( Reaffirmed 1996 )

IS : 2071 (Part II ) - 1974

Indian Standard
METHODS OF HIGH VOLTAGE TESTING
PART II TEST PROCEDURES

( First Revision)
High Voltage Techniques Sectional Committee, ETDC 19
Chairman Repwenting
SHRI H. WNARAYANA RAO Mysorc State Electricity Board, Bangalore
Members
SHRI N. S. S. AROKIA~WAMY Tami Nadu Electricity Board, Madras
SHRI M. K. SVNDARARAJAN ( Alkmak )
SHRI V. BALMVBRAUANIAN Hindustan Brown Boveri Ltd, Bombay
sHRl K. P. H. SHARY.4 (Al&d )
DR S. C. BHATIA Siemcns India Ltd, Bombay
DR D. P. SAHQAL( Alternak ) .
Drascr0~ ( CPRI ) C~I~~~W;~~ & POWcrCommission ( Power Wing ),

DEPUTY DIRECTOR( CPRI ) ( Alkmak )

SHRI M. L. DONORS J30m~m~~-ttie Supply 8t Transport Undertaking.

SHRI M. R. K. MENON( Alkmak )

ENGINEER SIJPERINTENDEN;~;~~ U. PtiG~arOmcat Pottery Development Ccntrc,
& Low Tsrtsto~ u
~BOR~TORY
ASSIST-ANT ELECTRICALENO~N~ER
( Alkmak )
DR H. V. G~PALAKRPHNA Indian Institute of Science, Bangalotc
DR R. S. N. RAU ( Alkmak)
SHRI M. M. MUKHERJEE Calcutta Electric Supply Corporation Ltd, Calcutta
DR K. S. DAPGIJPTA( &&rru& )
SHRI S. K. MWHERJEB National Test House, Cal&a
SWRID. V. NAPKE Heavy Elcctricals ( lndii ) Ltd, Bhopal
SHRI B. N. GHOSH( Alternate)
SHR~J. S. NEGI Jyoti Ltd, Baroda
SHRI C. S. N. RAJ~ ( AIkmak)
S,.,Rt SOHhN%hOH Punjab State Electricity Board, Patiala
SHRI A. K. CHOPRA( Alkrnak)
sHR1 P. J. WADIA The Tata Hydra-Electric Power Supply Co Ltd,
Bombay
SHRI G. K. TRAK~JR( Alfmurtc )
SHRI N. SRINIVASAN, Director General, IS1 (lG-05& Member)
Director ( Elec tech )
s?cretary
SHRI R. C. JAIN
Dcputv Dirrctor ( Elcc tech 1, IS1

INDIAN STANDARDS INSTITUTION

This publioltion is protected under the lndian C~qvrighr Acr ( XIV of 1957) and
reoroduction in whole or m part by any means except with wrttten permissian of the
publisher shalL be deemed to be an infringement ol’ copyright under the satd Act.
Is:2071 (PartlI)-1974

Indian Standard
METHODS OF HIGH VOLTAGE TESTING
PART II TEST PROCEDURES

(-First Revision)
0. FOREWORD *

0.1 This Indian Standard ( Part II ) ( First Revision ) was adopted by the
Indian Standards Institution on 17 January 1974, after the draft finalized
by the High Voltage Techniques Sectional Committee had been approved
by the Electrotechnical Division Council.

0.2 With a large number of high voltage and extra high voltage transmission
system being constructed in this country, expensive electrical equipments are
being put in service. It is necessary to ensure that such equipments are
capable of withstanding the over-voltages normally encountered in service.
This standard is intended to provide uniform methods of high voltage testing
of electrical equipments.

0.3 This standard covers various aspects of high voltage testing, such as
definition of terms, test procedures, requirements of the test objects and
approved measuring devices for dielectric tests with direct, alternating,
and impulse voltages, as well as impulse current. The Sectional Committee
decided to withraw IS : 2070-1962 which covered impulse voltage testing.

0.4 The revision of this standard was undertaken in view of the advancements
in high voltage technology.
0.4.1 This standard in addition, includes the procedures for carrying
out switching impulse voltage, artificial pollution test and the statistical
method of evaluating the test results.

0.5 This part ( Part II ) of the standard covers test procedures. The other
two parts are:
Part I General definitions and requirements
Part III Approved measuring devices.

0.6 In the preparation of this standard, assistance has been derived from
IRC Documents 42( Central Office ) 15, 20 and 20A Draft High-Voltage
Test Techniques, Test Procedures issued by the International Electrotechnical
Commission.

2
J
 IS:2971(PartII)-1974

9.7 In reporting the result of a test made in accordance with this standard
the final value, observed or calculated, is to be rounded off in accordance
with IS : Z-1960*.

1. SCOPE

1.1 This standard ( Part II ) covers the test procedures and methods for
generation and measurement of test voltage and currents for dielectric test
with direct voltages, alternating voltages, impulse voltages and impulse
currents. It also describes methods for the evaluation of test results and
specifies the criteria for acceptance,

2. TEST WITH DIRECT VOLTAGE

2.l Definitions
2.1.l R&de - Ripple is the periodic deviation from the arithmetic mean
value. The magnitude of the ripple is defined as half the difference between
the maximum and minimum values. The ripple factor is the ratio of the
ripple magnitude to the arithmetic mean value of the voltage.
2.1.2 Vahe of Test Voltage - The value of the test voltage is defined by
its arithmetic mean value.
NOTE -The maximum value of the voltage is approximately equal to the sum of the
mean valueand the ripple amplitude.

2.2 Test Voltage

2.2.1
Rtphments
2.2.1.1 The test voltage with the test object connected should be a direct
voltage with not more than a 5-percent ripple factor, unless otherwise speci-
fied in the relevant standards. Note that the presence of the test object
may af7ect the ripple factor.
2.2.2 Generation of Test Voltage
2.2.2.1 The test voltage is generally obtained by means of rectifiers,
or by electrostatic generators. The requirements to be met by the test
voltage source depend considerably upon the type of the apparatus to be
tested, and the test conditions. The requirements are determined maix-Jy
by the values and the nature of the test current to be supplied.
The important constituents of the test current are indicatedin 2.2.4.
The output current rating of’ the source and the magnitude of the storage
capacitance should be sufficient to charge the capacitance of the test object
*R&s for roundingoff numerical values (rtid).

3
IS : 2071 ( Part II ) - 1974

in a reasonably short time. ( Th is can take several minutes in some cases. )

The source should be adequate to supply the leakage and absorption currents,
and any internal and external partial discharge currents without excessive
voltage drop which might influence the test results.
In tests on internal insulation these currents are usually small, but when test-
ing wet insulators, leakage currents of the order of some tens of milliamperes
or predischarge pulses of the order 1Oaa coulomb may occasionally be
encountered.
For test on pclluted insulators, special circuit giving larger output
currents are necessary. Provisionally it is recommended that the output
impedance of the source should be such that when a pulse current of 500 mA
is drawn for a duration of 500 ms the voltage drop during each pulse or a
sequence of such pulses should not exceed 5 percent, the repetition rate of the
current pulses being one per second.

