appreciably greater than that given by ohmic resist-
ance. In such a case the construction shown in Fig. 81
may be applied. The ampere-load would be plotted on
the horizontal o\ by using a scale on which ol represents
100 ampere, and by projecting the corresponding points,
first to the circle and then to the vertical parallel to EA,
we find the terminal volt OA', OA", etc. This con-
struction, carried out for various loads, gives the
following results, the impressed E.M.F. being constant,
namely, 2076 volt.
Ampere in secondary o 25 50 75 100 200

Terminal pressure 103*8 103-2 102-35 101*3 100 92
The drop between no load and full load is thus 3*8 volt.
The drop between full load and 100 per cent, over load
(which the transformer is
perfectly able to stand for a
short time) is 8 volt more, or a total between no load
and double the normal full load of 1 1 *8 volt. This drop
is of course too
great for practical purposes. It is due
to the large inductance we have assumed, merely in order
to explain the graphic method.
Up to the present we have assumed that the load is

non-inductive. It remains yet to extend the investigation
to cases in which the secondary circuit has also self-
induction, or capacity, or both. Self-induction is intro-
duced, if the secondary current is used for feeding motors
or arc lamps, in which cases there is developed an E.M.F.
at right angles to the current. The pressure at the
secondary terminals must therefore have a component
equal and opposite to this E.M.F. of self-induction, and
this component must be in advance over the current by
90. Let in Fig. 82 OArepresent the secondary current,
OB the power component of the secondary pressure, and
i6o TRANSFORMERS
OC the counter E.M.F. produced by self-induction.
The secondary pressure is then represented by the vector
OD, which advances over the current by the angle <p.
We call cos the power factor of the motors or arc

<p
lamps worked by the secondary current, and BD

is the
wattless component of the secondary pressure.

It is conceivable that the secondary circuit contains
partly appliances having only resistance,

c and partly other appliances having resist-
ance and self-induction. This is the
case if to same leads are joined
the
glow-lamps and arc-lamps or motors.
In a distributing system of 100 volt the
J.
glow-lamps would be arranged in simple
FIG. 82. parallel and the arc-lamps in parallel
series of two or three lamps. In Fig. 83
OE is the vector of
secondary pressure, <f the lag of
that part of the circuit which has self-induction, and OA'
the corresponding current. The current taken by the
glow-lamps is in
phase with the pressure, and must be
represented by a vector which is
parallel to OE. Let
this vector be A' A, then OA
is the total current supplied
to the circuit and \[/ its lag behind the
pressure. A
glance at the diagram shows
that OA<OA' + A'A and that -fyf. If
we measure separately the current
then
supplied to the glow-lamps and to the
arc-lamps, and also the current in the
undivided part of the circuit, we shall
find that the latter is less than the sum
of the other two currents. Let for instance
the power factor of the arc-light circuit
be 071 (<p = 45), and let there be 5
parallel series of lamps, each taking 15 ampere, then
OA'=75 ampere. In addition to the arc-lamps let us
insert a number of glow-lamps, taking collectively 32
ampere. The total current is then, not 107, but only 100
ampere, as will be found on drawing Fig. 83 to scale.
The power factor of the whole circuit is then cos = 0*85.' \|/
The transformer is thus apparently loaded to 10 kw. in ;
reality, however, only to 8*5 kw. The reduction in the

GRAPHIC DETERMINATION OF DROP 161
load brought about by the phase difference between

is
pressure and current. We have now to investigate what

effect this phase difference has on the terminal pressure or
on the ratio of primary and secondary terminal pressure.
Let, in Fig. 84, OA
represent the secondary current
and OBthe terminal pressure. The E.M.F. induced in
the secondary winding must obviously contain the follow-
ing three components. First, OB secondly, BB', to
;
cover the ohmic loss; and thirdly, B'C, to counteract the

E.M.F. of self-induction. The vector of the secondary
E.M.F. is therefore OC, and by our assumption of an
equal number of turns in both coils, OC
is also the
E.M.F. induced in the primary coil. The E.M.F.
FIG. 84. FIG. 85.
impressed on the primary terminals must also have three

components namely, OC, CD, to counteract self-induc-
;
tion, and DE, to cover the ohmic loss. We thus obtain

the vector of the E.M.F. impressed on the primary
terminals OE. The inclination of the straight line BE
is, as before, determined by the ratio between resistance
and inductance.
With a mixed load of glow-lamps and arcs the angle
\|/
varies with the number of lamps of each kind in use at
any time. If, however, the load consists only of arc-lamps
which are switched in or out in series of two or three,
then the lag (which we will call f in this case, compare
Fig. 82) remains the same for all loads. The current
may, as before, be plotted on a line o\ drawn parallel to
BE (Fig. 85), the scale being so chosen that o\ repre-
ii
162 TRANSFORMERS
sents full load. The pressure on the secondary terminals
is for full load OB. For a smaller load ol-^ the construc-
tion shown in the diagram gives a greater pressure,
namely, OB'. By comparing this diagram with Fig. 82

it will be seen that the drop is greater when the load is
inductive. This may also more precisely be seen if we
extend the example previously given to a case where the
load consists of arc-lamps. To make the construction
clear we had to assume an exaggerated case where the
E.M.F. of inductance was as much as 20 per cent, of the
working pressure. A modified form of the graphic con-
struction is given later on in which the scale used may
be of much larger size, so that even with a very small
inductive drop the terminal pressure for any current and
any lag can be found with great accuracy. For the
present we retain, however, the previous example in
which the inductance was 20 per cent, and the ratio of
inductance to resistance was 10 i. Let us now connect
:
the same transformer to an arc-light circuit in which cos

^
= 071 (p
= 45), andconstruct the diagram Fig. 85 for
various loads. This gives the figures contained in the
following table. To facilitate the comparison the figures
of the previous table are repeated.
Secondary current in ampere o 25 50 75 100 200

Pressure at secondary (cos (f = i 103*8 103*2 102*35 IOI '3 I0 92
terminals \cos<f=o"]i 103*8 99*5 95*2 91 86*8 68
This transformer (2 per cent, resistance and 20 per

cent, inductance) has on an inductionless resistance a
full-load drop of less than 4 per cent. Such a design
is
barely tolerable even with a non-inductive load, but
with an inductive load it would be inadmissible, since the
drop would amount to as much as 17 per cent. To
render this transformer suitable for motor work, the
geometric arrangement of the winding would have to be
so altered as to considerably reduce magnetic leakage.
If tested on short-circuit, the full secondary current
must be obtained with a primary E.M.F., which at the
outside must not exceed 5 per cent, of the normal supply
voltage.
Our investigation thus far has shown that with a con-
stant E.M.F. impressed on the primary terminals, the

pressure on the secondary terminals drops as the load in-
creases, the drop being due to the ohmic resistance in
both coils, and to their inductance. We have also seen
that, other things being equal, the drop is the greater
the smaller the power factor of the circuit to which the
transformer supplies current.
The case where the circuit has capacity must now
claim our attention. -Capacity alsoreduces the power
factor,and it may perhaps be thought that also in this
case the reduction in the power factor would be accom-
panied by an increase of voltage drop. This, however,
is not the case. On the contrary, the introduction of
capacity into the circuit di-
minishes the drop, and may
under certain circumstances
even convert it into a rise of
pressure. To facilitate the
investigation, we assume
at
first that the
apparatus sup-
plied with current by the
transformer has only resist-
ance and capacity, but no
inductance. Let the capacity
be a shunt to the resistance FIG. 86.
(a concentric cable feeding
glow-lamps is a practical illustration of this case). If E
is the crest value of the E.M.F., C the in Farad,
capacity
then EC coulomb are charged into and discharged from
the condenser twice every period, the charges being alter-
nately positive and negative. Let us consider the moment
when the E.M.F. has attained its positive maximum
value, and commences to decrease. The condenser is
then completely charged by the positive current which
has up to that time been flowing into it. As soon as
the E.M.F. begins to decrease, the period of discharge
begins, the current being now negative, although the
E.M.F. still remains positive for a quarter-period longer.
The negative current attains its maximum value at the
moment when the E.M.F. passes through zero. It is
therefore in advance over the E.M.F. by a quarter-period.
1
64 TRANSFORMERS
In Fig. 86 OE
represents the position of the E.M.F.
vector at time t corresponding to the angle a, and e
the instantaneous value of the E.M.F. After the lapse
of an infinitely short time dt the instantaneous value has
increased by de = -%,E sin a.dt, and the charge of the con-
denser has been increased by the amount idt, i being the

current at time / which has flown under the potential
difference de. We have therefore
idt = Cde
de .
The differential quotient -r is determinate if the shape
of the E.M.F. curve is known. Let this be a sine

function, then
de da.
and since da. = iirvdt we have

de T,
-7,
= L27TV COS Ot
dt
i= CE27rv cos a
The condenser current attains its maximum value at
all values of a for which cos a = i ;
that is to say, for
a = o, a = TT, etc. It is zero for a= -, a= ^TT, etc. Since

2 2
the opposite is the case with regard to the instantaneous
value of the E.M.F., we see that the vector of the con-
denser current must stand at right angles to the E.M.F.
vector,and be in advance over it by 90. The maximum
value of the condenser current is
I = CE27TV,
and its effective value
.
CE
Since = is the effective value of the E.M.F., which

x/2
we may designate by e ; we have also
In this formula i is given in Ampere, C in Farad, and e

in Volt. The usual unit of capacity is, however, not the
Farad, but the Microfarad (a million times smaller), and
by adopting this unit we have
ik = Ce27rvio~
6
,
the index being added to the sign for the current, to

k
show is a condenser current, which, lead-
that the latter
ing over the E.M.F. by 90, carries no power. Let iw be
the watt or power component of the current, the phase of
which coincides with the phase of the E.M.F., then
W being the ohmic resistance of the apparatus to which

current is
supplied. By reference to
Fig. 87 it will be seen that the total i* g
current i is given by the expression
Oe the E.M.F. vector, Oik the

is
vector of the condenser current, and Oe m

the vector of the power current. It is
important to note that the diagram only FIG. 87.
represents the case where capacity and

resistance are in parallel, and the condenser circuit has
so little resistance that it may be neglected. This is
approximately the case of a concentric cable connected
at one end to the transformer and at the other to glow-
lamps. The two conductors form the inner and outer

coating of the condenser, which is being charged and
discharged by the condenser current ik If all lamps are
.
switched out W
= oo and z w = o. In this case $ = 90
and i=ik The cable takes only the condenser current.
.
As lamps are being switched on iw increases, $ decreases,

i increases, and the power factor increases also.
If arcs instead of glow-lamps are connected to the far
end of the cable, there will be inductance in addition to
resistance. The E.M.F. of self-induction is
and its phase lags by 90 behind the phase of the current.

1 66 TRANSFORMERS
The corresponding component of the E.M.F. impressed
on the circuit must therefore have this value, but be 90
in advance of the current. The watt component em is of
course co-phasal with the current. In Fig. 88 iw repre-
sents the current flowing through the inductive resistance,
e w the watt component of the E.M.F., and e s the E.M.F.
of self-induction. By a parallelo-
gram we find Oe, the vector of the
E.M.F. This not only drives the
current through the inductive re-
sistance, but it also produces the
condenser current ik leading by 90.
The total current supplied by the
transformer is the resultant of iw
and 4 and its vector is O/. Ac-
cordingly as inductance or capacity
FIG. 88. preponderates, the current will lag
behind or lead before the E.M.F.
In the diagram the relations are so chosen that the
current leads.
We
have up to the present assumed that the con-
denser forms a shunt to the apparatus absorbing power.
This is indeed the most frequent case, but it is also
possible that the circuit is interrupted by a condenser,
which is then in series with the apparatus
absorbing power. A case of this kind
occurs if a liquid resistance is used in
testing transformers. Abarrel containing
salt water or acidulated water, and two
sheets of lead as electrodes, or an iron
trough containing an alkaline solution,
FIG. for electrodes make very
and iron plates
efficient resistances, capable of taking
up large amounts of power. Liquid resistances are for

these reasons very often used in lieu of solid resist-
ances. It is well known that a metal plate dipping
into a liquid acts as a condenser, and in using a liquid
resistance we introduce therefore into the circuit a
capacity in series with the resistance. Let, in Fig. 89,
Oi the current passing through the liquid
represent
resistance, and Oe w the watt component of the E.M.F.
The E.M.F. required to produce the condenser current

= i\ C27TV in absolute measure.
i is e k It lags behind
the current by 90, as shown in the diagram. The
resultant E.M.F., that is, the E.M.F. which must be
impressed on the terminals of the liquid resistance in
order that the current i may flow through it, is Oe. It is
obvious that also in this case the current leads over the
E.M.F. by a certain angle, and that the power factor of
the liquid resistance is smaller than unity. In Fig. 89
the angle <p has been drawn large to make clear the dis-
tinction between the two components of the E.M.F. and
their resultant. In reality the capacity effect of a liquid
resistance is much smaller than represented by the dia-
gram, but such effects of considerable
magnitude may occur when a trans-
former is supplying high pressure current
through a long concentric cable to a
load consisting of a large ohmic resist-
ance. The cable acts as a condenser
which is in parallel with the resistance.
The total current flowing into the cable
has then two components one inversely
:
proportional to the resistance and in

phase with the E.M.F., and the other
proportional to the capacity of the cable FIG. 90.
and in advance of the E.M.F. by a

quarter period. The total current will therefore lead
as compared to the E.M.F. by an angle lying between
zero and 90, as shown in Fig. 88. The lead of
current will obviously also affect the working condition
of the transformer, and to find its effect we apply the
construction previously given. In Fig. 90 the current is
represented by the vector OA, and the terminal pressure
by the vector OB. The ohmic drop in the secondary
is
BB', and must obviously be parallel with OA. The

E.M.F. of self-induction in the secondary is B'C, which
is at right angles to OA. The E.M.F. of self-induction
in the primary is CD, and the ohmic drop in the primary
is DE. We thus obtain OE as the vector
of the E.M.F.,
which must be impressed on the primary terminals of the
transformer. Since BB', DE, and B'D are all propor-
1 68 TRANSFORMERS
tional to the current, the inclination of the line BE
remains the same for all loads, and its length may to a
suitable scale be made to represent the secondary current.
We assume for the present that the power factor is
independent of the current. In this case the pressure at
the secondary terminals may be found for different cur-
rents by the construction shown in Fig. 91, in which OA
represents the direction of the current vector, and Oo
that of the secondary terminal pressure. The inclination
of lo, on which we measure the secondary current, is
fixed by the electrical constants of the transformer, lo
must be parallel to EB
in Fig. 90. From O
as centre
we describe a circle with a radius equal to the E.M.F.
impressed on the primary
terminals (the transforming
ratio is
supposed to be unity).
To find the position of the
primary E.M.F. vector we

measure from o to the left
a distance representing the
secondary current, and find
thus the point I. From I
we draw a parallel to oO, and

FIG. 91.
find E. Drawing EB parallel
to lo we find B. Then OB
is the
pressure at the secondary terminals. For a smaller
current I' the same construction gives the pressure OB'.
As will be seen from the diagram, the terminal
pressure increases with the current. We
have therefore
with increasing load, not a drop, but a rise of terminal
pressure. The rise of pressure depends, amongst other
things, on the angle of lead (p. If this is smaller the rise
will also be less marked, as will easily be seen from the
diagram, and it is obvious that there must be a certain

phase angle, for which there is neither rise nor drop of
pressure, but exactly the same pressure at full load and
at no load. For smaller values of $ there will be a drop,
but this will still be less marked than would be the case
if the current
lagged behind the E.M.F. We
thus come
to the conclusion that the drop depends, not only on the
magnitude, but also on the nature of the load. It is

greatest for inductive loads, smaller for non-inductive

loads, and smallest (or even negative, i. e. a rise) on
loads having capacity. For this reason the use of a load
having capacity in testing transformers for drop is mis-
leading. The drop so measured is always smaller than
that which would be found on a lamp-circuit, and it

may
even happen that the drop is negative, that is to say,
that the pressure rises when the load is increased. This
will happen the
if transformer has much magnetic leakage.
Thus the worse the transformer the more favourable will
it
appear when tested on such a load.
Up to the present we have assumed that the angle
of lead or lag <p is the same for all loads, and determined
the pressure at the secondary terminals as a function of
the secondary current. In most cases it is, however,
only of importance to know the drop or rise at full
current, since the fitness of a transformer for practical
use must naturally be judged by its behaviour at full,
and not at some intermediate, load. On the other hand,
it is know how the secondary pressure
important to
depends on
the power factor of the apparatus to
which the transformer supplies current. The same
transformer may be intended for use in connection
with apparatus of different power factor, and it will
depend on the change in terminal pressure between
full load and no load whether such use is practically
possible.
The problem we have to solve is therefore the follow-
ing Given a transformer of known resistance and induct-
ance. The primary or supply E.M.F. is constant. Find
the pressure at the secondary terminals with full secondary
current, but for different values of phase angle <p. The
graphic solution of this problem is
extremely simple,
and follows as a matter of course from what has been
said in connection with Figs. 84 and 91. For a constant
ampere-load the length of the line KB
is constant, the
only variable in the diagram being the angle changes E
takes different positions on the circle of primary E.M.F.,
and the locus of B must therefore also be a circle of the
TRANSFORMERS
same radius, the centre of which has relatively to O the
same displacement as B has to E.
Let in the vertical represent the current
Fig. 92
vector, OS E.M.F. of self-induction at full current,
the
and So the ohmic loss at full current in both windings ;
then O0 is equal and parallel with EB

of Fig. 85, and
o is the centre of the second circle just mentioned.
For a positive phase difference (current lagging behind
E.M.F.) the secondary terminal pressure is smaller OB
than OE, its value at no load. For a negative phase
difference ^ (current leading before the E.M.F.) the
secondary terminal pressure OB
X is greater than its value
at no load. With a certain negative phase difference <p 2
the terminal pressure at full load is

exactly the same as
FIG. 92.
at no load, and we have neither a drop nor a rise of
voltage. This phase angle is given by the point of

intersection B 2 of the two circles. The drop or rise can
be directly measured in the diagram it is the length ;
included between the two circles. If the load is an

inductionless resistance having no capacity $ = o and the
drop given by the distance E 3 to B 3
is The diagram .
shows very clearly how the drop increases as the lag of

current behind E.M.F. becomes greater in consequence
of the greater inductance in the apparatus to which the
transformer supplies current.
The length of the line Oo is, as already pointed out,
a measure for the secondary current. If the load is
reduced o is shifted accordingly, and the same construction
can be made for every load. We can thus determine the
secondary terminal pressure for every working condition
of the transformer. Such a determination is to be pre-

ferred to a direct measurement of the drop, for the
following reasons. In the first place, to make the direct
measurement reliable, the apparatus serving as a load to
the transformer must have exactly the same power factor
as obtains in the apparatus to which the transformer will
have to supply current. A
liquid resistance can therefore,
as a rule, not be used in the test. Solid resistances and
lamps are not only more cumbersome, but their adjust-
ment to produce one particular lag requires the use of
choking-coils or similar appliances, and this complicates
the test. In the second place, the direct testing of large
transformers requires an amount of power which is not
always available and, lastly, if these difficulties are over-
;
come, we have to face the further difficulty of having to

determine a comparatively small difference between two
readings of large absolute value, which cannot be done
with great accuracy. For these reasons it is better not
to attempt any direct determination of the drop, but to
solve the problem indirectly, in the manner described
with reference to Fig. 92.* As an example to show the
application of the indirect method, we take a 60 kw.
transformer having a ratio of 3000 to 200 volt. Resist-
ance of primary 0*9 ohm ohmic loss 18 volt. Resistance
;
of secondary o'OO36 ohm ohmic loss i *o8 volt. By

;
reducing the transforming ratio to unity we would have
an ohmic loss in the primary of 18 - =1*20 volt.
3000
To determine the length Oo in Fig. 92 we have therefore
the following data
Ohmic loss in primary . i = 0*6 per

'20 cent.
i '08 =
,, ,, secondary .
0*54 ,,
,, ,, total . . 2*28 = 0*14 ,,
We
short-circuit the secondary through an ampere-
meter, and supply as much primary E.M.F. as will just
suffice to produce the full secondary current of 300 ampere.
This experiment shows that 255 volt must be supplied
1
See the author's paper on this subject (Elektrotechnische Zeitschrift,
1895, p. 260), and discussion thereon.
172 TRANSFORMERS
to the primary terminals. This corresponds to 1 7 volt
at the transforming ratio of i i. We
have therefore
:
in Fig. 92 Oo= 16*85 an d 80=2-28, whilst is 200. OE

We can now draw the diagram, Fig. 93. is the OA
current vector. On this we mark off the power factor
cos q>. The corresponding position of the vector of
E.M.F. is OE, and
the terminal pressure which we scale
off on OE is
187 In a similar manner we determine
volt.
the terminal pressure for all other values of cos (p. The
result is given in the table.
following
60 kw.-transformer 3000 200 : volt on open circuit.
Pressure at secondary terminals with 300 ampere in
FIG. 93.
secondary and power factors varying from TOO to 50 per

cent.
Power factor in per cent. 100 99 90 80 70 60 50

With leading current 197 200 205 207 210 212 213
.
With lagging current 197 195 190 188 187 186 185
.
used on a glow-lamp circuit this transformer would

If
at full load have a drop of only ij per cent. if used ;
on a circuit containing arc lamps or motors the power

factor of which is about 070 to o'So, the drop would be
approximately 6 per cent.
The diagram, Fig. 93, leads to some interesting deduc-
tions. In the majority of cases the circuit has, not
capacity, but inductance, and the following remarks apply
to these cases, that is to say, to the left-hand side of the
diagram. If we could build a transformer which has
absolutely no magnetic leakage, then OS would be zero,

and o would lie vertically above O. The inner circle
would then approach the outer circle more closely as
we go to the left. In other words, the drop would be
greatest for an inductionless, and smaller for an inductive,
resistance. This case is, however, unattainable in practice,
for we can never reduce magnetic leakage to zero. The
inductance produced by magnetic leakage can, however,
with a careful design, be made very small, especially
for low periodicities. Imagine that we have reduced the
inductance so far as to be equal to the resistance, then
OS = So, and Oo includes with OA an angle of 45.
The distance between the two circles would then be
approximately the same for all values of q>. We should
thus obtain a transformer which has approximately the
same drop for all values of the power factor.
As a rule, the reactance is, however, greater than the
resistance, and the two circles diverge towards the left.
As a consequence the drop increases as the power factor
decreases. If the same transformer is used for a high
and low frequency, the pressure at the secondary ter-
minals will at full current be lower in the former case.
The E.M.F. of self-induction is for both windings,
OS = 2 X
that is OS is proportional to v. The higher the

to say,
frequency, the greater is the divergence between the two

circles. It must also be borne in mind that the power
factor of the apparatus to which the transformer supplies
current (motors or arc-lamps) is lower at the higher
frequency, and in consequence the vector of E.M.F. in
our diagram is shifted the more to the left the higher
the frequency. Both causes conspire to increase the
drop. If then the transformer is intended to feed, not
only glow-lamps, but also motors and arc-lamps, the
frequency should be chosen as small as compatible with
the proper working of alternating current arcs (45 to 50
complete cycles per second). This frequency is also
advisable on account of certain reasons connected with
the design of non-synchronous motors.
174 TRANSFORMERS
Drop diagram simplified. The 60 kw. transformer
here chosen as an example has an inductive drop of over
8 per cent. This is rather more than usual in a good
design, but it was necessary to assume -so large a drop in
order to make the diagram, Fig. 93, distinctive. In well-
designed large transformers the inductive drop is generally
under 4 per cent., and then the graphic construction, Fig.
93, must be made on a very large-scale to get accurate
results. Even then the elasticity of the compasses with
which we draw the circles is a source of error. To
obtain the drop we can modify the construction so as
to make the drawing of
the circles superfluous.
If the sides of the tri-
angle OS0
are very small
as compared with OE,
then a line drawn from
o to E will be very nearly
parallel to OE, and the
drop will be very nearly
equal to the piece cut off
on 0E by a perpendicular
dropped from O on to it.
The triangle OS0 may

then be drawn to any
convenient scale, and the
drop found as shown in
FIG. 94 .
Fi g- 94- O<? is, as before,
the impressed primary
E.M.F. at short circuit current; So = e r is the
and full
ohmic drop, including eddy current losses, as explained
below, and OS = ^ is the calculated inductive drop.
Draw vector 0E under the given phase angle 
. . . .
(31)
Correction for eddy current losses. There remains
still a slight correction to be made. On p. 156 it was
CORRECTION FOR EDDY CURRENT LOSSES 175
mentioned that the watt component of the primary

impressed E.M.F. may not only be due to ohmic resist-
ance, but also to certain losses caused by eddy currents.
In consequence e r will be slightly greater than calculated
from the ohmic resistance. To find the true value of e r
we must use a wattmeter in the primary circuit, and
divide its indication by the secondary current. The
correction is small, and if a wattmeter is not available we
can approximate it by measuring, not only the secondary
current 2 but also the primary current z lt and calculating
z' ,
er from
CHAPTER IX
CALCULATION OF INDUCTIVE DROP THE IN-
FLUENCE OF FREQUENCY ON DROP THE
INFLUENCE OF FREQUENCY ON OUTPUT-
EQUIVALENT COILS THE SELF-INDUCTION
OF A TRANSFORMER WORKING CONDITION
REPRESENTED BY VECTOR DIAGRAM CON-
STANT CURRENT TRANSFORMER
Calculation of inductive drop. The inductive drop being
due to the interlinking of the leakage field with the
windings, we can approximately pre-determine it from
the drawing of a transformer by mapping out the leakage
field in relation to the coils. Such a method can, how-
ever, only yield qualitative, not quantitative results, as
we have no means of determining exactly wh'at the flux
density is in any given point. By applying in a general
way the laws of magnetic circuits we can compare
different arrangements and say what details will influence
the drop and in what ratio, but we cannot calculate the
absolute value of the drop. To get quantitative results
we must fall back on experiments. The method of
investigation is then as follows. First we determine
the general principles of interlinkage between leakage
flux and winding without assigning to the resulting
E.M.F. a definite value; then we apply the formulae
thus developed to definite cases investigated experi-
mentally and obtain coefficients by which the formulae
become applicable quantitatively.
We investigate first cylindrical coils and then sand-
wiched coils. Let, in Pig. 95, I and II represent the
cross-section of the two co-axial coils, the radial depth
of winding being a^ and # 2 respectively, and the length of
176
CALCULATION OF INDUCTIVE DROP 177
the coils Let the secondary coil 1 1 be nearest the iron.

/.
Leakage through the space b between the two

lines pass
windings, and are of the general shape shown by the

dotted lines. The lines surrounding I pass wholly
through air, and have therefore to overcome a greater
magnetic reluctance than the lines surrounding II, which
pass partly through iron. The ampere-turns in both
coils being practically equal, the stray field of II will
therefore be stronger than that of I. In order to be
able to treat the problem mathematically we shall make
the assumption that the two currents have a phase
difference of 180 and that the ampere-turns are equal.
Both assumptions are very nearly correct. shall We
further assume that neither the
yoke nor the other core has a
material influence on the shape
of the stray field, which we take
to be distributed symmetrically
round the axis of the coils.
There must then be a boundary
surface of cylindrical shape be-
tween the two fields, which
passes through the space b, and
is distant ^ from the inner sur-
face of coil I, and A2 from the
outer surface of coil II.Where
precisely this boundary is, we FIG. 95. Leakage of cylindrical
cannot tell. All we know is winding.
that b x <
4 because of the
presence of the iron on the right of II.
Let %
and ;/, be the number of turns, / the perimeter
of the boundary, and 7 the number of turns per unit
radial depth of II. The ordinates of the shaded area
represent, according to the scale chosen, either ampere-
turns or induction. In the space b both are maxima,
and at the boundaries of the coils both are zero. In
an elementary strip of II having the radial depth da the
number of turns is
dn = yda
With these turns are interlinked all those lines of
12
1
78 TRANSFORMERS
force represented by the shaded area between Bj and B.
The corresponding flux is
and the E.M.F.-
de = 'v l
The total E.M.F. self-induced in II is the integral

of this expression taken between the limits a = o and
a = # 2 or ,
Oo
= &,Btf 2 -
e.
+y (# 2
bince
BTJ
-=
i 15
or
T) a
B = T1
B 1
d #2 ^2
we can write
B+ B! i
2\ 2
B /
-(/ 2
Since ya.2 is nothing else than the total number of turns
in II, and B is proportional to- 2

- = -
2
(X 2 being the
crest value of the ampere-turns in II), we find the

following expression for the self-induced E.M.F. in II
where /
2 is a coefficient to be found by experiment. In
the same way we have for the E.M.F. of self-induction

in I-
The total induced E.M.F. due to the main flux <

being
E = 4'44v;z<
we find the ratio of leakae to useful E.M.F.
^_ 22
-- /
",
7
2 $
el _ -
/^
ET
It is obvious that /i</ 2 but as we are ,
only concerned
with the sum of the ratios of e and E we may introduce
a mean value for these coefficients and call it k. Since
X
Xj = 2 we then have the same fraction in both equations
and can add the terms in brackets. Thus 6 1 and b^ are
eliminated, and it is immaterial that we do not know their
relative values. All we know is that their sum is b, and
that suffices. We
thus find the following expression for
the leakage E.M.F. as a percentage
. . .
(.i,
if by e s we denote the total inductive

drop reduced to
one circuit, and by e the useful E.M.F. induced in the
same circuit, both being effective values.
From tests made on various transformers 1 I found
for k the average value io~ 3 if X is the effective value
of ampere-turns in one coil in units of 1000, and the <
main flux in megalines
Percentage dropô-i-h
12 (33)
$\ 3 //
In this formula the dimensions are to be inserted in cm.

