Testing of

Power Transformers

1. Verification of voltage Ratio

and vector Group or phase
displacement and polarity.

1. Turn Ratio Measurement

1.1 Purpose of measurement

The noload voltage ratio between two windings of a transformer is called turn ratio. The
aim of measurement is; confirming the noload voltage ratio given in the customer order
Specifications, determining the conditions of both the windings and the connections and
examining the problems (if any).
The measurements are made at all tap positions and all phases.

1.2 Turn Ratio Measurement

Bridge method
Measurement of turn ratio is based on, applying a phase voltage to one of the windings
using a bridge (equipment) and measuring the ratio of the induced voltage at the bridge.
The measurements are repeated in all phases and at all tap positions, sequentially.
During measurement, only turn ratio between the winding couples which have the same
magnetic flux can be measured, which means the turn ratio between the winding
couples which have the parallel vectors in the vector diagram can be measured.

Theoretical turn ratio = HV winding voltage / LV winding

voltage
The theoretical noload turn ratio of the transformer is adjusted on the equipment by an
adjustable transformer; it is changed until a balance occurs on the % error indicator. The
value read on this error indicator shows the deviation of the transformer from real turn
ratio as %.

(Measured turn ratio) (Designed turn ratio)

% Deviation =
* 100
(Designed turn ratio)
The % deviation of the turn ratios
should be
0.5 %.

1.2 Determining the Vector Group

Depending on the type of the transformer, the input and output windings of a multi
phase transformer are connected either as star ( Y ) or delta( D ) or zigzag( Z ). The
phase angle between the high voltage and the low voltage windings varies between 0
and 360.
Representing as vectors, the HV winding is represented as 12 (0) hour and the other
windings of the connection group are represented by other numbers of the clock in
reference to the real or virtual point. For example, in Dyn 11 connection group the HV
winding is delta and the LV winding is star and there is a phase difference of 330
(11x30) between two windings. While the HV end shows 12 (0), the LV end shows 11
oclock (after 330).
Determining the connection group is valid only in three phase transformers. The high
voltage winding is shown first (as reference) and the other windings follow it.
If the vector directions of the connection are correct, the bridge can be balanced.
Also, checking the connection group or polarity is possible by using a voltmeter. Direct
current or alternating current can be used for this check.
The connections about the alternating current method are detailed in standards. An
example of this method is shown on a vector diagram below.

Fig: Connection
group
representation
and measuring

The order of the measurements:

1)
3 phase voltage is applied to ABC phases
2)
voltage between phases (e.g. AC) is measured
3)
A short circuit is made between C and n
4)
voltage between B and B is measured
5)
voltage between A and c is measured
As seen from the vector diagram, in order to be Dyn 11 group, A.c > AB > B.b
correlation has to realized. Taking the other phases as reference for starting, same
principles can be used and also for determining the other connection groups, same
principles will be helpful.

1.3 Polarity test

1.3.1 Polarity test using voltmeters
Singlephase transformers
For singlephase transformers the polarity can be either additive or subtractive. The low
voltage winding is connected in series with the high voltage winding, either in phase or
in opposite phase. For additive determination of polarity, if the phase displacement is
correct, see figure 1.3.1.

Figure 1.3.1: Connection for additive polarity test

And for subtractive determination of polarity, see figure 1.3.2

Figure 1.3.2: Connection for subtractive polarity test.

Polyphase Transformers.
The vector group must be checked for threephase transformers. This is done by
connecting a terminal from the low voltage side to a terminal on the high voltage side,
see figure 1.3.3. When a threephase supply is connected to the high voltage winding,
potential differences appear between the open terminals and are used to determine the
vector group.

Figure 1.3.3: Polarity test and connection test on threephase transformer

Using one voltmeter

1.3.2 Polarity check using DC current.

This method establishes the polarity of single and threephase transformers by briefly
switching on a DC current source at the high voltage winding, see figure 1.3.4. The
polarity is shown on a polarized voltmeter connected to the low voltage side.

Figure 1.3.4: Basic analog ratio bridge circuit

Testing of
Power Transformers
2.Winding
Resistance
Measurement
.

2. Winding Resistance
Measurement.
2.1 Purpose of the test
Winding resistance serves a number of important functions like:
Providing a base value to establish load loss.
Providing a basis for an indirect method to establish winding temperature
and Temperature rise within a winding.
Inclusion as part of an inhouse quality assurance program, like verifying electric
continuity within a winding.

2.2 General.
Winding resistance is always defined as the DCresistance (active or actual resistance) of
a winding in Ohms [].

Temperature dependence
It should be noted that the resistivity of the conductor material in a winding copper or
aluminum is strongly dependent on temperature. For temperatures within the normal
operating range of a transformer the following relationship between resistance and
temperature is sufficiently accurate:

C+2
R2 = R1

C+1

Where:
R1
=
resistance
at
temperature
2
R2
=
resistance at temperature
1 = temperature in C
C = constant which is a function of material type
IEC [1] specifies: C = 235 for copper = 225 for aluminum

Principle and methods for resistance measurement:

There are basically two different methods for resistance measurement: namely, the
socalled voltmeter ammeter method and the bridge method.

Voltmeterammeter Method
The measurement is carried out using DC current. Simultaneous readings of current and
voltage are taken. The resistance is calculated from the readings in accordance with
Ohms Law. This measurement may be performed using conventional analog (rarely used
nowadays) or digital meters; however, today digital devices such as Data Acquisition
Systems (DAS) with direct resistance display are being used more and more.

Measurement with voltmeter and ammeter

The measuring circuit is shown in figure 2.1.
Resistance RX is calculated according to Ohms Law:

RX=U/I

The advantage of this method is the simplicity of the testcircuit. On the other hand, this method
is rather inaccurate and requires simultaneous reading of the two instruments.

Resistance measurement using a Kelvin (Thomson) Bridge

This measurement is based on the comparison of two voltage drops: namely, the voltage
drop across the unknown winding resistance RX, compared to a voltage drop across a
known resistance RN (standard resistor), figure 2.2.
DCcurrent is made to flow through RX and RN and the corresponding voltage drops are
measured and compared.
The bridge is balanced by varying the two resistors Rdec and RV, which have relatively
high resistance values. A balanced condition is indicated when the galvanometer
deflection is zero, at which time the following relationship holds:

Rdec
RX = RN
Rv

The influence of contact resistances and the connection cable resistances (even of the
connection between RX and RN) can be neglected.