2.2.3 Measurement of Test Voltage

2.2.3.r Measurement with approved devices -The measurement of the
arithmetic mean value, the maximum value or the ripple factor or any
transient drop of the test voltage should be made with a device specified in
Part III ofthis standard. Attention is drawn to the requirements on response
c!laracteristics when measuring ripples and transients.
2.2.3.2 Measurement with sphdrc-gap - The measurement of the maximum
value of the voltage only can be made with a sphere-gap.
The procedure usually consists in establishing a relationship between
the spacing at which disruptive discharge occurs and some other circuit
variable related to the test voltage. ( The relationship between the supply
voltage and the output voltage may not be sufficiently stable for measuring
purposes. ) In establishing this relationship, the procedure specified in
IS : 1876-1961* should be followed.
If this relationship is dependent on the presence of the test object,
the sphere-gap, the precipitation in wet test, etc, it is important that the
conditions should be the same during the calibration and the actual test.
The calibration is preferably mlde at 100 percent of the test voltage,
but if necessary, extrapolation may be made from a voltage not lower than
80 percent of the test voltage. Attention is drawn to the possibility of
erratic results due to pollution in the atmosphere.

~2.2.4 Measurement of Test Current

2.2.4.1 When measurements of current through the test object are made,
several separate components may be recognized which differ from each
*Methi for voltage measurementby means of sphere-gaps ( one sphere earthed ).

4
 IS t 2071( Pal-s-Ix) - 1974

otherby several orders of magnitude for the same test object and test voltage
as follows:
a) The capacitance current, due to the initial application of the test
voltage and to any ripple or other fluctuations imposed on it;
b) The dielectric absorption current, due to slow charge displacements
within the insulation and persisting for periods of a few seconds
up to several hours. This process is reversible, currents of opposite
polarity being observed when the test object is discharged and
short-circuited;
c) The continuous leakage current, which is the final steady current
attained at constant voltage after the components mentioned above
have decayed to zero; and
d) Partial discharge currents.
2.2.4.2 Measurement of the first three components necessitates the use
of instruments covering a wide range of magnitudes. From the observed
current variation with time, information may sometimes be obtained con-
cerning the condition of the insulation in non-destructive tests. Mcasure-
ments of partial discharge pulse currents are made with special instruments
which are dealt with in the IS : 6209-1971*.
2.2.4.3 The relative importance of each component depends on the type
and~the condition of the apparatus being tested, the purpose for which the:
test is made and the duration of the test. Accordingly, the measurement
procedures should be specified in the relevant standard when it is required
to distinguish a particular component.
2.2.5 Tesf Procedure
2.2.5.1 The voltage should be applied to the test object starting with a
value sufficiently low to prevent overvoltage due to switching transients.
It should be raised sufficiently slowly to permit accurate reading of the
instruments but not SO slowly as to cause unnecessary prolongation of the
stressing of the test object near to the test voltage. These requirements
are in general met if the rate of rise above 75 percent of the estimated final
voitage is about 2 percent of this voltage per second.
2.2.5.2 The polarity of the voltage or the order in which the voltage
of each polarity is applied, and also the rate of rise should be specified in the
relevant standard, in those cases where deviations are required from the
above specification.
2.2.5.3 Assured withstand voltq<e test - The test voltage should be raised
to the specified value as described in 2.2.5.1. It should be maintained for
thespecificd time and then reduced by discharging the smoothing capacitor
through a suitable resistor. The test is generally satisfactory if no disrupti\vc
discharge occurs on the test object.
- _II-
*Mcrhods for partial dischargemeasurements.
IS :-2071( Part II ) - 1974

2.2.5.4 Assured disruptive discharge voltage test - The test voltage should be
raised as described in 2.2.5.1 until a disruptive discharge occurs on the test
object. The value of the test voltage reached just prior to the disruptive
discharge is the assured disruptive discharge voltage.

3. TEST WITH ALTERNATING VOLTAGES

3.1 Definitions
3.1.1 Peak Value - The peak value is the maximum value except that
small high frequency oscillations arising for instance from partial discharges
are disregarded.
3.1.2 rms Value-The rms value of an alternating voltage is the square
root of the mean of the square of the voltage during a complete cycle.
3.1.3 Value of Test 1701tage- The value of test voltage is defined by its
peak value divided by dc
NOTE - The relc\ant eqtiipment standard may require a measurement of the rms
value of the test voltage instead of the peak value for the cases where the rms value
may be of importance. Such cases arc for instance when thermal effects are under
investigation.
3,2 Test Voltage
3.2.1 Requirements
3.2.1.1 Tile test voltizge should be an alternating voltage having a
frequency in the range 40-62 Hz or of an agreed value. Its form shall
approximate to a sinusoid with both the halfcycles closely alike and having a
ratio of peak/rms values equal to tiywithin A5 percent. The results of a
high voltage test are considered unaffected by deviations from a sinusoid if
these requirements are met.
3.2.1.2 There is insufficient information available to indicate whether
these requirements are generally met by tests circuits in common use; hence
somewhat greater deviation from a sinusoid may have to be accepted. Note
that the presence of the text object, especially those having non-linear impe-
dance characteristics, may considerably effect the deviation from a sinusoid.
NOTE - A useful measure, for the specification or acceptance testing of test equipment.
is the rms value of the harmonics. It can generally be assumed that the ahove require-
ments on deviations from a sinusoid will be met if this value does not exceed 5 percent of
the rms value of the fundamental.

3.2.2 Generation
3.2.2.1 Thetest voltage is generally supplied from a stepcup trans-
former but alternatively, it may be generated by means of a series resonant
circuit.
NOTE - Attention is drawn to the possibility that non-sustained external disruptive
discharges may cause a large over-swing between the terminals of the test object. The
phenomenon may cause failure or flashover of the test objccr or of the testing trans-
former. A cure can usually be effected by changing the natural frequency of the voltage
source, or by introducing some attenuation into the rystcm.

6
 Is:2071 (PartIx)-1979

3.2.2.2 Test transform - The voltage in the test circuit should be

stable enough to be practically unaffected ~by varying leakage currents.
Partial discharges or predischarges in the test object should not reduce the
test voltage to such an extent and for such a-time that the measured disruptive
discharge voltage of the test object is affected. Thisis usually achieved if:
a) the total capacitance of the test object and any additional capaci-
tance is not less than about 1 000 pF.
IVOTE -If any cxtcmdpro_tcctive suiuraistance doa not exceed 1 k&the trans-
?bbyby cfkctive tcmniwdcapac~tancemaybc regarded aa bcinginparaUclwiththetat

b) the steady state current delivered by the transformer, when the 1

test object is short-circuited at the test voltage, is not less than IA ,
rms. Exceptions are:
4 Tests on small samples of solid insulation, insulating liquids, or
combination of the two. For such tests, a short-circuit current
of the order of 6 1A rms may suffice.
ii) Tests under artificial pollution. For such tests, the requirements
of the following table should be met where Rs is the total series
resistance and Xs the steady-state series reactance at the test
frequency -of the supply circuit including the generator or the
network.
&l/X, Short-Circuit
Cu?Ymt r,
Arms
go.1 6
0.1-63 12
3.2.2.3 The series resonant circuit -The series resonant circuit consists
essentially of a capacitive test object or load connected in series with a variable
inductance to a medium voltage power supply. By varying the inductance
the circuit can be tuned to resonance at the required frequency under which
condition a voltage of a shape approximating very closely to a sinusoid is
applied to the test object. The stability of the resonance conditions and of
the voltage depend on the constancy of the test circuit impedances.
The method is especially useful when testing objects such as capacitors
and cable samples, when the external insulation effects are swamped by the
capacitance of the load. The circuit is unsuitable for tests when the resonant
conditions are seriously disturbed by non-linear leakage currents or discharges,
for example for tests on external insulation under wet and polluted conditions.
3.2.3 Measuwment
3.2.3.1 Measuremmt with approved devices-The test voltage and the
deviation from a sinusoid should be measured with devices specified in Part
III of this standard.