The symbols have the following meaning
1
Elektrotechnische Zeitschrift, 1898, No. 15.
i So TRANSFORMERS
X represents effective ampere-turns in one coil on one
limb in units of 1000.
<
represents megalines of useful flux passing through
both coils.
b is the radial space (copper to copper) between the
two coils.
#! and # 2 represent the
radial depths of the coils.
/ the mean
is
perimeter of the two coils.
/ is the length of the coils.
In the transformer designed in Chapter VI we have
for an output of 20 kw., that is 128
ampere in the
secondary
v ^
X=- -
x 128
=4*48 and (b= 1*027
1000
#! = 2*4 ;
a2 = i
'4 ;
b = 0*9 ;
/= 41 ;
p = 66.
In this transformer the inductive drop will be a little
over i
J per Since the
cent.
ohmic drop is only about i|-
per cent., the total drop with
128 ampere can under no
circumstance exceed 2 per
cent. This transformer may
therefore be used on a com-
bined service for light and
power.
In large transformers for
FIG. 96. Leakage of disc-winding. high pressure the inductive
drop may come out greater
than in the present example ;
to reduce it to some-
thing less than half we need only arrange the winding
of one circuit in two concentric cylinders and sandwich
the of the other circuit between them.
coil If the
windings are arranged in flat discs as shown in Fig. 96,

the same reasoning may be applied as with cylindrical
coils, but we must remember that for intermediate discs
the zero value of B lies in the plane which divides
each disc into two equal parts, so that the ampere-
turns producing maximum B are one-half of the total
ampere-turns in each disc. Only in the end discs which
lie against the two yokes will the zero value of B
coincide with the outer boundary of the disc, and in

these cases the full value of X
must be taken. To allow
for this circumstance we reason as follows. Let u be the
total number of discs on one core, that is, the sum of
primary and secondary discs, then u 2 discs will have

X X
to be counted with ampere-turns, and 2 discs with
ampere-turns. The average is
2 U
In the formula for the percentage drop we shall

therefore have to take for the coefficient, not o'i, but 0*05
times - ~, and only half the values for a^ and a z giving ,
-
Percentage drop 0*05 X (^-^\(b + -l^t
a
. .
(34)
/ being the mean perimeter of the coils and / their

radial depth.- X represents the effective ampere-turns
of one disc in units of 1000 and < the useful flux in
megalines.
The formula for the percentage drop has been
deduced for a core transformer, and it is not at first sight
obvious whether we may also apply it to shell trans-
formers. To test this matter experimentally, Mr. Moel-
linger has wound the same carcase in three different
l
ways the transforming ratio being in all cases i i and :
measured the inductive drop at the same load.

Case I. One primary and one secondary coil.
Case II. Two primary coils with one secondary coil
between them.
Case III. Three primary coils and two secondary
coils between them.
By using the formula

=
Percentage drop &
1
Elektrotechnische Zeitschrift, 1898, No. 15.
1 82 TRANSFORMERS
in each case we can find the value of k for intermediate
coils and end coils. The values found were 0*05 and
0*15 respectively. The agreement between experimental
and calculated results is
fairly good, as will be seen from
the following table
Case I Case II Case III
Percentage drop measured 430 196 327
Percentage drop calculated 426 191 34
According to Moellinger's experiments the end coils

produce three times as much leakage as intermediate
coils. According to our theory they should only produce
twice as much. Accepting, however, the experimental
results as substantially correct, we can, in order to
simplify the calculation, determine the coefficient k so
as to be applicable to all coils alike. According to
American practice the end coils are made half as thick
J
as intermediate coils, and they carry only ampere-turns.
The average value of k, which we will call ks can thus

,
be found from
V
2 x 3/ + (u - 2)/X = z// s
X
X being the excitation given by one intermediate coil
and u the total number of coils in both circuits.
u
We have thus
Percentage drop
= o-o 5 *t * '
Percentage drop = k, + *+') . . .

(35)
|(<}
where b is the distance ofcopper to copper between a
primary and neighbouring secondary coil a is the
its ;
thickness of a primary intermediate coil (or twice the

thickness of a primary end coil), a 2 the same for the
secondary circuit, and X
the exciting force of one inter-
mediate coil. The value of ks may be taken from the
following table
THE INFLUENCE OF FREQUENCY ON DROP 183
One primary full coil and two secondary

half coils .......... ks 0*067
Two primary coils, one secondary
full
full coil, and two
secondary half coils . k s = 0*060
Three primary full coils, two secondary
full coils, and two
secondary half coils .
^ = 0*058
The influence of frequency on drop. Imagine one
and the same transformer used alternately on circuits
,
of higher and lower frequency. For the same heating

limit in the copper the currents will remain unchanged.
The inductive drop is inversely proportional to the flux,
and therefore also inversely proportional to B, the useful
induction. Now we have to distinguish between two
cases. In one the E.M.F. remains unchanged, so that
the transformer gives the same output at the high and
low frequency and the other, where the E.M.F. at the
;
lower frequency is limited by the condition that the iron

heat shall be the same as at the higher frequency. In
the first case we have vB = constant, and in the second
case vB* = constant, where x is an exponent the value of
which may be found from
the curves Figs. 9 to ir,
representing the relations between frequency, induction
and losses by hysteresis and eddy currents. For a given
transformer and the same output the inductive percentage
drop is
or, since < and v are inversely proportional
D = cv
The drop is reduced in the same ratio as we diminish the
frequency. But the iron heat will increase. It is pro-
portional to vB^ or to x
= v
l -* = x
_
We may take 1*63 as an average value of x, so that if
PA is the iron heat at the frequency v, the iron heat at
V L will be
O'63
P
1 //i
-- P
A
M ( -
1
84 TRANSFORMERS
Thus halving the frequency whilst keeping the E.M.F.
constant will increase the iron heat by about 55 per
cent.
In the second case we assume that an increase of
iron heat is inadmissible. We
must then drop the
inductions at the lower frequency and be content with a
smaller output. From
vB* = ^ or B= (
we find
D = t#*
which may also be written in the form
='.D-. .
(36)
This expression is precisely of the same form as that
which gives loss in iron at constant frequency but
varying induction, namely
We
need not know the precise value of x. If we
represent lost power and induction by a curve, such as

Fig. n, the same curve with the scales suitably changed
will also represent the relation between frequency and
drop. Thus, let in a given transformer of v = 50 the

drop be 2 per cent., the induction 6000, and the loss 0*85
watt per kg., then we have
B = 6000 D=2 Ratio 3000 : i
W-o-85 v = 5o ,, i :
587
To half the frequency corresponds a loss of -5 =

5^7
0*425, and to this corresponds 6= 3750, which, divided
by 3000, gives D= 1*25. The curve of iron losses may
thus be used to determine the inductive drop at various
frequencies. Taking a 5o-frequency transformer with
2 per cent, inductive
drop as normal, its performance on
INFLUENCE OF FREQUENCY ON OUTPUT 185
circuits of other frequency found from the curve, Fig. 1 1,
will for the same heating be as under
Frequency .... 15 25 50 75 100

Inductive drop % . . o'88 1*25 2 2*6 3*1
Output ..... 60 77 100 113 125
of frequency on output.
Influence In this table a
line has been added to show how the output increases
with the frequency. For equal copper heat at all fre-
quencies the currents remain constant. The output is
therefore proportional to vB, and B is found from the
condition that the iron heat shall be constant. Let W
be the iron loss per kg. at v frequency and induction B,
then for the induction B! we have the loss 1} and for
W
equal heating we have
From the curve we find the corresponding value of

and the output is
If we assume
that the curve (Fig. n) may be
represented over short sections by
then for equal heating we have
V
0-385
v -.
The weight of iron, as far as it is influenced by the

frequency, will therefore be inversely proportional to the
'3 8 5 power of the frequency. As this is, however,
1 86 TRANSFORMERS
an inconvenient exponent, we may for frequencies
which do not greatly differ from the standard frequency
of 50 substitute the exponent 0*5, that is, the square root
of the frequency and this can be taken on the slide
;
rule. Thus we get the formula given on page 74 for

the weight of the carcase
10)
which may be used if the frequency lies between 40 and

60. The error introduced by substituting the square
root for the power 0*385 is only 2 to 3 per cent. The
ninth root of the power in the denominator is intended
to account for the better fill factor obtainable in larger
transformers. The formula is only valid for small and
medium size transformers say up to 100 kw. In larger
;
sizes the influence of the better fill factor and possibility

of better cooling result in a considerable reduction of
weight, as may be '
seen from the example given in
Chapter VII.
Equivalent Coils. The working condition of a trans-
former, such as can practically be built, may be repre-
sented by a perfect transformer to which are added
certain coils to represent the imperfections of the actual
transformer. These coils, having resistance and induct-
ance, may be
"
called "equivalent coils." By "perfect
transformer I mean a transformer having no losses, no
leakage, and requiring no magnetising current. The
object of introducing the conception of the perfect trans-
former worsened by equivalent coils so as to be reduced
to the condition of the practically possible transformer is
to simplify the analytical and graphical, treatment of the
latter. In Chapter VIII we have already made use of
this conception, though without specifically emphasising
it.
By combining vectors we have in reality combined
the E.M.F.s of the perfect transformer and its equivalent
coils. It will, however, be useful to investigate this
matter somewhat more in detail.
Let, in Fig. 97, T represent the perfect transformer.
EQUIVALENT COILS 187
Its terminals are i, i and 2, 2. Within these we have

on the primary side a coil R x having only resistance, and
a coil 1^ having only self-induction, both in series with
the primary winding. On the secondary side we have
similar coils. To
represent the constant iron losses we
have a R placed as a shunt to the
resistance-coil 7i
primary and to represent the wattless component of

the magnetising current an inductance-coil L^ placed
similarly.
If E! is the primary supply voltage, P /4
the iron loss,
and i^
the wattless component of the magnetising current,
and ih the watt component, we have
= Rt = or
R! and R 2 are the ohmic resistances of the windings,

and L! and L 2 the inductances, which can be found from
the inductive drop. Suppose we have by experiment
or calculation found the total inductive drop to be 2p per
cent, of the primary E.M.F., then by assigning one-half
of it to each winding we have
'
<
fi
- y F JLLi inrl
ailLl F - ^
-L>?9 E,
J-*l"
2
ioo ioo ;/!
T^
and
1
rL S 2 = (oL
T T
2 i }
'^1
n.2
and o)L 2 = -^ / (-
]
ioo 100 li\#i

188 TRANSFORMERS
Thus all the electrical constants of a given
transformer can be determined.
We
may still further simplify the conception of the
equivalent coils by assuming a transforming ratio of i i :
and the winding of the current receiving device so altered

that the primary voltage may be applied. The trans-
former may then be omitted from Fig. 97, and there will
only remain the equivalent coils, as shown in Fig. 98,
the primary and secondary being now combined into
one inductance coil L and one resistance coil R. The
apparatus receiving current is represented by the
inductance coil A and the resistance coil p.
The values of L^ and R 7 have not altered. The
,
new values for the two remaining equivalent coils

are
L=.32_Ei
100
and R=R 1 +
i!
The of a transformer. The self-induc-

self-induction
tion of any apparatus through which current is passing
may be defined as the reactance (>L), that is, the ratio
between the wattless component of the E.M.F. (-- -]

V 100 /
and the current.

To get the reactance we short-circuit the terminals
2, and measure the current supplied to i, i, and by
2
wattmeter the wattless component of the E.M.F. supplied
to these terminals. In this case the current I through
2, 2 be very nearly the same as that supplied to i, i,
will
so that we can also write
lOOl
for the reactance of the transformer proper. Since i^

and are very small as compared with I at any but the
ih
smallest loads, this expression will also hold good when
the transformer is under pressure. Only at very light loads
or on open circuit will the inductance L, and therefore
also the reactance o>L, be materially higher. In the latter
case we have I =o and l l = 7^ The electrical constants
THE SELF-INDUCTION OF A TRANSFORMER 189
of one equivalent coil containing both resistance and

inductance will then be
Take as an example a 2O-kw. transformer for 2000 volt

at 50 frequency on the primary side. Let it have 2 per
cent, iron loss and 2 per cent, copper loss, and let the
inductive drop be 3 per cent, and the wattless component
of the no-load current 0*5 ampere. have for this We
transformer under a moderate load
T 3 2000
100 10
314 L = 6 and L = 0*0192 Henry

The resistance of the equivalent coil is found from the
2 per cent, copper heat at 10 ampere primary current
from
R . 10= - .
2000, or R = 4 ohm
100
**
= -400- = 0-2 and ^ = , .
2000
This gives
2000
- = roooo
0*2
2
T 2OOO = 127 andIT/ / O'Z \
L = T
L =ft>L u
M ( )
= 10*0
314.0-5 vo'54/
We thus find that on open circuit the transformer has

a self-induction of 10*9 Henry, whereas on a moderate
load the self-induction is only 0*0192 Henry. On open
circuit the self-induction is 570 times as great as on load.
1 90 TRANSFORMERS
Working' condition represented by vector diagram.
We have already made use of vector diagrams to re-
present the relations between current, E.M.F., flux, exci-
tation and phase angles, assuming for sake of simplicity
that the transforming ratio is i i. These vector dia-
:
grams can, however, be made more simple by introducing

the conception of equivalent coils, which conception be-
comes possible for a transformation ratio of unity. In
this case the consuming device receives from the second-
ary terminals the current which has passed through the
equivalent main coils L and R (Fig. 98) with its full
strength, but attenuated as regards pressure by the re-

action of these coils and to the primary terminals has
;
to be supplied, not only this current, but also the currents

taken by the equivalent coils L^ and R/,, which, being
placed as shunts, have no influence on the pressure.
They only influence the primary current. As far as
the working current in the load X, p, and the effect of R
and L are concerned, we may disregard the shunt coils
altogether, but we must take them into account if we wish
to determine the primary current. It should be noted
that at constant primary voltage the current taken by
the shunt coils and its phase are constant, whatever may
be the working condition of the transformer and herein ;
lies the advantage of this method of treatment. can We

determine the working condition of the main circuit
of the transformer by the use of a very simple vector
diagram, and when this is done, acid vectorially the
currents in the shunt coils.
Let, in Fig. 99, MD 2 be the vector of the E.M.F.

supplied to the load, MI the current vector, then
and
DD 2 X
= RI
represents the loss of E.M.F. in R, and X
E= D
>LI the E.M.F. required to overcome self-induction.
The E.M.F. to be impressed on the primary terminals
is therefore ME. If the load consists of a group of con-
suming devices coupled in parallel, all producing the

same lag, its angle p will not change if some of the con-
suming devices are switched on or off. The length
of the vector MI will
change, but the phase angle p
will remain constant. The angle 2 D^D
also remains
constant, although the sides of the triangle will vary
WORKING BY VECTOR DIAGRAM 191
proportionately with the current. We can therefore

regard
as a measure of the current^ and by using a scale the

divisions of which are >L times as long as those of the
volt scale, we can scale off the current on the line X
E. D
We suppose the impressed voltage to be constant. ME
This is the usual case in
practice. To find the second-

ary E.M.F. and phase its
relation to the working cur-

rent I. if the latter be changed
by changing the load, we
reason as follows Since the :
angle DjED^j and the angle

<p
remain constant, the obtuse
angle at D 2 must also remain
constant, and D 2 must move
on a circle which passes
through E and M. Maximum
current will be obtained if
the load is short-circuited.
Then 2 D
coincides with M
and DjD is zero, because on
short circuit A is zero and D
(which also moves on a circle

by reason of the angle at
D always remaining a right FIG, 99. Constant power factor.
Variable current.
angle) has simultaneously

with D! moved to Its positionD . is
given by the con-
dition that
<D EM = <D ED 1 2
Let this angle be a, then the inclination of the tangents

in M and E to the circle through D
2 is in relation to the
line ME obviously a <p.

The centre of this circle
can thus be found. The centre of the circle through D

is in the middle of ME.
The radius of the inner circle is obviously
192 TRANSFORMERS
Radius =
ME
2 cos
(a + <p)
and the distance of its centre from the line ME is
We have now
all the data necessary to draw the two
circles. The
radius of the outer circle is half the im-
pressed voltage a, which depends on the equivalent
;
coils, is fixed by the condition that
Ip'a = R
Tf-
a>L
and f is given by the character of the load. To find the
working condition for any current I at phase angle <p
proceed as follows. Calculate ED 2 from
cos a
where ED^wLI, then strike a circle from E with radius

ED 2 and mark the point of intersection with the inner
circle D
The 2. triangle 2 X ED D
may now be drawn.
Prolong EDj to find D, the point of intersection with
the outer circle, then MD
is the direction of the current
vector, and the strength of the current can be scaled off

on ED using the ampere scale as previously explained.
l5
MD 2 gives the direction of the E.M.F. vector and the

E.M.F. itself if measured with the volt scale. To find the
primary current we have to add vectorially ih and i^. In-
stead of doing this to MI, the real current vector, we
may use the ampere scale with divisions of >L and rotate
the two components by 90, so that ^ falls in line with ME
and ih comes into quadrature with it. make We
then DjOi a measure of the primary current, and <p x is

is
its lag, whilstDjE is a measure of the secondary current,

and its lag in respect to the E.M.F. supplied to the load
is
<p.
Its lag in respect to the primary impressed E.M.F.
is
given by the angle DME
=\[/.
The power given to the main coils and load is
P= I . ME cos /
or, since I can also be scaled off on ED A ,
p=
ED 1ME CQS
,
If we join M and D
A (not clone in the figure to avoid
complication) we get a triangle MD
X E, the area of which
isobviously
EDi '
. MD = ED
A/rnk- -ME
lvrT7
cos \J/
I
= ME .
= M E TD 1
ME = I is an ideal short-circuit current which would
flow if the resistance of the equivalent coil R were zero.

For any given transformer this
ideal short-circuit current is a
constant. Measured with the
volt scale, D T is an E.M.F.X
We thus find that the power

given to the transformer (apart
from that used up in iron
losses) is
proportional to DjT,
namely
Fig- 99 s tn e general i
vector diagram of a trans-

former. For the special case
that p is zero it has first been
deduced by Mr. Heyland and
applied to the investigation of
asynchronous motors. An
asynchronous motor may be
considered a transformer in
FIG. loo. Unity, power factor,
which the secondary gives up variable current.
its power mechanically and
electrically, the former preponderating. To cause the
secondary to give up its power wholly electrically, we
need only hold it fast and insert as much inductionless
I
94 TRANSFORMERS
resistance in its winding as necessary to bring the
primary current to its normal value. Then D! and D
in Fig. 99 will coincide, and 2 M
will not be a broken
DjD
line as in Fig. 99, but a straight line. thus get the We
diagram shown in Fig. 100. The notation is the same
as in the previous figure. Since <p
= o, the radius of the
inner circle is
P A-
Radius = - ME
2 cos a
and a is found as before from the relation

R
(O.
R being the resistance of primary and secondary. The

transforming ratio is again supposed to be unity. The
power supplied to the primary apart from iron losses is
that supplied to the load or external resistance (or in

reality the power supplied mechanically) to the shaft is
P2 = I x DT2 2
The secondary current is DjE and the primary current

is DA-
In Chapter VIII was given a graphic method for
determining the drop for any current and any power
It is of interest to show that by
factor. using the
general vector diagram Fig. 99 we arrive at the same
result. The problem may be posed thus Let the :
primary impressed E.M.F. be constant, and let also the

secondary current be constant, but let the power factor
of the load vary what will be the E.M.F. impressed
;
upon the load ?
Let in Fig. 101 the triangle DjED 2 again refer to the

equivalent then
coils, since the current is constant, this
triangle will not alter in size, and D 2 can therefore not
move on the circle 2 ED M
of Fig. 99. It can only move
on a circle struck from E as a centre.
Let DiED 2 be the position of the triangle for p = o,
then MD
2 is the secondary terminal E.M.F. for unit
power factor. Reduce now the power factor so that Dj

takes the position D/ and 2 the position 2 '. The D D
wattless component of the E.M.F. in the load is now
DD/ and that in the equivalent coil is D/E. The total
E.M.F. impressed on the load is 2 '. The new posi- MD
tion of the triangle has been obtained by rotating it
counter-clockwise through an angle /3. Instead of doing
this we may also keep the triangle fixed and rotate the
EM
line in a clockwise direction through the angle ft so
M arrives at M P Obviously MB^M^.
that Since
MD-LDE and MD^DjE
we have B
<DMD 2
=3
and <D MD/ = ?-/3
2
If now we rotate D 'M 2
clockwise through the

angle ft so that D 2
'
comes
to D 2 and M comes to M lf
the angle included between

D.Mj and DM
2 will be-
so that
To get the terminal

E.M.F. at, say, <p
we strike
a circle with e 1 = from EM
E as centre, and draw from
D,the2
i
vector of secondary
r
power factor
i FIG. 101. Constant current, variable
.
E.M.F. under the angle <p

to 2 M. D Its point of intersection with the e1 circle is M 3
The terminal E.M.F, on the secondary side is therefore
the same construction as has been deduced in a

This is
different manner in Chapter VIII. It need hardly be

pointed out that the diagrams 99 to 101 are not drawn
to scale. In good transformers the effects of iron losses
(OOj), magnetising current (EO), leakage (DjE), and
ohmic losses (DgD^ are very much smaller than shown in
196 TRA NSFORMER S
the diagrams it was, however, necessary to exaggerate
;
these effects in order to make the diagrams clear.

Constant current transformer. The problem of getting
constant current at variable pressure from the secondary
terminals, if constant pressure is supplied to the primary
terminals, cannot be solved without some special device,
such as movable coils (Ferranti's). With fixed coils only
an imperfect solution is possible, though if some slight
variation of current is permissible, even an approximate
solution has some practical value.
may investigate We
this problem by using the general transformer diagram
FIG. 1 02. Constant current transformer.
given in Fig. 99. The carcase of the transformer must

be specially designed to produce much leakage ;
that is
to say, o>L of the equivalent coil must be large, and the

higher the frequency the better. Fig. 102 shows a
carcase specially designed to produce leakage.
The two coils are on two limbs of the carcase, and there
is a third limb without Its object is simply to
winding.
form a magnetic by-pass to the useful flux, that is to say,
to increase leakage. Since the winding space is not re-
stricted as in an ordinary transformer, the ohmic resist-
ance of the coils can be made very small, so that the angle
a in the triangle X
D ED
2 in Fig. 99 becomes very acute,
CONSTANT CURRENT TRANSFORMER 19;
and D E =a>LI
X
becomes large in comparison with DiD 2 .
We may assume a to be nearly zero then the centre of

;
the inner circle will be distant from by ME
The points D 2 and D will nearly coincide. DjE is a

x
measure of the secondary current, and is

very nearlyMD 1
the secondary E.M.F. On short circuiting the load

MD 2 becomes zero, and MDj so small that we may con-
sider ME a measure for the secondary current. The
smaller the power factor of the load the flatter will be
the inner circle (see Fig. 102) and the greater the varia-
tion of current DE
X for a given range of secondary
E.M.F. of, say, MD X
as a maximum down to zero.
If there be self-induction in the external circuit the
arrangement will therefore be very imperfect or only

applicable over a very small range of secondary E.M.F.,
the greatest attainable E.M.F. being many times smaller
than the primary E.M.F. The arrangement is therefore
useless for arc light circuits it can, however, ba used for
;
glow-lamp circuits. In this case <f is zero and D coin- x
cides with D. The variation between minimum current

DE at maximum voltage MD
and maximum current ME
on short circuit is much smaller, so that lamps connected in
series may be extinguished by short-circuiting their ter-
minals without risk of overstraining the rest of the lamps
remaining alight. The secondary circuit must, however,
not be interrupted, as that might put too great a stress on
the insulation of the secondary winding. With, say, 100
volt on the secondary at full load, and an equivalent
primary E.M.F. of about 500 volt, the transformer can
be so designed that on completely short-circuiting the
secondary the current will not increase by more than
about 4 per cent.
CHAPTER X
THE D YNA HOME TER THE D YNA MOME TRIG
WA TTME TER MEA S UREMENT OF THE
POWER CARRIED BY CURRENTS OF IR-
REGULAR FORM THE INDUCTION WATT-
METERMEASUREMENT OF POWER IN
THREE-PHA SE CIRCUIT THREE- VOL THE TER
METHOD THREE-AMPEREMETER METHOD
The dynamometer. We have up to the present gener-
ally supposed that current and E.M.F. follow a sine
law. This assumption, although very convenient for
the analytical or graphic treatment of alternate current
problems, is rarely in accordance with fact, and the
question arises what errors are introduced by assuming
that current or E.M.F. change according to sine curves,
when in reality the curves are of a more complicated
nature. This question will be answered by investigating
whether an amperemeter, if applied to the measurement
of a current of irregular curve, gives the true effective
value of the current, and a wattmeter gives its true
power. To investigate this matter, it is in the first place
necessary that we come to an understanding as to what
" To
is exactly meant by the term effective currrent."
some extent, this matter has already been treated in the
first chapter under the heading Fundamental Equation.
But there we have assumed that current and E.M.F.
are simple sine functions of time, and we have neglected
all the higher harmonics of the Fourier series. We have
now to consider whether our previous conception of the
term " effective current" will still hold good if these
higher harmonics are not negligibly small. To do this
we may again make use of the comparison of two glow-
lamps of exactly the same construction, and assume that
198
THE DYNAMOMETER 199
one of these is fed by a continuous current, and the other

by an alternating current. If both lamps are brought
to the same state of incandescence, and absorb therefore
an equal amount of power, then the effective strength
of the alternating current will obviously be equal to that
of the continuous current. Since both lamps must have
the same temperature, the resistances must also be equal.
Let this be R, then the work done in the time T is
'
2
I RT, if I is the strength of the continuous current.
The work clone on the other lamp is obviously /^PR^/,
if denotes the momentary value of the current, which

I
is,of course, a function of the time /, this function being

graphically represented by the current curve, which may
be of any shape. The effective value of the alternating
current is thus I and is given by the equation
,
i
- J i
/*<*
'.-Tf
The question we have now to consider is whether

the amperemeter will show this or some other value. All
the instruments used for measuring alternating currents
are based upon some electro-dynamic or some heat effect
produced by the current. In either case the momentary
action is proportional to the square of the momentary
value of the current and all these instruments will
;
therefore be equivalent as regards their suitability for

measuring alternating currents. We
may thus restrict
the investigation to one particular type of instrument ;
the result is valid for all other types.

For this purpose let us select the well-known dynamo-
meter invented by Weber. The movable coil is sus-
pended within the field of the fixed coil, and is thus
subjected to a deflecting force, which is proportional to
the square of the current, and which is balanced by the
torsional force of a spiral spring. In measuring a con-
tinuous current, I the deflecting force is given by the
,
equation I
2
2,
K being a constant depending on the
construction of the instrument. The balancing force
exerted by the spring is proportional to the angular
200 TRANSFORMERS
deflection D, which must be given to the torsion head to
keep the pointer at zero. We have thus the equations

I
2
=K D2
which is the well-known formula showing the relation

between current and reading on a dynamometer.
The question we have to consider is whether the
same formula is
applicable if the current is not continuous,
but alternating according to any harmonic function of
the time. Let this function be represented by a curve,
in which time is measured on the horizontal, and current
strength on the vertical. By squaring the ordinates we

obtain a second curve, and the area enclosed between
the axis of abscissae and this second curve represents
T
the expression /~ J V/, whilst the height of a rectangle of
equal base and area represents the square of the effective

current. The movable coil of the dynamometer is acted
on by two forces one being the tension of the spring,
;
and the other the electro-dynamic force of the current.

