Impulse Generator IMPULSE GENERATOR CIRCUIT

The complete discharge circuit of the impulse

generator, less charging equipment, is shown in
Circuli For u as Fig. 1, where, for convenience in analysis, it has
been segregated into 4 distinct parts.
Part I represents the equivalent discharge circuit
of the impulse generator proper. The capacitor
With impulse voltage testing of electrical units of the impulse generator (C) are usually of the
oil filled type commonly used for power factor
equipment becoming more widely used, correction. They are compact, reliable, and readily
it seems highly desirable to be able to procurable at reasonable cost. The discharge ca-
calculate the circuit constants required to pacitance of the generator is that resulting from
all the capacitors that discharge in series. The
prodxuce
produce the desired
the desired test waves and
test to
and to
waves
value of conductance G will depend primarily upon
calculate the waves that any given circuit the values of the charging resistors in the circuit,
constants will give. In this paper the but there may be also a small amount of leakage
over the supporting frames to ground. The inherent
solution oF the more commonly used im- resistance r of the impulse generator circuit, due to
pulse
pulse generator
developed and
generator circuits is developed
circuits is
and the resistance of the capacitors (0.01 to 1.00 ohm
summarized. Waves calculated by these each), leads, and gaps, is ordinarily small, probably
formulas check closely the waves actually being less than 100 ohms even under transient skin
effect conditions. The inherent series inductance
recordeby
recorded the cathode.ray
by the cathode ray oscillograph.
oscillograph * L is due to the inductance in the capacitors (0.01 to
3.50 ,Ah each) and to the inductive loops in the
discharge circuit, this latter inductance being pri-
By marily a function of the length and configuration of
J. L. THOMASON General Elec. Co., the discharge circuit. A method of calculating it is
ASSOCIATE A.I.E.E. Pittsfield, Mass, shown in Appendix I. The terminal capacitance K
includes the distributed capacitance of leads, ca-
pacitors, etc., to ground. Ordinarily it is too small to
exert an appreciable influence on the wave applied
| to the test piece.
MPULSE TESTING now has ex- Part II of Fig. 1 shows the circuit for controlling
panded from an experimental into a commercial field,
and it is desirable that there be available a reliable
and ready means of calculating the discharge or test
waves of impulse generators to facilitate testing in
IMPULSE WAVESHAPE
rGE-RA-, CONTROLCIRCUIT
TEST
PIECE
RECORDINGANDMEASURING
DEVICES
this broadening field. Insulator time-lag character- r ° CABLE-RESISTANCE
istics usually are obtained for 0.5x5- and 1.5x40- L GK T R2 c L/I S R0 DIVIDER
,usec (defined later) waves, whereas to develop fully C I i - -
the internal oscillations n transformers long waves L . R
with steep fronts may be necessary; or to study COMMON - GROUNDS -.
the possibility of cumulative oscillations, the applied Fig. 1. Equivalent circuit of impulse testing equip-
voltage should be oscillatory and of the same fre- ment
quency as that of the apparatus tested. It is thus
evident that to produce these specific waves and to C generator cdpdCitor units, arranged to be charged in
make sure. they are recorded on cathode ray oscillo- parallel and discharged in series
gramake
grams, itr
s e,
iS necessary, If cut and try methodss rare to G leakage conductance of charging resistors and over
generator frame (permits leakage of charge)
be avoided, to be able to calculate the circuit con- L inherent inductance in generator circuit (retards WdVe
stants required to produce these waves and to front and causes oscillations)
calculate the test waves . that1,s
these
A 4 circuit constants
"3'e *
r inherentbutresistance
front, in generator circuit (retards wave
not appreciably)
give. In several articles' wirtten at various K = terminal capacitance due to gaps, leads, etc.
times, the solutions for particular problems or L, = inductance to retard wave front or to introduce oscil-
particular circuits of the impulse- generator have lations
been shown. However, the rigorous <solutions
. ... < <. R2 =of ~~~~~~~~resistance control duration
RI = resistance toto control front of wave on and lilalso
of wave f'depending
many useful circuits probably never havre been C' = effective capacitance of test piece (affects voltage
developed; hence in this paper the solutions of the crest appreciably, depending on and 11)
more commonly used impulse generator circuits are Lr = effective inductance of test piece (affects wave dura-
developed and summarized, tion appreciably
S = sphere and introduces
gap (measures oscillations)
voltage crests, except for ex-
tremely short waves)
Full text of a paper recommended for publication by theA.I.EE. committee on Ro - resistance of divider ldetermine voltage
instruments and measurements, and scheduled for discussion at the A.I.E.E.
winter convention, New York, N. V., Jan. 23-26, 1934. Manuscript submitted Z = surge impedance of divider reduction
Sept. 28, 1933; released for publication Dec. 4, 1933. NWot publishced in pam- R = Z = cable grounding resistor (potentiometer For
phiel form. CR0. R = Z prevents reflections)
I. See bibliography at end of paper for numbered references. CR0 = cathode ray oscillograph