7
IS : 2071( Part II ) - 1974

3.2.3.2 .Vlea.rurenzentwith sfhere-gap - The measurement of the peak

value 01’ the test voltage only ~may be made with a sphere-gap. The pr*
cedure usually consists’ in establishing a relationship between the spacing at
\i.hich disruptive discharge occurs and some other circuit variable related to
the test voltage. In establishing this relationship, the procedure specified in
IS : 1876-1961* should be followed.
If this relationship is dependent on the presence of the test object, the
~sphere-gap, the precipitation in wet tests, etc, it is important that the condi-
tions should be the same during the calibration and the actual test.
The calibration is preferably made at 100 percent of the test voltage,
but extrapolation may be made from a voltage not lower than 80 percent of
the test voltage. Extrapolation may be unsatisfactory if the current of the
test circuit is not proportional to the voltage, or if modifications have occurred
in the voltage shape or in the frequency at the 100 percent voltage level.
3.2.4 Test Procedure
3.2.4.1 The voltage should be applied to the test object starting at a
value sufficiently low to prevent over-voltages due to switching transients. It
should be raised sufficiently slowly to permit accurate reading of the measur-
ing instrument, but not so slowly as to cause unnecessary prolongation of the
stress near the test voltage.
These requirements are in general met if the rate of rise above 75 per-
cent of the estimated final test voltage is about 2 percent per second of this
voltage. However, certain types of objects may necessirate a different rate
of rise which should be specified in the relevant standards.
3.2.4.2 Assured z&stand voltage test - The test voltage should be raised
to the specified value as described in 3.2.4.1. It should be maintained for
the specified time and then rapidly decreased. The voltage should not be
interrupted suddenly to avoid the possibility of switching transients which
may affect the subsequent test results. The test is~satisfactory if no disruptive
discharge occurs on the test object, unless other acceptance criteria are speci-
fied in the relevant equipment specifications.
3.2.4.3 Assured disrufitive discharge voltage test - The test voltage should be
i..Gscd as described in 3.2.4.1 until a disruptive~discharge occurs on the test
c~l)jrxt. The value of the test voltage reached just prior to the disruptive
discharge is the assured disruptive discharge voltage.

-1. TEST WITH LIGHTNING IMPULSE VOLTAGES

31.1 Deikxitions
4.1 .l Full Lightning Itnpulse - A full lightning impulse ( see Part I of this
standard ) is a lightning impulse which is not interrupted by a disruptive
discharge.

*hfLthi)d fcx lQlta$c measurementby means of sphere-gaps ( me sphere earthed ).

8
 IS : 2071 ( Part II ) - 1974

4.1.2 Chopped Lightning Impulse - A chopped lightning impulse is alightn-

ing impulse which is interrupted by the disruptive disc11 u-gc of an external
chopping gap or of the external insulation of the test object. causing a sudden
collapse of the voltage, practically to zero value. The collapse can occur on
the wave front, at the peak or on the tail.
4.1.3 Value of Test Voltage - The value of the test voltage is normahy the
peak value. With some test circuits oscillations or overshoot may be present
on the voltage. If the amplitude of the oscillations is not greater than 5
percent of the peak value and the frequency is at least O-5 MHz, or alter-
natively, if the amplitude of the overshoot is not greater than 5 percent of the
peak value and the duration not greater than 1 ps, a mean curve -may be
drawn for the purpose of measurement, the maximum ampiitude of which is
defined as the value of the test voltage ( see Fig. 1A ). For ot!lcr cases ( see,
for instance, Fig. 1B ), the relevant standards should define the value of the
test voltage taking into account the type of test and test object.
NOTE- In what follows, the term peak value’ includes the term ‘virtual peak
l

value ’ unless otherwise stated.

4.1.4 Virtual Front Time T, - The virtual front time It is defined as
I.67 times the time interval I between the instants when the impulse is 30
percent and 90 percent of the peak value ( see points A, B in Fig. 2 to 4 ).
If oscillations are present on the wave front, the points A and B should be
taken on the mean curve drawn through these oscillations.
4.1.5 Virtual Origin 0, of an Impulse -The virtual or’gin 0, of an impulse
is defined as the instant preceding that corresponding to the point A 1 see
Fig. 2 to 4 ) by a time 0.3 T,. For oscillograms having linear time sweeps,
this is the intersection with x-axis of a straight line drawn through the
reference point A and B on the wave front.
4.1.6 Virtual Time to Half Value T, - The virtual time to half value ‘I-,
of an impulse is the time interval between the virtual origin and the instant
on the tail, when the voltage has decreased to halfof the peak value.
4.1.7 Dejnitions .4pplicable Only to Chopped Impulses - Ideally t!lc chopping
of an impulse is characterised by an initial discontinuity which dccrcascs the
voltage. The voltage then continues to decrease to zero or ne.u-ly zero,
either directly or via oscillations.
NOTE ---With some impulse measuring systems, test objects or test arrangements,
‘oscillograms of chrpped impulses showa rounding off of the voltage or a strp which may
bc followrd by a partial recovery, or even may cause a double peak before the final
collapse of the voltage. Such cases are not~dralt with in this standard.

4.1.7.1 Virtual time of voltage collapse during chopping - The virtual time
of voltage collapse during chopping is 1.67 times the time interval between
points C and D ( see Fig. 3 and 4 ).
4.1.7.2 Instant of chopping - The instant of chopping is the instant when
the initial discontinuity occurs.
IS:2071(PartII)-1974

0 i i i i ps
TIME - TJME u

TIME- '

FIG. 1-A CONSTRUCXION FOR DERIVATION OF VIRTUAL PEAK VALUES

r-- P-
t , , , I , , , ,
0 2 4 6 8 lIS
I

0
, ,

2
, ,

4
, ,

6
, ,

8 PS
TIME - TIME --

I
TIME -

FIG. IB IRRECWLAR I~ULSE SHAPES

10

c
 IS:2071(PartII)-1974

4.1.7.3 Virtual time Tc to chopping - The virtual time TC to chopping is

the time interval between the virtual origin 0,, and the virtual instant of
chopping.
4.1.7.4 Virtual steepness of voltage during chopping - The virtual steepness
of voltage during chopping is the quotient of the estimated voltage at the
instant of chopping and the virtual time of voltage collapse.
NOTB -The use of points C and D are for defmition purposed only; it is not implied
that the steepness of chopping can be measured with any degree of accuracy using
conventional measuring circuits.