The former is constant, and the latter variable, but always
opposed to the former. If I, is the current at time /,
the force acting at that time upon the movable coil is
DK 2
If. This force produces an acceleration given by
the expression
DK
-
2 2
1
m we denote the mass of
-, if by
the movable coil reduced to the radius at which the force
acts. If mass were sufficiently small, then, this
this
acceleration, which is
alternately positive and negative,
would produce a visible oscillatory movement of the coil,
the velocity of movement being represented by
m
whereby we assume that at time / = o the velocity is also
zero. The mass of the movable coil is, however, very
large in comparison with the forces acting on it, and the
frequency with which the accleration changes from a
positive to a negative maximum is so great that no visible
oscillation is produced in the coil. This, moreover, is
THE DYNAMOMETER 201
an obvious condition for an exact measurement. The

velocity v must at all times be infinitely small, that is
to say
m
From this condition we obtain
The expression on the right-hand side of this equation

is
obviously nothing
else than the height
of the rectangle previ-
ously mentioned, that
is to
say, the square
of the effective value
of the current. We
have therefore
DK = IQ FIG. 103. Dynamometer adapted for measuring
power given by generator.
and this is a proof
that the really measure the true
dynamometer does
effective value of the current, whatever may be the
shape of the current curve. The constant of the K
instrument may once for all be determined by calibration
with a continuous current, and then used for all measure-
ments with alternating currents, provided always that
the instrument does not contain any metal masses in
such proximity to the movable coil that a disturbing
effect through eddy currents is produced.
The dynamometric wattmeter. The dynamometer
invented by Weber (commonly called the " Siemens
Dynamometer ") may also be used for measuring power
if it be so
coupled up that one of its two coils is traversed
by the working current or a constant fraction of it, and
the other coil by a current proportional to the E.M.F.
This arrangement of the instrument is shown in Fig. 103.
202 TRANSFORMERS
C is the movable, and c the fixed coil. G
is the alter-
nator supplying current to the apparatus T, which may

be a transformer.
The dynamometer, as usually constructed, has two
terminals, A and B (or three if the fixed coil is arranged
in two parts, so that the range may be increased). If
the instrument is intended for measuring power, then the
connection between the movable and the fixed coil must
be provided with a third terminal, D. The current of
which the power is to be measured is then led through
the fixed coil, whilst a shunt containing an inductionless
resistance, W, is arranged between the terminal B of the
movable coil and the return circuit. The shunt resist-
ance may consist of a ribbon of platinoid, constantan, or
other alloy having a small temperature coefficient, and
the ribbon should be wound zigzag fashion, so as to
reduce the inductance to a negligible quantity. A
series
of glow-lamps may also be used instead of a metallic
resistance, if the resistance at the given E.M.F. has been
previously ascertained. In the shunt circuit we insert
an amperemeter, a although this is not absolutely
} ,
necessary.
If the resistance W
is very large, we
may without
sensible error assume that the shunt current is in phase
with the E.M.F. impressed upon the circuit, which is
shown by the voltmeter v^. In other words, the current
flowing through the movable coil has no lag, and is pro-
portional to the E.M.F. The main current in the fixed
coil may laof or lead as compared to the E.M.F. If the
r i
apparatus receiving power is a transformer supplying

glow-lamps, there will be practically no lag if it supplies
;
arc lamps there will be an appreciable lag. Let I be

the main current (maximum value !,), and i the shunt
current (maximum value im ), then if E.M.F. and current
follow a sine law, the turning moment acting on the
movable coil is proportional to
l m sin (a-<p)i.m sin a
a representing the phase of the E.M.F. at the time to
which the expression refers, and <p the angle of lag of the
current passing through c. If by W
we denote the re-
THE DYNAMOMETRIC WATTMETER 203
sistance of the shunt, including that of the amperemeter

a,! and connections,
we have
=
m
W
and the turning moment may also be considered pro-
portional to
\,n sin (
a sin a
-<p)^
twisting the torsion-head of the instrument we

By
apply a twisting couple to the movable coil, which is
given by the expression
2
DK D
being the angular deflec-
,
tion, and K
the constant of the instrument. Let be m
the mass of the movable coil reduced to the radius at
which the forces act, then the acceleration at the time to
which the phase angle refers is
m
and the velocity attained after the time t is
=~ - -
mjf\DK*
v !,_%W sin (a <p)
sin a)dt
=
provided at time / o the velocity is zero. The mass
of the coil is very great as compared with the forces, and
the changes in direction of the electro-dynamic force
occurs with great rapidity, so that the movable coil
cannot follow these changes, and remains at rest. In
other words, v is at all times zero, and this requires
that
/DK - f\m^ sm
2
a ~ sm a
( f)
This equation may also be written as follows

204 TRANSFORMERS
5
The value of the integral is i and we have there-
fore
W 2
or by introducing the effective values of current and

E.M.F.
DK 2
= ~lcos<?
It has already been shown that the product \e cos $

is the power carried by the current I flowing under a
potential difference e, and having a lag p. The product
i 1 i
/WWVWM/V-I
o
FIG. 104. Measurement of FlG. 105. Measurement of
main current. shunt current.
WDK 2
is therefore the power of the alternating current.
The symbols have the following meaning
D is the angular deflection of the torsion-head,
K is the constant of the instrument which has been
determined by calibration with a known con-
tinuous current ?',
so that K=
W
the total shunt resistance in ohm.
is
We
may use the instrument as above explained with
reference to Fig. 103, and also as an ordinary dynamo-
meter, using the terminals and B, but not D. A We
measure first the main current I, arranging the con-
nections as shown in Fig. 104, and then the shunt current
,
arranging the connections as shown in Fig. 105. If
it is
permissible to neglect the self-induction of the
MEASUREMENT OF POWER 205
movable compared with the sell-induction of the

coil as
apparatus T, and if it is also

permissible to neglect the
resistance of the fixed coil as ccmpared with the resist-
ance of the shunt circuit W, then the dynamcmtter dees
not increase the lag nor waste power. Let the reading,
when arranged according to Fig. 104 be D and when 1}
arranged according to Fig. 105, be D 2 then ,
I
= I and
By multiplying these equations we obtain
Previously we obtained the expression
which, combined with the above, gives-
cos <p
=
By thus taking three measurements with the same

instrument (one power measurement and two current
measurements), we may find the lag <p. It is important
to note that only the enter into the
readings equation for
the lag. In order to determine the latter, it is therefore
not necessary to know the constant of the instrument,
nor need we know the shunt resistance.
Measurement of the poiver carried by currents of
irregular form. Up to now we have assumed that
current and E.M.F. follow a sine law. If they follow-
any other law, the measurement taken by the wattmeter,
as explained with reference to
Fig. 103, will still be per-
fectly correct. The work done by the current in time
T is
The main current I, as well as the shunt current /, may

206 TRANSFORMERS
be any periodic function of the time, the only condition
being that the frequency must be the same for both.
This is obviously the case, since both are produced by the
same E. M. F. The turning moment acting on the movable
coil is DK
2 =
I z, and varies
continuously. If the frequency
were sufficiently small (T very large), and the mass of
the coil also very small, the coil would assume an oscil-
latory movement. But the frequency is great, and the
mass of the coil is so considerable that no movement
takes place in other words, the integral of the acceler-
;
ation multiplied by the differential of time, that is the

velocity taken over the time of a complete period, is
zero.
J/-
T
It was shown above that W f \idt is the work done
by the current in time T. Since power is work divided

by time, we have
WDK 2
=P
that is to say, the reading of the wattmeter gives the
true power P, whatever may be shape of the current
and E.M.F. curves.
To make a power measurement by means of a
dynamometer, we must know the constant of the in-
strument, and the exact value of the shunt resistance W.
If the latter consists of platinoid or other alloy of small
temperature coefficient, this condition causes no diffi-

culty. If, however, a series of glow-lamps is taken as
the shunt resistance, the value of the latter is not con-
stant, but varies with the E.M.F. To find we may W
proceed in either of two ways Read
: the deflection D
on the wattmeter, the current i on the amperemeter a 1}
and the E.M.F. e on the voltmeter vl9 The two latter
readings give W= .,
and the power is then given by the
expression
The difficulty of this method lies in the necessity to take

three readings simultaneously. To obviate this, we may
do away with the reading of the current by calibrating
the resistance beforehand, which may be done with con-
tinuous currents. A
curve may be plotted, giving as W
a function of e, and then it suffices to take the readings of
D and e, W
being taken from the curve. The ampere-
meter ctj may be cut out of circuit, whereby the conditions
of an inductionless shunt
is more easily fulfilled.
instrument
If the is
connected as shown in
Fig. 103, it shows the

power supplied to the
transformer and that
wasted inthe shunt,
that isthe total power
FlG I0 6. -Dynamometer adapted for measuring
.
Supplied by the gener- power absorbed by transformer.

ator. If only the power
given to the transformer is to be measured, the watt-

meter must be connected up as shown in Fig. 106,
where the shunt is taken from the wire which leads to
terminal A. In this case, the fixed coil c carries only
the current passing through the primary of the trans-
former, and the power wasted by the current in the
movable coil C is not measured. The expression
P= WDK 2
gives then the power supplied to the transformer. If

the load on the latter consists entirely of glow-lamps,
then the product of secondary current and secondary
terminal pressure is the power delivered by the trans-
former. We
measure on the amperemeter a.2 the second-
ary current 2J and on the voltmeter v.2 the secondary
z'
208 TRANSFORMERS
pressure e>2 . The efficiency of the transformer is then
given by the expression
WDK'
If the supply
voltage is high, it is advisable to so
connect the wattmeter that in the instrument itself no
great potential difference can arise. Otherwise there is
the risk of breaking down its insulation. Fig. 107 shows
the arrangement of connections which should on that
account be avoided. Theoretically Fig. 107 is equivalent
with Fig. 1 06 ;
the latter arrangement is, however, from
a practical point of view, preferable, because the highest
potential difference which can arise between the fixed
and movable coil is only that due to the resistance and
inductance of the latter, and is
therefore only a small fraction of
the total pressure. On account
of safety in handling the instru-
ment, it is also advisable to earth,
if possible, that terminal of the
generator which is directly con-

9 nected with the terminal
the wattmeter.
A of
FIG. 107. Incorrect method
of connecting. The theory of the ordinary
dynamometer used as a watt-
meter given above is only correct if the assumption of
an inductionless shunt circuit is justified, by reason of
the inductionless resistance being very great as compared
with the reactance of the movable coil in series with it.
There will then be almost no lag of the current in this
coil. To reduce the lag absolutely to zero is, of course,
impossible, since the action of the instrument pre-supposes
the existence of a mechanical force, which can only be
obtained by means of a coil producing a magnetic field of
its own, that is to say, having a certain amount of induct-
ance which must produce some lag. If this inductance

is not negligible, a correction to the reading must be
made as shown in the following theory.
Let Oe in Fig. 108 represent the vector of the total
supply pressure, and O/ that of the main current through

the fixed coil, which has a lag <p,
so that
where 27rvL is the reactance, and W the resistance of the

apparatus to which power is
supplied. The power is
Of* 60.
If there were absolutely no self-induction in the watt-
meter, the shunt current would be in phase with Oe. As
there is some self-induction, it lags behind by an angle
which we denote by \//, and assumes in the diagram the
position O?o. If w is the shunt resist-
ance, and / its inductance, we have
2 TT vl
tan \b =
w
The wattmeter will then not indicate
the true power Oi x Oa, but the larger
power Oi x O^. To find the true power,
we must therefore multiply the reading
by the ratio , Let P' be the apparent
power shown by the instrument, and P

the true power, then FIG. 108. Correction
for inductance of
pressure coil.
Since Oa = Oe x cos <p

and Oc = O6 x cos((p i//)
= O6 x 1
cos \jt
cos (^ i//)
we have also
COS I//
COS ((f I//)
Since is a constant of the instrument, it may be

\fj
determined once for all. The phase difference between

main and shunt current may be found from
D
as previously explained. Since i// is known, we know

also <p,
and can determine the correcting factor.
14
2io TRANSFORMERS
The above expression may also be written in the
form
2
p = p'
i + tan \f
i + tan 
if the power circuit has capacity, or only resistance, whilst
the wattmeter has too much self-induction), then the

true power is larger than that read off on the wattmeter.
There are two values of for which the reading of the
i//,
wattmeter gives absolutely the true power. This will

obviously be the case if the correcting factor equals
unity namely, if
;
= o, and also if $ = $> In the former
i//
case we have an instrument with negligible self-induction,

and in the latter case the self-induction in the shunt
happens to be of such a value as to produce the same
lag as in the main circuit. The case most frequently
occurring in practice is that there is some
lag in the
main circuit, and that i//.
The correcting factor is
then slightly less than unity, and attains its minimum if
<p happens to be equal to 2\fj. This gives the greatest
error possible, and the correcting factor is then --*.
cos?
2
For all p, either greater or smaller than
other values of
the error is less, and the
2i//, correcting factor approaches
more nearly to unity. The following table gives the
greatest possible error for different values of i/;.
* o
2*5
5
7*5
10
o
I2'5
15
In modern instruments, not ordinary dynamometers

adapted as here explained for measuring power, but
THE INDUCTION-WATTMETER 211
wattmeters designed as such on the dynamometer prin-

is so small that no correction need be
ciple, the angle\jj
made when using the instrument. This is attained by

making the moving small and light, pivoting it in
coil
jewels, and including very large induction-

in its circuit a
less resistance. Sometimes the latter is subdivided into
two or three parts, and a switch is provided, so that the
range of the instrument may be increased.
In the ordinary dynamometer the moving coil has
about the same dimension as the fixed coil, and any
deviation from a right angular position of the two coils
relatively to each other must be avoided. This is the
reason why a torsion-head is necessary, so that the
controlling spring may be set by hand for each reading.
In a wattmeter proper the moving coil is so much smaller
than the fixed coil (or coils) that even when deflected it
does not get into a sensibly different
field. Setting by
hand by means of a torsion-head is therefore no longer
necessary, and we thus obtain an inspectional instru-
ment a pointer connected to the moving coil indicating
;
the power on a fixed scale.

The Induction-wattmeter. In wattmeters con-
structed on the dynamometric principle no iron may
be used to increase the turning moment, but there is
another type of instrument in which the use of iron is
not only permissible but necessary. This type acts by
reason of an induced current in the moving system, and
is therefore called induction wattmeters. In the most
simple form of instrument the moving system is a disc of
aluminium or copper, and the currents induced in it are
eddy currents. Fig. 109 is a diagrammatic representa-
tion of the principle. D is the disc pivoted in jewels
and controlled by a helical spring. M
represents the
end view of the iron core of a magnet excited by the
main current, and S that of a magnet excited by
the shunt current. B is a permanent brake magnet, the
poles of which embrace the disc on both sides. Its
object is
merely to make the instrument dead beat. The
flux from both electromagnets traverses the disc at right
angles, and each flux creates round its own magnet eddy
currents which cut through the flux of the other magnet.
212 TRANSFORMERS
In consequence of this interaction between flux and eddy
currents the disc receives a turning moment which is
current.
proportional to the power of the alternating
It
will revolve until the tension of the spring balances

the turning moment, and a
pointer attached to the disc
will indicate the power on a
suitably graduated scale. It
may here be mentioned that

the same
principle is also
used the construction of
in
energy meters. In this case

the disc is not controlled by
a spring, but its movement
FIG. 109. -Principles of induction
is retarded by the brake
wattmeter.
magnet. Its speed is a
measure of the power, and
the number of revolutions made a given time, as
in
recorded on a counter, is a measure of the energy passed
in that time.
The action of the instrument may be explained by a
vector diagram. We
suppose
E that the shunt circuit has only
reactance but no resistance, and
that the disc has only resistance,
but no reactance. Both supposi-
tions are only approximately
true. Let in Fig. no E be the
E.M.F. and I the current vector.
E produces a shunt current round
S (Fig. 109), which lags by 90,
and this shunt current produces
the field fa.
The main current
I
produces the field </>,.
Each
produces eddy currents
field
FIG. no. Vector diagram of
induction wattmeter. lagging by 90, the strength of
which is proportional to the flux.
We thus get im as the eddy due to <f> m or I and is as
the eddy due to fa or E. The turning moment exerted
is therefore proportional to
COS tp(l m (f)s -J- ts
<j> m)
THE INDUCTION-WATTMETER 213
and since fluxes and eddies are proportional to I and E,

we have
Torque = cos (^lE +
<p c^ El)
= ÎE cos $
That isto say, the torque, and consequently the angular
deflection, is a true measure of the power. The power
consumed by the wattmeter itself is less than in an
instrument of the dynamometric type, since for equal
torque the loss in the iron is smaller than the power
wasted in the inductionless resistance.
The condition that fa shall be in quadrature with
E cannot be completely realised in the simple arrange-
ment of Fig. 109 because the resistance of the shunt coil
is not quite
negligible as compared with its resistance,
but it is not difficult to eliminate this source of error.
Various arrangements for this purpose have been devised.
In one of these the coil of S is shunted by a dead
resistance, and a small inductance is put in series with
the two parallel circuits so formed.
In the new type of induction wattmeter designed by
Dr. Sumpner the error due to the resistance of the
pressure coil is not only made exceedingly small, but
the instrument is made independent of frequency, wave
form and iron losses. Dr. Sumpner also uses a shunt
magnet, but instead of a disc he employs a moving coil,
which receives current through a resistance from a special
"
type of transformer which he calls a quadrature trans-
former." The latter is a transformer having either no
iron at all, or a core with a large air space, so as to
require a very large magnetising current. The primary
of this transformer is traversed by the main current, and
the secondary supplies current to the moving coil. The
vector diagram of a quadrature transformer may be
obtained from Fig. 99, page 191, by drawing this figure
for a transformer having much leakage (and therefore a
very large magnetising current) and a very light load on
the secondary terminals. This means EO very large
and ED X
and OC^ very small, so that the angle OjD^
becomes very nearly a right angle.
Fig. 1 1 1 shows cliagrammatically a Sumpner watt-
2I 4 TRANSFORMERS
meter with its connections, and Fig. 112 its vector
diagram. Between the poles and armature of an electro-
FIG. in. Sumpner wattmeter.
magnet with laminated iron is suspended the movable

coil, which is controlled by a spring. The electro-
magnet is excited by a shunt

current i taken from the mains,
whilst the movable coil receives
the current i from the quadrature
transformer QT
through a large
resistance R. If the secondary of
the quadrature transformer had
absolutely no inductance, /r would
lag behind I by exactly 90, but
in consequence of the inductance
the lag is slightly greater than 90.
Let this increase be /3. If the
shunt coil had absolutely no resist-
ance the flux produced in the
interpolar space would lag behind
FIG i
E by exactly 9 o, but as there
i2.-vector diagram of
must be some resistance, the actual
wattmeter.
lag is a little smaller than 90.

Let the difference be a. Then if we put 7 = a + ft we
have, as will be seen from Fig. 112, between ^ and the </>
phase angle <p + y,

whilst that between I and E is only (p.
The indication of the wattmeter will therefore be propor-
tional to cos (<p 4- y), whereas the true power is propor-
tional to cos The frequency, iron losses, and the shape
(p.
of the current and E.M.I 7 curves are practically without

,
influence on the indication, and the angles a and ft are so

small that the error clue to them is almost negligible. It
is however,
possible to eliminate this error completely if
the instrument is always used on circuits of the same
frequency. For this purpose we need only make ft = a
so that y = o. This may be done by taking i through
the secondary of another small transformer of ordinary
(not quadrature) type, and inserting its primary into the
main circuit in such sense that an E.M.F. cophasal with I
is
injected into the moving coil circuit. The phase of
z\ may thus be advanced, and ft made negative.
Measurement of power in three-phase' circiiits. If
a three-phase circuit be symmetrically loaded, and the
electrical centre is accessible, the total power can easily
be found by measuring that of one phase by means of
any wattmeter and trebling it. If the loading is un-
symmetrical and the electrical centre is not accessible, two
wattmeters connected in a special way are required, the
sum of their readings being the total power of the three
phases. This method of measuring is due to Behn-
Eschenburg, but as it was first applied by Prof. Aron to
electricity meters, it is also called the Aron system. The
analytical proof originally given by Behn-Eschenburg
is rather complicated, and for this reason I give here
another and simpler proof. In Fig. 113 are shown
diagrammatically the various steps by which we arrive
at the Behn-Eschenburg system of
connecting up the
wattmeters. We from the case that the apparatus
start
whose power is measured is star-coupled, so that
to be
its electrical centre is accessible. Let A, B, C be the
star-voltages of the three phases, and a, b, c the corre-
sponding currents. Then by putting three wattmeters in

circuit connected as shown in I, we can measure the
power of each phase separately. The thin lines represent

connections to the E.M.F. coils of the wattmeters.
We can now alter this arrangement to II, where the
216 TRANSFORMERS
one wattmeter has been removed from c phase and
replaced by two other wattmeters placed in the a and b
IE
12"
FIG. 113. Behn-Eschenburg method of measuring power.
phases. This is permissible, since in a concatenated

system the current in one wire is at any moment equal
MEASUREMENT OF POWER 21;
to the sum of the currents in all the other wires taken

with opposite sign. The sum of the readings of the
four wattmeters in II will still give correctly the total
power. This arrangement may be simplified as shown
in III. Since the main coils of the two wattmeters
in each phase carry the same current, we can combine
the two wattmeters into one instrument having only one
main coil, but two shunt coils, the latter being connected
as in II. The dynamic effect of the two shunt coils,
which have then to be wound upon the same supporting
frame, is at any moment the sum of the dynamic
effects of
each coil if it could be taken singly. Thus
the shunt coil
belonging to a, and which is connected with c, is traversed
by a current flowing under the star-voltage C, whilst the
other shunt coil belonging to a is traversed by a current
flowing under the star-voltage A. Now, since in a con-
catenated system two star-voltages acting on the same
system can always be replaced by the corresponding
mesh-voltage (the latter being the resultant of the two
star-voltages), it is obvious that the duplex coil in a, with
its two shunt currents, can be replaced by a single coil
which carries a current produced by the mesh-voltage

at C. Wethus arrive at the Behn-Eschenburg con-
nections shown in IV. Two ordinary dynamometric
wattmeters are inserted, one with its main coil in each
phase, whilst the shunt coil is connected to its own and
the third-phase wire. The total power is given by the
sum of the two readings. It is, of course, also proportional
to the sum of the two turning moments, and if it were
not for the difficulty of mutual disturbance the two watt-
meters could be combined in one instrument with the two
shunt coils connected mechanically to the same axis. In
such a combined instrument the main coil of a would of
necessity be fairly close to the shunt coil of b, and vice
versd, so that two additional disturbing couples would be
introduced which would make the reading inaccurate.
This difficulty has been overcome in an ingenious
manner by Dr. Franke, 1 who has devised a system of
coupling up two instruments by which mutual disturb-
1
EUktrotechnische Zeilschrift, 1907, page 268.
218 TRANSFORMERS
ance is eliminated. It is therefore possible to combine
two wattmeters by placing them so close together that the
two shunt coils form mechanically one moving system
and the total power can be read off on one dial only.
Fig. 114 shows cliagrammatically this system of con-
necting the two wattmeters. Let A be the current in
the main coil and a the current in the shunt coil of the
dynamometric wattmeter in one line, and B and b the
respective values in the other wattmeter. The shunt
coils are connected on one side to their own line and on
the other through a resistance to an artificial electrical
centre o, which is also connected through a third resist-
-T- -i TT
e, ri ^cv'ffinnnM T'
a
-
I J^f
<
s/v
f
E,
i%/îs
FIG. 114. Siemens' method of measuring power.
ance to the third line. Let the star-voltages be e Lj

then the shunt currents are
and since these currents meet in the artificial centre o we

have
Let /3 be the torque experienced by the shunt coil of

wattmeter I in consequence of the interaction of A and
a and let 7 be the torque due to the disturbing influence
;
of B, so that
fi
= Ci/\a and y c^pâ
the coefficient/ expressing the disturbing influence. For
wattmeter II we have similarly
&= B< and e =
THREE-VOLTMETER METHOD 219
q expressing the disturbing influence of

'
A on the shunt
coil of II.
The total torque indicated on the dial is
Since a= (6 + c) and b (a + c), we have

a A (câ c,q(a + *)) + B(^ - c^p(b + c)}
= -
a = A(a(c! - c.,q] - c$c) + B(%, - erf) - crfc)
The resistances r may be chosen so that
c\
- W= c-i-c\P = ctf = erf =
r\ r>2 rz r^
where k is a constant. We have therv
a= A (e^k
- e. k] +
6 B(^ - e^k)
Now, e^ e^ is nothing else than the mesh voltage

between A and C, and e.2 ez is similarly the mesh voltage
between B and C. Call the former E ac and the latter
E 6c The formula for a may then be written thus
.
Now, we know from the investigation previously carried

out on Behn-Eschenburg's connection that the expression
in brackets is a measure for the total power P, so that
The shown on the dial of Franke's duplex

deflection
wattmeter a true measure of the total power trans-
is
mitted by an unsymmetrically loaded three phase line.

Three-voltmeter method. Profs. Ayrton and Perry
have devised a method of measuring the power transmitted
to a consuming device which can be applied in all cases
where the available pressure is sensibly greater than that
required by the consuming device, and the necessary
reduction in pressure can be made by the interposition
of an induct ionless resistance. In this method it is
220 TRANSFORMERS
necessary to measure three voltages, hence the name.

We measure the total voltage and its two components.
The method will be understood from Fig. 115, where
~ is the generator, a an amperemeter, an induc- W
tionless resistance, and T the transformer absorbing the
power which we wish to measure. A
voltmeter is placed
between the main leads let its reading be e. Another
;
voltmeter is used to show the potential difference e.,

between the terminals of the inductionless resistance, and
a third instrument shows the potential difference between
the terminals of the primary of the transformer. Instead
of using three separate voltmeters, we may of course use
the same instrument for taking all three readings by a
suitable switching arrange-
ment, as shown in Fig. 117.
This arrangement is pre-
ferable on account of its
greater simplicity, and be-
cause slight errors in the
FlG. I i
5 .-Ayrton and Perry three-volt- .
meter method of measuring power. Calibrations of the instru-

ment have less influence on
the result. The clock diagram of the method is shown
in Fig 116. OI is the current, OE^^ is the E.M.F.
impressed on the transformer, and EjE is the E.M.F.
absorbed in the resistance W. Since the latter is induc-
tionless, its vector EjE must be parallel to the current
vector OI. OE=e
is the total E.M.F. The watt-
component of the =
impressed E.M.F. e l is e w OA, and
the energy is OI x OA. Since W
has no inductance, we
have AE1
= BE=J the E.M.F.
,
of self-induction, due to
T, and the following equations obtain
from which we find
The power is
given by the formula
THREE-VOLTMETER METHOD 221
^
To find the power we must take four readings, namely,
three voltmeter readings and one amperemeter reading.
If the resistance is W
accurately known, the last reading
may be omitted and the power calculated according to
the formula
2\V
This is the power actually supplied to the apparatus

T, in our case a transformer. If
we want to know the power
2
supplied by the generator _ :? >
the power lost in the resistance

must of course be added, and we
obtain
Instead of calculating P, we can

find the watt-component of e l
graphically by drawing circles

with radii e and e, and shifting
l
Fl0 *""* 8
a vertical line parallel to itself '
until a position is found in which

"Sl^^*
the piece contained between the two circles is exactly
equal to c z This gives the position of the point E l in
.
Fig. 1 1 6, and therefore also the length of the vector

OA = e w The power is then
.
The diagram shows at a glance that a small error in the

volt-measurements will produce the larger an error in
the determination of the power the nearer the circle
of e l is to O or e, and that the error will be least if e 1
is
midway between O and e. To obtain an accurate
measurement of power by method we must, therefore,
this
so choose the resistance W
that e 2 does not sensibly differ
from e lt that is to say, that about the same pressure is
lost in the resistance as is used in the
apparatus under
test. The total voltage e must then be considerably
222 TRANSFORMERS
greater (from ij to 2 times) than that required by the
apparatus under test.
Another difficulty the necessity of using up in the
is
ballast resistance W
approximately the same power as
in the consuming device itself. The method is therefore
specially applicable in cases where the power to be
measured is small such, for instance, as an open circuit
test of a transformer. The ballast resistance may con-
veniently be a lamp-board, and in order to use the same
voltmeter for all three readings the special switch shown
in Fig. 117 may be used. When testing small trans-
formers the power to be measured (being practically the
FIG. 117. Three-voltmeter method of measuring power.
the power
equivalent of the iron losses) is so small that
a voltmeter can in com-
required by dynamometric
not be considered as negligible. In such a case
parison
the three-voltmeter method is specially useful. To
avoid the necessity of making a correction for the power
taken by the voltmeter, we may use an electrostatic
instrument.
Ais an amperemeter, V
a voltmeter having about
twice the range of the voltage necessary for the trans-
former, and S is the special switch. It is advisable to
make the supply voltage adjustable so that the test may

be made under the condition giving greatest accuracy, as
explained above. Where the supply voltage
is no
higher
THREE-AMPEREMETER METHOD 223
than that required by the consuming device the three-

voltmeter method is not applicable, but then we may
use the
Three - amperemeter method devised by Professor
Fleming. Current at fixed pressure is supplied at the
terminals K, Fig. 118, and is taken through an ampere-
meter a, at the other side of which it is "divided into two
circuits, one containing the transformer T to be tested,
and the other an inductionless resistance W. These
two currents are measured on the amperemeters a x and
a.2 the pressure is measured on the voltmeter e.
;
The
clock diagram of this combination is shown in Fig. 119,
where OE
represents the pressure of the supply current, i
the primary current of T, and iw its power component ;
i2 is the current flowing through the resistance and W ;
^^ I\WVWWV
r
__r__/WVVWW\A
,m
FT i'2
FIG. 118. Fleming's three-amperemeter FIG. 119. Vector diagram
method of measuring'power. to Fig. 118.
its vector must of course be parallel with OE. From

the diagram it will be seen that the following relation
obtains
The power is
given by
If the resistance is W
accurately known, the reading for
e need not be taken, and the power may be calculated
from
Also in this method accuracy

depends upon the proper
choice of the resistance. It should be so
adjusted that
/2 is not sensibly different from /, the total current i will
;
224 TRANSFORMERS
then be from i^ to 2 times the primary current zi taken
by the transformer. Fig. 1 20 shows an arrangement of
switches and connections whereby the same amperemeter
is used for all three
readings. I is a
single-lever and
1 1 a double-lever switch.
In considering both methods, we have tacitly assumed

that current and pressure follow a sine law the question ;
now arises, whether these methods will give accurate re-

sults if this condition is not fulfilled, that is to say, if the
curves representing E.M.F. and current are of irregular
shape. That the wattmeter gives correct indications
also in such cases has already been shown, and since
simultaneous measurements by means of a wattmeter
FIG. 1 20. Three-amperemeter method of measuring- power.
and one or the other methods here described are always

in accord, we naturally conclude that these methods
must also be generally applicable. Apart from such
experimental proof, this can also be shown by theory.
For this purpose we shall consider the three-voltmeter
method, the application to the analogous case of the
three-amperemeter method will then be self-evident.
Let in the following the letters e and i denote the
instantaneous values of E.M.F. and current respectively,
then the expression
is valid at any time in the cycle. We also have at al!

times
THREE-AMPEREMETER METHOD 225
and the power at any moment is
f)
= i^ l
__f j
or e
72 ,,2 i ^ ^, x, i ^2
The work done in the time T of a complete r

cycle \sf' pdt,
and the effective power is
1 2
~W o o o
It has been previously shown that the expression

= r r
e^dt is simply the square of the effective pressure
indicated by the voltmeter ;

if now we denote these
effective pressures by e, e lt e 2 respectively, we have
Since in arriving at this result (which is exactly the same

as that reached by the graphic method), we have made
no assumption whatever as regards the shape of the
E.M.F. curve, it follows that the three-voltmeter method
is
applicable to currents of any form.
CHAPTER XI
TES TING TRA NSFORMER STES TING SHEE T-
IRON SPECIAL IMPLEMENTS BY DOLIVO
DOBROWOLSKY, KAPP, EPSTEIN, RICHTER,
E WING THE BALLIS TIC METHOD THE
FL UXOMETER SCOTT S METHOD KAPPS
METHOD
--
Testing transformers. By means of the various
methods above explained the output and efficiency of
transformers can be determined. It is of course neces-
sary to have a source of current capable of supplying all

the power wanted, and a load capable to absorb the full
output of the transformer. To obtain by this direct
method anything like a reliable figure for the efficiency,
input and output must be measured with extreme accuracy,
the reason being, that the two are not very different, and
a small error in the determination of one or the other
causes a great error in their calculated ratio. Let for
instance the real input be 100 and the real output 97
kw., and let there be an error of i per cent, in each
measurement, the error being negative in the measure-
ment of the input and positive in the measurement of the
output. The measurements would then be 99 and 98
kw. respectively, and the calculated efficiency would be
99 per cent, instead of 97 per cent., which it really is.
To reduce as much as possible the magnitude of the
error in the determination of the efficiency, it is advisable
to make this determination by an indirect method in the
following way. The test is made simultaneously on two

equal transformers, which are so connected that the out-
put of No. i forms the input of No. 2, and the output of
this, supplemented by an external source of power, the
input of No. i. We obtain thus a circulation of power

through the two transformers, and need only supply as
226
TESTING TRANSFORMERS 227
much power as is wasted in both. This is a small

amount, and need only be measured with a moderate
degree of accuracy. The power circulating is also
measured, and it will be obvious that small or moderate
errors in both measurements cannot seriously affect the
accuracy of the result. The arrangement of apparatus
is shown in Fig. 121. D and B are the two equal trans-
formers, and C is a small auxiliary transformer which
supplies the waste power and thus keeps the total power
in circulation. Into the primary of C we insert an in-
ductionless rheostat R, for the purpose of adjusting the
pressure supplied to C, so as to obtain in the ampere-
WWWW\A/W
FIG. 121. Testing transformers.
meter A the normal full load current of the big trans-

formers. The connections between the latter must,
of course, be so arranged that their E.M.Fs. oppose
each other. If the large transformers were only con-
nected to C, the full current could be obtained in
them, but not the pressure. To ensure that also the
right pressure is maintained in B and D, we connect
their primaries, shown in Fig. 121 as thick wire coils,
with the generator. The connections are taken through
the wattmeter W\ and through the electrical centre of the
auxiliary transformer. The object of the latter arrange-
ment is to ensure that the voltage on the primary of one
transformer shall be raised by the same amount as that
of the other is depressed, so that the induction in both
228 TRANSFORMERS
shall be as nearly alike the normal value as possible. If
the centre of the auxiliary transformer is not accessible,
the connection may be made on one of its terminals, and
then there will be some inequality in the working condi-
tion of the two transformers, but the error thereby intro-
duced is not very serious, since the difference in primary
voltage is comparatively small. The wattmeter 1 is W
introduced to measure iron losses. The copper losses
are measured on the wattmeter 2 W
If we short-circuit
.
the rheostat of C, then the generator has to supply only

the no-load losses of B and D, which will be indicated
on the wattmeter lt W
Since both transformers are equal,
no current will be indicated in A. Now let us insert C
and adjust the rheostat until A
indicates the full-load
current. Then the large transformers are both working
under full load, and the wattmeters 1 and W
2 measure W
all losses.
The voltage on each primary is measured by the volt-
meter V, which is provided with a change-over switch s.
Let e be the average of these two readings and i the
current indicated on A, then the combined power of both
transformers is iei and the total loss is the sum of the
two wattmeter readings
W = Wj + W 2
The efficiency of each transformer is therefore
ei
- W
2
To get the truecopper losses switch S should be

opened. If remains
it closed whilst the contact on the
rheostat R is put down to the lowest contact so as to
short-circuit C, there will still be a small current (namely
the magnetising current of transformer D) flowing through
the wattmeter 2 W
If now the contact is raised so as to
.
produce a main current, this small current will, according

to the connection, either increase or diminish the current
flowing through the wattmeter, and to this extent 2 will W
indicate either a little more or a little less than the true
TESTING TRANSFORMERS 229
copper loss. The error may be avoided by repeating

the test with W
2 inserted in the primary
of B and taking
the mean of the two readings. In practical work for
measuring efficiency this correction need, however, not
be made, since the error is very small whilst for measur-
;
ing copper losses only, the simple expedient of opening

switch S is sufficient to avoid the error.
The test illustrated in Fig. 121 can also be used to
determine the drop by opening S and moving the contact
of R to such a position that a predetermined (preferably
the normal working) current flows through the primaries.
The voltage must then be read on a second voltmeter
(not shown in the diagram), which is connected to the
primary terminals. Let e Q be this voltage, i the current,
and w the reading on 2 W
then the equivalent resistance
,
r of one transformer is
Its ohmic drop is
and its inductive drop is
(O U= /5L-
)U^)being the reactance of the equivalent coil.