JANUARY 1934 169

 the shape of the wave applied to the test piece. also the small capacitance of the cathode ray oscillo-
Were it not for this circuit, the wave shape would graph deflector plates will not have an appreciable
be determirned by the constants of the impulse effect on oscillographs, except for chopped or peaked
generator and the test piece, and a desired wave waves.7) 8 The cable may damp peaked waves con-
shape (such as a 1.5 x 40-jusec wave) might not be siderably, the attenuating effect increasing with
obtained. Both the series inductance L, and the the shortness of the peak and with the length of the
series resistance R1 tend, in general, to retard the cable.8
length of the wave front. The inductance also will It is evident that the complexity of the complete
introduce oscillations in the wave and for this circuit of Fig. 1 precludes a rigorous solution thereof,
reason is usually made as small as possible. The but fortunately the circuit may be simplified greatly
shunt resistance R2 controls to some extent the dura- for analysis and without sacrificing necessary ac-
tion of the wave; in general for small loads, the curacy. The extent to which simplification may
duration varies directly with the shunt resistance. be carried depends, of course, upon the relative values
Part III represents the test piece or load. Where of the circuit constants of a given test arrangement.
the load is a bushing, an insulator string, or a gap, Moreover, the reason for ignoring certain constants
it may be represented by a capacitance. Ordinarily, may depend on test results or engineering judgment
the test piece capacitance will be between 50 and rather than on calculation. Thus, in its simplest
400 ,u,uf; to this must be added the stray capacitance form the discharge circuit of an impulse generator
of the leads which, depending upon the shape of with a capacitance load can be represented by 5
the test circuit, may vary between 50 and 400 ,/uf.
An ungrounded transformer also can be represented
by a capacitance, and a single-winding transformer C. capdcitance of impulse gener-
with one end grounded can be represented by a dtor Rl L l
capacitance in parallel with an inductance and a L1 total series inductdnce 1
resistance in series. From the kilovolt (kv) and RI total series resistdnce 1 c2V C2