Tl = 1*67t

FIG. 2 FULL IMPULSE VOLTAOR

TIME --

FIG. 3 IMPLLSE VOLTAGE CHOPPED ON THE FRONT

11
IS : 2071 ( Part f I) - 1974

TIME -
FIG. 4 IMPULSEVOLTAGE CHOPPED ON THE TAIL
4.1.8 Linearly Rising Front-Chopped Impulses - A voltage rising with appro-
ximately constant steepness, until it is chopped by the occurrence of a dis-
ruptive discharge, is described as a linearly rising impulse ( see Fig. 5 ).

I---- --

At
AL+

F:G. 5 LNEAIZLV RI~INQ IMPULSB

12
 IS : 2071( Part II ) - 1974

The parameters of the impulse are determined by two points E and F,

located on the wave front of the impulse or, if oscillatrons are present, on a
mean curve through them at 50 percent and 90 percent of the peak vaiue.
The impulse is defined by:
- the peak value at the instant of chopping as measured on the
oscillogram.
- the riqe time Tn. This is the time interval E-F ( see Fig. 5 )
multiplied by 2.5.
- The virtual steepness, 8. This is the slope of the straight line
E-F, usually expressed in KV/~.LS.

To judge whether the wave front can be considered approximately

linear, two straight lines are drawn parallel to the line E--F such that they
enclose the entire impulse Corn 50 percent up to the instant of chopping.
If the time difference corresponding to these two lines is not more than
O-05 times the rise time, the impulse can be considered approximately linear
( see Note ).
NOTE--To avoid excessive deviation from linearity it is recommended that the
prospective peak value of the imp&c should not be les than 1.1 times the peak value
at the instant of chopping.

4.1.9 Curves Relating the Disruptive Discharge Voltage Peak to the Times to the
Disruptive Discharge ( Voltage/Time Curves )

4.1.9.1 Voltage/time curvefor wavefronts rising linearly - The voltage/time

curve for impulses with wave fronts rising linearly is the curve relating the
disruptive discharge voltage of a test object to 7’~

The curve is obtained by applying~impulses with approximately linear

wave fronts of different steepnesses.
4.1.9.2 Voltage/time curve for impulses of constant prospective shape - The
voltage/time curve for impulses of constant prospective shape is tile curve
relating the disruptive discharge voltage of a test object to the virtual rime to
chopping which may occur on the wave front, at the peak or on the tail. The
curve is usually obtained by applying impulse voltages of constant shape but
with different peak values ( seeEig. G ).

4.2 Test Voltztge

4.2.1 Standard Lightning Impulse -The standard lightning impulse is a
full lightning impulse having a virtual front time of 1.2 us and a virtual time
to half value of 50 ps. It is described as a 1*2/50 impulse.

4.2.2 Tolerances - If not otherwise specified in the relevant stadnd~,

the following dEerences between specified values for the standard imp&e

13
IS : 2071( Part II ) - 1974

and those actually recorded are permitted when the measuring system meeti
the requirements of Part III of this standard.
Peak value f 3 percent
Front time & 30 percent
Time to half value f 20 percent

Overshoot or oscillations in the neighbourhood of the peak value are

tolerated provided their single peak amplitude is not larger than 5 percent
of the peak value. Oscillations on the first part of the impulse ( below 50
percent of the peak value ) are tolerated provided their single peak amplitude
does not exceed 25 percent of the peak value.
NOTE 1 --In specific cases, such as during tests on low impedance objects, it may br
difficult to adjust the shape of the impulse within the tolerances recommended, to keep
the oscillations or overshoot within the specified limits, or to avoid a polarity reversal.
Such cases shall be dealt with separately in the relevant equipment standard.
NOTE 2 - It is emphasized that the tolerances mentioned above constitute the permif-
ted differences between the specified values and those actually recorded by measurements.
These differences should be distinguished from measuring errors which are the difference
between the true values and those actually recorded.

No corrections are accepted for measuring errors. Note however that the measure-
ment of impulses chopped at the front at times shorter than 0’5 PS, especially those of
high voltage, presents considerable technical difficulties. The same applies to the
measurement of oscillations on the wave front especially at frequencies above one MHz.

TIME -

FIG. 6 VOLT/TIME CURVE FOR STANDARD IMPULSES

4.2.3 Standard Chopped LightninE Imp& - A standard chopped lightning

impulse is a standard impulse which is chopped after 2-5 ps. Because of
incomplete technical information, -the virtual time of voltage collapse has not
yet been standardized. This characteristic is of importance only for some
specific tests.

14
 Is:2071 (PartIIj-1974

4.2.4 Generation and Requirements qf Test Voft_age- The impulse is usuall?

generated by an impulse generator which essentially is an apparatus in which
a number of capacitors are charged in parallel from a direct voltage source
and then discharged in series via a circuit which includes the test object.
The values of the circuit parameters should be such that the required impulse
shape is obtained with the test object in the circuit.

4.2.5 Determination of Impulse Shape - The impulse shape should be adjusted

with the test object in circuit. For this purpose, an approved divider and
oscillograph should be used ( see Part III of this standard ). The impulse
shape should be measured in accordance with 4.1.4 to 4.1.8.
In general, the impulse shape should be checked for each test object.
Where a number of test objects of the same design and size are tested under
unchanged conditions, the shape need -only be verified once.
NWXlS- A determination of the impulse shape by calculation is not satisfactory~

4.2.6 Measurement of Test Voltage

4.2.6J Measurement with approved devices-The test voltage shouId be
measured by a device specified in Part III of this standard.
4.2.6.2 Measurement with sphere-gap - The, measurement of the peak
value of the test voltage may be made with a sphere-gap. The procedure
usually consists in establishing a relationship between the spacing at which
disruptive discharge occurs and some other variable in the circuit related to
the test voltage, for instance the maximum charging voltage of the first stage
of the impulse generator. In establishing this relationship, the procedure
specified in IS : 1876-1961* should be followed. If this relation is dependent
on the presence of the test object, the sphere-gap, etc, it is important that the
conditions are the same during the calibration and the actual test except that
the sphere-gap shall be opened sufficiently to prevent sparkover during the
test.
For tests on insulation which is not likely to be damaged by repeated
voltage applications or flashovers, calibration is made up to nearly 100 per-
cent of the test voltage.
For tests on other objects, extrapolation may be unavoidable but such
extrapolation should not be greater than 60 percent of the test voltage. The
extrapolation is only permissible if it can be shown that the test voltage is
proportional to the related quantity.

4.3 Test Procedure

4.3.1 The test procedure applicable to particular items of equipment,
for example the polarity to be used, the preferred order if both polarities are
*Method for voltaTe measurement by means of sphere-gaps ( one sphere earthed ).

15

e
IS : 2071-( Part II ) - 1974

to be used, the number ofapplications, and the interval between applications,

s;l,-luld be specified in the relevant equipment standard; having regard to such
factors as:
a) the accuracy of test results,
b) the statistical nature of the observed phenomenon and the Polarity
dependence of the measured characteristic, and
c) the possibility of progressive deterioration with repeated voltage
applications.
The test procedures recommended below are applicable to test objects
which are not subject to progressive deterioration by repeated voltage appli-
cations. For other types of test objects acceptance criteria other than
those specified in 4.3;2 to 4.3.5 may be needed.