Figure 2.1: Voltmeter ammeter method measuring circuit

Figure 2.2: Kelvin (Thomson) Bridge method

Value of the DCcurrent of measurement

Maximum value:
To avoid an inadmissible winding temperature rise during the measurement, the
DCcurrent should be limited to a maximum 10% of the rated current of the
corresponding winding.
Minimum value:
The lower limit of the DCcurrent is given by the following considerations:
The measuring circuit for all resistance measuring methods consists of a DCsource and
a transformer
winding fixed around an iron core as represented by the following equivalent circuit,
Winding inductance
is strongly dependent on current and displays the following
characteristic for
transformers, see figure 3.4.
As the measuring circuit timeconstant is given by the relation L/R, the currenttime
characteristic differs quite significantly when switching on the DCsource, depending on
the measuring current value (magnetizing current).
Therefore, the DC measuring current should be at least 1.2 times higher than the crest
value of the magnetizing current to be sure to saturate the iron core

Figure 3.3: Equivalent circuit of aFigure 3.4: Inductance of transformer winding

Winding
as a function of the current.

Testing of
Power Transformers
3.Magnetic
Balance and
Magnetising
Current
Measurement.

Magnetic Balance Test on 3phase Transformers

This test is conducted only in threephase transformers to check the imbalance in the
magnetic circuit. In this test, no winding terminal should be grounded; otherwise results
would be erratic and confusing. The test shall be performed before winding resistance
measurement. The test voltage shall be limited to maximum power supply voltage
available at site.

Evaluation Criteria
The voltage induced in the center phase is generally 50% to 90% of the applied voltage
on the outer phases. However, when the center phase is excited then the voltage
induced in the outer phases is generally 30 to 70% of the applied voltage.
Zero voltage or very negligible voltage with higher excitation current induced in the
other two windings should be investigated. The voltage induced in different phases of
transformer in respect to neutral terminals given in the table below.
Left side
phase

Central
phase

Right side
phase

AN

BN

CN

Voltage applied at left

side phase

230 V

180 V

50 V

Voltage applied at central

phase

115 V

230 V

115 V

Voltage applied at right

side phase

50 V

180 V

230 V

EXCITING/ MAGNETISING CURRENT MEASUREMENT

This test should be done before DC measurements of winding resistance to reduce the
effect of residual magnetism. Magnetising current readings may be effected by residual
magnetism in the core. Therefore, transformer under test may be demagnetised before
commencement of magnetizing current test.
Threephase transformers are tested by applying Singlephase 10 kV voltage to one
phase (HV terminals) and keeping other winding open circuited and measuring the
current at normal, minimum and max. tap positions.
Keep the tap position in normal position and keep HV and LV terminals open. Apply
1phase 10kV supply on IV terminals.
Measure phase to phase voltage between the IV terminals and current on each of the IV
terminals. The set of reading for current measurement in each of the tap position should
be equal. Unequal currents shall indicate possible short circuits in winding. Results
between similar singlephase units should not vary more than 10 %
.The test values on the outside legs should be within 15 % of each other, and values for
the centre leg should not be more than either outside for a threephase transformers.
Results compared to previous tests made under the same conditions should not vary
more than 25%. If the measured exciting current value is 50 times higher than
the value measured during precommissioning checks, then there is likelihood
of a fault in the winding which needs further analysis. The identical results
confirm no damage due to transportation. The availability of test data of normal
condition and faulty condition results help us to analyze the problem in future.

Measurement of Magnetization Current at Low Voltage

For 3phase transformers, the test shall be conducted either with 415 V, 3phase (neutral
grounded) or 230 V, 1 phase (preferred). For single phase transformers, the test shall be
conducted with 230 V.
This test is performed to locate defect in magnetic core structure, shifting of windings,
failures in turn insulation or problem in tap changers.
The acceptance criteria for the results of exciting current measurement should be based
on the comparison with the previous site test results or factory test results. The general
pattern is two similar high readings on the outer phases and one lower reading on the
center phase, in case of three phase transformers. An agreement to within 25% of the
measured exciting current with the previous test is usually considered satisfactory. If the
measured exciting current value is 50% higher than the value measured during pre
commissioning checks, then the winding needs further analysis.

Testing of
Power Transformers
4. MEASUREMENT
SHORT
CIRCUIT IMPEDANCE

OF

MEASUREMENT OF SHORT CIRCUIT IMPEDANCE

This test is used to detect winding movement that usually occurs due to heavy fault
current or mechanical damage during transportation or installation since dispatch fro the
factory.
Ensure the isolation of Transformer from High Voltage & Low voltage side with physical
inspection of open condition of the concerned isolators/disconnectors. In case tertiary is
also connected, ensure the isolation of the same prior to commencement of testing
The measurement is performed in single phase mode. This test is performed for the
combination of two winding. The one of the winding is short circuited and voltage is
applied to other winding. The voltage and current reading are noted.
The test shall be conducted with variac of 0280 V, 10 A, precision RMS voltmeter and
ammeter. The conductors used for shortcircuiting one of the transformer windings
should have low impedance (less than 1mohm) and short length. The contacts should
be clean and tight.
The acceptable criteria should be the measured impedance voltage having agreement to
within 3 percent of impedance specified in rating and diagram nameplate of the
transformer. Variation in impedance voltage of more than 3% should be considered
significant and further investigated.
The formula for calculating the percentage impedance with current and frequency correction
is:

Where:

Vtest = Test
voltage Vrated

=Rated voltage
Itest = Test current
Irated = Rated
current ft = Test
frequency fr =
Rated frequency

Testing of
Power Transformers
5.Measurement of
dissipation factor
(tan) of the
insulation system
capacitances.

Measurement of dissipation factor (tan) of the insulation

system capacitances
The purpose of the measurement

The insulation powerfactor test, similar to the insulation resistance test, allows certain
conclusions to be drawn concerning the condition of the transformer insulation.
The significance of the power factor figure is still a matter of opinion. Experience has
shown, however, that the powerfactor is helpful in assessing the probable condition of
the insulation when good judgment is used.

General
IEC defines the power factor as the ratio between the absorbed active power to the
absolute value of the reactive power. This corresponds to tan. IEEE [51], on the other
hand defines the insulation powerfactor as the ratio of the power dissipated in the
insulation in watts, to the product of the effective voltage and current in voltamperes
(corresponding to the apparent power) when tested using a sinusoidal voltage. Insulation
powerfactor is usually expressed in percent [51].
Measurement of powerfactor values in the factory is useful for comparison with field
powerfactor measurements and assessing the probable condition of the insulation. It
has not been feasible to establish standard powerfactor values for the following reasons:
There is little or no relationship between powerfactor and the ability of the
Transformer to withstand the prescribed dielectric tests.
The variation of powerfactor with temperature is substantial and erratic.
The various liquids and insulation materials used in transformers result in Large
variations in insulation power factors [51].

