A METHOD FOR TESTING CURRENT TRANSFORMERS

By
Francis B. Silsbee

CONTENTS
Page
t

Introduction

317

General principles Null method

318 321 325

method

Experimental results

28

Summary

328

INTRODUCTION
Several precise laboratory methods are now available for the determination of the ratio and phase angle of current transformers. 1 These, however, all require a considerable amount of special

apparatus, such as carefully calibrated noninductive shunts and very sensitive alternating-current detectors, and are therefore not suited for use under shop or central-station conditions. The task of comparing the constants of one transformer with those of a second transformer taken as a standard is much less difficult. The standard transformer should, of course, have the same nominal ratio, and its constants should have been determined by one of the precise laboratory methods. A method for such a comparison of two voltage transformers has been published by Brooks, 2 and another method applicable to either
voltage or current transformers

The method developed in the first of these and will

by Agnew 3 and by Knopp. 4 this paper is somewhat analogous to be found rather more rapid than the

As

in the other

less sensitive
1

comparison methods the detector may be much than in the absolute methods, and it is therefore

Agnew and

30, p. 435, 466, 1909; E. T.

Fitch, this Bulletin, 6, p. 281, 1909; Electrical World, 54, p. 1042, 1909; E. Orlich, E. T. Z., Robinson, Trans. Am. Inst. Elec. Eng., 28, p. 1005, 1909; F. A. Laws, Electrical

World, 55,

p. 223, 1910;

Sharp and Crawford, Trans. Am.

Inst. Elec. Eng., 29, p. 1517, 1910;

Silsbee, Trans.
2, p. 263, 1914.
2
3 4

Am.

Inst. Elec. Eng., 31, p. 1635, 1912; Schering

and

Alberti,

Agnew and Archiv fur Elcktroteclmik,

H. B. Brooks, this Bulletin, 10, p. 419, 1914 (Scientific Paper No. 217); Electrical World, P. G. Agnew, this Bulletin, 11, p. 347, 1914 (Scientific Paper No. 233). O. A. Knopp, Electrical World, 67, p. 92, 1916.

317

3i8

Bulletin of the

Bureau

of

Standards

[Vol. 14

possible to use a

more rugged type

of instrument, such as a

com-

mercial wattmeter.
useful

Multiple-range transformers are particularly

show practically proportional ratios and identical phase angles on the various primary connections. Two or three such transformers, the ratio and phase angle of which have been accurately determined, suffice for testing
as

standards, since

a considerable range of transformers.

GENERAL PRINCIPLES

method is illustrated by Fig. i 5 and and the unknown transformer, respectively. The primary windings are connected in series and supplied with current from a suitable source. The secondary windings are also

The

principle of the

are the standard

Fig.

connected in series, with such polarity that both tend to send current in the same direction, and any desired impedance loads such as zs and z x complete the circuit. A suitable detector D is

then connected so as to bridge across between the transformers. It is evident that the current AI through the detector (which we shall assume for the present to have a negligible impedance) is necessarily equal to the vector difference of the secondary currents Consequently, if the magnitude I s and / x of the transformers. and phase of AI are measured, the difference in performance of the two transformers can be computed. This measurement of AI may be made either directly by using as a detector one winding of a separately excited wattmeter, as is described below as the "Deflection method," or the measurement may be made indirectly by using an additional compensating circuit between .4 and B, as indicated by the dotted lines. By proper arrange-

ment

of the

impedances

all

of

AI may be made to flow

in this

susbee]

Testing Current Transformers

319

compensating circuit, and the current in the detector reduced to zero. This is described below as the "Null method." There are, of course, a great many other possible arrangements for measuring AI, but the two here described in detail will be found the most
convenient.
siderable extent

The performance of a current transformer depends to a conupon the impedance of the apparatus and wiring
circuit.

connected in the secondary

It is therefore of

what impedance is introduced into the circuit transformer by the arrangement of circuits shown in Fig. 1
transformer

importance of each
.

carrying a current I x at a terminal voltage Ex, and the equivalent impedance load is
is

Now

= 7- =
^x

f ^s

= z x j^x

(vectonally)

In the deflection method

Ed

(the voltage

between the terminals

E d = AIz d
And
hence

Z x = z x z d-f
* X

(vectorially)

and

similarly

Z = z + z d ~y~
s s

(vectorially)

It is therefore

evident that
is

when the secondary

currents are

nearly equal, so that -y-

very small, a detector of considerable

impedance and therefore of high-current

may

be used.