The advantage of this method of testing is not only
great accuracy, but also economy of power. The latter
point is of importance when testing for temperature rise,
since the final temperature is only reached after many
hours, and in the case of large transformers some days
of working at full power. The cost of power and the
difficulty of using it up in an artificial load become thus
serious obstacles, so that a test which only requires the
supply of the power wasted, as that shown in Fig. 121,
is also commercially advantageous.
It is, however, not always
possible to test two equal
transformers together for temperature rise. In this case
the transformer may be preliminarily heated in a drying
room (most electrical engineering works are provided
230 TRANSFORMERS
with such a room), and then put to work under normal
load, whilst thermometric or resistance readings are taken
from time to time to find out when the final temperature
rise has been reached. Or the transformer may be
worked alternately on open circuit to heat the iron, and
have continuous currents passed through both coils to
heat the copper. This preliminary period of heating
may be shortened by working at increased voltage and
current density. When the probable final temperature
has been reached the transformer is put to work normally,
and kept at work until the final temperature has been
actually reached.
Another method is to heat the iron by alternating
current sent through one winding, and the other winding
at the same time by continuous current. This winding
must, of course, be opened in the middle, and the two
halves must be coupled up in opposition so that no alter-
nating E.M.F. is produced at the two free ends. By
keeping a record of the continuous current and E.M.F.
supplied to this winding the rise of its ohmic resistance,
and therefore the rise of its temperature, may be graphic-
ally represented as a function of the time. From this
curve the time constant for heating may be found, and
from that the final temperature rise and the time in which
it would be reached may be computed.
The insulation of a transformer should be tested when
hot. It is also advisable to flash the transformer, so that
any weak spot in the insulation may be found out and

remedied before the apparatus is set to work. For this
purpose, temporary connection should be made between
(a) a primary and secondary terminal (b) a primary
;
terminal and carcase (c) a secondary terminal and car-

;
case. Care must of course be taken that during these

tests both poles of the generator are well insulated from
earth, or the carcase must be insulated from earth.
Testing sheet-iron. An obvious way of testing any
particular batch of plates intended to be used in the
manufacture of transformers, is to select at random some
of the plates sufficient for the carcase of a small trans-
former (preferably a stock size), wind it in the usual way
and test for iron losses. If the test of this sample is
SPECIAL IMPLEMENTS 231
satisfactory the whole batch can' be passed as suitable.

The drawback to this method is that the building up of a
complete transformer takes too much time. What is
required is a method of testing samples which does not
involve the winding of coils, and where the test pieces
are of a simple form, so that but little time is required in
the preparation of samples and not too much material is
wasted.
Special implements. One of the oldest instruments
is the iron tester of Dolivo Dobrowolsky, shown in Fig.
It has now only historic interest.
1
122. It consists of
two |shaped
|
cores of sheet-iron, which can be laid
together either directly or placed on
either side of the sample to be AA
tested. The sample is composed of
rectangular sheets and forms the
common yoke to the electro-magnets
n, s. When the magnets are placed
directly in contact, the direction of
the current through the coils is such
that both drive the induction in the
same sense when the sample is
;
inserted, the connections are charged

by means of the switch B, in such
manner as to produce the polarity
indicated in the diagram. The flux
FlG. 122.
now passes from both magnets iron tester.
Dobrowolsky
through the yoke. The current is

measured by a dynamometer marked EL Dyn. in the
figure, and the pressure by a Cardew voltmeter marked
Card. The power is measured by a wattmeter inserted
as shown. In using ,the apparatus the magnets are
laid together and the switch is put into the position
which produces circular magnetisation. The power corre-
sponding to various values of the induction is then
measured, the induction being calculated from the fre-
quency, the pressure and the known data of the coils and
magnet cores. The sample is then inserted, the switch
changed over and the measurements repeated. The
1
First published in 1892 in the Elektrotechnische Zeitschrift, from
which Fig. 122 is copied.
232 TRANSFORMERS
sectional area of the sample should be about double that
of the magnets. The difference between the two sets of
measurements is then the power wasted in the sample at
the various values of the induction. A drawback of this
method is the difference in magnetic leakage with and
without the sample. If the magnets are laid together
directly, and magnetised circularly, there is hardly any

leakage, and B can be calculated from E with great
accuracy. If the sample is inserted, the magnetic resist-
ance is increased, and leakage produced which diminishes
the value of B in the sample. At the same time there is
a difference the value of the induction along the
in
magnet cores, the induction being a maximum in the

centre of each core. E can therefore no longer be
regarded as an exact
measure for B, and an
error is thus introduced.
^
To avoid this diffi-
| culty, the author has con-
structed the apparatus
shown in Fig. 123. The
i sample consists in this
apparatus also of a batch

FIG. 123. Kapp iron tester. of rectangular plates, and
forms one of the two
longer sides of a rectangular frame, the three other
sides being formed by shaped plates of known
\ \
magnetic quality. Both longer sides are surrounded

by coils, the upper one being large enough to admit
the insertion of the sample without difficulty. The
connection is made for circular magnetisation, so that
only very little
leakage place, and this is the
takes
same for all samples. The sample must have approxi-
mately the same cross-section as the magnet. To
calibrate the instrument, a sample is
prepared from
the same iron as the magnet, and after
weighing the
total amount of iron in the the loss of power is
magnet,
determined for different values of B. This loss is then
allotted between magnet and
sample according to their
relative weights, and a curve is plotted
showing the loss
in the magnet as a function of B. If now another
sample
is inserted, total loss measured, we have only

and the
to deduct from the loss as found from the curve for
it
the particular induction observed, and the rest is the loss

in the sample.
The objection to this method of measuring the loss
in the sample is that the loss is obtained as the difference
between two measurements, both of which are larger
than the result desired, A
slight error in the measure-
ment of the total loss
may therefore mean

a large error in the
result. This draw-
back has been over-
come in the apparatus
shown in Fig. 124,
which has been de-
"
signed by the Hys-
teresis Committee" of
the German Associa-
tion of Electrical
Engineers, and offici-
ally accepted by this
Association in 1902,
after having been on
trial in various works
for some years. The
instrument also
is
"
known as the Ep-
stein Iron Tester,"
Prof. Epstein having
FIG. 124. Epstein iron tester.
been chairman of the
Committee. In this
method of testing no foreign iron is used, the whole of
the magnetic circuit being made up of sample plates in
the form of a square. Each side of the square is a bar
made up of strips with tissue-paper insertion. The bars
are 50 cm. long, and have a cross-section of 30 mm. by
about 25 mm. Each bar contains 2*5 kg. of plates, so
that for testing each batch a little over 10 kg. of plates
(allowing for waste) have to be cut up. Each side of
the square is surrounded by a magnetising coil, the four
234 TRANSFORMERS
coils being fixtures of the apparatus and wide enough to
admit the samples. The samples abut at the corners,
where they are pressed together by screws, a thin sheet
of fibre being placed in the butt-joint to avoid loss of
power by eddies. Fig. 125 shows a diagram of con-
nections. The terminals k, k of the apparatus are con-
nected to the supply terminals K, K, from which the
magnetising current is taken. To find the "figure of
"
loss the frequency to be used is 50 and the induction
10,000, which corresponds to about 85 volt. Each coil
has 1 50 turns, and the total resistance of the four coils
is 0*18 ohm. Since the magnetising current is only a
few ampere, the correction for copper loss is very small
in comparison with the figure of loss, which for alloyed
FIG. 125. Epstein iron tester.
iron is of the order of

magnitude of 2 watt per kg. or
20 watt for the whole sample. It is advisable to use
a frequency indicator (not shown in the diagram) when
an accurate test is required. The magnetising current
is
adjusted by the rheostat, and the power is measured
on the wattmeter W. The net area of the bars A is
calculated from the weight and density (about 777), and
their total length (2 m.) and the induction from the
formula
e being the E.M.F. induced in the four coils. This is

very nearly also the E.M.F. indicated on the voltmeter
V, but for very accurate work e may be found by correct-
ing the voltmeter reading for copper loss, the correction
being of course made vectorially. It is important to
open the voltmeter switch s when reading the wattmeter,

since the power taken by the voltmeter would otherwise
be counted as part of the iron loss. When measuring
the current it is advisable to open the switch s of the
pressure coil of the wattmeter so that only the true
magnetising current may pass through A. The machine
used as a source of E.M.F. should give as nearly as
possible a sine wave of E.M.F. The curve shown in
Fig. 1 1 gives total loss with alloyed plates of English
manufacture obtained by the author 'with this implement.
The power measured on W
represents the combined
hysteretic and eddy current loss occurring in the sample,
and as the weight of the sample is known the loss per
unit weight can easily be found. We
thus get the
quantity, which of immediate interest to the designer,
is
but it is also possible to get hysteretic and eddy current

losses separately. From equations (8) and (So) it will be
seen that the total loss for any given sample of iron is
given by an expression of the form
P = /*vB*+/(vB) 2
(38)
where h and /"are constants depending on the quantity

and quality of iron under test and the thickness of the
plates. of B is usually taken as 1*6, but
The exponent
to keep the investigation general we call it x. B is
found from the determination of e, as above explained,
and v is read on the frequency meter. have thus We
three unknown quantities, namely, h, and f
x, and by
making three tests under different conditions we can
obtain three equations from which the three unknown
quantities are determined. The operation can be a little
simplified if we make all the tests for the same induction,
for which the condition is
e
- = constant
v
1
We need only vary the speed of the generator and
i i
keep excitation as nearly constant as is required by

its
this condition. These two tests suffice. In these B

and therefore B^ and B 2 will remain constant, and the
236 TRANSFORMERS
equation for the lost power in the cases where the

frequency is v l and v 2 may be written in the simple
form
from which H and F may easily be found. We have

then for the fixed induction B
Hysteretic loss at frequency vx . . . .
v2 . . . .
Eddy current loss at frequency v .
Toget complete curves of losses the tests have to be

repeated with different values of B. To any two values
of B, say B and B', and the same frequency v correspond
two hysteretic losses, P^ and P/, so that
P AB
from which x may be found by taking logarithms
Having x, we find h from
and the coefficient^ may be found from
/ -Bl
The Epstein iron tester thus be used not only

may
to find the total loss for any values of B and v, but also
to separate hysteretic and eddy current losses and deter-
mine the coefficients in their formulae.
Mr. Richter has devised an implement for the testing
of complete sheets as they are produced by the rolling-

mill. His object is to avoid the labour and waste of
material when cutting up sheets into sample strips. The
magnetising coils are long' and narrow rectangles, held
in a wooden frame with their long sides parallel to each
other and arranged circularly around the axis of the
frame. The coils are placed evenly round the axis, with
sufficient space between to allow the sheets to be slid
through their openings so as to form a closed cylinder,
FIG. 126. E wing's iron tester.
which is circularly magnetised. From the weight and

dimensions of the sheets the cross-sectional area of the
magnetic circuit can be calculated, the test being made
by voltmeter and amperemeter, as in the Epstein
apparatus.
An implement in which samples of only a few ounces'
weight can be tested has been devised by Prof. Ewing.
Its principle is the
purely mechanical determination of
the hysteretic loss alone in a sample of very small
dimensions, namely 6 to 8 strips of 3 in. length and
238 TRANSFORMERS
-|
in. width. The
apparatus consists of a permanent
magnet e, Fig. 126, which is
suspended on knife-edges, f t
and weighted by a screw, g. For transport the magnet

can be raised off the knife-edges by means of a rack and
wheel, h. A dashpot below the magnet serves to steady
its swing, and a pointer moving over a scale at the top
shows the deflection produced when the sample a is

rotated between the poles. The sample is fastened by
screw-clamps b, b to a carrier, which can be rotated by
means of a handle, and the friction wheels d, c. The
screw i serves to level the instrument. The reversal of
magnetism in the sample is produced by the rotation
of the sample, and the work lost in hysteresis and eddy
currents per revolution is 2 TT x torque. The torque is
indicated on the scale by the pointer, and since 2?r is a
constant, we find that the deflection of the pointer gives
directly a measure for the loss per cycle, the speed of
rotation having no influence as long as it is not so high
as to sensibly augment eddy current losses.
The sample sheets are prepared to a gauge, the
length being sensibly less than the polar gap of the
magnet, so that the magnetic resistance of the air gap
preponderates over that of the sample itself. The object
of this arrangement is to avoid the error which might
otherwise be introduced when samples of widely different
permeability are tested. The magnet produces in the
sample an induction of about 4000 C.G.S. units, but this
can be slightly raised or lowered by taking less or more
sample plates. Prof. Ewing found that an accurate
adjustment as regards the weight of samples is not
required, since the deflection varies but slightly if the
number of plates making up a sample batch is varied. It
suffices to adjust the weight of the batch roughly to that
which corresponds to seven strips of 0*37 mm. thickness.
When testing armature plates, which are usually stouter,
a correspondingly smaller number of strips would be used
to make up the sample batch.
The apparatus is calibrated by using samples, the
hysteresis of which has previously been accurately
determined by the ballistic method. Two such standard
samples are supplied with the apparatus, together with
THE BALLISTIC METHOD 239
tables giving the of ballistic tests.

results In testing
other samples, a reading also taken
is with one of the
standards, and the ratio of the readings is taken as the
ratio of hysteretic losses between standard and sample.
By this method of testing, the accuracy of the instrument
is rendered independent of any possible change that
may have occurred in the strength of the permanent
magnet.
The ballistic method. The methods of testing iron
above described suffice for the immediate requirements
of the designer, but special circumstances may arise
when is desirable to know not only the power lost in
it
hysteresis and eddies, but also the shape of the hysteretic

loop. This cannot be found by any of the methods
hitherto described, and to get a complete knowledge of
the magnetic qualities of any brand of iron, some method
must be used which gives the relation between exciting
force and induction throughout a complete magnetic cycle.
Such an investigation is also necessary for the calibration
of certain workshop implements, such as the Ewing iron
tester.
To find the B-H curve we may use a ballistic
galvanometer, and make use~of the well-known physical

law that the deflection of the moving system of such
an instrument is proportional to the total quantity of
electricity which has been suddenly discharged through
it. The moving system may be a little magnet or a coil
as first used by Deprey D' Arson val. In the first case
the system is only slightly, in the second more effectively,
damped a certain amount of damping is unavoidable,
;
and, indeed, necessary for rapid working.

The elongation of the spot of light of a damped
galvanometer is given by the well-known formula
where a and b are constants, and VQ is the initial

/,
velocity with which the spot of light leaves its position

of rest. Counting the time / from x = o, then the interval
of time between two successive passages through zero
positions in the same direction is given by the condition
240 TRANSFORMERS
that the sine must be zero for both. We find thus the
periodic time
27r
T_ :
According to the theory of damped vacillations we have
where the controlling force,

c is the mass, and 8 a m
coefficient which, multiplied by the velocity, gives the
damping force, all values being referred to the spot of
light. In a perfectly undamped galvanometer 8 = o and
b = ,
so that its periodic time is
'Y
= Im
27TA/
1
the well-known pendulum equation. It is obvious that

T>T that is to say, that damping lengthens the periodic
,
time. As a further result it will be seen that damping

reduces the first and all subsequent elongations.
Before entering into the question how the effect of
damping can be allowed, for let us assume that it were
possible to make a perfectly undamped instrument, and
consider how such an instrument could be used for testing
iron. The correction for damping can then be considered
later.
The
first
(and indeed every subsequent) elongation
X Qthen proportional to the initial velocity z/ and this
is ,
again is proportional to the quantity discharged through

the galvanometer from a coil through which the flux <
is reversed or annulled. The first elongation is thus a

measure for the flux passing through the sample which
is surrounded by the coil. If the area of cross-section A
be known, and the magnetising force H be measured, we
can obtain the relation between B = ^ and H. Instead
A
of annulling
t> O the flux
or reversing we can also change
O it
suddenly, but by small increments (by changing H), and
thus get step by step the relation between B and H, that
is, the hysteresis loop. It is, of course, necessary to
THE BALLISTIC METHOD 24!
determine once for the ratio between B and .% or in

all
other words the galvanometer, and this
to calibrate
may be done in a variety of ways, which we shall now
consider.
Take a straight solenoid whose length / is at least
twenty diameters, and which has n^ turns, and place into
the centre a small co-axial pilot coil of area A and n turns,
then by sending l l ampere continuous current through
the solenoid there will be created within the pilot coil a
flux <
= AH, where
H
The same holds good for a ring-shaped coil when /
is the mean circumference. The turns of the pilot coils
may be distributed all round the ring or placed in one
part only, but they should be below those of the magnet-
ising coil. The total flux AH
which passes through the
n turns of the pilot coil is therefore known. If then we
observe the elongation, if the magnetising current is inter-
rupted (or reversed, which gives it twice as great), we can
determine the ratio between elongation and linkage flux.
As will be shown below, this ratio is constant for a given
resistance r in the galvanometer circuit, and we thus get
the equation
"
where b is the If we include in this
ballistic constant."
circuit not only the pilot coil of the standard solenoid,
but also that of the sample to be tested, the ballistic
constant need not be determined. All we need do is to
determine alternately the deflection obtained with the
two pilot coils and calculate from this the linkage flux of
the sample. This method of comparative observation
can be conveniently carried out by using the arrange-
ment shown in Fig 1
27.
S is thestandard, A x an amperemeter to measure its
magnetising current ly a reversing switch, Rj a

l \J 1
rheostat, and B a battery. The sample is prepared in
form of a ring wound with two coils, one the magnetising
16
242 TRANSFORMERS
coil receiving current from the same battery through a
rheostat R 2 and reversing switch 2 and the other a U ,
pilot coil in series with the galvanometer G, and the

pilot coil of the standard S. A
resistance, r, is inserted
to reduce the deflection of the galvanometer to a con-
venient amount, and s is a damping key by which after
vWvWWv I
/vWA/WV
FIG. 127. Ballistic test.
each reading the moving coil can be quickly brought to

rest.
The E.M.F. produced by a change of flux in the
standard is -^- 5f microvolt, and the corresponding

100 dt
current is . ^>JL *X .
microampere. The quantity

r 100 dt
discharged through the galvanometer is, on reversal
of Ij
2n
,
=
XQ ioor
We thus calibrate the galvanometer for micro-

may
coulomb or any other convenient unit of quantity. But
this calibration is not necessary to use a

if we wish
merely -comparative method, as may be seen from the
following. Let the various quantities in the sample be
denoted by the same letters as in the standard, but
distinguished by a dash, thus
loor
<' = ^
f
If the flux linkages n$ and n <f> are not very widely
different, so that the deflections are of the same order of
magnitude, this method of testing is convenient, as the
constant of the instrument need not be known but if the ;
flux linkage of the sample is either very much greater

or very much smaller than that of the standard, it
becomes necessary to adjust r so as to get convenient
deflections, and then the simple proportionality between
quantity and deflection is lost. It is no longer admissible
to use a comparative method, and it becomes necessary
to determine the ballistic constant. One way of doing
this has already been shown. We
found from a test on
the standard for microcoulomb
(39)
ioor
Another and very obvious method is to determine b
by
discharging a condenser through the galvanometer and
observing the deflection. The arrangement is shown in
Fig. 125, where C is a standard condenser, Cl a Clark,
or other standard cell (*= 1*4323 5 volt at 15 C. for
the Clark, or 1-0196 volt at 15 C. for the cadmium cell)
and S a two-way key. The quantity discharged is then
with a condenser of C microfarad Ce microcoulomb, and
if a deflection of x scale dimensions is
produced by this
discharge we have
b = (40)
244 TRANSFORMERS
A third method of finding b is as follows. Let k be
the constant of the galvanometer for steady currents, so
that
i = yx microampere
The deflecting force is proportional to the current and

also proportional to the deflection or
ex = at and a =
r
The acceleration on starting is produced by the force at,
and we have therefore
dv
00
y
/ aidt
-
= mv
aQ = mv or abx = mv
From v'
2
= C-3? we find v=x*\ ,
and this inserted
171
gives
,
ab = Jem
, ,
or - b
c
= Jem , b
or -
./m
=V
7 7 c
but for a completely undamped instrument the periodic
time is 2irV
m = T,
i
so that we can also write
27T X 27T
By sending a known fraction of a known current through

the galvanometer and observing the steady deflections
we determine 7, and by taking the current off and timing
the oscillations we determine T. From these two ob-
servations b can be found.
It should be noted that b depends on the damping
force, which may be considered constant, but not on the

resistance r, which may have to be raised between wide
limits so as to get convenient deflections at all values of

n(f>.
Wehave the general formula
thus to express a
sudden change in flux through the sample
lOOo:
n
In this formula x
is the deflection which would have
been observed the galvanometer had been absolutely

if
undamped. In reality the deflection is (because of damp-

ing) a little smaller, say x& have from the theory We
of harmonic motions for an undamped oscillation
and for a damped oscillation

~ an
7t
*>*>'&*
since the time t of a quarter period is . The ratio be-
air
n IT
tween the two deflections is e~ . Write 8 instead of T ,
20
then
XQ = xe~ ft
or x = xtfP
Let x be the next elongation in the same direction, x2
the second next, and so on, then subsequent swings take
place in intervals of
T =
27r
and the exponent of e in the equation for the deflections

x& x x.2lt
xn becomes
. . .
for x .......
.;> ....... y8
-j8-4)8
....... -p-sp
246 TRANSFORMERS
From this follows
=
4^ *
.... (42 )
Since ft is a very small number we can in the series
neglect the third and subsequent members and write
* = *-o(i+) (43)
x$ is the first elongation actually observed with a
moderately damped galvanometer, x is that elongation
which would have been observed if the galvanometer had
been absolutely undamped. It is this value x and not XQ
which has to be used in the determination of the ballistic
constant and the calculation of the change of flux from
b i vorx
n
The number /3 is called the logarithmic decrement ; a
convenient value of it is 2 or 3
per cent.
The fluxometer. The use of a ballistic galvanometer
presupposes the ability to change the flux very suddenly,
for the whole discharge from the pilot coil must be
completed before the moving system has appreciably

changed its position of rest. The condition of a very
abrupt change of flux is not difficult to fulfil if dealing
with a sample of moderate size, but if we attempt to take
the hysteresis loop of the carcase of a large transformer in
this way we find that unless an enormous resistance is
put into the magnetising circuit the change of flux does

not take place rapidly enough for the galvanometer, and
it is
preferable to use a method of investigation which is
independent of the time rate at which the flux changes.
Such methods have been devised by Mr. C. F. Scott, the
Author and Mr. Grassot, the latter using a special
instrument, termed by him a fliixometer. I take this
first, as being more akin to the ballistic galvanometer.
THE FLUXOMETER 24;
A delicately pivoted coil not subjected to any con-

trolling force swings in the strong field of a permanent
magnet, Fig. 128, the arrangement being similar to that
used in D'Arsonval instruments, but without a controlling
spring. The terminals, T, of this coil are connected to
the coil encircling the flux which is to be measured, say
the low-tension coil (or part of the low-tension coil) of a
transformer. The fluxometer coil is provided with a
pointer on one side and a mirror on the other, so that its
angular displacement may be observed either directly or
by a beam of light. The coil is set mechanically into its
zero position when its plane is
parallel to the polar axis
FIG. 128. Principle of fluxometer.
of the permanent magnet and no flux threads it. If a

current is sent through it a torque is exerted and the
coil takes an angular position, the flux now threading it
being proportional to the angle of deflection.

The principle underlying the action of the instrument
is the physical law that any electric circuit has a tendency
to maintain its total linkage flux. If this quantity is
forcibly diminished inone part of the circuit another part

Thus, if a pilot coil has been placed
will try to restore it.
over the middle of a bar magnet and is then stripped off
the linkage flux through the pilot coil is reduced to zero.
The fluxometer coil will then set itself at such an angle
as to thread the same linkage flux as that which has
vanished in the external part of the circuit.
Let M be the mass of the moving system,
c\ be the force exerted by the current through
the fluxometer coil,
248 TRANSFORMERS
Let Dz> be the damping force at speed v,
sv be the E.M.F. in the fluxometer coil generated
when moving at the speed v through the field
of the permanent magnet ;
the quantities M, c, D, and v referring to a point on the

indicating needle i cm. distant from the axis of rotation.
Let, further, L be the inductance and r the resistance
of the fluxometer coil, then an E.M.F. E applied to its
terminals will produce a current I and a displacement.
We have
at
E - ev - TL a
T? ^
di
dt
*''-
dt
w- L
dt
dv Ê csv ^L dfl
di~~ r r r dt
r \r ) r
Integrating this equation and remembering that both

I and r are zero at the beginning and at the end of the
process, we get
r /"x* //~ \ x* 00
r /
-- E* - (- + D )/\r J
vdt
Now the integral of vdt is simply the excursion of the

pointer 0. The damping coefficient is so small that D
it
may be neglected, and we thus get
/*
= J Edt
o
E is produced by the
Since change of the flux
through the pilot coil of n turns, we also have
THE FLUXOMETER 249
=n
di
e
Yl~~Y^
= ~9 '
/
(44)
\
The change of flux is proportional to the deflection

9, and the latter is independent of the rapidity with which
the change takes place. We have defined s as a
coefficient which, multiplied by the speed of the point
to which all quantities refer, gives the E.M.F. induced in
the fluxometer coil. Since we have assumed this point
to be cm. from the centre of rotation, 9 is not only a
i
length but also an angle, and v is not only a linear

speed but also an angular speed. The dimensions of
vz are therefore those of an E.M.F. L f M*T~
2
whilst ,
those of v are T" 1

,
giving for s
L M*T-
f 1
the dimensions of a magnetic flux this is in accordance ;
with the above formula (44), for 9 and n are simply numbers
having no dimensions. The factor e is therefore a
constant for each instrument, and is equal to the
product of the flux produced by the permanent magnet
with twice the number of turns in the fluxometer coil
divided by the angular length of the arc spanned by
each pole-piece. The calibration of the instrument is
done empirically, and in one specimen in the Author's
possession 5=12300 when 9 is reckoned not in radians
but in scale divisions. For this particular instrument
one scale division represents therefore a flux of
lines of force. The smallest number of turns we can

have exploring or pilot coil is n= i, so that the
in the
total range (the scale has on either side 100 divisions)
is a little over a
megaline. The method of using the
instrument for taking the hysteresis loop of the iron of
250 TRANSFORMERS
a transformer is shown diagrammatically in Fig. 129.
Two terminals of the transformer T
are joined to the
supply terminals, K, K, of a source of continuous current
through a reversing switch A, amperemeter A, and
regulating resistance R. P is the
exploring coil, con-
sisting of a simple loop, and this is connected with the
fluxometer F. If the transformer be very small, it may
be possible to use one of the windings as an exploring
coil. To take the hysteretic loop proceed as follows.
Set R to zero E.M.F. and put the fluxometer to zero
mechanically. Then shift R
so that a current I is
indicated on A and observe the deflection of the fluxo-
FIG. 129. Testing transformer by fluxometer.
meter. Put fluxometer to zero again and then take another

step on R, observing again the new value of the current
and the new deflection on the fluxometer. Proceeding
thus step by step we get the positive rising branch of
the loop, the descending branch is found by bringing R
back in steps to zero. Then S is reversed and the whole
process repeated for the negative part of the loop.
When the loop is
plotted take the area f\d$ by plani-
meter. The energy wasted in one cycle is obviously
T/jicryiflty, where n^ is the number of turns of the
magnetising coil. If there be butt-joints the hysteretic
loop will be sheared over, but its area will not be altered,
since air has no hysteresis. Let A
be the area of cross-
THE FLUXOMETER 251
section of carcase, and K its

weight in kg., then the
hysteretic loss at frequency v is for the whole carcase in
watt-
area of loop x IGT^/Z! = aior 8
vn 1
and the loss in watt per kg. is
for the induction
To calculate the loss per kg. is, of course, only possible

if the cross-section is con-
stant throughout the mag-
netic circuit.
The value of is taken<
from the loop as a check

;
it
may also be taken directly
by reversing S and observ-
ing the fluxometer. The
deflection will then be twice FlG -
130. Varying range of fluxometer.
that corresponding to (, and
the direct method is with this particular instrument there-
fore only applicable for values of <j> under o'6 megaline.
It is, however, possible to increase the
range of the
instrument so that it may be used for measuring the
flux passing through any, even the largest transformer.
Let in Fig. 130 P be the pilot coil of one turn
encircling the flux <. Join its leads to an inductionless
resistance Rj, and from a small fraction of this resistance
take leads to the fluxometer whose resistance is r. In
the instrument mentioned above r is a little under 20
ohm. If we make R about o'l ohm then only ^ per
cent, of the current going through the coil will be shunted
through the fluxometer, so that practically the same
current will flow through the whole of the resistance R 1?
and the E.M.F. impressed on the fluxometer will be to
the total E.M.F. generated in P as R x The resist- R : .
ance of the fluxometer need not be accurately known.