The measuring circuit / the measuring procedure [51]

Measurement using a bridge
The method is based on comparing the capacitance CX (transformer under
test) with a wellknown capacitance CN (standard capacitor).

Conventional ScheringBridge
Figure shows the measuring circuit for the insulation powerfactor
measurement of a twowinding transformer using a conventional Schering
bridge.

Instrumentation
The ScheringBridge test circuit consists of three main parts:
The unknown capacitance CX, which represents the transformer under test whose
powerfactor (or tan) and capacitance are to be measured.
The standard capacitor CN, which must be a HV capacitor with very low dielectric
losses. Normally its capacitance is between 100 pF and 10 nF.
The ScheringBridge casing contains resistors R3, R4 and r, adjustable capacitor C4
and galvanometer G. In order to reduce the influence of external disturbances,
coaxial cables must be used for the connection between CX (the transformer under
test) to the bridge and also between standard capacitance CN and the bridge.

Figure: Measuring circuit for the measurement of power factor and winding
Capacitances.
When the bridge is balanced, the unknown capacitance CX and tan can be calculated
using the following equations:

In most bridges the following resistance values are used for R4, to simplify the
calculation: 100/, 1000/ or 10000/ etc. in ohms.
For a 50 Hz measurement, with R4 = 1000/ and C4 in nF, the insulation power factor tan
will be:

A modern tan bridge with current comparator and

microprocessor
This bridge uses basically the same measuring principle as described above. Figure 18.1
b shows the measuring circuit for dissipation factor and capacitance measurement with a
modern tan _ measuring bridge with incorporated microprocessor.
The currents are balanced in a comparator (morewinding differential transformer) and
quadrature current is injected to balance the losses.
For the unknown capacitance Cx, the standard capacitor CN and the connections
between transformer and bridge are the same as mentioned above for the conventional
Schering Bridge.

MEASURING METHODS:
CAPACITANCE AND TAN MEASUREMENT OF BUSHINGS
C & Tan measurement of bushings shall be done at 10kV with fully automatic test kit
so as to have reliable test result.
For 3Ph autotransformer, short together all 400kV, 220kV and Neutral (isolated from
earth) Bushings. Also short all 33kV Bushings and earth the same.
Measurement of C1 Capacitance and Tan: Connect the crocodile clip of the HV
cable to the top terminal of the shorted HV/IV bushings. Unscrew the test tap cover,
Insert a pin in the hole of the central test tap stud by pressing the surrounding contact
plug in case of 245 kV OIP Bushing and remove the earthing strip from the flange by
unscrewing the screw (holding earth strip to the flange body) in case of 420 kV OIP
Bushing. Connect the LV cable to the test tap (strip/central stud) of the bushing under
test to the C & TAN KIT through a screened cable and earth the flange body. Repeat
the test for all Bushings by changing only LV lead connection of the kit to test tap of the
Bushing which is to be tested.
Measurement of C2 Capacitance and Tan : HV lead to be connected to the test tap
of the bushing under test (if required additional crocodile type clip may be used) and LV
of the kit to be connected to the ground. HV of the bushing is to be connected to the
Guard terminal of the test kit. Test to be carried out in GSTg mode at
1.0kV.
For measurement of 33kV Bushing Tan Delta, earth HV/IV Bushings (already shorted).
Apply HV lead of the Test kit to shorted 33kV Bushings and connect LV lead of the test kit
to Test tap of the Bushing under test.
Measurements shall be made at similar conditions as that of a previous measurement.
The oilpaper insulation combination of bushings exhibit fairly constant tan delta over a
wide range of operating temperature. Hence, effort is to be made for testing at
temperature near to previous test and Correction factor need not be applied.

Do not test a bushing (new or spare) while it is in its wood shipping crate, or while it is
lying on wood. Wood is not as good an insulator as porcelain and will cause the readings
to be inaccurate. Keep the test results as a baseline record to compare with future tests.

It is to be ensured that C& Tan measurement of bushings and testing of turrets carried out
before installation. This will prevent installation of bushings having C& Tan values beyond
permissible limits.
It is to be ensured that Test Tap points are earthed immediately after carrying out the
measurements for
that particular Bushing and earthing of test tap to be ensured by carrying out continuity test.

CAPACITANCE AND TAN MEASUREMENT OF WINDINGS

The combination for C & tan measurement of winding is same as that of measurement
of IR value. The summery of probable combination is given below:
Auto
Transformer
windin
g)

(Tw
o

Test
Mode

Shunt
Reactor

HV + IV to LV

UST V

HV + IV to E
LV to E

HV to
E

Tes Mod
t
e

3
Test
winding Mode
Transform
er

GST HV to LV1

UST

GSTg

HV to LV2

UST

GSTg

LV1 to LV2

UST

HV to
Ground

GSTg

LV1 to
Ground

GSTg

LV2 to
Ground

GSTg

Table: Combination for C & tan measurement of winding for various

Transformers/ Shunt Reactor.
Ensure that test specimen is isolated from other equipments. Removal of Jumpers from
Bushings is Pre
Requisite for C & Tan Measurement of Windings.
For ICTs (AutoTransformers): Shorting of all three phase Bushings (400kV&220kV)
and neutral to be done. In case of single phase, 400kV, 220kV and neutral Bushings to
be shorted Capacitance and Tan measurement of windings should be done in following
combinations:

Winding
Test Combination
No.

Test
mode

Cap
Symbol

Test lead
Connection

Remarks

1.

HVIV/LV

UST

CHL

2.

HVIV/ LV+G

GST

CHL +CHG

3.

HVIV / LV
with Guard

GSTg

5
6.

HVIV/LV

LV/

HVIV +G

LV/ HVIV
with Guard

HG

UST

CHL

GST

CHL +CLG
LG

GSTg

LG

lead of test kit

HV
to
HV/I bushings
V
of
transformer
LV lead of test kit to
LV
bushing of
transformer
do
do

LV to be
Guarded

lead of test kit

LV to
HV/IV
bushings
of
transformer
HV lead of test kit to
LV
bushing of
transformer

do
do

HV to be
Guarded

Table :.Winding combination for C & tan measurement for auto transformer.
Measurement intercheck can be done by calculating C1= C2C3 & C4= C5 C6 &
DF1=C2DF2C3DF3 / C2C3= C4DF4C5DF5 / C4C5 Where C stands for capacitance
and DF for dissipation factor or tan and attached suffix (16) denotes the sr. no. of
test in above table.
For Reactors: All 400kV and neutral Bushings to be shorted. HV of the test kit to be
connected to shorted Bushings and LV of the test kit to be connected to Earth
connection. Measure the Capacitance and tan Delta in GST mode. Neutral connection
with earth/ NGR to be isolated before the test.