4.3.2 Rated Withstand Voltage Test

4.3.2.1 Tests on nonself-restoring insulation - Unless otherwise specified
for tests on nonself-restoring insulation, three impulses of the specified shape
and polarity are applied at the rated withstand voltage level. The require-
ments of the test are satisfied if no indication of failure is obtained, using the
methods ofdetection specified in the relevant equipment standard.
4.3.2.2 Tests on self-restorin,o inslllation and on combinations of self and non-
scj:!-restoring insulation - A procedure usually recommended for test ob.jects
leaving combinations of self-restoring and nonself-restoring insulation is gtven
below :

Fifteen impulses of the rated withstand voltage and of the specified

shape and polarity are applied. The requirements of the test are satisfieJ
if’ no disruptive discharge occurs in the nonself-restoring insulation and not
more than two disruptive discharges occur in the self-restoring insulation.
In the cases of tests objects having self-restoring insulation only, the
test object can also be deemed satisfactory if the 50 percent disruptive dis-
charge voltage, determined according to 4.3.3 is not less than 1/( 1 - 1.3 o )
times the rated impulse withstand voltage, where o is the per unit standard
deviation of the disruptive discharge voltage. Unless otherwise reco-en-
ded in the relevant equipment standard the following values should be used
for air insulation:

- Lightning impulse tests, cr = O-03

- Switching impulse tests, e = 0.06
NnTE - In Case ofdoubt? relevant equipment standard is to deride in which of
the
the above rategories a particular tyw of equipment should tx pIa&.

4.3.3 Assured Disruptive Discharge Voltage Test - This test is intended as a

clleck that a specified test voltage will cause a disruptive &&arge wit11 a
given high probability ( see also Appendix A ). Five impulses of the specified

16
 IS : 2071 ( Part II ) - 1974

prospective peak value, shape and polarity are applied. If each impulse
causes a disruptive discharge, the requirements of the test have beensatisfied.
Ifmore than one impulse fails to cause a disruptive discharge, the rcquire-
ments of the test have not been satisfied. If one impulse fads to cause dis-
ruptive discharge, ten additional impulses are applied. If each of these
additional impulses causes a disruptive discharge, the requirements of the
test have been satisfied.
4.3.4 Fajiy Percent Disrujdive Disdiurge Voltage Test -The voltage corres-
ponding to a 50 percent disruptive discharge probability U, 50 percent may
be determined as follows ( see also Appendix A ) .
4 At least ten impulses are applied at each of the several test voltage
levels, the steps in voltage being 2-5 percent of the expected 50
percent disruptive discharge voltage. The value of the 50 percent
disruptive discharge voltage is found from a curve of disruptive
discharge probability versus prospective test voltage. The accuracy
of the determination increases with increased number of voltage
applications but there is usually no need to go above 20 applications
at each level.
The same method can be used to determine the standard deviation
of the flashover voltage. See Appendix A, which also gives some
information on confidence limits both for the 50 percent voltage and
for the standard deviation.
b) The following is an alternative procedure to determine the 50
percent value. A voltage Uk is chosen which is considered approxi-
mately equal to the 50 percent disruptivedischarge level. A voltage
interval n U, approximately 3 percent of uk, is also chosen. One
impulse is applied at the level & If this does not cause a disrup-
tive discharge, the next impulse should have the level uk + A U.
If a disruptive discharge occurs at the level &, the next impulse
should have the level Uk - AU.
This procedure is repeated, a number of times each impulse having
a level determined by the effect of the proceeding impulse. ‘I’hc
number of impulse nv at each voltage level Uv, is counted and the
50 percent disruptive discharge voltage is determined as:
c nv 179
U,, percent =
B NV
In this equation, the first level taken into account should be one at
which two or more impulses were applied. This partially corrects
for the error which may be introduced if C$ is much too low or much
too high. For normally distributed flashover probabilitir, t!le
indicated value of US0 percent determined ,as abo\ e 6~
ZTZV - 220 will generally lie within the range from 30 percent to
70 perLent of the true disruptive discharge probability.
IS:2071(PartII)-1974

4.3.5 Determination of Vety Low or Vcly High Disrujtive Discharge

Probabilities - Some guidance for the determination of voltages correspond-
ing to very low and very h&h disruptive discharge probabilities is found in
Appendix A. These procedures are not intended for conventional accep-
tance testing but may be useful in other cases for study related to insulation
coordination.

5. T)WI?S WITH SWITCHING lMPULSE VOLTAGE

5.1 Definitions
5.1.1 Switching Impulse - A switching impulse which should be
distinguished from a lightning impulse is defined in Part I of this standard.
5.1.2 Value of Test Voltage - If not otherwise specified in the relevant
standards, the value of the test voltage is the prospective peak value.
5.1.3 Time to Peak Te, - The time to peak Z-,, is the time interval between
actual zero and the instant when the voltage has reached its peak value.
5.1.4 Time to Hdf V&e Tb - The’ time to half value Zi, is the time
interval between actual zero and the instant on the tail when the impulse
has first decreased to half the peak value.
5.1.5 Time Above 90 Pexent ‘2-a - The time above 90 percent T,t is the
time interval during which the impulse voltage exceeds 90 percent of its peak
value.
Note to clauses 5.1.3 lo 5.1.5 - Practical reasons and theoretical consi-
derations have caused several different time parameters to be used to chara-
cterize a switching impulse. Two such parameters related to the wave
front are the time to peak and the virtual front time. The former is defined
in 5.1.3, the latter is defined either by analogy with virtual front time for
lightning impulses ( see 4.1.4 ) or by using some different reference points and
multiplying factors. The difference between wave front times defined accord-
ing to these various measures is usually small. For switching impulses with
time parameter as given in 5.2.1 and 5.2.2, the time to peak is between
1.4-l .8 timer the virtual wave front time.
Two other time parameters are related to the duration of the impulse;
they are the time to half value and the time above 90 percent. The latter
measurement is found practical in the cases where the wave form of the
impulse is dictated, for instance, by the saturation phenomena in the test
object or the test circuit, or when it is felt that the stress of the test object is
highly dependent on this parameter as for some kind of internal insulation.
These time parameters are defined in 5.1.4 and 5.1.5. It should be empha-
sized that, when defining a switching impulse, only one parameter related to
the wave front and one to the duration should be used.

18
 rs:zo71(PartII)-1974

5.1.6 Time to Chopping Tr - The time to chopping T, of a switching

impulse is the time interval between the actual zero and the moment when
the chopping occurs.
5.1.7 Linear+ Rising Switching Impulse -The linearly rising impulse is
defined in 4.1.8.

5.2 Test Voltage I

5.2.1 Standard Switching Impulse - The standard switching impulse is an

impulse having a time to peak of 250 ps and a time to halfvalue of 2 500 ,US.
It is described as a 25012 500 impulse.
5.2.2 Special Impulses - In the cases where the standard impulse alone is
not considered to be appropriate or adequate, impulses of 100/2 500 and 500/
2 500 are recommended. Also an oscillatory impulse has been proposed,
which has a duration of the first half period between 2 000 and 3 000 ps and
the peak value of the second half period as high as possible up to about 80
percent of the peak value of the first one.
5.2.3 Tolerances - If not otherwise specified in the relevant equipment
standards the~following differences between recorded and specified values are
accepted both for standard and special impulses when the measuring device
meets the requirement of Part III of this standard.
Peak value * 3 percent
Front time f 20 percent
Time to half value f 60 percent
NoTEI - In specific cases, such as during tests on low impedance objects, it may be
difficult to adjust the shape of the impulse within the tolerances recommended, to keep
the oscillations or overshoot within the specified limits, or to avoaid a polarity reversal.
Such cases shall be dealt with separately in the relevant equipment standard.
NOTZ 2 -The measured results of tests using switching impulses with linear wave
fronts are usually present as curves showing the times to disruptive discharge, or aitcr-
natively virtual steepness on one axis and disruptive discharge voltage on the other. For
this reason, and because the range of practical interest is very large, there is no necd~to
specify any particular wave front time or stecpncss to be used, nor any tolerances on
these values.