On

the other hand,

when

-j

is

large,

then the use of a detector of

may shift the load from the transformer of higher having the lower and thus make the difference in ratio appreciably less than the correct value. In the null method, on the other hand, E d is zero when a balance
high impedance
ratio to that
is

reached, so that

79 ^X ^X
of each transformer
is

and the performance

the same as

if

the

by a short-circuiting link across Although very evident theoretically, this fact was also tested experimentally, and the behavior of the transformer was found to be the same as when the detector was replaced by a
other transformer were replaced
C.

wire of negligible resistance.

320

Bulletin oj the

Bureau

oj

Standards

\Voi.i4

The principal limitation on the sensitivity attainable in any method of testing current transformers is due to the condition that the measuring circuit must not affect the performance of the transformer. This means that the measuring apparatus must not increase the equivalent connected secondary load by more than a certain resistance which we may denote by r. The gain in sensitivity of a comparison method, such as is described in this paper over an absolute method, is seen by comparing the power available for operating the detector in the two cases. In the usual absolute method the secondary current / is passed
through a noninductive resistance which is preferably equal to r. Ir is balanced by the drop due to the primary current flowing through a proportionately smaller resistance. With such an arrangement the impedance of the circuit external to the detector is approximately r, and for the maximum sensitivity the detector itself should also have this same impedance. If under these conditions the secondary current should differ by a small amount 81 from that required for an exact balance, then the unbalanced voltage acting in the detector circuit will be only blr. Since the resistance of the complete circuit is 2r, the current flow-

The voltage drop

big in the detector will be

and the voltage across the detector

will

be

X T

Consequently the power available

2 is
5 T

r.

On

the other hand, in the case of the deflection method outlined

in this paper,

the total difference AI between the two secondary

currents flows through the detector and produces a voltage at

terminals equal to AIz d

Here

z d is the

and

is

limited

by the

fact that this

impedance of the detector, voltage AIz d must not exceed

the permissible value Ir for the

largest value of

Conse-

we have

z &<iri Y

and

if

we use

a detector having this im-

pedance, the volt amperes available for a difference hi in the currents

_2 / 5/ ~Zjr.

Since the two transformers will usually differ by

only a few per cent the factor

and the detector

50 or 100 times less sensitive than in the precise laboratory methods. It is this relation which brings the method within the range of sensitivity obtainable with commercial pivoted
instruments.

may be

susbee\

Testing Current Transformers

321

NULL METHOD

of null

The connections of what is probably the most satisfactory form method based on the general arrangement outlined above
2.

ABCD is a slide wire of about 0.2 ohm total a mutual inductance of about 600 microhenry s, the primary of which can carry 5 amperes without excessive r is a resistance of about 30 ohms, preferably capable heating. of being set at several other values down to 2 ohms. As the value of r t appears in the denominator in the equations below it is convenient to have the total resistance between F and C, including the resistance of the secondary winding of M, an exact integer M, r t and r2 need be calibrated only to the for each setting of r v per cent accuracy which is desired in the difference of the ratios; that is, to about 1 per cent. The detector shown in the figure is a separately excited electrodynamometer instrument. A commercial wattmeter of lowcurrent range may be adapted for this work by bringing out taps directly from the moving coil without using any of the series The moving coil is connected as shown and the curresistance. rent coil excited by its full rated current in either of two phases, which preferably are in quadrature, through the double-throw switch G. Any other form of alternating-current detector sensitive to 0.00005 ampere might be used. 5 As a precaution a 10-ampere ammeter should be connected in parallel with the detector, so that the transformers will not be damaged if they have inadvertently been connected in opposition instead of aiding. If on closing the circuit this ammeter shows
are given in Fig.
resistance.

is

no appreciable current the polarity of the transformers and the ammeter should then be disconnected.

is

correct,

The procedure is to adjust the position of the slider C and the value of the mutual inductance until no deflection is obtained on closing G in either direction. When a balance is thus obtained

all

of the differential current

A/

is

flowing through r 1} and also

the difference of potential between points

and

F is

zero.