All that we require to know accurately is the ratio
252 TRANSFORMERS
of R
Rj. Let Rj = ;;zR, then on reversing the magnet-
:
ising current and thus producing a change in the linkage

of P amounting to 2 (/> we have
...... (45)
Since only the ratio of to R

x
R
and not their absolute
,
values, are of importance, temperature does not affect the

accuracy of the method, and copper may be used for the
resistance coil, which should be wound bifilarly. Making
m = 2O the range of the fluxometer can be increased to
about 12 megalines.
Scotfs method. Mr. C. F. Scott l has devised a very
ingenious method for plotting the curve connecting excit-
ing current and flux in any magnetic circuit by making
use of the law that constant E.M.F. in the pilot coil means
proportionality between flux and time. In its simplest
form the test is carried out as follows Through the fine
:
wire winding of a large transformer, which we will call

the primary winding, we send a current which can be
regulated between a positive and negative maximum
at any time rate that may be required to
keep the
E.M.F. in the secondary winding constant. This
winding is connected to a sensitive voltmeter, and the
deflection must be kept constant whilst the
primary
current is made to change from a positive maximum
value to an equal negative maximum value. The test
requires three observers one;
watches the voltmeter and
operates the regulating appliance for the. current, the
second marks time, and the third reads on an ampere-
meter the primary current and books it against the time.
The observations thus yield in the first instance merely
a time-current curve, but as by reason of the secondary
E.M.F. being constant B and t are proportional, the
curve may by a suitable change of scale, also be made to
represent the relation between induction and exciting
force, and if the carcase contains no butt-joints also the
true hysteretic loop.
Various appliances can be devised for current regula-
"
1
On Testing Large Transformers," by I. S. Peck, EL World and
Engineer, 1901, p. 1083 ante.
SCOTT'S METHOD 253
tion, but I have found two filaments of mercury as shown

in Fig. 131 a very convenient form of rheostat. The
contacts are attached to a block of wood, which is provided
with a handle, and can be moved longitudinally on the
board ; the latter has two grooves planed out for the
reception of the mercury. Current is supplied by a few
secondary cells and measured in A. If the sliding
block is in the middle no current flows if it be shifted ;
to the right the magnetising coil of turns receives %

current in one direction, and if shifted to the left in the
other direction. V
is a millivoltmeter connected to the
secondary winding of n turns. In the diagram the two

windings are shown on different limbs this is merely ;
FIG. 131. Scott's method.
done to avoid complication. In reality both windings

are on both limbs. Both instruments have central zero.
Assume a carcase without butt-joints and let A be
the section and / the length of the magnetic circuit, then
the current passing through the millivoltmeter of resist-
ance R will be in ampere
. nA dE
corresponding to the E.M.F. in volt
-
at
r
R may be considered to include the resistance of the

254 TRANSFORMERS
secondary winding, which is, however, generally very

small in comparison with the 100 ohm or so of the
voltmeter. If, however, a milliamperemeter be used to
indicate the E.M.F., then its resistance is much lower

(with Siemens' instruments exactly i ohm),
and the
resistance of the secondary winding must be included in R.
The ratio for changing the scale of / in seconds to B
in C.G.S. units is therefore
i :
The current i necessary to work the voltmeter is very

small, and when we are testing
a large transformer negligible
in comparison with the mag-
netising current I lf but when

testing a small transformer and
when for V a milliamperemeter
O/d Current
is used the current i may
produce a sensible magnetising
effect on the carcase, and has
to be considered in determin-
ing the ratio between excit-

ing current, which is directly
FIG. 132. Scott test.
plotted, and the magnetic force
H, which we require, if we
wish to plot the hysteresis loop. We have
TT __ .
(#1 1 + ni
I
where I values between the maxima

takes all I .
Let in
Fig. 132 dWd be the time-current curve
plotted from the original observations. Before moving
the slider the magnetising current is + \ Q = oa. Im-
mediately the movement begins the current passes to
T = n
oa =
t T
h 1 1
where the second term is the pilot current i reduced to

the primary winding. In the diagram it is a a = iQ If
f
.
we wish to leave off the process exactly at the same

SCOTT S METHOD 255
negative induction as it was before we must continue to

move the slider until A indicates I =
(I + / )- As 2
soon as the movement stops becomes zero, and we have

z'
I2
= -I From a preliminary test we find io. If then
.
we start the test with a current + Ii, we must continue

until the current is I = (Ij + 2/o). 2 This gives equal
positive and negative induction.
Shifting the curve
a'b'c'
originally plotted to the right
by the distance z'
we get the curve abc, and symmetrically to this the curve
cda to complete the loop. This gives as yet only the
relation between time and current corrected for the dis-
turbing effect of the pilot current. To get the true
hysteretic loop H-B we must alter the scales as already
stated. The operation may be shown by the following
example. In a transformer having 670 primary and
100 secondary turns is 70 A
sq. cm. and / 136 cm.
The resistance of the secondary is 0*038 and that of
the milliamperemeter used for V
in Fig. 128 is i ohm.
The total time taken to perform the change from
4- B to B is 34 seconds, the instrument V
showing
steady 30 millivolt or 0*03 ampere. This gives
0-03 =
1-038 at
Since t is one half 34 we find
To make Fig. 132 represent a true hysteretic loop we

must use such a scale for the ordinates that #0 = 7514.
To find the scale for the abscissae we determine
/o
= 100 0-03 = 0-0045 ampere
^
The test is started with 0#' = o*28 ampere and finished

- We
with (0-28 + 2 x 0*0045) ampere. have
I = o*28 + 0*0045 =0*2845
and
i
H
TJ --
= 1*2^.670.0*284^
^ '
^L=\-^^
136
256 TRA NSFORMERS
The scale for the abscissae must therefore be so chosen as
tomake oa= 175. The area of the loop divided by 4?r
gives the energy in erg which is used up in hysteresis
by the whole carcase if this is taken through a complete
cycle between 6 = 7514.
This energy may also be found as follows. The
reactance E.M.F. in the primary is
obviously
=e-
n
and the power absorbed by the iron at any moment is
0J. The total

energy the integral of e^dt taken
is
between the limits shown in the time current loop that ;
is to
say, the energy is e l times the area of the loop. As
FIG. 133. Scott test as altered by Morris and Lister.
the latter is obtained in coulomb and ^ is given in watt

the product will be joule.
If the carcase has joints the loop obtained will not
be the true hysteretic loop, but its area will still give the
energy wasted in hysteresis, since air is not a hysteretic
substance.
The necessity of making a correction for the magne-
tising force of the pilot current can be avoided by adopt-
ing an arrangement proposed by Messrs. Morris and
1
Lister, whereby the E.M.F. in the pilot coil is balanced
by an externally provided E.M.F., so that no pilot
current flows. The magnetisation of the carcase is then
due to the primary current only. In Fig. 133 the
external source for balancing the E.M.F. in the pilot
coil n is a battery B
sending a heavy current through
1
Journal Inst. EL Eng., 1906, vol. 37.
KAPPS METHOD 257
the fixed resistance W

and rheostat R. As soon as the
switch S is closed a P.D. will be maintained between the
terminals of W, and this may be read off on the milli-
amperemeter M. G
is a detector
showing whether
current is
flowing through the pilot coil or not. During
the test the operator of the sliding contact watches G
and keeps its needle at zero. L is an inductance in-
serted to make this task easier by steadying the primary
current, and u is a reversing switch so that the operation
may be repeated in the opposite sense.
Kapp's method. In the author's method l a time-
current curve is also taken, but the current is not regu-
lated by an operator. It is simply allowed to flow under
a constant impressed E.M.F. Let be the flux in
<
megalines produced by a continuous current of I amperes

through n turns of winding under an E.M.F. of e volts,
then
If now ebe suddenly reversed, then I will pass from its

initial value I
through zero to the final value -f I -
Any intermediate value of the current must obviously

satisfy the equation
=*.^+RI
TOO at
..... (46)
By observing t and I a time-current curve may be plotted,

and from this curve and the known values of e and n the
hysteresis loop giving as a function of I may be drawn.
<
The arrangement of the test is shown in Fig. 134.

B is a source of current 2 capable of giving from 50
to 100 times the magnetising current I which is passed,
through the transformer coil T. This current is taken

off on the heavy shunt resistance S, between whose
1
EL Eng., 1907, vol. 39.
Journal Inst.
2
In a modified arrangement due to Mr. Dennis Coales, two equal
batteries are used coupled up in opposition. One of the batteries is
shunted by a rheostat so that its terminal E.M.F. becomes lower than
that of the other. The balance being thus disturbed, the resultant
E.M.F. of the two batteries is no longer zero. It can be adjusted by
the rheostat to the same value as that obtained in the Author's original
arrangement between the terminals of S.
258 TRANSFORMERS
terminals the E.M.F. e is maintained and indicated on
the voltmeter V. A is an
amperemeter with a central
zero and ^ a reversing switch. Care must be taken to
have the contacts of this switch in good order, so that
its resistance may be exactly the same in either position.
S may conveniently be the shunt belonging to V, so that

this is instrumental in indicating the main current given
by B. All connections should be of sufficiently stout

wire, and A should be of sufficiently low resistance to
FIG. 134. Kapp's test.
reduce the loss of E.M.F. between S and T as much as

possible.
To make the test, regulate r so that A
indicates the
desired magnetising current I and note the E.M.F. e.
Then knock s sharply over, starting at the same time a
stop-watch and noting the current indicated by as a A
function of the time. The movement of the needle for
values of I lying between I and zero is fairly quick, so
that in this region only single observations can be taken
by stopping the watch at the moment that the pointer
passes a predetermined point on the scale. After the
zero has been passed the movement becomes sufficiently
slow for a continuous series of co-ordinate values of
KAPPS METHOD 259
current and time to be noted. For transformers of

similar type the speed of the needle is approximately
proportional to the f power of the output. Thus, if
with a lo-kw. transformer zero is reached in 4 seconds,
it would be reached in about
6^ seconds with a 2O-kw.
and in about 16 seconds with an 8o-kw. transformer.
The shape of the time-current curve is of the character
shown in Fig. 132. If there were no hysteretic loss, it
would be a true logarithmic curve, but owing to the
influence of hysteresis there is a depression in the upper
part as shown.
From (46) we have
looR/j
n
Now I I is the length of the ordinate between the

curve and the + I line, so that f(\Ql)dt is the area
enclosed between the curve and this line. Integrating
between the limits and + to which correspond
< <
,
the times o and t^ we find
i I OO -LV x"x / v
2 fa= Qo
-^ (47)
if
by Qo we denote the whole area between the curve and
its
asymptote.
Integrating between the limits < and +<, to which
correspond the times o and /, we find
)o-Q) . '. . .
( 4 8)
By combining (47) and (48) we get

260 TRANSFORMERS
Q the shaded area in Fig. 135.
is
Having fixed on a
value of I, we find by planimeter the corresponding area
Q, and from (49) the corresponding value of the flux (.
It is thus easy to find by means of a planimeter corre-
sponding values of I and <, and to plot these as shown

in Fig. 134. The hysteretic energy per cycle is
obviously
n
E= x area of loop
100
If there are no joints in the .carcase, and its cross-
FIG. 135. Kapp's test.
sectional dimensions are such as to make the induction

the same in any part, the true B-H loop can, of course,
be plotted, and the permeability as a function of the
induction may In most cases, however,
also be found.
a knowledge of the exact shape of the B-H loop and of
the permeability is of secondary importance what we ;
require is a knowledge of the hysteretic loss in the whole

transformer, and this may be found graphically from
Fig. 135 without even drawing the loop.
The hysteretic energy absorbed by the carcase in one
half-cycle is obviously the difference between
l
r
efldt
KAPPS METHOD 261
the total energy supplied, and
the energy lost in copper heat. The latter quantity may

be expressed in the form
'i T ^
Rio +s
ft
f \\-dt *0
or e fl'dt
+J
where I' = I can be determined graphically by the con-
struction shown by dotted lines in Fig. 135. The

hysteretic energy for one half-cycle is, therefore
o 4
i- = el (I
- V\dt watt-second
2
<
The integral is the area (expressed in coulomb) between
the original time-current curve, and the new I' curve
shown in a dotted line. The area is to be taken with
reference to the sign of the current that is to say, ;
negative up to the point I = o and positive for I > o. By

plani metering the two areas and deducting that which
is
negative, we find
This construction applies to any transformer, whether it
has joints or not, and whether the induction is the same

throughout the magnetic path or not.
CHAPTER XII
SAFETY APPLIANCES FOR TRANSFORMERS
SUB-STATION AND HOUSE TRANSFORMERS-
REDUCING IRON LOSSES TRANSFORMER FOR
THREE-WIRE SYSTEM BALANCING TRANS-
FORMERSA UTO - TRANSFORMERS SERIES
WORKING BOOSTERS SCOTT'S SYSTEM
Safety appliances for transformers. The reason why
we use transformers is that we may carry the power
under high pressure, and distribute it under low or
moderate pressure. It is, however, an essential condition
that the insulation between the transmission circuit
(primary) and the distributing circuit (secondary) be
absolutely perfect. If this condition be not fulfilled, the
use of transformers may even become dangerous on
account of an unjustified feeling of security. The two
windings in a transformer must necessari-ly lie in close
proximity, and thus an injury to the insulation may

cause a leakage of current and a transfer of pressure from
the primary to the secondary coil. Since in a widely
distributed network of primary conductors their insula-
tion cannot be absolutely perfect, it will be obvious that
any leak between the primary and secondary coil of any
particular transformer may raise the absolute potential of
the secondary to a dangerous amount. This potential
will depend on the position of the leak in the transformer,
on the position of the equivalent leak in the general
system of high pressure or primary circuits, and on the
insulation of the secondary circuit. It may be a few
hundred volts only, or it may be equal to the full primary
voltage. If in the latter case a person touches any part
of the secondary circuit he will receive a dangerous or
fatal shock. To avoid this danger several expedients
are possible. One very obvious preventive is to place
262
SAFETY APPLIANCES FOR TRANSFORMERS 263
between the two windings a metallic dividing-sheet which

is well earthed. If the insulation between the two
windings is damaged, contact is not made between the

primary and secondary direct, but through the interven-
tion of this dividing sheet, and thus the potential of the
secondary is prevented from rising to any dangerous
extent. This appliance ensures safety only in so far as
regards a leak from one winding to the other, but it is
useless against a leak in any other part of the trans-
former for instance, between the primary and secondary
;
leading-in wires, or between the terminals of the two

circuits. Even if by good workmanship the danger of a
leak in the transformer itself, or its terminal boards, could
be completely eliminated, there still remains the possi-
bility of a contact or leak between the supply wires. An
obvious case is that where both the high and low
pressure circuits are overhead, and so near each other
that a branch of a tree blown across them by the wind
bridges the two circuits. It is, of course, not good
practice to use the same posts for both circuits, but in

certain localities for instance, immediately outside a
transforming station proximity is sometimes unavoid-

able and to exclude any danger from such causes it is
;
wise to act on the principle that the line, rather than the
transformer, should be fitted with the safety appliance.
The transformer itself is the least vulnerable part of the
system, and requires protection less than the line, but if
the line is protected, the transformer is also protected.
One way of protecting line and transformer simul-
taneously is to earth some point of the secondary circuit,
preferably the middle of the winding in a single-phase
or the star point in a three-phase transformer, since then
the potential difference of the secondary mains to earth
becomes a minimum, namely, equal to half, or very little
more than half the line voltage. If contact takes place
anywhere between primary and secondary, the former is
thereby connected to earth, and all danger of a fatal
shock is avoided. The danger as regards fire is, on the
other hand, increased by this expedient. If the whole
of the secondary circuit is insulated from earth, a fault
must occur at two places of different potential before a
264 TRANSFORMERS
danger in respect of fire can arise, but if one point of

the secondary circuit is permanently connected to earth,
a fault occurring in one place only is sufficient to create
danger. The margin of safety is therefore reduced by
one-half if we
earth a point of the secondary winding.
There is danger of damage by atmo-
also increased
spheric electricity. A
system completely insulated from
earth is less liable to be struck by lightning than one
which has somewhere an earth-connection. Finally,
there is the objection that such a system may give rise
to capacity currents (the coils in the transformer as well
as those in the generator have capacity to each other
and to earth) which disturb telephonic work. For all
these reasons the simple expedient of permanently earth-
ing one point of the secondary
circuit cannot be considered
a generally applicable, or even
when applied, a satisfactory
way of protecting the low-
pressure circuits of trans-
formers against the infiltration
of high-pressure.
The trouble about in-
FIG. 136. Protection against rise
of pressure.
creased fire and lightning
danger and telephonic dis-
turbance can be overcome if the earth-connection is not
permanent, but only established at the moment when it
is wanted. This was the leading idea in a safety
appliance introduced quite early in the history of trans-
formers by the Thomson- Houston Company in America.
The appliance consists of an earth-plate and two
metal knobs, a, b, Fig. 136, which are connected to the
secondary mains. Between the knobs and the earth-
plate is inserted a thin sheet of insulating material
(paraffined paper or mica). As long as no fault between
primary and secondary occurs, the potential difference
between the knobs and the earth-plate remains within
the limit of the secondary voltage, and this is not sufficient
to break down the insulation between knobs and earth.
If, however, through a fault in the insulation between
secondary and primary, the secondary assumes the poten-

tialof the primary, the insulation between a and earth

and b and earth is broken down, thereby short-circuiting
the secondary winding. The primary current then rises
to such an amount that the safety fuses s, s go, and
the transformer is thereby automatically cut out of
circuit.
The same principle has more recently been revived
in an improved form by Prof. Goerges in his safety-plug,
which is being manufactured by the Siemens-Schnokert-
Werke, Berlin. Here, also, one electrode of the plug is
connected with the line to be protected, and the other
with earth, but the mica insertion is only used as a
distance-piece and not as a body which must be pierced
by the discharge to earth.
Externally the safety-plug resembles the well-known
fuse-plugs commonly used on the Continent for pro-
tection against excessive rise of current, but instead of
a fuse embedded in emery powder, the plug contains
two metal electrodes insulated from each other by the
body and screw of the plug, which are of porcelain, and
a thin mica disc pierced with four holes of 3*5 mm.
diameter. The electrodes are perfectly smooth circular
plates, and their distance apart is determined by the
thickness of the mica disc. In case of undue rise of
pressure between them, a spark passes through one or
more of the holes, and this welds the two discs together,
thus providing an efficient connection to earth, or the
other circuits similarly protected.
It is important to notice that in this plug even a very
small current suffices to produce the welding together of
the electrodes through the holes in the mica, so that
even an incipient fault will be detected and rendered
inocuous by this plug. The fact that welding takes
place already with a very minute current makes the
action independent of the goodness of the earth-con-
" "
nection. It is well known that an earth good enough
to carry off large currents is generally very difficult to
provide, but so good an earth is not required for the
l
Goerges plug, since a current as small as 0*0345 ampere
1
Elektrotechnische Zeitsckrift, 1905, p. 314.
266 TRANSFORMERS
is sufficient to produce welding. If then the earth is not
good enough to carry off sufficient current to lower the
pressure, the other plug will come into action, thus short-
circuiting the low-pressure leads, and causing the fuses
on the primary to blow, and thus removing all danger.
The pressure at which the plug acts depends on the
thickness of the mica insertion. With 0*12 mm. the
pressure is 800 volt. After a plug has acted it can be
put into working order again by cleaning off the welded
parts with emery paper, and turning the electrodes so as
to bring parts of the original surfaces facing each other
through the holes. The right length of spark-gap is

obtained by simply screwing the plug down tight on to
the mica.
Asafety device invented by Major Cardew is shown
in Fig. 137. In this
arrangement the
action depends on
electrostatic attrac-
tion between a plate
E connected to the
secondary, and an
aluminium foil lying
on a plate connected
Earth to earth. The alu-
minium foil has the
FIG. 137. Cardew's safely device.
form of two discs
connected by a narrow bridge, and is together with
the two plates enclosed in a box, provision being made
by means of a screw thread in the cover of the box
to accurately adjust the distance between the plate
E and the aluminium foil. The latter is permanently
kept at the potential of the earth (zero), whilst the plate
E has under ordinary circumstances a potential not
exceeding the secondary voltage. The electrostatic
attraction corresponding to this potential difference is
insufficient to raise the foil if, however, a fault occurs
;
between primary and secondary, the potential difference

immediately rises to such an amount that the electrostatic
attraction suffices to raise the foil and bring it into contact
with the plate E, thereby earthing the secondary wind-
ing. In the safety device first described by Cardew 1

a fuse S was provided and arranged to hold up a weight
which, if the fuse melted, would short-circuit the primary
leads, and thus cause their fuses s, s to go, and the trans-
former to be cut out of circuit. It has, however, been
found that when a good earth is obtainable this is a
superfluous refinement, since the short produced on the
secondary by the lifting of the aluminium foil is in itself
sufficient to make the primary fuses go. The apparatus
can be set to come into action if the potential of the
secondary rises to 400 volt.
Hence even an incipient
fault in insulation between
the two circuits is sufficient
to automatically disconnect
the transformer from the
circuit.
Ferranti's safety device
is in Fig. 138. The
shown
secondary mains are con-
nected to the primaries
of two very small trans-
formers coupled in series,
whilst their secondaries are
coupled in parallel. The
secondaries are connected
to a fuse carrying a conical
weight over a correspond- FIG. 138. Ferranti's safety device.
ing set of terminals. The

connection between the two primaries is joined to earth,
as is also one of the terminals, the other two being
joined to the secondary mains. As long as the insulation

between the primary and secondary circuits of the main
transformer is perfect, there is absolute balance between
the E.M.Fs. of the secondary windings of the two small
auxiliary transformers, and no current passes through the
fuse. If, however, a fault occurs, the balance is disturbed,
a current passes through the fuse and melts it, and the
weight falling between the terminals short-circuits the
1
Journal Inst. EL Eng., Vol. XVII, p. 179.
268 TRANSFORMERS
secondary mains, and puts them to earth. The primary

fusess, s are thereby caused to blow, thus cutting the faulty
transformer completely out of circuit. It is important

to note that this safety device is a protection, not only
against a real short between primary and secondary, but

even against an incipient fault of insulation between the
two circuits.
The
safety devices here described and others on
similar principles are quite reliable where the secondary
mains are fed by one transformer only, and being within
a building or underground, are not subject to disturbance
from atmospheric electricity. The Goerges plug is even
applicable on an underground network fed by several
transformers in parallel, but when we have to protect
overhead low-pressure mains, all these devices, although
still
ensuring safety, are liable to come into action by
reason of a passing disturbance through atmospheric
electricity. What is required is a safety device which
will discriminate between a rise of pressure due to an
atmospheric cause, and therefore lasting only a very

short time, and a permanent rise of pressure due to a
leak or short-circuit between the low-tension and high-
tension series. The device should not be of the nature
of a delicate physical apparatus, but rather of the nature
of a substantial appliance fit to be put into an engine-
room or sub-station, and should require no attention.
Up to the present no such implement has been put on
the market.
Sub-station and house transformers. It is convenient
to make a distinction between transformers placed into
a secondary distributing centre and large enough to
supply current to a number of distinct consumers and
transformers placed on the premises of each consumer.
In the first case we speak of sub-station transformers,
and in the second of house transformers. If we except
consumers of large powers where the pressure supplied
to the installation need not be limited by other considera-
tions than appertain to the wiring, switch-gear, and the
motors themselves, we may take it that for a general
lighting and power service, whether given from a sub-
station or a house transformer, considerations of personal
SUB-STATION AND HOUSE TRANSFORMERS 269
safety as well as the nature of glow-lamps impose a limit

on the pressure in the secondary circuit of the transformer.
With carbon filament lamps as at present made 220 volt,
or at most 250 volt may be considered as an upper limit
of working pressure, whilst with metallic filament glow-
lamps a still lower limit is as yet usual. Thus a low or
moderate pressure in the distributing circuit is a necessity,
whilst a high pressure in the transmission circuit is an
economic advantage, and, indeed, also a necessity, if the
transmission has to be effected over a considerable
distance. The transformer is then the intermediary
Li
FIG. 139. Distribution from sub-stations.
apparatus by which the two conditions, cheap mains and

moderate supply voltage, can be simultaneously fulfilled.
The typical arrangement of transformers for a sub-station
system is shown in Fig. 139. C denotes omnibus bars
in the central station S, s the primary transmission mains
;
or feeders T, T are transformers at two sub-stations, and

;
V, V the supply mains. Measuring instruments, switches,

and fuses are of course also required, but have been
omitted from the diagram to avoid complication.
The diagram shows each transformer supplied with
current by its own feeder, whilst on the secondary side
each transformer supplies a network of distributing
2;o TRANSFORMERS
mains, which lattermay be either separate from each
other, or theymay be inter-connected, as shown by the
dotted lines. The inter-connection of secondary mains
has the advantage that a more nearly constant pressure
can be maintained throughout the secondary network,
and that at times of small demand some of the trans-
formers may be disconnected from the primary and
secondary mains, whereby the power wasted by them
when working an open circuit is saved. On the other
hand, there is the danger that a defect in one part of
the network may affect the whole system, and to mini-
mise this danger it is advisable to insert fuses into all
the important junctions of the secondary network.
Instead of using separate feeders to the different sub-
stations, we may also provide a primary network to
which the primary terminals of all the transformers are
connected in parallel.
When a district is supplied on the house-transformer
system acomplete network of high-pressure feeding
and distributing mains conveys current to a large num-
ber of small transformers, each placed as near as possible
to the place where the low-pressure current is required
one transformer to each house), so that no network
(i. e.
of secondary or low-pressure street mains is required.

The weight of copper in the street mains is thereby
much reduced, which is an advantage. On the other
hand, there are some drawbacks. Owing to the greater
length and the many junctions in the system of high-
pressure mains, the insulation is more difficult, the
high-pressure must be brought into the houses of the
consumers, and the loss of power in the transformers is
greater. Single transformers cannot be disconnected,
thus increasing the light-load loss, and even at heavy
load the loss of power is greater, since small transformers
cannot have as high an efficiency as large transformers,
and the total capacity of the transformers connected must
be greater than in the sub-station system. The constant
losses are therefore also greater. One house wired for
100 lamps may have sometimes 80 per cent, of its lamps,
or say as many as 80 lamps, alight, though this will not
happen very often. Twenty houses wired collectively
SUB-STATION AND HOUSE TRANSFORMERS 271
for2000 lamps will never use simultaneously 80 per cent,

that is, 1600 of the installed lamps but at most 1000,
or 50 per cent. In some cases, especially if the houses
are of widely different character (shops, offices, dwelling-
houses, restaurants), the maximum simultaneous load
may be even considerably less than 50 per cent, of the
total installed load. The ratio of the total maximum of
power supply observed in a given district to the sum of
the maxima observed at different times in each individual
house is called the diversity factor, and it is due to the
circumstance that this diversity factor is larger than
unity, that the total capacity of a sub-station may be
smaller than the collective capacity of house-transformers,
were the same districts supplied on the house-transformer
system.
Take a district in which 1000 kw. in motors and
lamps are installed. If it be supplied on the house-
transformer system the aggregate capacity of trans-
formers would be about 800 kw., made up of mostly
small sizes of, say, 2 to 10 kw. The aggregate iron
losses will be about 2^ per cent, or 20 kw., and the
copper losses about i|- per cent., or 12 kw. the latter ;
taking place, however, only during a short time daily.

The iron losses are going on all the year round, and the
energy wasted per annum is about 175,000 kw.-hrs.
The total copper loss is very much smaller we may ;
roughly estimate it at 5000 kw.-hrs. This is almost

negligible in comparison with the iron losses. If the
same district were supplied by two or three sub-stations,
the total capacity of sub-station transformers with a
diversity factor of 2 would only be 400 kw., and the
transformers at the sub-stations would be so large that
the iron losses need not exceed i per cent., or 4 kw.
The total annual iron loss would therefore be only 35,000
kw.-hrs. The copper losses will also be reduced, though
not in the same proportion, because, owing to the diversity
factor, each transformer will be working at a fair load
for a longer time daily. The annual copper loss may
be taken at about 3000 kw.-hrs. We
thus find-
Energy wasted in house transformers .