5.2.4 Gemvation and Reguiremcnts of Test Voltage -Switching impulses are

usually generated by an impulse generator ( see 4.2.4 ). They can also be
generated by discharging a capacitor through the primary winding of a
testing transformer. The basic circuit arrangement for the latter method is
shown in Fig. 7 where C, is the capacitor to be discharged and K is a
switch. The circuit elements within the dotted area used for adjusting the
impulse shape, R, and C, merely being shown as one possible arrangement
for this purpose. Other methods are also in use involving current through a
transformer winding.
NATE -When testing a transformer with switching impulses, the transformer itself
may be used for generating the impulse by the method described for testing a transformer.

19
IS:2071(PartII)-1974

K r-------‘-r TTEST OBJECl

t
I
R2
f I
VOLTAGE
I
I DIVIDER
I
r; I -‘-II’ I
CI I 6 I Id
I
I t
I I
I
i . I . 0
I
I I TRANSFORMER
L--------a
FIG. 7 GENERATIONOF SWITCHINO IMPULSES

The elements of a circuit for the generation of switching impulses should

be so chosen as to avoid excessive distortion of the impulse shape due to pre-
~discharge currents in the test object. Such currents may reach considerable
values esoecially during pollution tests on external insulation at high voltages.
In test c’ircuits of high internal impedance, distortion of the voltage due to
such currents may prevent the occurrence of disruptive discharge.

5.2.5 Iktcasurement of Test Voltage - The test voltage should be measured

with a device specified in Part III of the standard, Alternatively, it may be
measured with a sphere gap ( see IS : 1876-1961* ). However, it may be
noted that this standard gives no information specially related to switching
impulses.

5.2.6 Test Procedure-The test procedure is the same as for lightning

impulse testing ( see 4.3 ).

6. IMPULSE CURRENT

6.1 Definitions
6.1.1 Impulse Currents--Two types of impulse currents are dealt with.
One has a shape which increases from zero to the peak value in a short time
and thereafter decreases to zero approximately exponentially or in the manner
of a heavily damped sine curve. This type is designated by the wave front
time and the time to half value, see 6.1.3 and 6.1.7.

The other type has an approximately rectangular shape and is designa-

ted by the virtual duration of the peak and the total virtual duration
( see 6.1.6 and 6.1.7 ).

6.1.2 Value of Test Current - The value of the test current is normally
defined by the peak value. With some test circuits overshoot of oscillations
*Method for voltage measurement by meam of sphere-gaps ( one sphereearth&).

20
may he present on the current. It will depend on the type of the test and also
whether the value of the test current shall be defined by the actual peak or a
smooth curve drawn through these oscillations.
6.1.3 Virtual Front Time T, - The virtual front time T, is defined as
l-25 times the interval between the instants when the impulse current is 10
percent and 90 percent of the-peak value.
If oscillations are present on the wave front, the 10 percent and 90
percent values should be derived from the mean curve drawn through these
oscillations.
NOTE-The difference between wave frant tima measured according to this de&&
tion and the one in 4.1.4 is generally less than 10 percent.

6.1.4 Virtual Origin O1 -The virtual origin 0, of an impulse current

is defined as the instant preceding that corresponding to the time at which
the current is 10 percent of the peak value by 0.1 T1,.
For oscillograms having linear time sweeps this is equivalent to the
intersection with the x-axis of a straight line drawn through the 10 percent
and 90 percent points on the wave front.
6.1.5 Virtual Time to Half Value, It - The virtual time to halfvalue, ‘I-,
is the time interval between the virtual-origin and the instant on the tail,
where current has first decreased to half the peak value.
6.1.6 Virtual Duration of the Peak of a Rectangular Impulse Current -The
virtual duration of the peak of a rectangular impulse current is defined by the
time during which the current is greater than 90 percent of the peak value.
6.1.7 Virtual Total Duration of a Rectangular Impulse-The virtual total
duration of a rectangular impulse current is the time during which the
amplitude of the impulse is greater than 10 percent of its peak value. If
small oscillations are present on the wave front, a mean curve should be drawn
to determine the time at which the 10 percent value is reached.

6.2 Test Current

6.2.1 Standard Imjnhe Currents - TWO standard impulse currents are used.
One has a virtual wave front time of 8 ps and a time to half value of 20 ps;
it is described as an g/20 impulse. The other has a virtual wave front time
of 4 ps and a time to half value of 10 t~s; it is described as 4/10 impulse.
Rectangular impulse currents usually have virtual durations at the peak
within 2 000 to 3 500 vs.
6.2.2 Tokranccs
6.2.2.1 The following tolerances on the actual recorded values are
permitted when the measuring system meets the requirements of Part III of
this standard.

21
I.S:2071(Partn)-1974
For 8120and4110 impulses
Peak value * 10 percent
Virtual front time, Tr f 10 percent
Virtual time to half value, 7s f 10 percent

Small overshoot and oscillations are tolerated provided that their single
peak amplitude in the neighbourhood of the peak value of the impulse is not
-more than 5 percent of the peak value. Any polarity reversal after the
current has fallen to zero, should be not more than 20 percent of the peak
value.

For rectangular impulses

Peak value + 20 percent
- 0 percent

Virtual duration of the peak + 20 percent

- 0 percent

Overshoot and oscillations are tolerated provided that their single

peak amplitude is not more than 10 percent of the peak value. The total
duration of a rectangular impulse should not be more than 1.5 times the
virtual duration vf the peak and the polarity reversal should be limited to 10
percent of the peak value.

6.2.3 Measurement of Test Current - The test current should be measured

by a device specified in Part III of this standard.

6.2.4 Measurement of Voltage During Tests with Impulse Cunents

6.2.4.1 It is sometimes required to measure the voltage developed across

the test object during tests with high impulse currents. Anyone of the app-
roved devices listed in Part III of this standard can be used for the purpose.

The impulse current lnay induce high voltages in the impulse voltage
measuring circuit, thereby giving measuring errors. To check this, it is
recommended that the conductor which normally joins the voltage divider to
the live end of the test object, should be disconnected-from the live end and
connected to the earthed end of the test object but maintained in approxi-
mately the same loop. Alternatively, the test object is short-circuited or
replaced by a metal conductor. The voltage measured under this condition
when the generator is discharged should be substantially zero at least during
the part of the impulse, which is of importance fos evaluating the results.

NOTE -This short-circuitcheck may be ma& at a reduced currem.