Testing of
Power Transformers
6. Measurement
Insulation Resistance.

Measurement of insulation resistance

Insulation resistance tests Megger tests are performed to determine the insulation
resistance from individual windings to earth or between individual windings. Knowledge
of the insulation resistance is of value when evaluating the condition of the transformer
insulation.
Insulation resistance is commonly measured in megohms, (M).
It should be stated, that variations in insulation resistance can be caused by numerous
factors including: design, temperature, dryness, and cleanliness of parts, especially of
bushings. When insulation resistance falls below specified value, it can often be brought
back to the required value by cleaning and drying.
Insulation resistance varies with the applied voltage. Any measurement comparisons
should always be carried out at the same voltage.

Figure: Principal measuring circuit for the insulation resistance measurement

IEEE Std. C57.12.00 [50] also specifies the insulation resistance measurement between
core and earth. It shall be measured after complete assembly of the transformer at a
level of at least 0.5 Kv DC for a duration of 1 minute.
The test is conducted with the help of megaohmmeter. IR is proportional to the leakage
current through/over the insulation after capacitive charging and absorption currents
become negligible on application of DC voltage. Insulation resistance shall be measured
after the intervals of 15 sec, 60 sec and 600 sec. The polarization index (PI) is defined as
the ratio of IR values measured at the intervals of 600 and 60 seconds respectively.
Whereas, the dielectric absorption is the ratio of IR values measured after 60 sec and 15
sec. IR is normally measured at 5 kV DC or lower test voltage, but the test voltage should
not exceed half the rated powerfrequency test voltage of transformer windings.
Polarization index (PI) is useful parameters for logistic interpretation of IR test results.
This ratio is independent of temperature and gives more reliable data for large power
transformers. A PI of more than 1.3 and dielectric absorption factor of more than 1.25 are
considered satisfactory for a transformer when the results of other low voltage tests are
found in order. PI of less than 1 calls for immediate corrective action. For bushings, an IR
value of above 10000 Mohms is considered satisfactory.
The IR value of transformer is dependent on various factors such as configuration of
winding insulation structures, transformer oil, atmosphere condition etc. therefore,
present trend is to monitor oil characteristics

for judging the condition of dryness of the transformer and not to rely solely on absolute
values of IR. It may
be note that no national/international standards specify minimum insulation resistance values of 25 transformers. The
value of IR may be very low under heavy fog or humid conditions.

During IR measurement, we must ensure following conditions:

Transformer is disconnected from other associated equipment
Bushings are cleaned and free of moisture
Transformer tank and core are properly grounded
Both ends of winding under test are shortcircuited.

Measuring Methods:
IR measurements shall be taken between the windings collectively (i.e. with all the
Windings being connected together) and the earthed tank (earth) and between each
winding and the tank, the rest of the windings being earthed. Before taking
measurements the neutral should be disconnected from earth. Following table gives
combinations of IR measurements for autotransformer, three winding transformer &
Shunt Reactor.
For Auto
transformer
HV + IV to LV
HV + IV to E
LV to E

For 3 winding
transformer
HV + IV to LV

For Shunt
Reactor
HV to E

HV + LV to IV
HV + IV +LV to E

Where HVHigh voltage, IVIntermediate voltage, LVLow voltage/Tertiary

voltage windings, E Earth
Unless otherwise recommended by the manufacturer the following IR values as a thumb
rule may be considered as the minimum satisfactory values at 30C (one minute
measurements) at the time of commissioning.

Insulation resistance varies inversely with temperature and is generally corrected to a

standard temperature (usually 20 C) using table (Source: BHEL instruction Manual) as
given below.

PI= R10 / R1 (dimensionless),

Where PI is Polarisation Index and R is resistance
The following are guidelines for evaluating transformer insulation using polarization index
values:

A PI of more than 1.25 and DAI of more than 1.3 are generally considered satisfactory for
a transformer when the results of other low voltage tests are found in order. PI less than
1 calls for immediate corrective action. For bushings, an IR value of above 10,000 M is
considered satisfactory.

Testing of
Power Transformers
7. CHECK LIST FOR
ENERGISATION
OF TRANSFORMER.

CHECK LIST FOR ENERGISATION OF TRANSFORMER/

REACTOR
PRELIMINARY CHECKS
1.Release air at the high points, like oil communicating bushings, buchholz petcock, tank
cover and the cooling devices including headers, radiators, pumps, expansion joints etc.
of the transformer. Air release should be resorted from low points to high points.
2.Check the whole assembly for tightness and rectify where necessary.
3.Check the general appearance and retouch the paint work if need be.
4.Check that the valves are in the correct position:

Tank: valves closed and blanked

Cooling circuit: valves open
Conservator connection: valves open
Bypass: valves open or closed as the case may be.
Onload tap changer: valves open

5.Check that the silica gel in the breather is blue and that there is oil in the breather cup (oil
seal)
6.Check that CCCLG are shorted
7. Check the oil level in the main conservator and the conservator of onload tap
changer, bushing caps, flanges, turrets, expansion bellows as per manufacturers
recommendation. Level should correspond to 35 C mark on oil level gauges for BHEL
transformers
8.Check the bushings:
Oil level (bushings fitted with sightglasses)
Adjustment of sparkgaps /arcing horn gaps, if provided
Conformity of connection to the lines (no tensile stress on the terminal heads)
Bushing CT secondary terminals must be shorted and earthed, if not in use.
Neutral bushing effectively earthed
9.Check the onload tap changer:
Conformity of the positions between the tap changer control cubicle and the tap
Changer head
Adjustment of the tapchanger control cubicle coupling
Electric and mechanical limit switches and protective relays
Step by step operation local and remote electrical operation as well as manual
Operation and parallel operation, if any
Signaling of positions
10. Check the quality of the oil:
Draw off a sample from the bottom of the tank
Carry out DGA and oil parameters test (i.e. BDV, Moisture content, resistivity & tan at
90C and IFT )

before energisation.
11. Check the oil of OLTC chamber, if not good, drain and fill with filtered oil
upto desired level. 12 Check that equalising link between OLTC tank and
Main tank is removed
13 Extraneous materials like tools, earthing rods, pieces of clothes, waste etc.
should be removed before energisation.