Con-

C and F and

between

and

one another. From this relation the differences in the ratios and phase angles of the two transformers may be computed.

must be equal and

in phase with

for this

A new type of vibration galvanometer has been recently developed by Agnew which is very well suited work. A description of this sensitive yet rugged instrument is to be published shortly in this

Bulletin.

322
If

Bulletin oj the

Bureau

of

Standards

\v0i.j4

and R x are the ratios of the transformers 5 and X, respectively, and <x B and a x are the corresponding phase angles, then we have 6 for the case where the slider C is to the right of B (Fig. 2)
a

R
and
6

tan (a x

- a8) = 6 + ac
as follows:

The equations Riven below may be deduced

For the connections as drawn in Fig.

we have
and

A/=/-/ x
A!(.ri+juLi)+hu.\f=/ x r-:
laws,

(i)

(2)

by KirchhofT 's

where the currents are to be regarded

as vector quantities.

Hence, eliminating A/, we get

lAn+MLi + M}=U(n~r:+^Li)
T%

(3)

U
n
r= -^
T\

'

V n
,

Substituting
,

=
l

-"
/,

= ^^
T\

Tl

= l?pin

+ a -bc-V...+j{-b-ac-ob....)

(5)

This quotient must

now be determined
But
since

The

ratio of a current transformer

to that of the secondary.

we
A

terms of the current ratios and phase angles of the two transis simply the ratio of the magnitude of the primary current, are now regarding the currents as vector quantities, we must
as )
(6)

write for the ratios

R = ^(cos cti+j sin

i?i=-p (cos a x +>sin a x )

(7)

where the complex factors in the parentheses have been introduced to make the ratios themselves (#, and Rx) simple numbers instead of vector quantities. (The complex factor in each case merely turns the vector I p through the angle a, into coincidence with the secondary current.) It is to be noted that this assumes that the phase angle, a, is positive when the reversed secondary current leads the primary current. From (6) and (7) we have

R x (cos <zs+j sin _/g R (cos ai +> sin a x )~I x

(8)

Of rationalizing

^|cos(a 9 -a)+y

sm(aa-*)}=^
get

(9)

Equating

(5)

and

(9)

and separating the

real

and imaginary terms, we

*cos(a

Ra

+ a be

2
.

(11)

and

Ra

p >sina( x)=-b ac ab.

..

(12)

Solving these equations simultaneously,

we have
l/5

^
and
which are the formulas

= {(i+a-&c-62)2+(_6_ at o6)*} -i + fl-^.-6c

(13)

(14)

desired.

The deduction

for the case

when the

slide

is

to the

left of

(Fig. 2)

is

similar to the

above with

slight

differences in the second-order terms.

Silsbee]

Testing Current Transformers

323
as transformer

For the case where

is

on the same side of

X,

^=
tan (a x
-

a 4- a

be

-a )
s

=^b

ac ab
ajL t

where
^2
^1

and

oj=27tX frequency.
fi

= the

total resistance

between

C and

through

rt

and

(Fig. 2), in

ohms.

= the L = the
r2
x

resistance of the slide wire between

B and C, in ohms.

self -inductance of the secondary coil of the mutual inductance M, in henrys. = the value of the mutual inductance, in henrys. (This is to be taken as positive if a x is greater than a 9 as shown by the test described below.)

Ammeter

fhoaell

L*

&h>a&el

Fig. 2

324.

Bulletin of the

Bureau

of

Standards

[Vd. i 4

The second-order terms are usually very small and to an accuracy of a tenth per cent we have in either case
R* = Ra (i a) x = a* + 3438 b
if

the angles are expressed in minutes.

is obtained with the slider between B and D then (provided the second-order terms are less than a), the standard transformer is supplying the greater secondary current and consequently has the smaller ratio. The question which secondary current leads the primary current by the greater angle can best be determined once for all by adding some resistance This will always make 7 X lead to the load on transformer X. more than before and if the value of required for a balance is increased then I x was leading originally and and b are to be taken as positive with this connection of the mutual inductance. In the case when the slider C is to the right of B (Fig. 2) and the part A B of the slide wire is omitted, then the measuring circuits introduce no impedance whatever into the equivalent secondary circuit of transformer X. It is therefore possible by this arrangement to test a current transformer under the condition of zero external load. The load on transformer 5 or, in case C is to the left of B, on both transformers, is practically the same as the impedance of the slide wire and the primary winding of M. The slight change in this, due to the fact that r, in effect shunts r 2 is of the order ar 2 and its effect on the performance of either

the balance

(Fig. 2),

transformer
slider

is

entirely negligible.