180,000 kw.-hrs.
,, ,, sub-station transformers 38,000 ,,
272 TRANSFORMERS
If the installed load is equally divided between
total
lighting and power, the energy sold per annum will be

about 400,000 kw.-hrs. for power, and 200,000 kw.-hrs.for
lighting, or 600,000 kw.-hrs. in all. The annual efficiency,
allowing 2 per cent, loss in the mains, will therefore
be
With house transformers ... 75 per cent.
With sub-station transformers .
92 ,,
The cost of house transformers, including terminal

boards and fuses, and a provision for housing them
safely that is, beyond the reach of unauthorised persons
is abouta kw., whilst sub-station transformers, in-
^4
cluding accessory apparatus and their housing, may
all
be taken at half this amount. The initial outlay will
therefore respectively be ^3200 and ^800, showing a
saving of ^2400 in favour of the sub-station system.
At 10 per cent, for interest, repair, and amortisation, this
means an annual saving of ^240, to which must be
added the saving energy wasted, which amounts to
in
1
42,000 kw.-hrs. Taking the engine-room cost at id. per
kw.-hr., this amounts to another ^590, making the total
saving ^830 annually. Against this has to be set the
increase in annual working expenses due to our having
to provide a secondary net-work. If the capital outlay
on this account exceeds % 6s. per installed kw., the
system of house transformers will be economically better ;
if the
outlay for cables is less, it will be better to use
sub-stations.
calculations have not been given as hard-and-
These
fast rules, but merely by way of example, how the com-
mercial advantages of the two systems may be compared.
A definite conclusion in any special case can, of course,
only be reached if all the conditions (such as annual
energy required per installed kw., diversity factor, extent
of district, cost of transformers and cables, and engine-
room cost of energy) are known, but from what has been
shown above we may postulate as a general principle
that the use of house transformers is commercially jus-
tified if power derived from a cheap source has to be
distributed in small parcels amongst widely scattered
REDUCING IRON LOSSES 273
customers in country districts. In towns customers are

fairly close together and the power has a greater value,
hence sub-stations are economically preferable.
Reducing iron losses. There are cases where a
consumer is not within easy reach of a secondary
network and yet requires low-pressure current. In such
cases (schools, hospitals, asylums lying outside the town),
the general system of supply from sub-stations has to be
supplemented by house transformers. If the supply is
given mainly for lighting, the load factor that is, the ratio
between the' total energy in kw.-hrs. actually supplied
during the year to the energy represented by the product
of maximum demand in kw. multiplied by the 8760 hours
of the year is
very small. For a purely lighting load
the demand factor that is, the ratio between maximum
demand observed and lamps installed will seldom exceed
So per the load factor for an isolated
cent., whilst
installation be from 4 to 6 per cent.
will only If the
transformer capable of supplying the maximum demand
has 2 per cent, iron loss and is kept under pressure all
the year round it will consume annually 175 units
magnetising energy per kw. capacity, whilst the energy
actually delivered to the lamps is only from 350 to 520
units. The annual efficiency of the transformer is there-
fore only f to f .A better efficiency would be obtained
by installing two or more transformers and switching them
into and out of circuit in accordance with the demand
actually existing at any time. This, however, would
entail an amount of personal supervision which only large
establishments could afford to provide, and then there
would be the danger of overloading a small transformer
ifthe attendant forgets to switch in the big transformer
at the right time. For this reason it is safer to work the
switch-gear automatically. Several such systems (some of
them also applicable to sub-stations) have been suggested
from time to time, and by way of example I give here
the latest device, designed by Mr. A. F. Berry. This
inventor uses two transformers coupled in series, one
small, the other large. As long as the demand does not
exceed the current capacity of the small transformer, this
is
doing the greater part of the work, but if the demand
18
274 TRA NSFORMER S
rises beyond a given limit, the primary and secondary of
the small transformer are simultaneously short-circuited,
and the large transformer is doing all the work alone.
The arrangement is shown
diagrammatically, Fig. 140.
Tj is the
small, and T the large transformer. The
secondary current is taken through on electromagnet E,
the armature of which rests on a lower contact as long
as the current is not If the
strong enough to raise it.
demand has exceeded a predetermined limit, the attraction
of E is sufficient to raise the armature so that it comes
against the upper contact. The two contacts are

connected with the two coils of the solenoidal magnet E 1?
as shown, and accord-
ingly as one or the
other is in touch with
the armature the core
of this magnet is either
pulled down (light
load) or pulled up
(heavy load). When
in the latter position
the switch Si short-
circuits the primary,
and the switch S 2 the
secondary, of the small
transformer. A
snap-
lock (not shown in the
diagram) is connected
FIG. 140. with the core of the
Berry's system of automatic control.
solenoid which not only
holds the latter in position after each movement, but also
interrupts the current through the coils of the solenoid.
This is to prevent waste of current through these coils
during the time that they are not required to act. Mr.
Berry claims that the extra cost of this switch-gear and the
small transformer is compensated by the saving in capital
outlay on the large transformer. Since the latter is most
of the time only very slightly magnetised, it enters on the
period of its full load at a low temperature, and being
worked intermittently with long spells of rest between
short periods of load, it
may, as was shown in Chapter VI,
TRANSFORMER FOR THREE-WIRE SYSTEM 275
be made smaller than a transformer continuously under

pressure.
Transformer for three-wire system. The well-known
system of continuous current distribution by three wires
can also be employed in connection with transformers.
We need only connect the middle point o of the secondary
winding, Fig. 141, to the zero wire, and the outer
terminals m, n of this winding to the two outer wires.
The primary leads s consist
of two wires only connected
to the primary terminals^,
q. The lamps a, b are con- &AAAA A
| T T T T
nected between the outer
I
q-n,
Wires and the zero wire 0. _ Se three-wire
Fig< I4I< condary system.
The pressure between m
and n is double the lamp voltage, and we are thus
able, exactly as in the ordinary three-wire system, to
effect considerable economies in the cost of the dis-
tributing mains. Care must, however, be taken to group
the different coils of the secondary winding in such way
that the ampere-turns produced by the two secondary
currents have the same value in all parts of the magnetic
circuit. If this is not done, the leakage or inductive
drop would be greater on the more heavily loaded part
I 9 <*>
T <><>
FIG. 142. Secondary three-wire system with balancing transformer.
of the system, and the supply voltage would be unevenly

divided between the two groups of lamps.
Balancing transformers. It may happen that the sub-
station must be placed at some distance from the district
to be lighted. In this case the middle wire need not be
brought back to the sub-station transformer T, Fig. 142,
if a balancing transformer Tj is established in some point
of the district to be lighted. The output of the balancing

transformer need not be larger than half the maximum
2 ;6 TRANSFORMERS
differencebetween the loads on the two sides a b of the y
Let ia be the maximum current in a and ib the

system.
current which simultaneously obtains in b, then one coil
of the balancing transformer must take up the current
"7 __ 7
-
and its other coil must give off an equal current.
If the lamp voltage is e, then the output of the balancing
transformer is
given by the expression ( --b
the out-
je,
put of the sub-station transformer at the same time being

= (ia + tb
)e.
Since it is, however, possible that
both sides of the system may occasionally carry the

maximum current, the sub-station transformer must be
If by / we denote the
designed for an output of 2t a e.
ratio of load difference between the two sides to the
maximum load on one side, we have
The output of the balancing transformer must therefore be
2**
Since iae is half the output of the sub-station trans-

former, we have the ratio between its size and that of
the balancing transformer given by the fraction 4 p. :
Thus for a load difference of 100, 50, 20, 10 per cent,

the balancing transformer would be respectively , ,
4 8
,
the size of the sub-station transformer. These
20 40
figures show that a comparatively very small balancing
transformer may render it superfluous to carry the middle
wire of the system back to the sub-station.
Another application of balancing transformers may be
made in adapting a single continuous-current generator
to a three-wire system. Let in Fig. 143 the outer circle
represent the armature of an ordinary continuous-current
generator supplying current to the outer wires a, b of a
three-wire system. Then by taking from two tapping
A UTOTRANSFORMERS 277
points connections to the slip-rings (represented in the

diagram by the two inner circles) we obtain at their
brushes an alternating voltage whose crest value is equal
to the voltage on the outer mains. The brushes of the
slip-rings are connected to a balancing transformer such
as is shown in Fig. 142. From what has been explained
in connection with this diagram it will be obvious that
the middle point, o, of the winding of this trans-
former divides the pressure between a and b equally, pro-
vided both windings are as intimately mixed as the
primary and secondary of an ordinary transformer. The
zero wire may then be connected to the point o.
Thebalancing transformer must, of course, be de-
signed for the
frequency
corresponding to the con-
tinuous-current machine.
This is
v
=pu
where p is the number of
pairs of poles and u the
speed in revolutions per
second. As an instance
Fig ^--Balancing transformer.
"
take a six-pole generator

running at 120 revolutions
per minute. The frequency will be 6. Let the output be
200 kw. at 500 volt, then the full-load current will be 400
ampere. Let the greatest out of balance current be
10 per cent., then the transformer will have to take in
20 ampere on its primary and give out 20 ampere on
itssecondary side, the terminal pressure on each side
being 250 volt crest value, or 180 volt effective value.
A transformer wound for an output of 3*6 kw. at 6 fre-
quency and transforming ratio of i : i will therefore
suffice for this purpose. It is important to design the
transformer for a very small copper loss so as to ensure
equal division of but it will be seen from this
pressure,
example that the balancing transformer, even if designed
on a very liberal scale, will be only a small accessory to
the generator.
also
Autotransformers. Balancing transformers may
278 TRANSFORMERS
be used for subdividing a given supply pressure between
a number of circuits, so that lamps requiring a lower
pressure than that supplied may be used individually on
these circuits. Originally used for arc lamps, this method
of subdividing pressure has, with the advent of the
metallic filament lamp, acquired additional importance.
Balancing transformers arranged for this purpose are
generally called autotransformers, because part of the
winding is traversed by the difference of the two currents,
and only the rest of the winding is traversed by the high-
pressure current only. This arrangement is instrumental
- --
in a certain saving of material, so that an autotransformer
is smaller and cheaper than a
I r 5 r 1 2 transformer with two distinct
2
windings. The extent to which
j
material may be saved can be
seen from the following con-
sideration-
_ I
b
Let in Fig. 144 oa be that
part of the winding which is
transversed by the difference
! 2 - Ii of the two currents, and
FIG. 44 -Autotransformer.
i .
ob the remainder which is tra-

versed by the current l l alone. Let oa consist of n^ and
ob of n 1 n z turns, and let the transforming ratio be
0tf
m= . Then I2 = m\ and
ly as equal current density gives
the best utilisation of the material we have qz = qi(m 0-

The total volume of copper will therefore be
m) ^ *
'm
v = m i
m
The volume of copper in an ordinary transformer is
v l = &2qn l
Or, taking v l as the standard, we have
v = v^ m i
m
A UTOTRANSFORMERS 279
Since in transformers of the same type, but different

sizes, the ratio of volume of iron to volume of copper is
approximately constant, the fraction *

indicates the
m
quantity of material required in an autotransformer rela-
tively to an ordinary transformer. The ratio of material
saved is the reciprocal of m, and hence for large trans-
forming ratios the auto-principle has very little advantage,
whilst the necessity of tying the two circuits electrically
together is a distinct disadvantage. This explains why
autotransformers are only used for low pressure and low
transforming ratio. They may be used with advantage
as starting devices for induction motors and for reducing
a moderate voltage to a still smaller value. Thus with a
transforming ratio of 1^5, 2, or 3 the weight of an auto-
transformer will be only 34 per cent., 50 per cent, or 67
per cent, respectively of an ordinary transformer.
Metallic filament lamps for no volt can now be
obtained. If, then, the supply pressure is 220 volt, we
could use such lamps by supplying them from an auto-
transformer, which need as regards weight and cost only
be equivalent to an ordinary transformer of half the out-
put. This is on the supposition that all the lamps are
fed from the same circuit, but if we can split up the lamps
into two circuits, each carrying half the number, the auto-
transformer need only be a quarter the size of an ordinary
transformer. Generally for m circuits each carrying th

m
of the total number of lamps at i th pressure we have
m
for the weight of the autotransformer
of the weights of an ordinary transformer. Thus, if at

220 volt supply pressure we wish to use osmium lamps
of 73 volt we can divide them into three circuits and use
an autotransformer which will only weigh one-third as
much as an ordinary transformer. Or, if we wish to use
arc lamps requiring a pressure of about 36 volt we can
280 TRANSFORMERS
group them in six circuits, and the autotransformer will
still be reasonably small,
namely, a little less than half the
size of an ordinary transformer. The lamps will be quite
independent of each other, as if they were all in parallel
on one and the same circuit.
Series working. Transformers may be advantage-
ously used if it be required to work a number of lamps
in series off a circuit in which an alternatino; o current of
constant strength is maintained. If we were to insert
the lamps themselves into such a circuit, the insulation
of the lamps to earth would have to be so perfect as to
withstand the full potential difference of the alternating
current, a condition not always easily fulfilled. If, how-
ever, we feed the lamps from the secondaries of series-

transformers, it is only necessary to provide perfect
insulation for the trans-
formers, which presents
no difficulty the insula-
;
/vwvwws tion of the lamps need

a~
only be good enough for
the voltage required by
each lamp. The arrange-
FIG. 145. Series working. merit is shown in Fig.
T, series-
145. T are
transformers supplied from a constant-current alternator,
and L, L are the lamps. The primary return circuit is
not shown. Since the current in the primary is constant,
the current in the secondary is also approximately con-
stant as long as the lamp is in circuit. There is, how-
ever, the drawback that if a carbon should fall out of
a lamp, or some other accident happen whereby the

secondary current is interrupted, the induction in the
core and the E.M.F. in the secondary of that particular
transformer (if this is of the ordinary construction for
parallel work) would rise very considerably. Since the
primary current must, on account of the other lamps, be
kept constant, the pressure at the generator has, in such
a case, to be increased. The transformer with open
secondary becomes magnetically overloaded and must
eventually burn out. To avoid this danger we must
make provision to give the secondary current an alter-
SERIES WORKING 281
native path in case the lamp circuit should become

interrupted. This may be done in two ways. may We
employ a kind of automatic "cut-in" as in a, or a
choking coil as in b. The cut-in consists of two
electrodes separated by a thin sheet of mica or paraffined
paper, which, under normal conditions, is sufficient to with-
stand the secondary voltage. If, however, the secondary
voltage rises considerably, in consequence of the lamp
circuit being opened, the insulation between the electrodes
breaks down, and the cut-in short-circuits the secondary
coil of the transformer. The choking coil, which may
be used instead of a cut-in, allows a current to pass
through its winding proportional to the lamp voltage,
but lagging by nearly 90 behind it. The power lost in
the choking coil is the sum of hysteresis and ohmic loss ;
and by a proper design of choking coil it is thus possible

to minimise the loss of power, although the presence of
choking coils must worsen the power factor. This may
best be seen by an example. Let us assume that the
lamp requires 10 ampere at 35 volt, and that its power
factor is 80
per cent. The power actually supplied to
the lamp is therefore 280 watt. Let the choking coil be
so constructed that it takes 5 ampere if the pressure is
35 volt, and that the loss of power in it is 5 watt. Its
power factor is therefore = 0*0285. If we now

- -
35 x 5
draw a vector diagram to represent these working con-
ditions, we find that the total secondary current supplied
by the transformer is 13-5 ampere. We
also find from
this diagram the power factor of the combination
lamp
plus choking coil is only 0*6. If now the lamp current
is
interrupted the choking coil must pass the whole 13*5
ampere, and the voltage must rise to
35 = 95 volt
This is an excess of 170 per cent, over the normal

voltage, and accompanied by a similar rise in the
is
magnetisation of the iron core. It is of course always
possible to so design the choking coil that it can stand

282 TRANSFORMERS
this increase of magnetic load without danger for any
length of time.
Sometimes it is convenient to use a transformer for
feeding a circuit of lamps which requires a

in series,
nearly constant current, although the number of lamps
inserted may be varied. This condition is of course
fulfilled the primary current is constant, but if the
if
primary voltage is constant a transformer for parallel

work (that is, a transformer of the usual construction
having as little magnetic leakage as possible) would be
quite unsuitable. Such a transformer keeps the secondary
voltage approximately constant, but not the secondary
current. When we have lamps in series it is the current
which must be kept constant, whilst the voltage must
vary as nearly as possible in accordance with the number
of lamps alight at any time. As was already shown in
FIG. 146. Constant current transformer.
Chapter IX, this condition can be met at least approxi-

mately by shaping the transformer in such way as to
produce a large magnetic leakage. A construction of
this kind is shown in Fig. 146. It is a core transformer
with primary and secondary coils on separate limbs and
with expansions a, b of the two yokes arranged specially
to produce magnetic leakage. The primary coil is joined
to the primary constant-pressure lead s; and the secondary
coil to the circuit
containing the glow lamps L in series.
It will be obvious that with an
open secondary or
lamp circuit the leakage field between a and b will be
very small, since the core of the secondary coil offers a
ready path for the magnetic flux. If, however, the lamp
circuit be closed, a current flows in the secondary coil,
pushing back part of the flux produced by the primary

coil, and the leakage field, not only between a and b but
all over the transformer, will be much increased. The
BOOSTERS : :
larger the secondary current the more lines are pushed

back,,and the lower will be the secondary E.M.F. If
a lamp is short-circuited the current will at first increase.
This increase produces more magnetic leakage, and
lessens thus the flux which produces E.M.F. in the
secondary. The increase in current strength will there-
fore be considerably smaller than would obtain with an
ordinary transformer, and in this way it is possible to
keep the current at least approximately constant when
lamps are put out of action by being short-circuited.
For the exact determination of the working condition
see the vector diagram given at the end of Chapter IX.
Boasters. If some of the feeders between the central
station and the sub-stations are very long, it is some-
times advantageous to allow a greater voltage drop in
them than in the shorter feeders, and to raise the
147.:
1
pressure at thehome end of these long feeders by an
amount corresponding to the extra drop. For this pur-
pose special auxiliary transformers, so-called "boosters,^
may be used. This system of boosting-up the pressure
at the home end of long feeders has been invented
simultaneously and independently by Mr.1
Stillwell in
America, and by the Author in England. It is shown
diagrammatically in Fig. 147.

C are the bus bars in the station, S is a feeder
supplying current to the transformer T
at a sub-station.
Y are the distributing mains connected to this trans-
former. The boosting transformer has its primary
permanently connected to the bus bars, whilst its
secondary is put in series with the feeder and is sub-
divided into sections, so that by using a switch , a
greater or lesser number of secondary turns can be
inserted. In this manner the additional voltage put into
11
British Patent, No. 4345, March 21,
284 TRANSFORMERS
the feeder at the home end may be varied from zero to
the full voltage given by all the secondary turns of the
booster. The full voltage is added when the feeder
carries its maximum load the switch is then placed on
;
its highest contact. As the load decreases the switch is

shifted to a lower contact, the intention being to boost
up by the amount corresponding to the drop in pressure
due to the impedance of the feeder. Since this drop is
proportional to the current, the adjustment of the switch
may be made in accordance with the readings of an
amperemeter in the feeder circuit, or pilot wires may
be brought back from the sub-station and connected to
a voltmeter. The switch is then adjusted so as to keep
the pressure indicated by the pilot voltmeter constant.
It is obvious that in either case the switch-lever can be
worked automatically by a
g ~small electro - motor con-
trolled by a relay. Since,
in passing from one contact
to the other, the switch-
lever, if it were made in one
solid piece, would short-
and possibly burn out
circuit,
FIG. 148. Booster. the section of the secondary
winding connected to the
two corresponding contacts, it is necessary to employ a
lever consisting of two parts, each smaller than the width
of the gap between two contacts, and having an insulating
partition between them. The two parts must of course
be joined by a suitable resistance, or preferably by a
choking coil. With such a construction there can occur
neither a short-circuit in the booster nor an interruption
of the feeder current.
The necessity to send the whole feeder current through
the switch, and the drawback of a complete interruption
of the feeder current if this switch should get out of
order, has led the Author to design the modified arrange-
ment of booster in which the switch is connected, not with
the secondary, but with the primary circuit of the auxiliary
transformer. This arrangement is shown in Fig. 148.
The feeder circuit is permanently connected with the
BOOSTERS 285
bus bars through the secondary winding of the auxiliary

transformer, whilst the multiple contact switch is inserted
into its primary connection with the bus bars. The
primary winding is subdivided into groups a, b, c, etc.
According to the position of the switch-lever, more or
less of these groups are active, thus causing the magnetic
flux and the E.M.F. in the secondary to be smaller or
greater respectively. The first group a must of course

contain a sufficient number of convolutions to prevent
the auxiliary transformer from being magnetically over-
loaded. This kind of booster must therefore be larger
than that shown in Fig. 147, but as in any case the cost
of a booster is very small as compared with the saving
in the cost of the feeder thereby rendered possible,
the extra outlay is insignificant,
whilst the possibility of keeping
up the supply, even if the switch
should become deranged, is a
distinct advantage.
In a third type of boosting
apparatus there is no switch of
any kind, either in the secondary
or primary circuit. This type is
shown in Fig. 149. The con-

structionresembles that of a
two-pole dynamo with shuttle- FIG. 149. Booster.
wound armature. The field is

built up of sheet-iron plates, and is provided with the
primary winding P, P, whilst the armature carries the

secondary winding S placed over a core of sheet-iron discs
in the usual manner. Both windings are permanently
connected, the primary with the bus bars, and the
secondary with bus bars and feeder as in Fig. 148. By
means of worm gearing, the coil S may be placed at
various angles with reference to the polar surfaces. If
the coil S is turned into a vertical position, the flux -of
force passing through it is a maximum, and the E.M.F.
generated in this coil is a maximum. If the coil be
it is ineffective, whilst in intermediate
placed horizontally
positions any desired boosting effect may be obtained.
By turning the coil beyond its horizontal position the
286 TRANSFORMERS
action may also be reversed, that is to say, we can reduce
the E.M.F. at the home end of the feeder. The advan-
tages of this type of booster are that no switches of any
kind are used, and that the adjustment of the boosting
effect is made, not by definite steps, but as
gradually as
we please, by means of the worm gear.
Abooster constructed on the same principle may
also be used to regulate the alternating pressure
supplied
to a rotary connector. In these machines the ratio
between the alternating pressure supplied to and the
continuous pressure derived from the armature is con-
stant whatever may be the excitation, so that no
adjust-
ment of continuous pressure can be made by means of
a rheostat in the exciting circuit as is done in an
ordinary continuous-current generator. Yet it may be
necessary to adjust the pressure at which the continuous
current is delivered. This is done by adjusting the
alternating pressure of the
driving current, a special
type of booster being used for the purpose.
This booster is a three-phase transformer with mov-
able secondary winding in construction it resembles an
;
ordinary induction motor, the primary being wound on

the stator to produce a rotating field, whilst the secondary
is wound on the part which usually is the rotor, but which
in this case is not allowed to rotate. The arrangement
is shown in Fig. 150.
U the converter with its commutator K, from
is
which the continuous current is delivered, and its slip-

rings s, by which it relieves the alternating three-phase
current. Between the slip-rings and the source of alter-
nating current in this case a three-phase transformer T
is
placed the adjustable booster B. Its primary wind-
ings are connected to the source, and produce a magnetic

field of constant strength
revolving round the rotor with
a velocity corresponding to the frequency. The winding
of the rotor is represented by the three coils inside the
inner circle. For the sake of simplicity these are shown
parallel, but it must be understood that they are placed
with an electrical angular displacement of 120 to each
other, so that by being successively cut by the revolving
primary field, the E.M.Fs. induced in them follow each
BOOSTERS 287
other at intervals of a third period. The phase in rela-

tion to the primary at which the voltage of the second-
ary is injected must therefore depend on the angular
position at which the rotor is set, as shown by the little
vector diagram below the figure. In this E is the E.M.F.
of the source, e the E.M.F. induced in the secondary
coils of the booster, and E] the E.M.F. supplied to the
slip-rings of the converter. The position of the vector e

depends on the position to which the rotor is set, so that
E! may be made either larger or smaller than E. The
use of a booster of this kind alters slightly the power
factor, but as there are always two positions of e for each
FIG. 150. Booster applied to converter.
required value of E we may

1? choose that by which the
power factor is increased.
Since a considerable torque is exerted on the second-
ary, it is necessary to use worm gearing for setting the
rotor, and in large boosters it is advisable to couple two
mechanically together, the electrical connections being
made in such sense that the two torques eliminate each
other. The latter arrangement has been first used by
Messrs. Siemens, Schuckert Werke in the Paderno power
transmission. In this case the boosters were not used in
connection with converters, but simply for the purpose
of compensating the drop in long and heavy feeders.
TRA NSFORMER S
An interesting application of trans-
Scott's system.
formers the conversion of a two-phase into a three-
is
phase system, and vice versa, invented by Mr. C. F.

Scott.
1
The arrangement is diagrammatically repre-
sented in Fig. 151, where is a G
two-phase generator
supplying current to the primaries of two transformers
Tj and T 2 The secondaries of these transformers are
.
joined together, as shown in the figure, leaving three

terminals, A, B and C, free for connection to the secondary
circuit. Since the primary currents in T l and T 2 have a
phase difference of 90, there is also the same phase
difference in the E.M.Fs. generated in the two secondary
coils. The E.M.F. between terminals A
and B is there-
fore the resultant of two components, one being the full
/wvwwwx
FIG. 151. Scott's system. FIG. 152. Vector diagram of

Scott's system.
E.M.F. generated in the secondary of T and the other 1}
half the E.M.F. generated in the secondary of T 2 the ,
latter component being moreover displaced by 90 as
regards the former component. Let, in Fig. 152, be OA

the E.M.F. of T\ and OB half the E.M.F. of T 2 then ,
BA is the resultant E.M.F. which we measure between

the terminals A
and B. In the same manner we find
CA as the resultant E.M.F. produced by Tj. and the left
half of T 2 whilst CB is the E.M.F. produced by both
,
halves of T 2 It will be obvious that, by a proper choice

.
of the number of turns in the secondaries, we may so
arrange matters that OB = lAB. Then AB = BC = CA,

and OA = AB->/ 3 ,
or OA = 0-867 AB-o'86; BC. The
1
The Electrician, April 6, 1894.
SCOTT S SYSTEM
winding must therefore be such that the secondary volt-

age of T! is 0*867 of the secondary voltage of T 2 In .
the clock diagram the vectors of terminal pressure pass

then through zero at intervals of 60, or in the same sense
at intervals of 120, which characterises a three-phase
current. We obtain thus from the terminals A,B,C a
three-phase current.
The advantage claimed by Mr. Scott for this system
is that the
generation and utilisation of the current may
be effected by two-phase machinery, whilst the trans-
mission may be made in three phases. The former
condition he considers to be an advantage as regards the
independent working of motors and lamps, and especially
their regulation, whilst the latter condition is, of course,
1)
FIG. 153. Scott's system.
conducive to economy in copper on long lines of

transmissions.
A complete plant arranged according to Scott's
system is shown in Fig. 153. G is a two-phase generator
producing 100 volt, which pressure is transformed up to
2000 and 1730 volt in the two transformers shown. To
the three free terminals are joined the line wires, and
between each pair there is a pressure of 2000 volts. At
the points of consumption the three-phase current is
either transformed down and converted into a two-phase
current for working motors (A) or supplying light (B), or
it may be used as a three-phase current for working
motors (D). Although the circuits are inter-connected,
the regulation for constant pressure in the lamp circuits
causes, according to the inventor, no more difficulty
than if the lamps were connected directly with the
generator.
19
290 TRANSFORMERS
The mechanical construction of the carcase of the
loo-k.v.a. transformer is shown in Figs. 216 and 217. It
will be noticed that the frame is constructed in the form

of a grid, so as to allow the cooling medium direct access
to the plates of core and yoke.
CHAPTER XIII
THE TRANSFORMER IN RELATION TO ITS CIR-

CUITSEQUIVALENT COILS IN PARALLEL
AND SERIES CONNECTION RISE OF PRES-
SURE THROUGH RESONANCE DETERMINA-
TION OF THE DANGEROUS CONDITION RISE
OF PRESSURE ON LOADED LIGHTING SYS-
TEM IS SMALL INFLUENCE OF POWER
FACTOR BREAKDOWN OF CABLES IN LARGE
NETWORKS
The transformer in relation to its circuits. Up to the
present we have considered the transformer as an appar-
atus by itself, receiving energy from a source not in-
fluenced by -its presence and giving up energy to some
receiving device which, apart from its ability to absorb
energy, has no influence on the transformer. In other
words, we have assumed the primary current to be derived
from an inexhaustible source and the secondary current
to be given to an apparatus which can only absorb, but
not return energy. These conditions obtain in the
use of transformers. It has been shown in
ordinary
Chapter IX that the reactance of a transformer working
under load is extremely small, and for this reason any
reactive effect of the consuming device is transmitted to
the source of current much in the same way as if the
transformer were not interposed. The transformer is
simply a means of linking the two circuits together and
adjusting the pressure, but is otherwise inert. There are,
however, cases when a transformer may cease to play
this passive role, and by reason of an interaction between
its inductance and the capacity of the circuit cause a rise
of pressure sufficient to break down itself or a cable,
generally the latter. Such special circumstances may
arise under the two extreme cases of a transformer work-
291
292 TRANSFORMERS
ing either at no-load or under short circuit. In the first

case the reactance is large because only the equivalent
coils representing excitation are acting in the second ;
tbe reactance voltage is large because the current has

enormously increased.
Equivalent coils in parallel and series connection.
When drawing the vector diagram of a transformer
under load we have made use of the conception of
equivalent coils in parallel across the primary terminals,
one of these coils having such an inductance as to let
pass the magnetising current, the other having such
a resistance as to let pass a current which, multiplied by
the primary pressure, represents iron losses. When
considering the effect of capacity in the circuits it is
convenient to substitute for these two parallel coils,

two coils in series with each other, and with the capacity,
as was already done in Chapter IX under the sub-
heading "The Self-induction of a Transformer." Let
R and >L be resistance and reactance of the two
equivalent coils in parallel, and Rj and ^Lj the respective
equivalent values for the coils in series, then we have
R
R sin <f
cos L sin 2
L sin f cos = R cos

L and R, discussed previously
when introducing the conception of equivalent coils, we
can now substitute one coil containing wLj and Rj in
series, and the diagrammatic representation of a trans-
former having the transforming ratio i i will be as shown :
in Fig. 154, where coLj and Rj represent reactance and

RISE OF PRESSURE THROUGH RESONANCE 293
resistance of the exciting coil and L the load which is
supposed to be switched off.
Let B represent the bus bars at the station, and let

the transformer be joined to them by a concentric cable.
As long as both conductors remain connected to the bus
bars the pressure at the terminals of the transformer
cannot rise above the station voltage, but if the switch to
the outer combustion has opened the cable as well as the
transformer may, under certain circumstances, be subjected
to an excessive pressure due to exact or approximate
resonance between the inductance L x and the capacity of
the outer conductor to earth. This capacity is indicated
in the diagram by C, whilst the capacity of all the outer
conductors of other cables in the network fed from the
same bus bars is indicated by C On opening the switch
.
s the circuit remaining is as follows From the upper bus

:
Gable
nWWP-AAAAOi
coLi Hi
%%M%Z%Zf^
Earth
FIG. 154. Rise of pressure through resonance.
bar through the inner conductor to the equivalent coil

wLjRj, then to the outer conductor, from there through
C to earth and finally through C to the other bus bar.
We have thus two capacities and an inductance in series.
The two capacities in series are equivalent to a single
capacity c of the value
Now ina large distributing system the aggregate

capacity, C all the feeders and network connected with
,
of
them is enormously greater than the capacity of the
so
single feeder to the transformer under consideration,
that we can write c = C, and we have thus a circuit as
shown in Fig. 155, whose natural frequency is
i
\~--
27T
i ooo
..... /
(53)
_ \
294 TRANSFORMERS
where Lj is given in Henry and C in microfarad. If
v happens to be not very different from the frequency at
l
the bus bars we have approximate, if it happens to be

equal to this frequency we have exact, resonance,
and the
current flowing through the circuit will be nearly or
exactly given by
the resistance of the cable being neglected because it is

very small as compared to R T .
This current is larger than the normal magnetising

current and produces a terminal pressure also larger than
the normal. How much larger will depend on the
relation between wLj and Rj. In a transformer having
little iron loss, but a large
magnetising current, that

Ki is to say a low power factor
at no load, the excess pres-
sure thus
produced by
approximate or complete
FIG. ^.-Resonating circuit.
resonance may be several
times the normal working
pressure,and may cause a breakdown in the cable either
between the two conductors or between the outer
conductor and earth.
Determination of the dangerous condition. An
example will make
the foregoing clear. Let the primary
feeders and networks of an electricity works have a total
length of 100 km. (63 miles), and let there be one feeder
leading to an isolated transformer. Assume a bus voltage
of 3000 and a frequency of 45. If concentric cables are
used, the capacity of outer conductor to earth will vary
according to the size of cable between 07 and 1*5 micro-
farad per km. Let in our case the capacity be i
microfarad per km. and let the cable feed a 2O-kw. trans-
former, which has i J per cent, or 300 watt iron loss, and
let the magnetising component of the no-load current be
3 per cent., or 0*196 ampere. The reactance of the

magnetising coil will then be 3000 0*196= 15,300 ohm,
:
and the resistance of the parallel coil representing iron

DANGEROUS CONDITION 295
losses will be 3000 o* i = 30,000 ohm. For the equivalent

:
series arrangement we find from (50) and (51)

)L!= 1 2 100 ohm
1^ = 6200 ohm
Lj = 43 Henry
We have then a circuit consisting of a capacity C, a

resistance of 6200 ohm, and an inductance of 43 Henry,
all in series. It should be noted that Rj has no
physical
existence it is a fictitious resistance corresponding to
;
the iron loss at normal excitation, that is to a no-load

current of ^/O'ig6 + O'i = o'22 ampere.
2 2
This current
will flow if on the terminals of the transformer 3000 volt
is
impressed. To produce this E.M.F. the pressure on
the bus bars must be
where C is given in farad. For a certain value of C