CHECKING OF AUXILIARY AND PROTECTIVE CIRCUITS

1.Check temperature indicator readings and their calibrations
2. Check the setting and working of the mercury switches of winding and oil temperature
indicators and presence of oil in the thermometer pockets
Follow the same procedure for the thermal replicas
3.Check the direction of installation of buchholtz relay.
4.Check the operation of the buchholz relay and the surge protective relay of the tap
changer :
Alarm and tripping
Protections and signals interlocked with these relays
5.Check the insulation of the auxiliary circuits in relation to the ground by 2 kV megger
for 1 min.
6.Check the earthing of the tank and auxiliaries like cooler banks at two places.
7.Measure the supply voltages of the auxiliary circuits
8.Check the cooling system :
Check the direction of installation of oil pumps
Check the direction of rotation of the pumps and fans
Check the working of the oil flow indicators
Check the setting of the thermal overload relays
Go through the starting up sequences, control and adjust, if necessary, the relay
time delays
9.Check that there is efficient protection on the electric circuit supplying the
accessories and tightness of all electrical connections
10.Check the differential protection, overcurrent protection, restricted earth fault
protection, overfluxing protection etc. are in service and settings are as per CC/Engg
recommendations. After the inspection / tests are completed, the transformer may be
energised from the incoming side on NO LOAD. The initial magnetising current at the
time of switching will be very high, depending upon the particular moment in the cycle.
The transformer should always be soaked for few hours under constant care i.e. keep it
energised for twelve hours. Excessive vibrations of radiator parts etc. should be located
and corrected. The transformer hum should be observed for any abnormality. After that it
may be checked for gas collection. Should the gas prove to be inflammable.

Testing of
Power Transformers
8. PRECOMMISSIONING CHECKS
/TESTS FOR TRANSFORMERS.

PRECOMMISSIONING CHECKS /TESTS FOR TRANSFORMERS &

REACTORS
Once oil filling is completed, various precommissioning checks/ tests are performed to
ensure the healthiness of the Transformer/ Reactor prior to its energisation. Various
electrical tests to be performed and their significance are given below.
Nam
Sr. e
of Test

Purpose of test/ check

No. Check point

3.1
3.2

3.3

Core insulation
tests

To check the insulation between Core (CC&CL) and

Ground

Operational
Checks
on protection
System

Operational Checks on cooler bank (pumps & Fans),

Breathers
(silica gel or drycol), MOG, temperature gauges
(WTI/OTI), gas
actuated relays (Buchholz, PRD, SPR etc.) and
simulation test of
protection system
Test reveals the condition of insulation (i.e. degree of
dryness of
paper insulation), presence of any foreign contaminants
in oil and
also any gross defect inside the transformer (e.g.
Failure to remove
the temporary transportation bracket on the live
portion of tap
changer part)

Insulation
Resistance(IR)
measurement

3.4

Measurement of C1 & C2 Capacitance and Tan in UST

mode.
Changes in the normal capacitance of an insulator
indicate
abnormal conditions such as the presence of moisture
layer,short
circuits or open circuits in the capacitance network.

Capacitance
and
Tan
measurement
of bushings

3.5

3.6

3.7

Capacitance and
Tan
measurement
of
windings

Dissipation factor/Loss factor and capacitance

measurement of
winding is carried out to ascertain the general condition
of the
ground and interwinding insulation

Turns ratio
To determine the turns ratio of transformers to
(Voltage
identify any
measureme
ratio)
ntabnormality in tap changers/ shorted or open turns etc
Vecto
r
Group

To determine the phase relationship and polarity of

& transformers

Polarity
3.8

3.9

Winding
resistance
measurement

To check for any abnormalities due to loose

connections, broken
strands and high contact resistance in tap changers

Magnetic
Balance
test

This test is conducted only in three phase transformers

to check the
imbalance in the magnetic circuit

This test is conducted to ascertain possibility of short

3.10 Floating Neutral circuit in a
point
measurement
winding.
This test is used to detect winding movement that
3.11 Measurement of usually occurs
due to heavy fault current or mechanical damage
Short Circuit
during
transportation or installation since dispatch from the
Impedance
factory.
Exciting/Magnetis To locate defect in magnetic core structure, shifting of
3.12 ing
windings,
failures in turn to turn insulation or problems in tap
current
changers.
These conditions change the effective reluctance of the
measurement
magnetic
circuit thus affecting the current required to establish
flux in the
core.
To measure the vibrations of core /coil assembly in the
tank of the
3.13 Vibration
measurement of reactor. Movement of the corecoil assembly and
Oil
shielding
structure caused by the timevarying magnetic forces
immersed
results in
Reactor
vibration of the tank and
quipment. These
ancilliary
vibrations
have detrimental effects such as excessive stress on
the corecoil
assembly
Operational
3.14 check
on OLTCs

To ensure smooth & trouble free operation of OLTC

during
operation
.

This test is performed to

operatio
3.15 Stability of
check
the proper n
of
Differenti
&
protectio
Transforme
Reacto
Differential, REF
al
REF
n
of
r
& r
by
of
simulating actual conditions. Any problem in CT
Transformer/
connection, wrong
cabling, relay setting can be detected by
this test.
Reactor
3.16 Tests/ Checks on
To ascertain the healthiness of bushing current
Bushing Current
transformer at the
Transformers
(BCTs)
time of erection

Frequency
3.17 Response
Analysis (FRA)
measurement

mechanic
transforme
To assess the al
integrity of the
r.
Transformers while experiencing severity of short circuit
current
looses its mechanical property by way of deformation of
the
winding or core. During precommissioning this test is
required to
ascertain that Transformer active part has not suffered
any severe
impact/ jerk during
transportation.

Oil sample for DGA to be drawn from transformer main

3.18 Dissolved Gas tank before
Analysis (DGA) of commissioning for having a base data and after 24 hrs.
of charging
oil
subsequently to ensure no fault gas developed after
first charging.
sample
DGA analysis helps the user to identify the reason for
gas formation
& materials involved and indicate urgency of corrective
action to be
taken
Thermovision
3.19 Infra

A thermo vision Camera determines the temperature

distribution
on the surface of the tank as well as in the vicinity of
red scanning (IR the Jumper
connection to the bushing. The information obtained is
thermography) useful in
predicting the temperature profile within the inner
surface of tank
and is likely to provide approximate details of heating
mechanism.
Thermovision scanning of transformer to be done at
least after 24
hrs. of loading and repeated after one week.