It is evident that

of

if 5 always had a smaller ratio than the would always be on the standard side of B and the part AB the slide wire would be unnecessary. This arrangement is very

resistance of

away with the necessity of allowing for the making the total impedance load on transformer equal to the desired value. The ratio of the standard transformer could be reduced by omitting a few turns from the
desirable since
it

does
in

AB

secondary winding in the process of manufacture. In completed transformers of the hole type this can still be done by passing a few secondary turns through the hole in a direction to oppose the main secondary winding. Since at the point of balance no current flows through the moving coil of the detector the self -inductance of this coil has no The mutual inductance between the two effect on the results.

Silsbee]

Testing Current Transformers

325

coils of

the wattmeter, however, does produce an electromotive

F even when no current flows and This source of error can easily be eliminated by shifting the position of the control springs so that the normal zero of the instrument corresponds to the position of
force

between the points

and

therefore affects the setting.

zero mutual inductance. 7

the current in Phase I of the exciting circuit

is

in phase with

the current in the transformers and Phase II is 90 from this, then the slider C may be adjusted with G closed to the right, with G to the left, and the two settings will be independand It is not at all necessary, however, that these ent of each other. phase relations be exact and Phase II may be 6o from Phase I,

s-fv

1mm a
Load

i5pr^

PV

Am.

C^D
W7?.
I//TX

QBSl

haseX fi

I
Fig.

PMsel

a polyphase supply is not available, current for Phase I may be drawn through a lamp bank and Phase II from the same supply through a reactance coil or transformer winding. With this arrangement it is well not to work the iron of the core much above its normal rating or the harmonics in the
If

as in a 3-phase system.

exciting current will

become

excessive.

DEFLECTION METHOD
In cases where a variable mutual inductance and slide wire
are not available, but a polyphase supply
tion
7

is

at hand, the deflecin Fig. 3)

method

(the connections for

which are shown

may
it

This position can be determined by short-circuiting the moving;

better

by connecting
full

to a large inductance of low resistance, such as the 220-volt winding of a transformer)

and passing

rated

current through the fixed coil. Under these conditions the pointer deflects toward the position of sero mutual inductance, and by suitably moving the control springs the desired position can soon be located. An alternative method is to connect a telephone receiver to the moving coil and pass full-rated current through the fixed coil. The springs can then be adjusted until no sound is heard in the telephone.

59467 18

11

326

Bulletin 0} the

Bureau

of

Standards

iva.14

The detector in this method must be of the dynamometer type, and arranged so that one coil may be supplied from two circuits, one giving a current in phase with the current in the transformers and the other giving a current having a known
be used.
phase relation, preferably quadrature with the former. If Di and D q are the deflections in divisions, observed with the excitation in phase, and in quadrature, respectively, we have

R x=
R*
tan (a x <x 3 )

KDj
^3

K D3

approximately,

where
If

is

the wattmeter
coil,

the constant of the instrument in amperes per division. is excited by passing a current I through its curk K = IqRv

rent

where k is the constant and R Y is the resistance

in

watts per division for a given range

of the voltage circuit of the

same range.

the

w attmeter
r

is

excited

by applying a voltage

EQ to the voltage

circuit,

then

In case the second phase available as a source of excitation is not in quadrature with the first, but gives exciting current which lags behind the current in the transformers by an angle 6 and produces a corresponding deflection D d) then we have

R*_
R*~
, tan (,-,)
,

J+
,

KD
h

= K f PeDi
J7

(-^

cos 6 \

As before, the question which secondary current is the greater and which leads the primary current by a greater angle can best be answered by changing the load on one transformer and noting the effect on the deflections, remembering that an increase in secondary resistance makes the secondary current smaller and also makes it lead the primary current more.