kilovoltampere (kva) ratings ,of the

,.
transformer
r .
2 of R2 shunt resistdnce
~~~~~~~~~~Capacditance of test piece
__1
these constants can be approximated by simple
relations. The inductance relation Fig. 2. Equivalent circuit oF impulse generator
10 (% IX) (kv)2
2irf (kva) henries
circuit constants as shown in Fig. 2. The circuit
gives the equivalent inductance for the normal constants L1 and R, include both the added and the
leakage reactance between 2 windings of a trans- inherent inductance and resistance while the stray
former. This inductance will have to be modified capacitance to ground of the connecting leads has
to suit the test arrangements, such as when the been considered part of the load capacitance.
low voltage side of the transformer is connected to
other apparatus for excitation. The resistance SOLUTION OF CIRCUIT
R -
10 (% PIR)
(kva)
(kv)2I ohms The shape of the voltage wave applied to the
test piece can be obtained by calculating the im-
is usually of negligible magnitude even when multi- pedance drop across the test piece. For illustrative
plied by some factor as large as 10 to compensate purposes the circuit of Fig. 2 can be simplified into
for skin effect. The capacitance may be con- the circuit for eq 1 in Table I, by considering the
sidered in 2 parts: that of the bushing, which varies load capacitance negligible; by Heaviside's notation
between 100 and 350 ,u,uf; and that of the trans- the voltage drop across the resistance R2 of that
former winding, which varies between 200 and 1,000 circuit is given by eq 1. This equation can be
1.4Af for ordinary transformers and is probably 10 derived readily as follows:
times higher for shielded transformers.5 The other 1
capacitances to ground must be added, of course, E =I( + pL1 + R1 + R2)
to obtain the total load capacitance. Ep1
Part IV represents the recording and measuring ER2 = IR2 = p2 + 1
devices. Both the sphere or other type of gap used + Alp + A2
to measure the test voltage and the divider resist- The constant terms are defined in Table I. The solu-
ance will have some capacitance to ground, which tion of this equation can be obtained from a table of
will add to the total load on the generator. The equivalent operators such as Bush's ;4 the particular
divider resistance also will offer another shunt resist- form of the solution that will be used will depend
ance path to ground. The accuracy with which the upon whether the roots of the denominator are real
test wave is recorded by the cathode ray oscillograph or complex.
will depend to some extent upon how well the divider As more constants are placed in the impulse
lead-in constants are balanced.6 The cable ground- generator circuit, the solultions will become more
ing resistance R should be equal to the surge imped- complex. In the circuit for eq 2 of Table I the- series
ance Z of- the lead-in cable to prevent reflections inductance has been neglected, but the capacitance
between the 2 when the divider resistance R0 is of the load has been considered and another resist-
large in comparison with Z, which is the usual case.7 ance included. Ordinarily, most of the resistances
The small capacitance of the divider resistance and that are in the series path with the load can be con-

170 ELECTRICAL ENGINEERI:NG

>I Table I-Summary of Formulas for the Solution of Impulse Generator Circuits
Circuits Equations Constants

CDl R2 ER2E-IR-
RR2R2 ~~EAopi+ A2
+Alp (1) AO = 1?2
Li ARI±R2 A2
Li CIL,
j7T~~~~TL EC~~IC2 EBop 1 (2 R=1? ± R2) + C2(R2 + R?3)1
:IyRLJC2 pC2 p2 + Bip + B2 C2(RIR2 + RRIR + R2R3) 1 = C1(R1
ClC2(R+R2 + Rd?3 + R2R)3)

E2=IC2 R2 1 + R,
Li + C2R,2R;
r~ Lg RR 4R3J4 PC2 p3 + D1p2EDopl
+ D2p + D3 (3) Do CLR2 + C2R4 1+ C2L1(R2 ± R3)
RC1L1±
CIL, C R4)±± R3)+
+ ± ~~C2LlR4(R2 -) C1C.2L1(R2 ± Rg) +CIC2LRR2
IC2 EFopl14 F R2
-C PG2 p3 + F1p2 + F2p + F3 (4) Fo C2L1(R2 + R3)
Ici EF8(pC2R2 + pC2R3 + 1)1 L1 + C2(R1R2 + R1R3 + R2R3)
LiIR3 E2 Cpc p3 +FlpF + F2p+F ( C2LF(R2 + R3)
ELI= ILIPLi =lcipLi1 R3+R_ 1
EF4(p3C2R2 + p3C2Rs + p2)1 (6) F ILC + C2LI(R2 + R3) F CIC2L,(R2 + Ra) 4 C2(R2 + Ra)
p3 + F1p2 + F2p + F3

cL,WR,IX R3g
p EGo(p2L2 + pR3)1 R2 RI R2 R2 G2=CL+
1 RIR2 + RR3+R2R3
LGL2 =L
A,I aR2 & ER2 GpI2 + G2PG (7) Go=L-L lLV+LJFL =R2+R3
EHop( 1 R- 1
rI RI 2 Ecs 4 +
+ Hp3+Hp2+H+p (8) Ho
H0 H0
C2C= HLlR2Rl-2++ ____ + K+ t ++
H3 - C2C3L R2R2 + C1L R H4 C1C2CL,R2R3 3 2 +C3R
EC2 p ± EKo(p2L2 + pR6)1 R2 K
K RI + K+ K6 R6
W-+ Kip3 + K2p + K3p+K (9 + + R3)
0 = C2LL2(R2 Li ± 2
R1R2 g R4 R5
R5 1
= CILl 2LI
1 [REC2 +4R2)9
R2{ +R+3 +K+
RK6+ CL2+ + R+ R6 +K
K6 R++KR,+

L2 c=p+
+N R+Np+N o
K2RNo=CL + RR2R3+
L2L3) RIK +
C2L2+ R3 2
LL CR4)+1 L
K-K6 C1 C2L, ±RLIK R4L+AK
C2L1L i L i1 1l +
±RIK6
C2 C2R4(R2 + RO
K4 lC1C2L1L25 R2Rs, K6R2+R3+R4
C+R1C2 K02 R2 +R3 R4(R2R3
EC, = N(22 + pR2) 1 (10) No = 1 N LI + R3) ±L(1+]3
L(R2 SR 2
R3 ~~P4'+ N1p3 + N2p2 + NSp + N4 C,(L1L2 + LlL3 + L2L,) L12+LL3+LL
LI Lz =R1R2 + RiR3 + R2R3 +
~~~~~~~~~N2 C1(L1 + L2) + C3(L2 + La)
3 3 CLIL2 + L1Ls + L2L3 C4C3(LlL2 + LjL3 + L2L3)
T $ | TC3 N = ~~~~~~C1(R1+ R2) + C,(R2 ± R,) N4 = C13LL 1L+LL)
1- C N, C1Cs(LjL2 + LIL3 + L2L,) 1
sidered equal to zero if desired. As the denominator extracting approximated roots is applicable. A
of eq 2 is of the second degree, its solution is similar method that is similar to Horner's method, but which
to that of eq 1. The circuit for eq 3 is nearly the seems to contain less chances for error and which
same circuit as that in Fig. 2, except that resistance seems easier of solution, is shown in Appendix II.
R3 is in series with the load and resistance R4 shunts In higher than third-degree equations which may or