22
 IS:2071(Part10()-1974

APPENDIX A
( Clauses 4.3.3, 4.3.4 and 4.3.5 )
STATISTICAL PROCESSING OF THE TEST RESULTS

A-l. GENERAL
A-l.1 Statement of the Problem - As disruptive discharges are a random
phenomenon, statistical procedures are important to obtain more precise
information from the tests. It is assumed that, to each test voltage level V, a
probability p is attached, for~a disruptive discharge to take place at that level.
The aim of the tests is to determine the probability distribution function
p ( U ) relating p to U.
Usually the probability function is expressed in terms of certain para-
meters out of-which the mean value U, and the standard deviation o are
dealt with below. These parameters and the form of the function can be deter-
mined from a very large number of voltage applications if successive voltage
applications do not change significantly the characteristics of the test object,
or if a fresh test object is used for each application. In certain cases, such as
impulse tests on external insulation, the probability distribution function is
found to be approximately Gaussian from a few up to 95 or even 98 percent
flashover probability. The standard deviation usually is between 2 and 8
percent. Very large values of the standard deviation have been found
during impulse tests in internal insulation.
Little information is available on the character of the probability dis-
tribution function for internal insulation for any type of voltage and fol
external insulation for direct and alternating voltage.
In practical tests, the number of voltage applications is limited. Then
the parameters are determined only within certain limits with a certainty
which can be determined. The values so determined are called U and s.
A brief summary of some simple methods which have been found useful
in the analysis of high voltage test results are given below. The methods are
based on the above assumption of undisturbed test objects.
NOTE -Attention is drawn to the fact that the distribution function is not ncccssarily
symmetrical with reference to the average. Thus, it is necessary to distinguish bctwcen
the terms ’ the average value ’ and the ‘ the most probable value ‘,

A-l.2 Classification of Test Resdts -Test results can be subdivided

into two categories to permit statistical evaluation.
Class I -This class comprises tests made bv repeated application of
voltages of substantially constant shape in which, for each voltage ICYCI:
the proportion of voltage applications causing disruptive diszhargcxs
is recorded. It mainly applies to impulse tests but certain alternating
and direct voltage tests belong to this class.

23
IS: 2071 (PaMI)-

Class 2 - This class comprises procedures in which each application of

the test voltage causes a disruptive discharge. The tests are made by
applying continuously increasing voltages to the test object, and by
measuring the actual disruptive discharge voltages obtained. Such
tests can be made with direct, alternating or impulse voltages. Tests
causing disruptive discharge on the wave front of an impulse belong to
this class.
A-1.2.1 Analysis of Class 1 Test Results - The results of Class 1 tests can be
plotted on Gaussian linear paper ( probability paper ) with the voltage on
the-linear axis. If they lie approximately a straight line, then the distribu-
tion is approximately Gaussian. The voltage corresponding to p = 50 can
be used as determination of the mean value U and the voltage range between
P - 0.5 and p = 0.16 as an estimation ofs.
Other more accurate methods to determine U and s are found in the
literature.
A-1.2.2 Analysis of Class 2 Test Results - The results of Class 2 tests appear
as a series on n voltage values Uv from which the mean flashover voltage U
and the standard deviations can be obtained:

$2 = -
n-l
l ;(uv-c)’
1
Alternatively, nv/n+ 1 can be plotted on a probability paper as a function
of Uv where nv is the number of &shovers up to and including the voltage
Uv and n is the total number of voltage applications. The curve permits a
determination of U and s in the same manner as in A-1.2.1 but does not
necessarily give the same result as the numerical method described above.
A-l.3 Confidence Lb&s - F rom any set of n measurements, statistical
checks may be made to define the limits between which the true mean value
Um and the true standard deviation u may be stated to be with a given
probability PC of being correct. These limits are commonly expressed for
Pe = 0.95 and are then termed ’ 95 percent confidence limits ‘.
The limits for the-true mean value U, are then given by:

urn= f&p<
where tp is a function of& and n, n being the total number of voltage applica-
tions bothin the case of Class 1 and Class 2. The function t, for any value of
pc can be obtained from standard text books on statistics, but for pc = 95
percent and n = 20, the value tp = 2 may be used as an approximation:

24
 IS:2071(PartII)-1974

Confidence limits for the value of the true standard deviation e of the
measurements may be found but with less accuracy from the expression:

o=S[l*@L-]

For a more exact calculation of the confidence limits, the distribution function
shall be known.
NOTE - The assumption ofa Gaussian distribution leads to asymmetricd confidence
limits for a.
A-l .4 Determination of Voltages Corresponding to Very Low or Very
! High Disruptive Discharge Probabilities - In specific cases, a deter-
mination of very low or very high flashover probabilities of the test object is
described.
Conventional assured withstand and assured &shover ( SCG4.3.2
and 4.3.3 ) tests are not suitable to provide such information. The probability
P of passing a conventional withstand test is:
P=(l--p)5+5p(l-_li)14
where P is the probability of a flashover and account is taken of the additional
10 voltage applications if flashover occurs during the first 5.
Wii a test voltage corresponding to p = O-05 the object will have 90
percent probability P of passing the test. With p = O-01 the corresponding
value is P = O-99.
Thus, the fact that a test object has passed or failed such a test gives
little information concerning the actual flashover probability.
Even the test method described in A-1.2.1 is not particularly suited for
_the determination of low or high probabilities. A modified test procedure,
therefore, described below for the particular case of low flashover probabili-
ties. For high probabilities, see Note given at the end of~this clause.
In a first approach, several sets of three voltage applications are made,
starting at a level U. below the estimated withstand voltage. After every
set, the voltage level is increased by a constant amount E of between 2 percent
and 5 percent of Uo. This series of tests is finished at the level U, when the
first flashover occurs.
The test is then continued in a similar manner but with 25 voltage
applications eat each voltage level, starting at the level U, = U, - 3 L.
If no flashover occurs, then the voltage level is increased by successive steps
of E until a flashover occurs. If a flashover occurs during the series of
applications at U,, then the voltage level is reduced successively by steps of
2 E until 25 applications cause no flashover. Thereafter, the level is increaced
by steps of E until a flashover occurs once again.

25
iS:2071(PartII)-1974

The test result is taken as the highest voltage value which has not given
flashover during any series of 25 voltage applications. This value corres-
ponds to a flashover _probability of about one percent and is below eight
percent with 98 percent probability. Usually, the complete test requires
about 75 voltage applications.
NOTE- The above procedure applies also to the case of high probabilities; it is
s&‘icient to replace the words ‘withstand ’ and ‘prohability ’ of ‘withstand ’ by
‘ flashover ’ and ’ probability of flashover ‘. The voltage is decreased from a high value
instead of being increased from a low value.