Stisbee]

Testing Current Transformers

327

The principal limitation of this method is, as mentioned above, the effect of the impedance of the detector in shifting the load from one transformer to the other. The change in the equivalent
impedance load on either transformer due to
this is given

by

AI ZdK'ylDi y - zd AZ j j
Since this varies with the deflection
for this

+D

2
<l

it is

impracticable to allow

impedance in arranging the secondary loads on the transformers and it is necessary to reduce AZ not to a comparatively small and definite value, as in the null method, but to a value which can be entirely neglected. If the moving coil of a commercial wattmeter is used as a detector, it must therefore be shunted by a rather low resistance and the sensitivity correspondingly reduced. The current circuit of a 1 -ampere wattmeter has about the desired impedance and will be found satisfactory as a detector
for this work.
is

The use

of the voltage circuit for the excitation

also convenient, since

avoids the necessity of a separate

resistance.

The mutual inductance between the

introduces an

equivalent load on the transformers, but

over the part of the

used the

coil is

reasonably near the position of zero mutual

inductance, this error will be less than that due to the impedance.

As a

specific

example of the various factors entering into the

choice of a detector, let us consider the commercial wattmeter

used in the experiments described below, which has the following constants
Nominal current
2.

Maximum Maximum

current
voltage

3.

amperes. amperes.
volts.
volts.

Nominal voltage
Resistance of fixed coil

75.

no.
o.

27

ohm.

Resistance of moving coil Resistance of voltage circuit

42.

178.

ohms. ohms.

Watts per division

.5

When
circuit

this instrument is used with excitation

on the voltage

we have

K=
2d

= 4.5 X

io~ 3 amperes per division,

=o.27 ohm, approximately.

Bulletin oj the

Bureau

of

Standards

[Voi.i 4

When
coil

used with excitation on the current circuit the

mo v

must be shunted to give a

ohm

so that

K = X
3
1

178

= 6.0 X io
i

per amperes r r

division,

zd

ohm, approximately.
1

If

the two transformers differ by

per cent in current, then at

the 5-ampere point

and
in

we have in the first case the D = 8.3 divisions AZ = 0.003 ohm,

the second case

D=
and

1 1

divisions

AZ = o.oi ohm.
further error arises
if

the exciting voltages are not in the

correct phase.

The magnitude

the difference in constants of the two transformers.

is 1 per cent in ratio and 30 minutes in phase angle, a phase of 7 degrees in the exciting voltage will produce an error of about 0.1 per cent in ratio and 3 minutes in angle. An auxiliary 5-ampere wattmeter connected as shown at in Fig. 3 is a convenient means of determining these phase relations.

Wm

EXPERIMENTAL RESULTS
methods described above have been tried experimentally and the results are plotted on a large scale in Fig. 4. Curves .4 and B give the ratio factor and phase angle, respecof the

Both

a 2 5 -ampere, 40 volt-ampere portable current transformer at 60 cycles. The curve is drawn through values obtained by a precise laboratory method. The crosses are points observed by the null method using as a standard a similar transformer carrying considerably less load. Curves C and D give the ratio factor and phase angle of another transformer which was tested by the deflection method. As before, the curves are drawn through the points obtained by a precise laboratory method and the circles show the values obtained by the method here proposed.
tively, of

SUMMARY

general

method has been outlined

for the determination of

the ratio and phase angle of current transformers in terms of the

constants of previously calibrated standard transformers of the

Stlsbee]

Testing Current Transformers

329

same nominal ratio. It has been shown that such methods are more sensitive, or conversely may be used with much less sensitive instruments, than the laboratory methods now in use for the absolute determination of the ratio and phase angle Two of the most convenient of the many of a single transformer. possible modifications of the general method are described in
essentially

W24

W2D

\
1

>

N.

<

i\*
<

m
"*""*-*w
"~-"~-~
t

//)/Z

__
<

SO

a
I

\
(

Pef/a

Si

/JJO/7

40

MA
f

^L____
> e

Q
/

5COVPAY <\MP&5
7.

Fig. 4

The experimental results obtained on two pairs of transformers by these methods agree very satisfactorily with the values obtained by absolute methods. It is hoped that the methods
detail.

be found useful in commercial plants where delicate laboratory equipment is not available, and where large numbers of transformers must be tested rapidly and with moderate accuracy.
will

Washington, July

1,

191

7.