1.0 TRANSFORMER AND REACTOR

1.1 Following points to be checked After Receipt
of transformer / reactor at Site:
1.1.1 N2 pressure and Dew point to be checked after receipt of transformer at Site. It
should be within permissible band (as per graph provided by manufacturer & given
below in Fig1)
1.1.2 The data of impact recorder shall be analyzed jointly in association with the
manufacturer. In case the impact recorder indicates some serious shocks during
shipment, further course of action for internal inspection, if necessary shall be
taken jointly. Impact Recorder should be detached from the Transformer/ Reactor
preferably when the main unit has been placed on its foundation.
1.1.3 Oil Samples shall be taken from oil drums/ tanker received at site and sent to our
Lab (CIOTL / IOTL) for oil parameter testing. The copy of test certificate of routine
testing at oil refinery should be available at site for comparison of test results.
1.1.4 Unpacking and Inspection of Accessories to be carried out taking all precautions so
that the tools used for opening do not cause damage to the contents. Fragile
instruments like oil level gauge, temperature indicators, etc. are to be inspected
for breakage or other damages. Any damaged or missing components should be
reported to equipment manufacturer, so that the same can be investigated or
shortage made up as per the terms/ conditions of the contract.

1.1.5

Core Insulation Test shall be carried out to check insulation between Core (CC&CL)
and Ground. (Not applicable for Air Core Reactors)

1.1.6 After receiving the accessories at site same should be inspected and kept ready for
immediate erection:
If erection work can not start immediately due to some reasons, then accessories
should be repacked into their own crates properly and packing list should be
retained.
All packings should be kept above ground by suitable supports so as to allow free air
flow underneath. The storage space area should be such that it is accessible for
inspection; water does not collect on or around the area and handling/transport
would be easy. Proper drainage arrangement in storage areas to be ensured so
that in no situation, any component get submerged in water due to rain, flooding
etc.. Immediately after the receipt of main unit and also the accessories, same
should be inspected and if found satisfactory, the unit should be erected
completely and filled with dry transformer oil as per the instruction.
It is preferable to store the main unit on its own location/foundation. If the foundation
is not likely to be ready for more than three (3) months, then suitable action
plan has to be taken from the manufacturer regarding proper storage of
the Main Unit.
If the transformer/ Reactor is to be stored up to three (3) months after arrival at site, it
can be stored with N2 filled condition. N2 pressure to be monitored on daily basis
so that chances of exposure of active part atmosphere are avoided. In case of drop
in pressure, dew point of N2 has to be measured to check the dryness of the
Transformer/ Reactor.

In case of storage of transformer in oilfilled condition, the oil filled in the unit should be
tested for BDV
and moisture contents once in every three months. The oil sample should be taken
from bottom valve. If BDV is less and moisture contents are more than as given for
service condition then oil should be filtered.

1.1.7 During erection, the exposure of active part of transformers should be minimized.
Further either dry air generator should be running all the time or dry air cylinders
may be used to minimize ingress of moisture. The transformer should be sealed off
after working hours. It is practical to apply a slight overpressure overnight with dry
air or N2 inside less than 300 mbar (30 kPa or 0.3 atmospheres). Next day the
pressure is checked and suspected leaks may be detected with leak detection
instruments , with soap water or with plastic bags tightened around valves (being
inflated by leaking air) For oil filled units whenever oil is drained out below the
inspection covers, job will be treated as exposed. Other exposure activities are as
below:
1)
Bushing erections
2)
Jumper connections of Bushings
4)
Fixing bushing turrets on cover
5)
Fixing bushing turrets on side
6)
Core insulation checking
7)
Buchholz relay pipe work fixing on cover.
8)
Gas release pipes/equaliser pipe fixing.
9)
Entering inside the tank for connections/inspection etc.
For oil filled units depending upon the level up to which the oil is drained decides the
exposure time. All such exposure time should be recorded in a log sheet to decide the oil
processing (drying) and oil filling of transformer. For transformers with a gas pressure of
2.5 3 PSI, the acceptable limits of dew point shall be as under:

TABLE Variation of Dew Point of N2 Gas Filled in

Transformer Tank w.r.t
Temperature.

Parameters of Transformer Oil

The oil sample from the transformer tank, after filling in tank before commissioning
should meet the following specifications as per IS: 1866 2000 (latest Revision) given
in table below:

Testing of
Power Transformers
9. Partial Discharge
Measurements.

Partial Discharge
Measurements
9.1 Purpose of
measurement
A partial discharge measurement (PDmeasurement) is a nondestructive tool used to
establish the condition of a transformer insulation system. The goal of partial discharge
measurement is to certify that no harmful PD sources exist. A PDmeasurement makes it
possible to detect and localize areas within the transformer which are exposed to
elevated dielectric stresses, i.e. stresses which in the long run can be harmful to safe
transformer operation.
Partial discharge measurements are explicitly specified in standards or in customer
specifications. They are to be carried out in conjunction with dielectric tests in high
voltage laboratories using ACvoltage in the power frequency range.
For HVDC transformers PD measurements are also carried out on dielectric tests with DC
voltages
For onsite PD measurements (for example on repaired transformers) other types of PD
free excitation may also be carried out [221].
Partial discharge measurement should generally be the last dielectric test conducted on
the transformer.

9.2 General
Partial discharge is a partial voltage breakdown within a series of insulating elements
between two electrodes of different potential, (capacitances C2 and C3, see figure 9.1).
During a typical PD measurement, the magnitude of the detectable value of partial
discharge activity is recorded as a function of the applied voltage. A partial discharge
can be interpreted as the rapid movement of an electric charge from one position to
another. For very fast changes, or during the first instant after charge movement, the
individual insulation links in a series of connected links between two line terminals can
be regarded as a number of series connected capacitors.

BU = bushing
HV = high voltage
NT = neutral terminal
C1, 2, 3 = active part of transformer (including oil)
C1 = weak region

Ct

= test object capacitance (C2 and C3)

Figure 1 : Schematic representation of a part of the transformer
insulation.

If the two line terminals are connected together via an external capacitor Ck, see figure 9.2,
the charge movements within the seriesconnected insulation links (capacitances C2 and
C3, see figure 9.1) will also be reflected in the charge of external capacitor Ck. The charge
movements can be detected as circulating current impulses i(t) in the parallelconnected
capacitors Ck and Ct, see figure 9.2.

Ct = test object capacitance

Ck = coupling capacitor
G = voltage source
i (t) = PD current pulses
i~k,~t = displacement currents
Z = voltage source connectors
Q = transferred charge
Ut = voltage at parallelconnected capacitors
Zm = measuring impedance
Figure 9.2: Equivalent circuit for PD measurement.
Two requirements must be fulfilled to initiate a partial discharge (i.e. electric breakdown)
within the weak region of an extended insulating system:
Local electric field stress E in the weak region must be greater than the inception
electric field of the PD Source.
Free electrons must be available to initiate the electric breakdown, see clause A 9.1.
Excessive stress in the weak region can result from design flaws, contamination or
deviation from permissible tolerances in the manufacturing process, insulating material
flaws, etc. Another possibility is hidden damage to the insulation caused by preceding
tests.