the load as would a divider resistance. This resist- may not contain more than one pair of complex roots,
ance is omitted in the circuit for eq 4; if R3 of that the root-squaring method of Dandelin, Lobachevsky,
circuit be considered equal to zero, there is obtained and Graffel of extracting roots is applicable.
the circuit for Fig. 2 which is probably the most
condensed circuit describing an impulse generator WAVES RESULTING FROM CIRCUIT CALCULATIONS
with a capacitance load. The solutions for eq 3
and 4 are similar and take the forms shown by The effect that each circuit constant has on the
equivalent operators 1 and 3 of Appendix III, the voltage wave applied to test piece can be shown best
form of the solution depending upon whether the 3 by varying each constant separately. The circuit
roots obtained by equating the denominator to zero constants of Fig. 2 representing an impulse generator
are all real or 1 real and 2 complex. The denomina- with a capacitance load, were given practical values
tors of both eqs 3 and 4 are of the third degree. and by making a series of calculations their effects
As there were no available published equivalent were shown. For an illustrative set of curves a
operators above the second degree, third-degree practically smooth 1.5x40-pysec wave was used as the
equivalent operators were developed; these are reference wave.
listed as Nos. 1 to 5 in Appendix III. In this paper the first term of the impulse wave
In eq 5 of Table I, only the transient terms Of designation, as the 1.5 term, is the time in micro-
the numerator should be used if the voltage drop seconds from zero time to the time of maximum crest;
across the generator capacitance be required; but and the second term, as the 40 term, is the total
if the total voltage on the generator or the total duration in microseconds of the wave from zero time
current through it be desired, the total numerator to the time of half crest amplitude on the tail or de-
of the right hand term should be used. The com- creasing part of the wave. This "2-point" method
plete denominator iS used, of course, in both cases. of wave designation gives a good general description
Equation 6 will give either the voltage drop across of the wave. The second term is used as the total
the series inductance or the current through it, duration of the wave from zero time to the time of
according to its arrangement and solution. half crest because it is determined easily when waves
The circuit for eq 7 has an inductive load rather are calculated, and is easy to read from transcribed
than a capacitive load. As the denominator is of oscillograms because of the single time scale. When
the third degree, its solution is similar to that of calculations are made the voltage equation takes
eq 4. the form of eq 3 in Appendix III, for example,
The circuit for eq 8 has an extra shunt capacitance
to represent the stray capacitance of the generator E = A [6-a:-E_-t {cos wt + B sin ct} I
and leads to ground. The denominator of this where A and B are constants. Suppose E, is the
equation is of the fourth degree. As with the crest voltage; then, since the oscillatory term
third-degree equations, there were no available usually becomes negligible long before the time to
published equivalent operators; therefore, fourth- half crest is reached, the time t or the duration of
degree equivalent operators were developed, these the wave to half crest from zero time is given by
being listed as Nos. 6 to 16 in Appendix III.
The circuit for eq 9 shows an impulse generator Ae-ait
with a transformer load, where the inductance of
the transformer is in the discharge circuit to ground. When the 2 terms of the wave are expressed as deci-
In the circuit for eq 10 each part of the circuit mals, their exact values can be easily expressed and
contains inductance so that its position effect can be typed; and when the 2 terms are separated by a
analyzed. The solutions of both these circuits are smaller letter "x," there is no confusion as to their
fourth-degree equivalent operators as shown in meaning or limit, whereas a dash for separation might
Appendix III. be taken as a range of values. This method of im-
These fourth-degree equivalent operators take pulse wave designation is in accord with recom-
3 forms depending upon whether the 4 roots obtained mendations made by the lightning and insulator
by equating the denominator to zero are all real, 2 subcommittee of the A.I.E.E. committee on power
real and 2 complex, or all complex. The solving of transmission and distribution in a report presented
these third- and fourth-degree equations to obtain atthe 933 A.I.E.E.winterconvention(see"Reco-
the exponents shown in the final solutions in Appen- mendations for Impulse Voltage Testing," ELECTRI-
dix III may require appreciable time. However, CAL ENGINEERING, V. 52, January 1933, p. 17-22).
these roots can be determined to as high a degree Figure 3 shows several waves containing different
of accuracy as desired by various methods. values of series resistance. As the series resistance
For a third-degree equation, the exact values of is increased the oscillations are damped out and the
the 3 roots in terms of the circuit constants can be magnitude of the crest is decreased. The front of
obtained by Cardan's formula. For equations of the wave first decreases up to some critical value of
the third or higher degree and where only 2 roots series resistance, and then increases; but the drura-
of the equation are complex, Homner's method of tion of the wave seems to vary nearly directly with