26
 AMENDMENT NO. 1 FEBRUARY 1977

IS : 2071( Part II )-197tOMETHODS OF HIGH

VOLTAGE TESTING

PART II TEST PROCEDURES

( First Revision )

Alterations

4.1.7.3) - Substitute the following for the existing

(Page11, clause
clause:
’ 4.1.7.3 Time to chopping TC - Time interval between the virtual
origin and the instant of chopping.’
( Page 17, clause 4.3.4 )
a) Line 2 - Substitute ‘ Us0 'for ’ iJ, 50 ‘.
b) Item (b), paragraphs 2 and 3 - Substitute the following for the exis-
ting paragraphs:
‘ This procedure is continued, the level of each impulse being thus
determined by the result of the previous one, until a sufficient number
of observations has been recorded. The number of impulses nv applied
at each level .VV is then counted and the 50 percent disruptive
discharge voltage is given by:
us0 = _“-;$L

In this equation, nv should bl > 20, the first level taken into
account being one at which two or more impulses were applied. This
avoids any appreciable error if Uk is chosen much too high or much
too low. If a better accuracy in the determination of the 50 percent
disruptive discharge voltage is required, the number of voltage appli- L

cations should be increased but generally need not be higher

than 40.’
c) Last but one line- Substitute the following for the existing matter:
‘Cny >20’
( Page 21, xlausc 6.2.1,tart sentence ) - Substitute the following for the
existing sentence:
‘ Rectangular impulse currents have virtual durations of the
peak equal, within the specified tolerances, to 500 ps, 1 000 ~LSor
2 000 PS or between 2 000 l.~sand 3 200 ps. ’

Price Rs 1-z 1

c
 (Page 24, clause A-1.2.1,Ens 4) -Substitute‘p = 0’5 ’ for
‘p = 50 ‘.

( Page 24, clauses A-1.2.1 and A-1.2.2 ) - Substitute * u ’ for ’ U ’

wherever it appears.
C,auSL Page 24, clause A-13 ) - Substitute the following for the existing I
.

“ A-l.3 Confidence Limits - From any set of n measurements, statistical

checks may be made to define limits between which the true mean value
u m and the true standard deviation Q may be stated to lie with a given
probability pCof this being correct.
I These limits are commonly expressed for PC= 0’95 and are then
termed ‘ 95 percent confidence limits ‘.
For Class 2 test results, and on the assumption that-these have an
approximately Gaussian distribution, the limits are given by the
confidence limits for an arithmetic mean value. These are:
z7-stp/v/ u,g a+stp/l//n
~s/(n-l)/Xsp/2<~<~~((n-l)/Xs(l-~~)
where 6 is the arithmetic mean of n voltage values, J is the standard
deviation of n voltage values, tp , x* p/2 and x’ ( 1 - p/2) are variable
in the student’s ( tp ) and in the &i-square ( xs) distribution for a- 1
degrees of freedom and with p a 1 -PC . If no standard statistical
table is available, the following values may be used for pC= 0’95.
n t, It/Y-- V (n- 1) /xsp/z t&I’- l)/x’( l---p/2)
( Lower limit ) ( Upper limit )
5 1’24 0’60 2’87
10 0.72 0.69 1’83
0’55 0.73 1’58
El 0’47 0.76 1’46
30 0.37 0’80 1’34
40 0.32 0.82 1’28
50 0.28 0.84 1’25
NOTE - That the confidence limits of Q are asymetrical.

For II = 20, the 95 percent confidence limits thus are given by:
8-@47s<um<~+@47s
0.76 s < o i; 1.46 s

In the case of Class 1 or Class 3 tests, other methods for the calculation of
the confidence limits shall be used, for which reference should be made to the
jjterature.”

2
 (Page 25, clause A-1.4 ) - Substitute the following for the existing
clause:
’ A-1.4 lktermination of Voltages Corresponding to Very Low or
Very High Disruptive Discharge Probabilities - For some purposes, )
a determination is desired of voltage levels corresponding to very low or
very high probabilities of disruptive discharge of the test object.

Conventional rated withstand tests or assured disruptive discharge tests

as described in 4.3.2 and -4.3.3 are neither intended, nor suitable for provid-
ing such information. The fact that a test object has passed such test
procedures in itself gives little information concerning actual disruptive
discharge probability.

Analysis of Class 1 test results as described in A-1.2.1 gives more

information on the probability distribution, but this is still not adequate for
determining the test voltage levels giving defined low or high disruptive
discharge probabilities, A modified test procedure is therefore described
below for determining the voltage level corresponding to a very low
disruptive discharge probability. ( For high probability, see Note below ).

As a preliminary test, several sets of three voltage applications are

made, starting at a level UO below the estimated withstand voltage. After
every set, the voltage level is increased by a constant amount Z of between
2 percent and 5 percent of UO. The series of tests are finished at the level
VI when.the first disruptive discharge occurs.

The test procedures me &II con&i& iii a similar manner, but with
25 voltage applications at each voltage levei and starting at the level Uz= U,
-3 8. If no disruptive discharge occurs, the voltage level is increased after
eachseries of 25 by successive steps of the same amount I;, until a disruptive
discharge occurs. When a disruptive discharge does occur, say during the
series of applications at level Us, the voltage is then reduced in successive
steps equal to 2 Z until a full series of 25 applications has been applied with
no disruptive discharge. Thereafter, the level is increased in steps equal to
x until one more disruptive discharge is obtained.

The test result is then taken as the highest voltage value which has not
given disruptive discharge during any series of 25 voltage applications. This
vallie corresponds to a disruptive discharge probability of about 1 percent
and there is about 98 percent certainty that it will be less than 8 percent.
Usually, the complete test requires about 75 voltage applications.

NOTE - Almost identical procedures are used to determine the voltage corres-
ponding to very high probabihtics; the foregoing description applies if the words
‘ withstand ’ and cprobability of withstand ’ are replaced by ‘ disruptive discharge
and ‘ probability of disruptive discharge ‘. The voltage is first decreased from a high
value instead of being increased from a low value.

t.
 Addenda

(Page 2, clause 0.4;1, 2 )- Add the word ‘ test ’ after ‘ switching

line
impulse voltage ‘.
( Page 24, clause A-1.2 ) - Add the following matter after the first
paragraph:
‘cIass 3- Class 3 comprises tests made by repeated application
of voltages of substantially constant shape in w.hich the level, for each
voltage applied, is determined by the result of the preceding voltage
application, the first voltage applied being roughly equal to the esti-
mated 50 percent disruptive discharge value.’
( Page 24, clause A-1.2.2 ) -Add the following new ctause
after A-1.2.2 :
‘ A-1.2.3 Analpis of Results from Clars 3 Tests- A test procedure far
making Class 3 tests and the method of analysis of test results SO obtained to
determine the 50 percent disruptive discharge voltage are described in 4.3.4
(b). For other methods and applications reference should be made to the
literature.’
( Page 26, clause A-1.4 ) -Add the following new clause after A-l .4:
‘ A-1.5 Conventiona’r Rated Witi&iind and Assured Disruptive
Discharge Tests -The probability P of passing a conventional rated
withstand test ( see 4.3.2.2 ) is given by the binominal distribution:

P=(1-~ji’5+15~(l-_P)r4+ !+( 1 -.p)”

where p is the probability of a disruptive discharge application at the test

voltage.
Using ?he test procedures described in 43.2, therefore, and for a
test voltage corresponding to fi = 0’10, the object will have 0.82
probability p of passing the %SL With P = O’OI? the corresponding
value is p = 0’_999 5.
Similarly, an assured disruptive discharge voltage test, accord- L
ing to 4.3.3 gives the probability Q of passing the test, where
Q=c5+5(l --P)Pl’
Withp = 0.9, the object will have 0’70 probability of passing the
test and with p = 0’99, the corresponding value is Q = O-994.

(ETDC 19)

4
._ - .._
Reproduced by Reprography Unit, ISI;‘New Delhi