9.3 Principle of PD measurement

All PD measuring methods are based on the detection of PD current impulses i(t)
circulating in the parallel connected capacitors Ck (coupling capacitor) and Ct (test
object capacitance) via measuring impedance Zm.
The basic equivalent circuit for PD measurements is presented in figure 9.2 [212].
The measuring impedance Zm can either be connected in series with coupling capacitor
Ck or with the test object capacitance Ct.
As discussed in section 9.3 General, PD current impulses are generated by charge
transfers between parallel connected capacitor Ck (coupling capacitor) and Ct (test
object capacitance).
Present IEC and IEEE Standards have both established rules for measuring and
evaluating electric signals caused by partial discharges together with specifications on
permissible magnitude.
The IEC approach to the processing of the recorded electric signal is different from the
IEEE approach. IEC transforms the signal to an apparent electric charge generally
measured in picocoulombs (pC), while IEEE transforms the signal to a Radio Interference
Voltage RIV, generally measured in micro volts (V).
The use of the RIVmethod for PDsignal detection will be abandoned, although the IEEE
standard has not yet been officially approved. The detection of apparent charge in pC is
the preferred method now in use in IEEE Std. C57.113 [56].
For the detection of apparent charge the integration of the PDcurrent impulses i(t) is
required.
Integration of the PD current impulses can be performed either in the time domain
(digital oscilloscope) or in the frequency domain (bandpass filter). Most PD systems
available on the market perform a quasi integration of the PD current impulses in the
frequency domain using a wideband or narrowband filter, see clause A 9.2.

Note:
For short duration currents (nsrange) the test voltage source is practically decoupled
from the PD measuring circuit (parallel connection of Ck and Ct) by the inductive
impedance Z (stepup transformer connections).
For the HVcomponents without any bushing an external coupling capacitor Ck must be
connected in parallel with the test object Ct, see figure 9.3.

PDS = PD system
Ck = coupling capacitor
Ct = test object capacitance
Z = voltage source connection
Zm = measuring impedance
Figure 9.3: Test circuit for measurement without capacitive tap

9.4 PD measurement on transformers

Circulating PD current impulses generated by an external PD source (in the test circuit)
or by an internal PD source (in the insulating system of the transformer) can only be
measured at the transformer bushings. Bushing capacitance C1, see figure 9.5a,
represents the coupling capacitor Ck, which is connected in parallel with capacitance Ct
(test object = total capacitance of the transformer insulating system). An example of a
typical transformer insulating system is shown in figure 9.4.
For power transformers the measuring impedance is generally connected between the
bushing measuring tap and earth, i.e. in parallel with capacitance C2, see figure 9.5a.
For bushings without a capacitive tap an external coupling capacitance Ck must be
connected in parallel with the bushing, see figure 9.5b.
There are some essential differences between the two Standards (IEC and IEEE)
regarding the evaluation of the PD current impulses.

PR = press ring
BI = barrier insulation
LV = low voltage
HV = high voltage
RW = regulation winding
Figure 9.4: Transformer insulating system; Barrier system with shielding rings and angle
rings towards yoke.

9.5.1 IEC Standard [3]

According to IEC, PD measurements are conducted by measuring the apparent charge,
q. In this context the apparent charge is obtained by integrating the PD current impulse
using wide band or narrow band filter, see clause A 9.2.
The PD measuring system is connected via a coaxial cable to measuring impedance Zm, see
figure 9.5.
The apparent charge q, measured in pico coulombs (pC), corresponds to the charge
transferred during the U voltage drop compensation process at one of the parallel
connected capacitances Ct (transformer insulation) and bushing capacitance C1 or
coupling capacitance Ck, see figure 9.2 and 9.5.
This voltage drop U may be caused either in the test object (internal partial discharge in
the bushing or in the transformer insulating system) or in the test circuit (external partial
discharge). If PD activity is detected during the test, the PD source must be investigated,
see section 9.7.
The magnitude of measurable apparent charge qm in pC must be defined by the
calibrating procedure for each test circuit.

Calibration
Calibration of the PD test circuit is performed using a battery operated calibrator. The
calibrator consists of a squarewave generator with adjustable amplitude U0 connected
in series with a small capacitor C0 (C0 should be less than 10% of Ck). For PD
measurements on transformers, the calibrator is connected across the bushing, or across
the coupling capacitor connected in parallel with the bushing, see figure 9.5. Calibration
must be performed separately for each bushing.

C1 = Ck = coupling
capacitor SE =
shielding electrode
PDS = PD system
CAL =
calibrator C2 =
capacitive tap
Zm = measuring impedance
Figure 9.5a: Calibration circuit for PD measurement on transformers; bushings with
capacitive tap
Under the assumption that C0 << Ck, the injected impulse from the squarewave
generator corresponds to the charge q0, which is set to predefined values (100 pC, 1000
pC etc.) by the adjusting the amplitude U0. IEC 60270 recommends that the rise time of
the injected impulse should be 60ns, amplitude U0 between 2V and 50 V, selectable
polarity, and the repetition rate 100 Hz.

q0 = U0 C0

where:
q0 = injected charge
U0 = adjustable voltage of the square
wave generator C0 = calibrator
capacitance
The measuring circuit, consisting of the test capacitance Ct of the test object, coupling
capacitor Ck, measuring impedance Zm, coaxial cable and measuring system, is now
calibrated, see figure 9.5.
During the PD test the measuring system values are read directly in pC. This pC reading
is only valid for the specific calibrated bushing.

PDS = PD
system CAL =
calibrator

Ck = coupling capacitor
Zm = measuring
impedance SE =
shielding electrode
Figure 9.5b: Calibration circuit for PD measurement on transformers; bushings without
capacitive tap

9.5.2 IEEE Standard [50], [51]

For routine PD measurements, IEEE Standards require the measurement of RIV (RIV =
Radio Interference Voltage). RIV is determined in V (interference voltage). A narrow
band filter performs quasiintegration of PD current impulses with quasipeak detection
at center frequencies between 0,85 MHz and 1,15 MHz. The narrowband pass filter is
applied to allow suppression of external noise in nonshielded laboratories by varying the
center frequency of the filter. The measuring system is called a RIVmeter or a radio
noisemeter.
The RIV in V depends on both, on the transferred charge and on the repetition rate of
the PD impulses (number of PD impulses per second). This is why it is not possible to
directly convert measured RIV values in V into values of apparent charge in pC, see
clause A 9.2.
The transferred charge (measured in V) is the result of a compensation process of the
voltage drop U at one of the parallel connected capacitances Ct (transformer insulation)
and bushing capacitance C1 = Ck or coupling capacitor Ck, see figure 9.5.