172 ELECTRICAL ENGINEERING

 R1401.=51 087 lL1I140
R= OM 0.l50.'f ______
120
20
.52x5OHM
L2x5C=OC2=0.010250..f
1 90ioh c
RI
0
C,
10C0002500f
= 0.012500 p.f LLRi Ci =oo900pAh
0.012500
Ri Li I.f C1 02
C
o2 R2 2900UfHMS
OL
C2
2
RI= 29000 OHMS
R 2,90 OHMS
-.-Ri,.-,
:ic-i:R
29000OHMS
,90OM
1. Ox 100 21.9
z <~~~~~ ~ ~ L......J....J 0099j.f IC2=250AAf
0.90 x39.O ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~U .50 2 40.0 C2l=I000ppAf
4.75 x 47.2
080R01=1,000 80z
w
~~
0 1.48x27.8 ...
z
Li =900pJh
x40.0 /2J75 x42.3
1~~~~~~~~~~~~~~~~~.50 LI.39000 LI= 6,000
4.6 5x 45.6

60 ol
_L~i.SE
cwi .60
0 2 3 4 5 6I 748
0-~~~~~~~~~~~~~SE

~2f~I2~I2 C20.0025pf 48~I2 C~0.0025-f C- ..v 208~?2 , 0

Z40 ol- z 4 44.5 -- -

j~~~~~~~~-3~2~( - - =4=

9002fJoO
0 3900 4900 03:~~~~~~~~~~~u L 22 X 43.32
OHMS- 0 900 2900 300 49000 0 0 020030040 50 60 70 8000000
0 1 2 3 4 5 6 7 8 9 0~~~~~~~~~~L ~ C IyL.L

Fig. 4. Efront,oduration,Rand crves Fig. 6.Efront,oduration, and crestv Fig. 8. Efront,oduration, and crves
mantdeohave vssantueohave vssantueohave v.C

C,=0.012500 jff 4 7 0
10C
q_2OIOJ20 Ce0.000250 A f
w LI= ~L900Li0R;
290 HM
,=00 0p
Li L, ..
-.10 LI.005
LII
R0,
200Ms =
002Op
L CI 8
In
::LR,?=
< 0--HM R, 2900 OHMS C_8 2 ,00M
_1. 1000000 OHMS _i 02= 2= 2,900 HMS Z R

02=2,900 il~ ~ ~ ~ ~ ~ ~t L2=4IOO =290OHS300ph 5946=10

u 80-
-4 - -

0 3 4237 4 090 1604 6 u9 042 2 3 4 66 8

DUAIONC RU.
ESEC
SECC

716 140 1j40 C1o C 00020I' Ca 80 O.I2SO

I.1 60 I2 -FRONTR ______IC 4212 c=
2I 40~ 01 20 0OM - c? 420 0020.fC
L4=0o
C z LUm
L=0 0120-'~
o u606 -l2
C.3 20------ --IrI LU
1 44 0DURATION 0 2C - -
0 o 2O0 40 0 3O 0 36 O
1~~ 20~~~'2C 4,000~~R 6,0IN00 1,0 O40MS00 io 2000 300 40021020 0 0050600-901
0
0 2900 02 IN OHMS 0 0.01 0.02 0.03~ ~ ~ ~ ~ ~0.0~ ~ ~L' 0.05h 0.0 0.0 0.8 1. SEC

Fig. 10. Front, duration, and crest Fi g. 12. Front, duration, and crest plotted tronatsmallrascale, showings
magnitude of wave vs. Rimagnitude of wave vs. Ci that iudtheisasame
wave ocilato
with th eisrssane hsi hwncery Wv1hae2acltdbymaso0omla ienti
inFig. 4. The~C crest25O vaitonas ifhwni igC.pae se ale1.Cici connection toehe itfale

JANUAY 193 1732of L R 0 2 2=0005P

for slightly different values of inductance, but the An inductance can be placed in parallel with the
crest magnitude does not seem to vary much for the load capacitance so that a transformer load can be
chosen variation in inductance. Although the series simulated. Figure 13 shows that as the load induct-
resistance is high, the oscillating effect is seen in Fig. 5. ance is decreased the voltage wave applied to the
What might be termed the load regulation of the load is decreased in crest, front, and duration, and
impulse generator is shown in Fig. 7, where the effects that the wave is a damped oscillation. This damped
of variations in the load are shown. Figure 8 shows oscillation effect is shown clearly in Fig. 14 where the
that both the front and the duration of the wave 100,000-nh wave of Fig. 13 has been drawn to a
have minimum points at about the same value of smaller scale.
load capacitance. It is interesting to note that for Besides showing the effects the circuit constants
different loads the duration of the waves may be have on the test-piece voltage, other characteristics
the same, but the front increases in general. Thus, can be learned by calculations about the generator
waves with durations of 41 ,Asec have a I-j.tsec front circuit during discharge. In Fig. 15 it is shown that
for a 60-,u,uf load and a 2-,usec front for a 350-,u,uf at the first instant of discharge the entire voltage
load. Figure 8 shows also that small variations drop is across the series inductance; but this voltage
in the load do not greatly affect the crest magnitude. drop soon decreases to zero and at that instant the
The damping effect that low shunt resistances have current from the generator capacitance is a maxi-
on the oscillations is shown in Fig. 9. In Fig. 10 it is mum. Figure 15 also shows that the charging cur-
evident that ordinarily the shunt resistance has rent of the load is zero when the load voltage is a
only a small effect on the front of the wave, but maximum. It is interesting to note that, although
that the duration varies almost directly with this the components of the voltage drops are quite vari-
resistance. Since it was shown in Fig. 4 that the able, the voltage of the generator capacitance de-
duration varies almost directly with the series resist- creases at a nearly smooth rate.
ance, it seems evident that a relation exists between One of the primary values of impulse generator
the total series and shunt resistance, R, in the circuit calculations is to prove that cathode ray oscillo-
and the duration of the wave, t. For small loads, grams actually record the test-piece voltage. Figure
calculated waves and measured oscillograms show that 16 shows a calculated wave for a capacitance load
the relation is t -0.7 RC. The constant C repre- and upon it is represented a transcribed oscillogram
sents the capacitance of the generator, and if ex- for the same circuit conditions. The 2 waves nearly
pressed in microfarads the time t is expressed in coincide.
microseconds. This same relation exists in the A calculated wave and a transcribed oscillogram