(in our case about 0*3 microfarad), there will be reson-
ance, and the term in brackets will become zero, so that
a bus-bar voltage of o'22R 1 =1360 volt will suffice to
produce the full voltage on the transformer. Since the
bus-bar voltage is not 1360 but 3000, it will be obvious
that there must be a rise of terminal voltage on the
transformer. The question is, how large a rise? It
would not be correct to assume that the rise will be
simply inproportion of 1360 to 3000. This would be
the case R! were
if a physical resistance, but as it is
only a fictitious resistance to represent iron loss, and as
the latter varies with the no-load current, it is obvious
that R! cannot be a constant. The problem is as follows :
Given a constant bus-bar voltage and a transformer of

known iron, find for various values of the capacity
between the outer conductor of cable and earth the
pressures between inner and outer conductor, and also
between outer conductor and earth.
The solution is as follows From the known quality
:
of the iron calculate the iron loss P as a function of the

terminal pressure, and plot this as shown by the dotted
2 96 TRANSFORMERS
curve in Fig. 156. Plot in the same diagram, also as
functions of the terminal pressure, the total no-load
current i and its two components, i^ and ih The copper .
loss, being exceedingly small, need not be taken into

account. Now assume any terminal voltage, larger than
3000, say, for instance, 3500, and draw its vector in
Fig. 157 to an arbitrary volt scale. Let this be OA.
From Fig. 156 we find the corresponding magnetising
2000/
1-0
0-9
0-8
P/
0-7
0-6
1000
gO-5
*
04
o|
0-3
0-2
0-1
0123456789
FIG. 156.
Terminal Pressure. in
Characteristic curves of transformer.
current ^ = 0*23, and the current corresponding to the

iron loss /A = o'ii. Let OBand BC be the vectors of
these two components drawn to an arbitrary ampere
scale, then = OC is the resultant or no-load current.
z*
Since this current is charging the condenser the terminal

E.M.F. of the latter must be at right angles to it. Draw
then from A the line AD
at right angles to OC, and
determine its points of intersection with a circle, the
radius of which represents on the volt scale the pressure
DANGEROUS CONDITION 297
at the bus bars, namely, 3000 volt, The condenser

voltage e can then be scaled off on
the line AD. It is either of the
two values
AE = 5700
No other value is possible at
the assumed terminal voltage of
3500. But in order that either
voltage may obtain the capacity C
must have a definite value, which
is found from
/o
= >*Cio- . .
(54)
C being given in microfarad. The
is
capacity
for e= 575 . . C = i'6 I
$7' Determination of
dangerous capacity.
,,
= 5700 . . C = o'i6i
By repeating the construction here explained for
other values of terminal voltage we find other values
for e and C, and we are thus able to plot the relation
0-5 1-0 1-5
FIG. 158. Voltage due to resonance in an unloaded system.
between capacity to earth of outer conductor (or what

comes to the same thing), length of feeder and cor-
responding pressures between the two conductors and
298 TRANSFORMERS
between outer conductor and earth, as shown in
Fig. 158. In this figure the curve I gives the terminal
pressure on the transformer, which is, of course, equal
to the pressure between the two conductors of the
cable, and the lower curve 1 1 gives the terminal pressure
between the outer conductor and the lead sheath, that is,
earth. As be seen, both pressures exceed 8000
will
volt if
capacity is 0*25 microfarad, which cor-
the
responds to a length of feeder of a quarter kilometer.
With a longer or a shorter feeder the excess of pressure
will be less than 5000 volt. We
may thus consider
250 m. a dangerous length of feeder. For the insulation
between the two conductors the danger is not very great.
A 3OOO-volt cable will probably stand 8000 volt also,
but the outer conductor is not very heavily insulated
against the lead sheath, and for this light insulation
8000 volt indeed a dangerous pressure, which in all
is
probability produce a breakdown. All danger can,

will
however, be avoided if the switch gear is either so con-
structed that the inner conductor must be switched out
first, and the outer conductor must be switched in first, or
if only solid connections without any switches or fuses are
used for the outer conductors.
In exemplifying the rise of pressure by resonance
for a definite case, I have assumed that concentric cables
are used, as this is the usual practice in single-phase
working, but the same argument also applies to stranded
cables for either single or multiphase working. In such
cases the "dangerous length of feeder" is by reason of
the smaller capacity much greater, and as all the con-
ductors are equally well insulated the danger for each
is no
greater than that for the inner conductor in the
case of a concentric cable. On the other hand, the
simple remedy of omitting all switches and fuses in one
of the conductors is no longer available, as it would
increase the danger for the others and displace the
electrical centre of the system. The remedy is, however,
simple enough it consists in;
arranging the switch gear
so that all the conductors of one feeder are switched on
and off together.
Rise of pressure on a loaded lighting system is small. It
PRESSURE ON LOADED LIGHTING SYSTEM 299
may be objected that this arrangement can only refer to

intentional switching, whilst the accidental blowing of a
fuse through overload may inter-
rupt one conductor and thus estab-
lisha dangerous condition. This
objection is not valid, because if
there be a load the rise of pressure

can only be very small. This
will be seen from Fig. 159, which
is plotted for the same transformer
as Fig. 158, but on the supposition

that the transformer has a load of
10 per cent, of its normal, and
that the power factor of the load
is90 per cent.If the load were
non-inductive the rise of pressure
would be almost imperceptible,
but even at 90 per cent, power
factor it is
quite moderate, and
the dangerous length of feeder is
now 600 m. For a feeder made
of stranded (instead of concentric)
conductors it would be 3 km. or
more and perfectly harmless.
In
Influence of power factor.
the case represented by Fig. 159
the slight rise of pressure is mainly
due to the fact that the power
factor of the load is only 0*9 instead
of unity, and the conclusion seems
plausible that a lower power factor
would result in a bigger rise of
pressure. This is indeed the case.
Let the transformer of the previous
example have an inductive drop of
4 per cent, and a copper loss of^
ij per cent., and let it be used to >
current to an induction FIG. Rise of pressure on
supply
Jri J
159.
loaded system.
! r
motor whose power factor at
starting at a pressure corresponding with 3000 volt on
the primary of a "perfect" transformer is 30 per cent.
300 TRANSFORMERS
with a starting current of 15 ampere in the primary. A
simple calculation, which need not be repeated here,
shows that the combination of motor and transformer
can be replaced by an equivalent coil of 79 ohm. resist-
ance and 230 ohm. reactance, or 244 ohm. impedance,
giving a starting current on the primary side of 12*3
ampere. Assume now that with this current one of the
fuses at the home end of the feeder happens to blow we ;
shall then have again a circuit as shown in

Fig. 155,
only that now reactance and resistance are much smaller,
FIG. 160. Voltage due to resonance in a loaded system.
requiring a greater charging current, and therefore a

greater capacity that is to say, a greater length of feeder
to produce a dangerous condition. In this case it is not
the inductance of the transformer, but that of the receiving
apparatus which produces the rise of pressure. The
transformer may, however, break down in consequence.
The graphic treatment of this case is the same as
that already explained in the case of an unloaded
transformer, and need not be set out in detail. The
result is given in Fig. 160, which shows that with a
capacity of about 13 microfarad the pressure on the

transformer terminals, and therefore between the two
BREAKDOWN OF CABLES 301
conductors of the cable, as well as the pressure between

the disconnected conductor and earth, will be about
10,000 volt. The pressure on the terminals of the
transformer is represented by the curve I, and that
of the outer conductor to earth is represented by the
curve II.
Breakdown of cables in large networks. It is a
common experience that if a " dead earth " occurs at
some place in a high-pressure network, and even if
the faulty place is promptly isolated by the fuses
blowing, the insulation of the cables in some other part
Earth
FIG. 161. Illustrating break-down in large network.
of the network breaks down. This is also due to the

interaction of inductance and capacity. Let in Fig. 161
I be the inner and O
the outer conductor leading from
the central station to a sub-station where a transformer
P, S supplies current to the secondary network. Ii are d
primary, and I2 2 O
secondary
,
cables joining this sub-
station to others, which, in their turn, also receive
primary current from the central station, the arrangement
being that commonly in use of a complete secondary and
a complete primary network, the two interlinked by
transformers, and the primary network fed at many
points by high-pressure feeders. Let all the cables be
302 TRANSFORMERS
concentric, and the outer conductors neither fused nor
provided with switches. Owing to the great capacity of
the outer conductors to earth, the potential to earth of
d, as well as that of O 2 will be very nearly zero, and
,
that of Ii to earth will be very nearly equal to the bus-

bar voltage, and that of 1 2 to the lamp voltage. A short-
circuit to earth on the outer conductor is therefore
unlikely, and, if it should nevertheless happen, inocuous,

but a short-circuit to earth of any part of the inner con-
ductor will result in a heavy current and blowing of the
" "
nearest fuses. Let, for instance, a dead earth be
developed by some failure of insulation at point E, and
fusing together of the inner conductor with a metal part
well earthed, say the case of the transformer, then the
fuses f,fi, and/2 will promptly blow, but the current going
to earth will not be interrupted thereby, for P receives
E.M.F. by induction from S from the other sub-stations,
and it is only after the fuse f* has also blown that the
earth-current ceases. But/s, being in the secondary, is
necessarily a heavy fuse, and requires some time to come
into action. During that time there exists a dangerous
condition, for we have now the inductance of the trans-
former due to its magnetic leakage in series with the
capacity of O and d to lead sheath, that is to say, the
capacity of the whole of the network to earth. This capacity
in a larger system may be enormous, perhaps 100 micro-
farad or more. The charging current will now flow from E
(equivalent to the lead sheaths of all the primary cables)
through P to the totality of the outer conductors. Let
C be the capacity to lead of all the outer conductors,
then we have again a circuit as represented by Fig. 155,
but with this difference, that E is now given, not by the
bus bars, but by the primary of the transformer, and the
equivalent coils to Li and Ri represent now the effect of
magnetic leakage and true copper resistance. The
inductance and capacity being in series, there will be a
rise of pressure, but whether this will be a dangerous
rise will depend on the electrical constants of the trans-
former and network. In the first place it should be noted
that E is smaller than the bus-bar voltage, the reduction
depending on the resistance of those cables which bring
BREAKDOWN OF CABLES 303
the current from the neighbouring sub-stations. If the
capacity be very large and the transformer very small,

the latter is almost in the condition of short-circuit. It
be burned but no rise of will be

may up, great pressure
produced on the cables, so that no breakdown
of the
cables is likely to occur. if the transformer is
Again,
very large, the charging current which the cables can
take will be insufficient to produce any considerable
E.M.F. of self-induction, and also in this case there is no
danger. Between these two extreme cases there may,
Volt
10,000
5000
10 20 30 40 50K.V.A,
FIG. 162. Rise of pressure in large network.
however, be others where the inductance due to

magnetic leakage and the capacity of the network are in
such proportion to produce perfect or approximate
resonance, and then the pressure of the outer conductors
to earth may rise sufficiently to break down the insula-
tion in one or more places.
After what has been already explained, the reader
will have no difficulty in determining for any given net-
work this dangerous size of transformers. As an example
I take a network of 100 km.
having a capacity to earth of
100 microfarad and transformers with \\ per cent, ohmic
304 TRANSFORMERS
and 4 per cent, inductive drop. The resistance of the second-
ary network is such that at full load the ohmic drop between
a sub-station and a consumer midway between two sub-
stations is 1 1- per cent. The frequency is 4.5, and the working
pressure 3000 volt as before. On making the calcula-
tion for various sizes of transformers, we find that one of
about 10 k.v.a. will produce a pressure of about 8000
volt, both on its own terminals and between the outer
conductor and earth. With smaller and larger trans-
formers the pressures are less. Fig. 162 shows the
relation between capacity of transformer and pressure
for this particular case. Curve I shows the pressure on
the terminals of the transformer itself, and curve II
shows the pressure between the outer conductor and
earth. Taking a rise of pressure up to 3000 volt on the
outer conductor of the concentric cables as just on the
verge of danger, we see that transformers below 7 k.v.a.,
and above 22 k.v.a. may be used, but not transformers
between these two limits. This is with concentric cables.
With stranded cables the capacity of a loo-km. network
would be barely 20 microfarad, and then even a lo-kw.
transformer would already be outside of the danger limit.
CHAPTER XIV
SOME EXAMPLES OF MODERN TRANSFORMERS
THE practical development of any new piece of machinery
or apparatus is generally a matter of trial and error.
At first, whilst the scientific principles underlying the
new application of natural laws are but imperfectly
understood, we have a period during which inventors,
groping more or less in the dark, seek success in
abnormal designs or the special development of some
detail which later on is seen to be of minor importance ;
then comes a period where the really essential details are

recognised and receive consideration, and as these are
perfected we get to the final stage, characterised not by
divergence, but rather by uniformity of design. The first
period in the development of the transformer has not
been dealt with in this book. To the technical historian
itmay be interesting to investigate the early designs of
Goulard and Gibbs, Lane Fox, Rankin Kennedy and
other pioneers, but such investigations will not help one
to either understand the working of a transformer or to
design one. The second stage, namely the conscious
improvement of details, has been treated in the previous
and it now
pages, only remains to give the reader a
general survey of the last stage by placing before him a
few examples of the best modern practice.
The Brush Electrical Engineering Co., Ltd. Figs.
163 to 165 show a loo-k.v.a. single-phase oil-cooled
transformer, and Figs. 166 to 168 a 5<D-k.v. a. three-phase
oil-cooled transformer. Both are of the core type with
the windings arranged concentrically.
The performance of these transformers is given in
the table on p. 308.
20 305
306 TRA NSFORMER S
^77777^777)^,.
iN.g 9 6 5 1FT. IN.1Z 9630 1FT.
FIG. 163. FIG. 165.
Brush loo-k.v.a. single-phase

oil-cooled transformer.
IN12 9630
FIG. 164.
EXAMPLES OF MODERN TRANSFORMERS 307
308 TRANSFORMERS
Type Single-phase Three -phase
Output, k.v.a. IOO 50
Frequency .
50 50
Primary voltage 2000 6500
Secondary voltage 4OO 440
Primary connection ... Mesh
Secondary connection Star with neutral
brought out
Efficiency per cent.

Full load 98-1 97*4
| load . 8-2 97'4
J load .
97-1
J load .
95*5
Full load, drop per cent,

for cos <f i .
1-65
for cos If = o'8 3'
Temperature rise of oil 4 6C. 43
FIG. 167. Plan to Fig. 166.
The Oerlikon Company. The design shown in Figs.

169 to 171 is of special interest, as it represents one of
the largest transformers made up to the present. It is a
three-phase core type 35oo-k.v.a. transformer cooled

by forced blast. The blast is supplied from below
through a ventilating culvert in the floor of the trans-
former house and ascends through the winding and
carcase, the latter being for this purpose subdivided
into thin packets. The transformer is provided with
6 ''S
6
3io TRANSFORMERS
wheels, as shown in Fig. 171, which run on rails laid
over the ventilating duct in the floor. By this arrange-
ment a transformer can easily be withdrawn for overhaul
or replaced by a spare transformer in case of breakdown.
The primary voltage is 27,000 volt, the secondary
3000 volt, the frequency is 50. The windings are
concentric, the high-pressure coil being outside and
completely shielded off from the low-pressure coil by a
micanite partition 7 mm. thick and extending top and
bottom some 15 cm. beyond the ends of the coils.
Each primary has 22 subdivisions, and each secondary
FIG. 170. Plan to Fig. 169.
12 subdivisions. The primaries are star connected, the

secondaries mesh connected.
The winding data are given in the following table
Oerlikon 3500- k.v. a. Transformer.
Primary outside star connected. Secondary inside mesh connected.

Turns per coil . .
344 Turns per coil . . 68
2 subdivisions of 12 turns 2 subdivisions of 4 turns
20 subdivisions of 16 turns 10 subdivisions of 6 turns
Strip . .20 x i '5 mm. Strip . .
44><3'5
Length .
2700 m.
.
Length . .
465 ni.
Weight . .
726 kg. Weight . .
645 kg.
The current density in both windings is 2*55 ampere per

sq. mm. The induction is 13,600, and the total flux
20*6 megalines. The carcase weighs 7 tons, and the
calculated iron loss with alloyed plates is 21 kw. The
iron loss is therefore only 0*6 per cent, of the output.
The calculated copper loss is 10*35 kw. in the three

primary and 9' 15 kw. in the three secondary coils. The
total losses at full load are 40*5 kw., or 1*14 per cent, of
the full- load output. This makes the efficiency at full
load nearly 99 per cent. The insulation has been
tested with 50,000 volt for the high-pressure, and 6000
volt for the low-pressure coils to earth during fifteen
minutes.
Messrs. Ferranti, Ltd. An air-cooled core type
transformer for moderate power is represented in Figs.
172 to 174. The core is of rectangular section, with the
end plates stepped so as to more nearly fill the rounded
space within the coils. Core and yoke are bolted up
with strong gun-metal flanking plates, those of the lower
yoke being provided with flanges for attachment to the
cast-iron base. For outside protection a non-perforated
steel shell is used, and ventilation is provided by holes in
the baseband cap as shown in Fig. 172. For a primary
pressure of 2000 volt at 50 frequency and a trans-
forming ratio of about 10: i the dimensions are as
follows
Fig. 175 shows the rise of temperature of the 30-

k.v.a. size,and Fig. 176 that of the 5O-k.v.a. size at
full load, whilst
Figs. 177 and 178 represent the regula-
tion. In each diagram two curves are shown, the
ordinates representing percentage drop at the secondary
terminals as a function of the power factor in the
i_
fe
secondary circuit, but with this difference, that the lower

curve refers to constant current and the upper to con-
Room Temperatur >
10.0 11.0 12.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Time in Hours
FIG. 175. Heating curve of Ferranti 3o-k.v.a. transformer.
stant power. In the latter case the current must

increase with decreasing power factor, which accounts
for the greater drop.
35
Room Temperature
9.0 10.0 11.0 12.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8-0
Time in Hours
FIG. 176. Heating curve of Ferranti 5O-k.v.a. transformer.
Fig. 179 shows the general design of single-phase

oil-cooled transformers for small and moderate out-
TRANSFORMERS
put. The transformers may be either placed on a level
floor or attached to a wall, for which purpose the case
1-0 0-95 0-9 0-85 0-8 0-75
FIG. 177. Regulation curve of Ferranti 3O-k.v.a. transformer.
is provided with lugs. The dimensions A to E in

inches, and the weight in cwt. are given in the following-
table.
The efficiency at full non-inductive load ranges from
1-0 0-95 0-9 0-85 0'8 0-75

> cos if
FIG. 178. Regulation curve of Ferranti 5o-k.v. a. transformer.
91*2 per cent, in the ^-k.v.a-. size to 97*9 per cent, in

the 5o-k.v.a. size; at quarter load from 777 to 97*1
per cent.
For the transformers here described iron and copper

losses are as under
The copper loss is given at full secondary current,

corresponding to the volt-ampere at which the trans-
former is rated, and after it has been at work sufficiently
long to have reached its final temperature.
TRANSFORMERS
179. Ferranti standard type

MFiG.
I oil-cooled transformer.
Fig. 1 80 shows the design of a

small transformer for
a very high primary pressure. To secure good insula-
tion the primary circuit is arranged in 8 distinct coils,
which may be more easily handled and tested before
being assembled. In the present case the transformers
are intended for three-phase io,ooo-volt circuits.
The dimensions for a 50- and 3O-kw. transformer, the
output being obtained at a power factor of 075, are as
follows
In star coupling each primary takes 5800 volt, or

725 volt on each individual coil. To protect the low-
pressure circuit an earthing shield is used.
A three-phase transformer, where the three phases
are combined in one apparatus, is shown in Figs. 181 and
182. It is a i2O-kw. transformer, star coupled for 5000
volt primary line pressure, and a secondary pressure,
which may be adjusted to either 370, 380, or 390 volt.
For this purpose tappings are taken on the secondary, and
by means of a three-pole switch the number of turns on
each limb may be changed from the normal of 49 to either
48 for the lower or 50 for the higher voltage. This is
shown diagrammatically in Fig. 183. The switch is seen
in Figs. 181 and 182 mounted on the top of the casing.
The core area is 350 sq. cm., the flux 4*45 megalines,
and the induction 12,700. The following table gives the
winding data
INS.1Z 9 6 3 3 FT.
FIG. 1 8 1. I20-kw. 5000 to 380 volt 4O-frequency three-phas e transformer

made by Messrs. Ferranti, Ltd.
320 TRANSFORMERS
SECONDARY TERMINALS
NEUTRAL
PRIMARY TERMINALS
FIG. 183. Diagram
of connections to
Fig. 181.
FIG. 182. End elevation to Fig. 181.

FIG. 184. 5oo-k.v.a. 3 1200/2200- volt transformer made by the Bullock El. Mfg. Co.
The Bîllock Electric Manufacturing Co. Figs. 1 84 to

188 are good examples of high-class American practice in
21
322 TRANSFORMERS
the design of transformers. Figs. 1 84 and 185 show a 500
k.v.a. oil-cooled shell-type transformer for 60 frequency
FIG. 185. 500-k.v.a. 3 1200/2200- volt transformer made by the Bullock

Electrical Manufacturing Co.
taking current at 31,200 volt, and delivering current at

2200 volt. On account of the high voltage, special
precaution has been taken to separate the coils by
insulating partitions, shown by thick lines, and the

terminals are also immersed in oil. The secondary
winding is in two
groups, so that the same type may be
,EP.
FIG. 186. Heating curves of 5oo-k.v. a. transformer.
coupled up for half the secondary pressure and double

current. The transformer is one of a group of three for
Load
FiG. 187. 500-k.v.a. 3 1 200/2200- volt transformer.
three-phase work, as is the usual practice in America, in

preference to building one three-phase transformer of
treble output. The use of three single transformers in
324 TRANSFORMERS
mesh connection has the advantage that a failure of one
transformer need not interrupt the three-phase service ;
the other two remaining at work are simply overloaded,

so as to do the work of three but as the time-constant of
;
large transformers is very great, no damage is done to

the overloaded transformers during the time required to
bring a spare transformer into service. This advantage
is, of course, lost if star coupling is adopted. The trans-
former here illustrated is, however, sufficiently well insu-
lated to be used in star-connection, when the line pressure
on the primary side is 54,000 volt. The oil is cooled by
FIG. 1 88. Plan to Fig. 189.
a cold-water worm placed in the upper part of the case,

and the terminals are brought through the cover by large
porcelain ferrules. The total weight of copper is 300 kg.
In Fig. 1 86 a heating test of this transformer is
recorded. To shorten the time, the test was started with
a 1 2 per cent, overload, and then the run was continued
with about full load. The final temperature-rise at
normal full load is 38 C. for the coils, and 35 C. for the
" "
carcase (marked case in the diagram). The great
difference of temperature between water inlet and water
outlet should be noted, as also the small difference
between the temperature of the carcase and the out-
flowing water. This indicates a very vigorous circulation

of the oil and efficient action of the worm.
r~
Fig. 187 shows the performance of this transformer

as regards efficiency and regulation.
Asmall transformer of the type frequently used in
326 TRANSFORMERS
America for private houses is shown in Figs. 188 to 190,
The secondary circuit is arranged in two groups, and
there are four secondary leads brought out of the case,
so that not only can the same type be used for full and
half voltage, but, by grouping the two windings in series
and joining the connecting joint to the neutral bar of the
distributing switchboard, a supply on the three-wire
system may be given. The dimensions refer to a 5-k.v.a.
transformer at 60 frequency for the voltage usual in a
house-to-house system of supply, namely, 1000 to 2000
volt on the primary and not over 200 volt on the
secondary.
The British Westinghouse Co., Ltd. Before entering
on a description of the various designs, it will be useful
to say a few words concerning the principles on which
this firm has standardised its transformers. Although
high efficiency is always desirable, there are cases where
it is especially important. Thus in a lighting transformer
at low loads is far more important than
high efficiency
in a power transformer, because the latter is not worked
for a very long time at light load, and power current is,
as a rule, cheaper than lighting current. On the other
hand, capital outlay for the very reason that power
current must be supplied cheaply, is an important matter
in power supply undertakings, and for these reasons it
may be good policy to sacrifice a little in efficiency if

thereby the fixed charges can be reduced. To satisfy
the various conditions of working the British Westing-
house standardise two types, one for high, the other for
medium efficiencies.
The relation between these may be seen from the
following table, which refers to the two types of 4O-k.v.a.
transformer
The high-efficiency transformer is, of course, more

costly and also heavier. In the 4o-k.v.a. size the
complete weight with tank and oil is 660 kg. for the
high-efficiency transformer, and 540 kg. for the medium-
efficiency transformer.
FIG. 191. FIG. 192.

4O-k.v.a. 2ooo/2OO-volt transformer made by the British Westinghouse Co., Ltd.
Another matter which requires attention when

standardising a line of transformers is the question of
heating. An actual working temperature up to 85 C.
may be permitted, but a knowledge of this limit alone
is not sufficient to determine the
cooling surfaces we ;
must also know the temperature of the room in which

the transformer will have to work and the character
of the load and the cooling conditions.
328 TRANSFORMERS
In large transformers this information is generally
available beforehand, and the designer can make his
calculations accordingly. Small transformers must, how-
ever, be made in quantities for stock, and the designer
cannot know in what localities and under what conditions
they will have to work. He must therefore design the
small transformer for a lesser temperature rise than might
be allowed in a medium-size transformer. If he designs
o
a large transformer with water-cooling he can calculate
still more closely, that is to say, allow a
greater tempera-
ture rise, because all the conditions are of such a nature
that no great departures from known averages are likely
to occur.
The smaller transformers are rated for a temperature
rise of40 C. over the surrounding air. An air tempera-
ture of 45 C., which might occasionally be reached in
badly-ventilated transformer chambers or pillars, would
then not cause damage. Larger transformers are rated
for a temperature rise of 45 C. as their load conditions
can more accurately be predetermined, and more care is
taken in placing them in properly-ventilated chambers.
Oil-insulated water-cooled transformers are normally
rated for 50 C. rise over the temperature of the
entering water as this will, under ordinary conditions,
;
not exceed 30 C., this rating should be safe.

Figs. 191 and 192 show a 4o-k.v.a. single-phase
oil-cooled shell transformer of the high-efficiency type.
Jt is designed for the standard frequency of 50 and a
primary voltage of 2000. The carcase weighs 220 kg.,

and the winding 125 kg. The coil area is 350 sq. cm.,
and the induction 6500 lines per sq. cm., giving a flux of
2*27 megalines. The window area is 216 sq. cm. The
quantity of oil required is 34 gallons, or 138 kg. No
cooling-worm is used, sufficient surface for air cooling
being provided by the corrugated sheet-iron case. In
smaller sizes up to 10 k.v.a. the case is of cast-iron,
and provided with lugs for fixing to hanger irons for
attachment to a wall.
In Fig. 193 is shown one of a set of three trans-
formers supplying current to a six-phase 1000 kw.
rotary converter. For this purpose each secondary
6
&
I
s
I
330 TRANSFORMERS
circuit must be in two parts, so that four secondary

terminals are required. To provide extra cooling
surface the heads of the coils are splayed out, the
carcase is built up of narrow packets separated by
circulation ducts, and the corrugations of the case are
very deep.
Fig. 194 shows a three-phase core-type oil-cooled
transformer made for the Castner-Kellner Alkali Works,
Newcastle-on-Tyne. The capacity is 1200 k.v.a. at
5750 volt on the primary side, and 40 frequency. The
voltage on the secondary side is 175, but by altering the
method of connecting up, the same winding may be used
for different voltages.
Fig"- J shows three different arrangements of
95
terminals, marked A, B, and C. The winding is sub-
divided into discs as explained in Chapter VIII. For
sizes larger than 1200 k.v.a. the cooling by a corru-
gated case is supplemented by a cold-water worm.

For pressures up to 20,000 volt cooling by artificial
blast without oil may be used. Fig. 196 shows a trans-
former of this type for 11,000 volt on the primary and
400 volt on the secondary side. The output is 550
k.v.a. at 33 frequency. A number of these trans-
formers have been installed at the Baker Street sub-
station of the Metropolitan Railway. They serve for
supplying current to rotary converters. The trans-
formers are kept cool by a strong air-blast sent in
through a duct in the floor. Gratings, the opening of
which may be separately regulated, are provided to
suitably sub-divide and direct the stream of air through
carcase and winding, and in order that the attendant
may see at a glance whether the ventilation is in order
a little wind-mill, indicated at a in Fig. 196, is fitted to
each transformer over the top grating.
The advantages of cooling by air-blast over cooling
by oil are greater cleanliness and convenience in case of
repair, but care must be taken to have the air free from
soot or dust. On the other hand, oil (unless it must be
supplemented by a water-worm) requires no accessory
apparatus, such as a fan, and no expenditure of power.
It has also the advantage of increasing the time constant.
332 TRANSFORMERS
ft
ft ff
t t r t
FIG. 196. 55o-k.v.a. 000/400- volt

1 1,
air-blast transformer made bythe

British Westinghouse Co., Ltd.
r
V F
.A.
J.
FIG. 197. Lamp transformer made by the British

Westinghouse Co., Ltd. *
334 TRANSFORMERS
In transformers connected to overhead lines which are

liable to atmospheric disturbances it is also sometimes
claimed for oil that it acts as a self-healing insulation,
in the solid dielectric
closing the small hole punctured
when a static discharge occurs. Whatever may be the
value of this view, the fact remains that experience
has led most makers to use oil for high-pressure
transformers.
With the advent of the wire lamp a demand has
arisen for very small transformers for single lamps or
groups of a few lamps. Wire lamps are at present
made for moderate voltages, 25 to no, but many of the
existing supply systems exceed these
limits, so that a transformer becomes
necessary to burn the lamps inde-
pendently. It is, of course, possible to
use one transformer for the whole of
the lamps, but then the iron loss would
reduce the yearly efficiency of the in-
and it is also questionable
stallations,
whether the existing leads would be
able to carry the larger currents with
a moderate ohmic drop. When this is
not the case, small transformers of the
type shown in Figs. 197 and 198 may be FIG. 199. Lamp
used. The switch must, of course, be put transformer on wall
bracket.
on the primary side of the transformer,
the latter with its wire lamp taking simply
the place of the previous carbon lamp. Fig. 199 shows
a lamp-transformer attached to a wall bracket.
Messrs. Brown, Boveri & Co. The largest trans-
former ever made is probably that supplied by Messrs.
Brown, Boveri & Co. to the Betznau Power Station
(Switzerland). Its capacity is 4600 k.v.a., and it serves
to transform up a machine current of 8000 volt to a
line pressure of 27,000 volt. It is a three-phase core
type star-coupled transformer. Its primary phase volt-
age is 4650, and its secondary 15,620 volt. Figs. 200

to 202 show the construction to a scale of i 20*6. :
Fig. 203 shows the method of winding, and Fig. 204 is a

general view of the transformer by the side of its tank.
336 TRANSFORMERS
FIG. 200. 46oo-k;V.a. three-phase transformer made by Messrs. Brown, Boveri & Co.
FIG. 201. 46oo-k.v.a. transformer made by Messrs. Brown, Boveri & Co.
22
338 TRANSFORMERS
The following particulars will be of interest
Frequency * . . . "r ..-.." 50

Core area, sq. cm. , . . . .
2,380
Induction lines per sq. cm. . . . .
11,300
Flux in megalines . ; .. .
26*9 .
Number of turns per phase in primary 78 .
Number of turns per phase in secondary 267 .
Area of primary conductor in sq. mm. 178 '".
Area of secondary conductor in sq. mm. 44 .
Current density in primary, ampere per sq. mm. 1*85

Current density in secondary, ampere per sq. mm. 2*14
Resistance of primary per phase in ohm 0*030 .
Resistance of secondary per phase in ohm 0^440 .
Iron loss measured in watt 40,500 . ... .
Total copper loss measured in watt .. .-."