_ G l l ll 120ll0Q4812I14
I2 C =0.012500 rf L- RI C
|= 0.023300 lf
< W100 , 0.4 0 Cl
2=0.00 OHM5C0Rpf
l | | ||X 00 LI= 900ph
120 L 100juh
- T50/1IN
O 42.5juSEC
RS_____ __0R232OHS
R - - 10
<IR CL I.OhoLIoRI l
LU2=
_60SO24
5> 1. x 40)P
8
D
XRR=2900 OHMS jC
RI= 670OHMS 20 65 OHMS
~o L2=op0.90x43.5
>.
--t - -
- -

40F 156i eCautw.of2590Mff

ui~ SEW
F 1 m o

20
_

1 SEC
z
L R
waves C2= 0.00025on pa OSCaLLOGRA
E
Z20z 0.08 V\
~~~~~~~~~~iI --K 6
LI 90ROTOCACLAE
R, 2,O00OOHMS
simple circuitf2= c
2900 OHMS thog
40
-
aCALCULATED i
73
-
_j 20 =
on e u
0-~~~~h cirui The cdcltdwv s154 s h rnfre od oprdwt

th cru
resistanc Calculathed aocn co wavee shapewfor
calculated
Fig. cirt cnshaps to FigTe calcultedwv isone1apac40ansec the transformer
load,load,
oscillogram represented by the crosses osr corardmwith oscillogram
apparently is the same
wavest impressedso difperentl eparsofen wavelhap with
te oscillograms Fae uing.17 com ercilmpsett
the is aFelcing.7 shuc
simple
isthecapacthne discharging throughwavuea
capacitance Thelscalculator.
circuit of aircuth of probably greater
1=x4 the0)e;load
iswhere
Cacltedwae
interest is shown in Fig.forh17
ance
resistance. The shunt resistance also has an effect transformer. Figure 17 shows:
on the crest magnitude, Fig. 10 indicating that the (1) the measured wave which is the average of 3

11eshow thaetsmll vari1 hi

bsrations
as
inan dotasfrealavLe=o)aend c4onsideringhow the
this capacitance

well expressed in the relation t 0.7 RC. importance of considering the constants of the load

174 ELECTRICAL ENGINEERING

added to the impulse generator, and the agreement Appendix IIl-Equivalent Operators
that; can be obtained between oscillograms and
calculations when the calculations are made cor- Where PI =-a, P2 -(3+ co, and P3 =- - 1
rectly. for n = O, 1, 2, or 3
pn 1 (-CY)n-1,e-at
Ps + K2P2 + K3P + K4 - (a - (3)2 C2
Appendix I-Inductance of Generator Circuit + (-(3 + , (..1.,g - ,,s)nB1c(tB + R (1)
2co(a-a-C4) + -2wo(a -_ -_)
By means of 2 formulas from the Scientific Papers of the U.S. 1
Bureau of Standards,'0 the inherent series inductance in the dis- R = - for n = 0; R = 0 for n = 1, 2, or 3
charge circuit of the impulse generator can be calculated. 4
The leads and the banks of capacitors will form V-shaped paths, Where P1 = a, P2 = -( + jco, and P3 = -(-jc
the inductance of which may be considered that of 2 parallel wires
of radius p centimeters, of length h centimeters corresponding to 1
the height of the V, and of distance d centimeters apart equal to
half the base of the V. The formula for each V is then
p + K2P2 ± KIP + K=
L = 4h ( l dogp~ + 1 -h/-d 1-
( -
1 [cat ~
+ 2 L- -
a-20 cos
32 COt +
X 10- microhenries ,s2 +
The inductance of these V's form about 2/3 of the inherent series (co2 + a8-B2) sin wt ] (2)
inductance of the generator. (0+co' ) K (2
The remainder of the inherent inductance of the generator is P
approximately equivalent to 5/8 of the inductance of a square of 1 =+ --_
- t
sides a centimeters in length. For ease in computation the con- P3 + K 2P2 + K3P + K4 (a.- )2' + 2

ductor is considered of square section and of a centimeters on a

side. Small variations in a do not affect the total inductance
Coicos ct +d sin cot _| (3)
greatly. The formula for the total inductance of the square is
P21 1
L = 8a logsa + 0.477 _ + 0.033 )X 10-3 microhenries P3 + K2P2 + K3P + K4 (a - (3)2 + W2
Fa O l+ co2 a-
(32 a4sin a J
Of course, approximately only 5/8 of this inductance will be added [ a cCO (4)
o

to the inductance of all the V's in order to determine the total

series inherent inductance of the impulse generator circuit. p3 1 12
P3 J K2P2 + K3P +K4 = (a-,B3)2
+W L
)~-~ ~ ~~~~o t+@-2a + OW2 + CtW2i
0

Appendix 11-Method of Root Extraction for e(2a (2 - 2) COS Wt + ( A + sin wt (5)

2 O (2

Equations Containing Not Over 2 Complex Roots Where P, = -a,P2=-, P3 =-y, and P4 =
-a for n = 0, 1, 2, 3, or 4
Suppose a third degree equation be considered.
pI + k2p' + k3p + k4 = 0 = (p - pl)(p - P2)(P - PS) P4 +K2Ps + K,P2+K4P
pn 1 + K5 ___+_K_A
At +
Arrange the coefficients of the equation so that the first root can be (_3)n-IE-5t ( -4E-yt
-lY (
obtained by a method similar to long division. A + +R (6)
1 + k2 + ks + k4 I !+
a -a(k2-a)
+ ±a[k3- a(k2-aC)]
+ R I-for n = 0; R = 0 for n = 1, 2, 3, or 4
1 + (k2- a) + ks- a(k2- a) +E K5
a +a(k2 -2a) A,= -a' + a2(( + y + 6) - a(/3y + (3 + 7yb) + (3yS
1 + (k2-2a) + [ks-a(2k2-3a)I A3= -,32 + 2( + y + 8) -(3(a + a6 +7) + a7y
The nearer that k, minus a[k3 -a(k2 - a)] or e approaches zero, A -y3 + a2(a + (3 + 6) - y(a( + a6 + (36) + a(36
the more accurate the roots of the equation will be determined.
If the difference, e, is zero, the remainder 1 + (k2 - a) + [ks - Where Pi = -a, P2 = -, P3 = -y + jw, and P = -y - jw
a(k2 - a)] may be written as 1 + (A) + (B). This corresponds
to the quadratic equation p2 + Ap + B = 0 with solutions 1 _ at + -Pt + --ft
+KP' KP2 + K4P + K -A± -A+A72 + As2
+
P2 - 2 -(-A7Y - Ao) cos set + (A7oo-A8y) sin cot +
Ly2 + w 2 J Ks
and
-A -\/A.2 - 4B Pi 4-at le f
2_c-t
P3- 2 P4 + K2P, + K3p2 + K4P + K, + A + A72 + As2
A
These roots, of course, may be real or complex. The first root [A7 cos ct + As sin ot] (8)
would have to be real, and as determined above would be P, Q- a
As long as there are not more than 2 complex roots in an equation p2 1 -za,.-at _.(3-,#t
of any degree usually the real roots can be extracted in a manner 2=++
similar to the method by which the root pl = -a was obtained. P4 + K,P' + K,P' + KR' + K, A5 A^,+
In the preceding development, should f not be equal to zero upon 2e-Y r-(-Ay+A)csx-Ac 8)sns 9
the first trial, the division process may be repeated and a second A7' + A,' L'J
remainder obtained. This second remainder [k, - a(2K, - 3a)lI
may be divided into e to obtain a correction for a. If the correction A, = -a3 +r a'((3 + 27y) - a(7' + co' +r 2(37E) + (3(72 + co')
be positive, it should be added to the first value of a to obtain a A, = -(B3 + (3'(a + 27) - (3C- + co' + 2a7r) + a(y' + co'2)
more nearly correct value of a If the correction be negative, a is A7 = -2co'(a + (3B-27)
too large by approximately that amount and should be reduced. A, = 2co(7' - co' + a3 _ a7 - 3w

JANUARY 1934 175

 PI 1 ai 2&E-at 02C-01 2 -yt 4. OPERATIONAL CIRCUIT ANALYSIS (a book), V. Bush. John Wiley and
=4+P+ +KA +K + A + 2 Sons, New York, N. Y.
5. SHUNT RESISTORS FOR REACTORS, Kierstead, Rorden, and Bewley. A.I.E.E.

iA &^Y' - o2) - 2A syw cosot + 2A7-YCO+As(ty2-w2)sinxt 8_,y )ol) sin wt (l)o_ TRANS., V. 49, 1930, p. 1161-77.
6. LABORATORY MEASUREMENTS OF IMPULSE VOLTAGES, C. M. Foust and
J. C. Dowell. A.I.E.E. TRANS., V. 52, 1933, p. 537-43.
P4 1 aI3-at 03_,6Ot 2EYt . 7. LABORATORY MEASUREMENTS OF IMPULSE TESTING, L. V. Bewley's dis-
P4 + K2P3 + KIP2 + K4P + K5 -A6 -A6 A72 + A82 cussion of paper. A.I.E.E. TRANS., V. 52, 1933, p. 555-7.