25,500
Efficiency at full load, per cent. . . . .
98*6
Maximum drop, per cent. . . . . 1*6
FIG. 202. 46cx)-k.v.a. three-phase transformer made by Messrs. Brown, Boveri & Co.
The total weight of iron is 12 tons, so that with full

non-inductive load the weight of iron is only 2*6 kg.
per kw. The copper weight is 2400 kg. or 0*524 kg.
per kw. The total weight of active material is only
ABOVE
The 40 top turns

Presspahn separated by
distance, pieces 1mm. presspahn
BELOW
FIG. 203. Method of separating the primary and
secondary coils.
3*124 kg. per kw. It will be obvious that with SO
large a reduction in the weight of active material per kw.

a very perfect cooling device becomes necessary, and for
this reason the usual plain cooling pipe has been replaced
340 TRANSFORMERS
ribs in the manner
by an elaborate cooler provided with
of a radiator as shown in Figs. 201 and 202. The
terminals are taken through long porcelain tubes, the
lower ends of which are well below the oil level.
FIG. 204. 46oo-k.v.a. three-phase transformer

made by Messrs. Brown,
Boveri & Co.
Messrs. Siemens Schuckert Werke. Figs. 205 and 206

illustrate an air-cooled transformer of the shell type for
50 frequency, 3000 to 220 volt, and 6'86 to 92 ampere.

The dimensions inscribed are mm. The carcase is built
up of 0*3 mm. alloyed sheets insulated to 0^33 mm.,
and has butt-joints. The iron weight is 165 kg. and
the measured iron loss at the induction of 9700 is
This
335 watt or at tne rate f 2 '3 watt P
>
er kg-
agrees within a few per cent, with

the curve given on
page 26.
-325
700
FIG. 205. 2O-k.v.a. 3OOO/22O-volt transformer made by Siemens Schuckert Werke.

342 TRANSFORMERS
The particulars of this transformer are as follows
Core area in sq. cm. . . ; . .
300
Flux in megalines . . . . .
2-9
Number of turns in primary . . V .
464
Number of turns in secondary . i .
34
Section of primary sq. mm. . . ".-'.
'
6
Section of secondary sq. mm. .. ..
. . 80
Resistance of primary hot in ohm . .
2*25
Resistance of secondary hot in ohm , / 0*0126
Ohmic drop in per cent. ., . , .
1*03
Inductive drop in per cent. . . . ; .
0*90
Maximum possible drop in per cent. .
j
1*37
Final temperature rise in iron, degree C. .
/ 55
Final temperature rise in copper, degree C. .
55
FIG. 206. Side view of Fig. 206.

FIG. 207. 14-k.v.a. three-phase transformer made by Siemens Schuckert Werke.
FIG. 208. Plan of Fig. 207.

344 TRANSFORMERS
The primary winding is subdivided into 4 coils of
116 turns. Each coil consists of 58 layers of 2 turns,
the section of copper being 1*3 mm. by 47 mm. wide.
The secondary winding is also subdivided into 4 coils, two
of which have each 8 turns, and the other two 9 turns.
The copper section is 10 mm. by 8 mm. wide. It is
FIG. 209. End view of Fig. 208.
made up of two strips, 5 by 8 mm. wound on together.

The weights are
Carcase . ... . ,
.165 kg.
Primary copper . V _, 37*3 ,,
Secondary copper . . -.; 36*1 ,,
Total active material . .

238*4 ,,
or at the rate of 11*92 kg. per kw. at full non-inductive

*
load. The efficiency is
97^ /o*
A three-phase transformer, also of the shell type,

with butt joints, is illustrated in Figs. 207 to 209. The
dimensions are mm. It is a I4~k.v.a. oil-cooled trans-
FIG. 210. looo-k.v.a. three-phase transformer made by Siemens Schuckert Werke.
former for 50 frequency, 3000 to 200 volt, and 2*8 to 37*4

ampere. Both circuits are star coupled. The windows
are 87 mm. square, and each contains 518 primary and
346 TRANSFORMERS
38 secondary wires. Each primary is arranged in two

coils, and each of these has 28 layers of 9 turns and one
FIG. 2ii. Side view of Fig. 210.
layer of 7 turns. The two coils are placed side by side,

and outside of them are placed the two secondary coils,
each containing 19 turns. The primary conductor is
round wire of 1*8 mm., the secondary is strip 3*4 mm.
by 9*4 mm.wide. The resistance cold is, for the primary

4*2, and fpr the secondary 0*0245 ohm. per phase. The
weight of carcase is 149 kg., and the measured iron
loss at an induction of 9600 is 276 watt or r86 watt per
kg., which corresponds exactly with the Author's tests

recorded in Fig. n. The primary copper weighs
38*8 kg., and the secondary 36*4 kg. The total copper
<$ j._ - .-.-.-.-

p.
- _
*._._ _.
1"- , -* -v ?n -
t.-f.^'l: fflr.f--tr.--tr i.-^
FIG. 212. Plan of Fig. 210.
loss at full load when

the transformer is hot is 250 watt.
The final temperature rise, both in iron and copper, is
55 C. The inductive drop is 1*8 % the ohmic drop is ;
i-5 %
A
.
three-phase transformer for 1000 k.v.a. at 50

frequency is illustrated in Figs. 210 to 212. It is also
of the shell type with butt joints, but with special
cooling appliances. The transformation ratio is 5000
to 10,000 volt, the iron loss amounts to 8*3 kw., and the
343 TRANSFORMERS
FIG. 213. 225O-k.v.a. three-phase transformer made by Siemens

Schuckert Werke.
copper loss to 6*82 kw., corresponding to an efficiency of

98*5 % at full non-inductive load. Oil is used for cooling,
and to facilitate the flow of heat from oil to case the
latter is
provided with internal ribs, as shown in
Fig. 212. There is no cold-water worm employed,

but to cool the case its external surface is played on
by water. For this purpose a water-supply pipe is laid
round the top of the case on the outside. This pipe is
provided with holes through which the water is squirted
against the outer surface of the case, and, flowing
o D D
FIG. 214. Carcase of three-phase Siemens Schuckert

225<D-k.v.a. transformer showing oil channels.
down, keeps itcool. The water is carried away by

the drain-pipe shown at the bottom. To keep this
irrigation system in good working order it is protected
by an outer envelope of canvas stretched over a re-
movable frame. With a water-supply of 1*5 Cbm. per
hour the final temperature rise of the uppermost layer
of oil is 33 C. over the temperature of the entering
water.
Fig. 213 is a general view of a larger transformer of
the same type. Its capacity at 50 frequency is 2250
k.v.a., the transforming ratio being 5000 to 50,000
volt. The efficiency at full non-inductive load is 98 %,
and the short-circuit test pressure is 3*5 % of the
primary voltage. Both ends of each of the three high-
350 TRANSFORMERS
pressure windings are taken to terminals, so that the

three phases may be either connected in star for 50,000
volt, or in mesh for 29,000 volt. The low-pressure
winding remains in either case star connected. The
transformer requires 1.7 Cbm. cooling water per hour,
which is passed through a copper worm in the upper
part of the case (not shown in Fig. 213),
and the final
temperature rise of the top layer of oil over the tempera-
ture of the entering water is 35 C.
In order to provide the additional cooling surface
necessary with an increase of linear dimensions, ducts
are left between the individual sections of the winding
and the insulating parti-
tions separating them,
and furthermore, the
carcase is
provided with
oil-channels, as shown in
Fig. 214.
Messrs. Johnson and
Phillips, Ltd. The fol-
lowing particulars refer

to a 2^-k.v.a. single-
phase transformer of the
core type for 50 fre-
quency, and a trans-
FIG. 215. Auto-transformer for wire lamps forming ratio of 2500

made by Messrs. Johnson & Phillips, Ltd. to 200 volt, and 1*065
to 1 2
-5 ampere. The
two cores and lower yoke are | shaped stampings,
|
and the top yoke is laid across with butt-joints. Width

of plates, 41-3 mm. (if");
length of cores, 152 mm. (6") ;
depth of cores, 92 mm. (3-f") width of window, 70 mm.

;
(2f"). Each core receives two high-pressure and two

low-pressure coils of disc form. The weight of the car-
case is 1 6*3 kg., the total copper weight 9*2 kg. The
primary has 2750 turns of 0762 mm. wire covered to
0*915 mm. the secondary 230 turns of 2*64 mm. wire
;
covered to 2*94 mm. The transformer is placed in a

cast-iron case and oil-cooled, the filling requiring four
gallons.
After i6|r hours' run at full load the tempera-
ture rise of the case was 34*5 C., and of the oil
O /"
47*5 C.
FIG. 216. Carcase of loo-k.v.a. transformer

made by Messrs. Johnson & Phillips, Ltd.
The exciting current and core losses at various

voltages applied to the secondary were as follows
Volt. Ampere. Watt.
90 0*2 16-8
100 0*3 20*4
I IO 0-36 24*0
130 0*46 28-8
150 O'5O 36-8
170 0-68 44*o
190 0-82 52*0
195 0-85 53-6
Fig. 215 is an external view of a small auto-trans-

352 TRANSFORMERS
FIG. 217. Carcase of loo-k.v.a. transformer made by Messrs.

Johnson & Phillips, Ltd.
former for halving the pressure in installations where

wire lamps are used.
FIG. 218. Shows arrangement and mounting ot coils in loo-k.v.a. transformer.
The of a single phase ico-k.v.a. 50-

particulars
-
frequency core-type oil-cooled transformer, illustrated in

Figs. 216 and 217, are as follows
Primary 6000 volt, 17 ampere

Secondary 2200 45-5 ,,
Weight of carcase, 460 kg. total copper weight,

;
206 kg. The cores are 140 mm. wide, and 303 mm.
deep; their length is 419 mm. The nett core area is
350 sq. cm., and the flux 3-02 megalines. The induction
is 8650. The primary has 810, and the secondary 300
turns. The coils are concentric, the secondary being
23
354 TRANSFORMERS
inside, but the primary on each limb is sub-divided into
five sections to facilitate insulation and provide for the
circulation of oil. In both windings strip and not round
wire is used, the section being 2-03x6-1 mm. for the
and 3*43 x io'i6 mm. for the secondary. When
primary,
excited on the low-pressure side the no-load current at
2200 volt is 1*28 and the iron loss is 1018 watt.
ampere,
FIG. 219. 6oo-k.v.a. three-phase transformer, made by

Messrs. Johnson & Phillips, Ltd.
To give the oil free access to all parts of the carcase

the cast-iron frame is made
form of an open grid,
in the
as will be seen from Figs. 216 and 217.
the subdivided
Fig. 218 shows the method of fitting
coils over the cores.
Fig. 219 is a general view of a three-phase 600-
k.v.a. transformer of similar construction.
APPENDIX I
Dimensions ofphysical quantities in the electro-magnetic system
Length L
Time . T
Mass . M
Surface . L2
Volume L3
Angle . a number
Angular velocity
Angular acceleration .
Linear velocity . LT- 1
Linear acceleration LT- 2

Force . LMT- 2
Energy . L MT -
2 2
Power . L MT -
2 3
Moment of inertia L 2
M
Magnetic mass .
Magnetic flux .
Magnetic induction L~
Magnetic force . L-
Permeability a number
Magnetic reluctance L- 1
Electric resistance LT- 1
Reactance, impedance . LT- 1
- 2
Electromotive force
Current
Electric quantity (charge)
Capacity L -1 T 2
Inductance L
Frequency a number
35.
APPENDIX II
used in this book

Unless otherwise stated in the text, the symbols
have the meaning as under
A . . .
= area through which a magnetic flux passes.
B . . .
= magnetic induction.
C . .
= electrostatic capacity.
E . . .
= electromotive force (generally crest value).
G . . * weight.
H . . .
= magnetic force.
I . . .
= current (generally crest value).
L , . .
= inductance (generally in henry).
N . V .
= flux.
P . r, .
= power.
Q .
"
. .
= quantity of electricity.
R . , .
= resistance.
S . . .
= surface.
T
'
* . -
= temperature rise.
X . . .
= exciting force in ampere turns.
e . . .
= electromotive force (generally effective value).
/ .
.
. .
= current (generally effective value).
n ; ". '.
= number of turns.
q / .
. .
= cross-section of a conductor in sq. mm.
r '.- .
'
.
= resistance.
/ . ,; .
= time.
a '. . :
.
= an angle.
or . . . .
= a phase difference.
i//
< . . .
= a magnetic flux.
o- I> .
. . .
= specific cooling surface.
w . . .
= an angular speed.
v . . .
= frequency, complete periods per second.
356
APPENDIX III
FORMULA USED
357
358 TRANSFORMERS
APPENDIX III 359
36o TRANSFORMERS
INDEX
Ageing, 23. Constant transformer, 196.
Alloyed iron, 22. Construction of carcase, 41.
plates, 22. Cooling ducts, 72.
Alternating current, 12, 88. medium, 52.
Amperemeter method, 223. and heating, theory of, 59.
Apparent power, 107. Copper, best distribution of, 116.
Arnold's rule, 121. Core transformer, 33, ill.
Aron, 215 Correction for eddy currents, 174.
Arrangement of coils, 6. Cotton covering, thickness of, 132.
Autotransformer, 277. Crompton, 43.
Ayrton and Perry, 219. Currents, alternating, 12, 88.
combination of, 79
Balancing transformer, 275, 277. constant, 195.
Ballistic constant, 242, 243, 244. correction for eddy, 174.
method, 239. exciting, 105, 109.
Barlow, 58. idle, 102, 106.
Behn-Eschenburg, 216. irregular, 95.
Berry, 274. no load, 102.
Best distribution of copper, 116.
distribution of losses, 119. Dangerous condition, 294.
Booster, 283. Demand factor, 273.
Breakdown of insulation in large Diversity factor, 271.
network, 301. Dobrowolsky, 231.
Brown-Boveri, 336. Drop, 156.
Brush Electrical Engineering Com- calculation of inductive, 176.
pany, 305. graphic determination of, 170,
Bullock Electrical Manufacturing 174.
Company, 322. influence of frequency on, 183.
Ducts, cooling, 72.
C. values of, 57, 58. Dynamometer, 198
Carcase, construction of, 41. Dynamometric wattmeter, 201.
proportions of, 46.
Cardew, 266. Economic importance of small
Characteristic, 98, 104. losses, 128.
curves of temperature, 296. Eddy currents, correction for, 174.
Choking coil, 105. Effective values, 12, 13, 14.
Coales, 257. Efficiency, weight and cost, 137.
Coils, arrangement of, 6. yearly, 126.
choking, 105. Energy stored magnetically, 98, 100.
equivalent, 186, 292. Enlarging a design, 138.
grouping of, 151. Epstein, 22, 233.
Combination of currents, 79. Equivalent coils, 186, 292.
electromotive forces, 81. Ewing, 237.
Commercial value of different types, Examples of transformers by
127. British Westinghouse Company
Constant current, variable power 327-333-
factor, 195. Brown-Boveri, 336-338.
362 INDEX
Examples of transformers by- Lighting transformer, 125.
Brush Electrical Engineering Load factor, 273.
Company, 305-307. intermittent, 64.
Bullock Electrical Manufacturing Logarithmic decrement, 246.
Company, 322-326. Losses, 15, 119, 128, 273.
Ferranti, Limited, 311-318.
Johnson and Phillips, 350-352. Magnetic leakage, 3, 5, 149.
Oerlikon Company, 308-310. Mordey, 44.
Siemens Schuckert Werke, 340- Morris and Lister, 256.
349-
Exciting current, shape of, 105, 109. Natural frequency, 87.
No load current, 102.
Ferranti, 43, 267, 311.
Figure of loss, 26. Oerlikon Company, 308.
Fill factor, 135. Output, influence of frequency on,
Fleming, 223. 185.
Fluxometer, 246.
Form factor, 31. Perry, see Ayrton and Perry, 219.
Francke, 217. Plates, alloyed, 22.
Frequency influences output, 185. thickness of, 24.
natural, 87. Power, apparent, 107.
Fundamental equations, 10, 14, factor, 147.
transformer, 123.
Goerges, 265. carried by current of irregular
Grassot, 246. form, 205.
Grouping of coils, 151. in three-phase circuits, 215, 217,
219.
Heat conductivity of stampings, 58. of an alternating current, 12, 88.
Heating, 48. of an irregular current, 95.
and cooling, theory of, 59. Pressure,- rise of, 292, 303.
Hedgehog type, 33. Proportions of carcase, 46.
Henry, 83.
Higher harmonics, 87, 94. Reactance, 84.
Hydraulic analogy, 4. Reducing iron losses, 273.
Hysteretic loop, 101. Resistance, approximate formulas for,
loss, Steinmetz formula, 17. 133-
Rise of pressure through resonance,
Idle current, 102, 106. 292, 303.
Impedance, 8,5.
Inductive type of wattmeter, 211. Safety appliances, 262.
Influence of E.M.F. curve on Scott'smethod, 252.
hysteretic loss, 27. system of changing two-phase to
higher harmonics, 87. three-phase, 288.
Influence of linear dimensions, 67. Self-induction and capacity, 82.
power factor or excess pressure, 299 of a transformer, 188.
shape, 31. Series working, 280.
Insulation, breakdown of, 301. Shell transformers, 33.
Intermittent load, 64. design of, 134.
Iron, alloyed, 22. Siemens Schuckert Werke, 340.
testing sheet, 230. Stalloy, 22.
weight of, 74. Stampings, heat conductivity of, 58.
Steinmetz, 17.
Johnson and Phillips, 350-355. Sub-station and house transformers,
268.
Kapp, 43, 232. Sumpner's wattmeter, 214.
method, 257.
Temperature coefficient, 57, 58.
Leakage flux, 6. rise, 55.
magnetic, 3, 5. Testing sheet-iron, 230.
INDEX 363
Testing, special implements for, 231, Voltmeter method, 219.

232, 237.
transformers, 226.
Theory of heating and cooling, Wattmeter, dynamometric type, 201
59.
Thickness of plates, 24. inductive type, 211.
Three-amperemeter method, 223. Sumpner's, 214.

-phase types, 40. Weight of iron, 74.
-voltmeter method, 219. Westinghouse, 42, 327.
-wire system, 275. Working condition graphically re-
Time constant, 61, 1 18. presented, 190.

diagram, 153.
Vector diagram, 142.
Vectors, use of, 76. Yearly efficiency, 126.
THE END
RICHARD CLAY & SONS, LIMITED,
BUNGAY, SUFFOLK.
( 4 )
Kapp's Electric Transmission of Maxim's Artificial and Natural

Energy, los. 6d. Flight, 5j, net.
Transformers, los. 6d. net. May's Ballooning, 2s. 6d.
Kingsley's Roses and Rose Growing, Maycock's Electricity and Magnetism,
6s. net. 2s. 6d.
Lamps, Electric, Maycock, 6s. Maycock's Electric Lighting and

Land Measuring Tables, Cullyer,2.r.6;/. Power Distribution, vol. i. 6s. net. ;
Surveying, Middleton, $s. vol. ii. 7-c. 6d.
Surveying and Levelling, Alternating Current Circuit and

Walmisley, 6s. net. Motor, 4^. 6d. net.
Field Work and Instru- Electric Wiring, Fittings,
ments, Walmisley, $s. Switches, and Lamps, 6s.
Landolt's Optical Activity and Chemi- Details Forms, 2s. 6d. net.
cal Composition, 45. (id. Electric Wiring Tables, y. 6d.
Lathes, Horner, 6^. Mazzotto's Wireless Telegraphy,
Leather Work, Leland, $s. 35-.
6d. net.
Leland's Wood-carving, 5-f. Measurements, Electrical, Crapper.
Metal Work, 5.$-. Electrical Arithmetic, Whittaker,
Leather Work, 5-r. is. net.
Drawing and Designing, 2s. Houston, 4.T. 6d. net.
Practical Education, 6s. Mechanical Engineer's Pocket-Book,
Lens Work, Orford, 3*. [net. Bjorling, 35-. 6d. net.
Lenses, Photographic, Taylor, $s. 6d. Tables, Foden, is. 6d.
Light, Heat and Sound, Ashworth, Refrigeration, Williams, los. 6d.
2s. net. Metal Turning, Horner, 35. 6d.
Lightning Conductors, Lodge, 15^. Work, Atkins, 6s. net.
Locomotive Engineer, McDonnell, is. Work, Leland, 5-r.
Locomotives, Cooke, 7s. 6d. . .
Miller, y. 6d.
Lodge's Lightning Conductors, 15.$-. Metric System, Wagstaff, is. 6d.
Logarithms for Beginners, Pickworth, Perkin, is. 6d. net.
is. Middleton's Surveying and Surveying
Loppe and Bouquet's Alternate Instruments, 5J-.
Currents in Practice, 6s. Mill Work, Sutcliffe, ioj. 6d.
McDonnell, How to become a Loco. Miller's Metal Work, 3^. 6d.
Engineer, is. Mine, Coal, Boyd, 3*. 6d.
Maginnis' Atlantic Ferry, 2s. 6d. Mineralogy, Hatch, 2s. 6d.
Magnetism and Electricity, Ashworth, Mining, Coal, Burns and Kerr, 10
2s. 6d. Parts, 2s. each net.
Bottone, 2s. >d. net. Model Engines, Alexander, 5^. net.
Houston, 2s. 6d. net. Moedebeck's Aeronautics, los. 6d. net.
Maycock, 2s. 6d. Motor Cars, Farman, 5^.
Manual Instruction, Woodwork, Ignition Methods, Hibbert,
Barter, 6s. is. 6d. net.
Drawing, Barter, 3^. 6d. Motors, Commutator, Punga, 4^. 6d. net.
Metal Work, Miller, $s.6d. Electric, Bottone, 3^.
Manures, Griffiths, fs. 6d. Hobart, 125. 6d. net.
Marine Engineers, Advice to, Roberts, Maycock, 4.$-.
6d. net.
31. net. Hydraulic, Bodmer,
Drawing and Designing, Moving Loads on Railway Bridges,
Roberts, 6s. Bamford, 4*. 6d. net.
Electric Lighting, Walker, Municipal Electricity Supply, Gib-
Jji. bings, 6s.
Marine Steam Turbine, Sothern, 65-. Nadieine, Drainage, is.
net. Neumann's Electrolytic Analysis, 6s.
Massee's The
Plant World, 2s. 6d. Optical Activity and Chemical Com-
Mathematical Tables, Hutton, I2s. position, Landolt, 4^. 6d.
Mathematics, Elementary, Hatton, Instruments, Orford, 2s. 6d.
2s. 6d. Optics, Geometrical, Blakesley, 2s.6d.
UNIVERSITY OI
Optics of P Rudorf's Periodic Classification and

net. Chemical Evolution, 4$. 6d.
Orford's Lens Work, 3*. Russell's Electric Cables, los. 6d.
ratt,
THIS BOOK IS DUE ON THE LAST DATE jymbols, Greene,
STAMPED BELOW
,ight Installations,
AN INITIAL FINE OF 25 CENTS

of Accumulators,
WILL BE ASSESSED FOR FAILURE TO RETURN
THIS BOOK ON THE DATE DUE. THE PENALTY
WILL INCREASE TO 5O CENTS ON THE FOURTH [.*?.;
Nadieine, is.
DAY AND TO $1.OO ON THE SEVENTH DAY Limbing, Sutcliffe,
OVERDUE.
1 Vocab., French-
et.
:al Work, Atkins,
FEB 27193 nutator Motors,
th, 2s.
ter, 6s.
?ards for Marine
net.
FEB 28 i i
Turbine, 6s. net.
ieat, Ashworth,
APR 17 1933
Neuman and
il edition, 2s. 6d.
[net.
8 1S34 Building Work,
&** icr,is. 6d.
looke, 75. 6d.

, Alexander, $s.
"ill Work, Sut-
Maginnis, zs.Gd.
neering, Stevens
let.
othern, 6.r. net.

^sis, Arnold &
net.
ctures, Twelve-
Ivetrees, 6^. net.
29 Twelvetrees,
> Steam Turbine

let.
rrents, $s.
;nts, 6s. net.
Farrow, $s. net.
Steel, Twelve-
LD 21-50m-l,'35Veying Instru-
Surveying and Levelling, Walmisley, Turner and Hobart's Insulation of
- 6s. net.
Field
Walmisley,
Work and
5-r. net.
Instruments,
Electric Machines, los. 6d. net.
Turning Metal, Horner, 31. 6d.
Twelvetrees' Structural Iron and Steel,
Sutclifte's Steam Power and Mill 6s. net.
- Work, ior. 6d.

Sanitary Fittings
$s. net. ..
"
and PJtembing,
*&
Concrete-Steel, 6s. net.
Concrete- Steel Buildings, icw.net
Vacuum Tubes, Experiments, Salo-
Switches, Electric, Maycock, 6s. mons, 2S.
Tables, Electric Wiring, Maycock, Vc/ ^tric Analysis, Coppock, 2s.
-
- 3*. 6d.
Mathematical, Hutton, 12s.
7 /Ps Metric System,
I Electricity in
is. 6d.
Homes and
- mann,
Mechanical, Foden, is. 6d.
Percentage, for Analysis, G*
3*. net.
shops, 5*. net.
.ctric
eers, 5*.
Lighting for Marine
Taylor's Optics of Photography, _ .id Hawkins' Dynamo, i$s.

net. sy's Field Work and Instru-
Teacher's Handbook jbf Metal 5-r. net.
Miller, 3j. 6d. id Surveying and Levelling,
Telegraphic Vocabulary
los. net. s Friction, its Reduction, 31.
- 3/.
Telegraphy, Herbert.
Wireless, Bot*
O .
:r's
.-Book,
electrical
Mechanical Engineer's
35. 6d. net.
Engineer's Pocket-
*
Telephone Sys* jok, 3^. 6d. net.
Office HP Arithmetic of Electrical Engin-
eering, is. net.
Williams' Mechanical Refrigeration,
in, los. 6d.
Wireless Telegraphy, Bottone, 31.
.lanual, Mazzotto, \s. 6d. net.
Wiring, Electric, Maylcock, 6s.
.s. 6d.
Traction"/ Electric, Rider, 105-. 6d.
net.
Transformers, Kapp, los. 6d. net.
Still, $s. net.
Trigonometry, Adams, 2s. 6d. net. Wood's Discount Tables, is.
Tubes and Pipes, Bjorling, 31. 6d. Woodwork, Barter, 6s.
net. Drawing, 3^. 6d.
Vacuum, Salomons, 2s. Woven Design, Colour in, Beaumont,
Turbines and Hydraulic Motors, 21S.
Bodmer, 15^. X Rays, Bottone, 3^. 6d. net.
Turbines, Steam, Stevens and Hobart, Yates' Textbook of Botany, zs. 6d. net.
2is. net. Yeaman and Gay's C. S. Electricity
Marine, Sothern, 6s. net. Supply, lew. 6d. net.
Complete Catalogue of Technical and Scientific Books, Post Free,
WHITTAKER & CO.
WHITE HART STREET, PATERNOSTER SQUARE, LONDON, E.C.

Transformers, Kapp, 1908 PDF

Uploaded by

Copyright:

Available Formats

Transformers, Kapp, 1908 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Transformers, Kapp, 1908 PDF

Uploaded by

Copyright:

Available Formats

A

Accur Voven Design,

Alten Motors and

Apot, nent Making,

Cabinet Making, Denning, 5$. Electric Ignition for Motor Vehicles,

Chemical Analysis, Coppock, 2s. Salomons, 7-r. 6d.

-- Composition, Landolt, ^s. 6d.

Chemist's Compendium, Thompson, 7*. 6d.

3-y. 6d. net. Measurements, Houston, 4-f. 6d.

Draughtsmen's Work, u. 6d. Electricity in Homes and Workshops,

Dynamo, Hawkins and Wallis, Houston, 2s. 6d. net.

- Salomons, 6s. net.

Engineer Fitting, Homer, 51. Hawkins' and Wallis's Dynamo, 15^.

English and American Lathes, Helical Gears, Horner, 5*.

Fitting, Engineer, Horner, 5^. Hobart and Ellis' Armature Con-

BARR, J. R. ... Direct Current Electrical Engineering. IDS. net.

GRAY, J Electrical Influence Machines. 53.

MAZZOTTO, D. . . Wireless Telegraphy. 33. 6d. net.

RUSSELL, S. A. . . Electric Light Cables and the Distribution of

j> Electric Lighting forMarine Engineers. 53.

WILLIAMS, H. . . Mechanical Refrigeration. IDS. 6d.

WHITTAKER & CO., 2 WHITE HART STREET, LONDON, E.G.

SINGLE AND MULTIPHASE CURRENTS

SECOND REVISED AND ENLARGED EDITION

WITH TWO HUNDRED AND NINETEEN ILLUSTRATIONS

WHITTAKER AND CO.

RICHARD CLAY & SON?, LIMITED,

years ago, enormous improvements have taken place in

the more general appreciation of oil as a cooling and

insulating medium, the advent of alloyed iron, and last,

but not least, the growing demand for large unit in

connection with power transmission. Whilst only a few

show how by a careful study of the thermic condition the

output of a transformer of given weight and cost may be

dealing with the testing of transformers and iron samples.

RICHARD CLAY & SON?, LIMITED,

years ago, enormous improvements have taken place in

the more general appreciation of oil as a cooling and

insulating medium, the advent of alloyed iron, and last,

but not least, the growing demand for large unit in

connection with power transmission. Whilst only a few

show how by a careful study of the thermic condition the

output of a transformer of given weight and cost may be

dealing with the testing of transformers and iron samples.

engineer need not trouble about the scientific principles

physicist. admitting the commercial value

thorough knowledge of their principles, and the

ability to do equal or better work by means of apparatus

space to a seemingly only theoretical subject.

to analytical methods, and where the latter were unavoid-

able only the simplest analysis and the elementary

Principleof Action Magnetic Leakage Arrangement of Coils

Losses in Transformers Alloyed Plates Determination of the most

Usual Types Construction of the Iron Part Proportions of the Iron

Design of a Shell Transformer Fill Factor Winding Efficiency,

Testing Transformers Testing Sheet-iron Special Implements by

Safety Appliances for Transformers Sub-station and House Trans-

The Transformer in Relation to its Circuits Equivalent Coils in

PRINCIPLE OF ACTION MAGNETIC LEAKAGE-