8. EFFECTS OF SHORT LENGTHS OF CABI.E ON TRAVELING WAVES, McEachron,
A7rY(3 _y2) + Asw(3y2 _ 0,2)
[FA7Y(3co2 W2 y2) + A8CO(3 y2 W2)
COS Wt
cot +Hemstreet, and Seelye. A.I.E.E. TRANS., v. 49, 1930, p. 885-94.
9. CALCULUS OF OBSERVATIONS (a book), Whittaker and Robinson. D. Van
(I A7,o(0w2 - 3-y2) + As-y(3w'2 - 'y2) i sin ctj (11) Nostrand Company, New York, N. Y.
10. SCIENTIFIC PAPERS OF THE U.S. BUREAU OF STANDARDS, NO. 169. For-
Where Pi -a + = jico, P2 = -oa - jco, Ps -
-3 + jQ, and mulas and tables for the calculation of mutual and self-inductance. Dec. 1916.
P4 = -,3 -jQ

p4 + K2P3 +K3p2 + K4P + K6

r-(axAg + coAlo) cos cot + (wAo - aAjo) sin cotl
L co(A92 + A102)(a2 + co2)

+ [e(iA
(OAl-( +
1 220+5 A12f2l(]82 + 2) Portable
Au)(A + S2A12) cos 52t + (52All- OA12) sin Slt]
ScIering
_9 cos _t
+ Alo sin o ridge ror Field Tests
+1 (12)

P4 + K2P3 +
P1
K3SP2 + K4P + K6 co(A92 + A102)
+ 4E_#t [iLt COS Qt + A12 sin Qt (13)
A portable Schering bridge which is ap-
p4 + K2P3 + K3P2 + KOP +
]'2 1
e Ks plicable for power factor, dielectric loss,
- ( aA9 - coA io) cos ct - (,A9 + aA io) sin ct t and capacitance tests on condenser bush-
L co(A92 + A102)i2t - J+ ings and other equipment of higher capaci-
[-(Mil O2A12) COS - (QAu, + MA12) sin f2t] (14) tance in the field is described in this paper,
Q(A 112 + A122) and operating procedure is given. Such
As 2(w(a - +()
=
tests are desirable for forestalling failures
A11 = 2Q(O - a) of equipment. Although earlier power
A12 = (,W2 Q2) + (at 0) 2 -

factor tests on bushings have been made in

pI I c-at the Field, this is probably the first applica-
P]4 + K2P3 + K3]p2 + KS] + Kr co(A92 + A 152) tion of the Schering bridge to Field tests.
[As(a2 - c2) - 2AioaoCw cos wt + 2A9aw + Aio(a22 Co)2) .
sin wot +f(A12 su12 Al+ 2
2) -2A1200 CO -t-+ By
]-+
Q(A +r-2 Al222[: 1() COSt + C. F. HILL
2AliOQ + A12(32- Q2) sin 2t] (15) MEMBER A.I.E.E.

PI 1 = -at T. R. WATTS
P4 + K2P'3 + K3p2 + K4P + K6 co(A92 + A102) ASSOCIATE A.I.E.E.
[ Ag(-a3 + 3ac2) + Alo(3a2co - W3) C coSt + G. A. BURR All of the Westing-
house Elec. and Mfg.
Ao(c3 - 3 a2c) + Alo(-a3 + 3ac2) sincwt] +
ASSOCIATE A.I.E.E. Co., East Pittsburgh, Pd.

Q2(A 12 + A122) AllA( -(33 + 3/ 2) + Ai2(332sl S3)

-
HE attention given to bushings in
cos S2t + A,,(S2 - 332Q) + A12(-03 + 30f2) sin Qt (16) service has not kept pace with the increasing ser-
vice demand throughout the years. This is not
negligence on the part of the operating companies,
as they have always serviced bushings by the best
available means. Generally a bushing has been put
Bibliography into service and operated for life, or until it failed,
wth only routine surface inspection. The change
1. THE EFFE;CT OF TRANSIENT VOLTAGES ON DIE3LECTRICS, F. w. Peek, Jr. of apparatus from indoor to outdoor operation, the
A.I.E.E. TRANS., 34, 1915, P. V. 1857-1909.__________________________________
2. IMPULSE TE1STING TE3CHNIQUE:, Foust, Kuelini, and Rohats. Gen. EleC. Full text of a paper recommended for publication by the A.I.E.E. committee
Rev., July 1932, P. 358-66. on instnimeiits and measurements, and scheduled for discussion at the A.I.E.E.
winter conVentionl, New York, N. Y., Jan. 23-26f, 1934. Manuscript suhmitted
3. CHARACTERISTICS OF SURGE GENQERATORS FOE TRANSFORMER TESTING, Oct. 23, 1933; released for publication Dec. 12, 1933. Not published in pam-
P. L. Bellaschi. A.l.E.E. TRANS., V. 51, 1932, p. 936-45. pIslet form.

176 ELECTRICAL ENGINEERING