Calibration
The PD test circuit is calibrated in the same way as the one for measuring apparent
charge in pC, see figure 9.5. Assuming that C0 << Ck, the applied sinusoidal voltage
corresponds exactly to the values defined by the adjustable amplitude of U0 in V (100
V, 1000 V etc...).
During the PD test the measuring system values are read directly in V. This V reading
is only valid for the specific calibrated bushing.

According to IEEE Standard C57.12.90 [51] PD activity within the transformer may be measured in
terms of apparent charge (picocoulomb). This approach should normally provide several advantages,
including less attenuation of signal.

9.5.3 Sensitivity of the PD measurement

The true charge q1 released during an internal electric breakdown in a weak region in
the transformer insulation is no measurable, see figure 9.1. Only the charge transfer
between the capacitance of the weak region C1 and capacitances of the insulating
system (C2 and C3 in figure 9.1) is detectable at the bushing (Ck). Capacitances C2
and C3 (winding and transformer insulating system) are directly connected to this
bushing. The relationship between true charge q1, apparent charge q and measurable
charge qm is discussed in clause A 9.3.
The sensitivity of the PD measurement (i.e. the measurable apparent charge qm in pC or
interference voltage in V) is strongly dependent on the components in the test circuit;
especially on the ratio of the test capacitance Ct (insulating system of the transformer)
and coupling capacitor Ck (capacitance of the bushing). The influence of the coupling
capacitor value on the sensitivity of the PD circuit is shown in figure 9.6 [212]. This
shows that calibration must be repeated if major changes are made to the test circuit
(connection of new coupling capacitor, etc.).

q = apparent charge
qm = measurable charge
qm =
Ck
Ct +
q
Ck
Figure 9.6: Influence of coupling capacitor Ck on the measuring sensitivity qm /q.
For an extended insulating system such as that of a transformer, the calibration
described above is only valid for PD defects that are close to the bushing.
The real PD impulse currents especially those generated by internal PD sources hidden
deep within the insulating system are heavily attenuated by the RLCM network of the
transformer [214]. (RLCM stands for resistances, inductances, capacitances and mutual
inductances.) Such slow PD impulses are only detectable at the bushings and may not
always be correctly measured by the applied PD system, see clause A 9.2.
The amplitude of apparent charge qm is therefore not always a meaningful criterion to
decide if the PD source is dangerous to the insulating system. A procedure for
investigating the internal PD source is described in section 9.7.

Note 1:
Normally Zm and the measuring system must be matched to correctly quasi integrate
the PD current impulses. Mixing measuring impedances Zm and detection systems of
different manufacturers is not recommended.

Note 2:
For delivery tests it is only specified to measure on bushings for Um > 72,5 kV (IEC) or on
bushings Um >115 kV (IEEE)
It is recommended that all bushings of the transformer under test should be equipped
with measuring impedances Zm to ensure that PD activity can be detected
simultaneously. If PD activity is detected in the test circuit, identification and localization
of the PD source is then much faster, see section 9.9.
To avoid external discharge (corona) in the PD measuring circuit all transformer bushing
tops should be covered with shielding electrodes (including earthed bushings), see figure
9.5. Shielding electrodes should also be used for all sharp metallic parts on top of the
transformer and in the voltage source connections, see clause A 9.8. The voltage source
in the HV laboratory must be PDfree. All objects in the test field close to the transformer
under test must be earthed. Typical external noise sources are discussed in clause A 9.4.

9.6 PD measuring procedure

The procedure for PD measurement is basically defined by the induced voltage test
procedure shown in section 8, Table 1. There are some differences between PD
measuring procedures according to IEC and those specified in the IEEE Standards.
Sometimes the customer specifies a special PD procedure according to his experience
and to the recommendations of the national technical committee.

9.6.1 IEC Standard [3]

According to the IEC Standards PD measurements shall be carried out in conjunction with
induced voltage test on all transformers with highest voltage for equipment (Um) above
a certain level. The PD measurement is mandatory for long duration induced voltage test
(ACLD) as well as for short duration induced voltage test (ACSD). Time sequences for
ACSD and ACLD are given in section 8, figures 8.3 and 8.4.
PD activity should be checked at all bushings where the system voltage is higher than
72,5 kV. PD activity is measured in pC. Any wideband pass filter or narrowband pass
filter can be used as a PD measuring system, see clause A 9.2.
The first PD measurement (values of apparent charge in pC) should be made at a low
testvoltage level (ca. 10% Ur). This value serves as a reference for the background noise
level in HV laboratory. According to the IEC Standards, the background noise level must
be lower than half of the required pC value of apparent charge for the specific
transformer. In a shielded HV laboratory, the background noise level is sufficiently low.
The following PD measurements should be made at each test voltage level indicated in
figures 8.3 and 8.4 (parts A, B, D, E) with the exception of the enhancement level (part
C). All measured pC values at all bushings of the transformer should be documented (see
example Table 1).

During the long duration test (part D in figure 8.4), the PD activity should be checked at least
every 5 minutes
at each bushing. The best way to check PD activity in the transformer insulating system is to
apply a multi channel PD measuring system capable of detecting PD activity at all bushings
simultaneously (8channel PD system).

Acceptance criteria for PD test

The PD test is considered successful if no continuous PD activity greater than the
specified apparent charge amplitude in pC is detected at any bushing, and if there is no
rising trend in the apparent charge amplitude during the long duration test.
The recommended acceptable values of apparent charge given in the IEC Standards are:
300 pC at 130% Um
500 pC at 150% Um
the level of continuous PDactivity does not exceed 100 pC at 1.1 Um
where:
Um = highest voltage for equipment according to IEC or in
other words highest r.m.s phasetophase voltage for which the
transformer winding is designed.

9.6.2 IEEE Standards [50], [51]

According to the IEEE Standards, the PD measurement should be performed at the one
hour level of the test voltage (defined by the customer), before the enhancement level
and during the onehour test after the enhancement test voltage, see figure 8.5 in
section 8. The duration for part A must be sufficiently long to initiate possible PD activity
in the oilimpregnated transformer insulating system (minimum 10 minutes). The physics
of PD activity in oilimpregnated insulating systems is discussed in clause A 9.1.
PD activity should be checked at all bushings where the system voltage is 115 kV. PD
activity is measured in V. A narrowband pass filter should be used as a PD measuring
system. The recommended center frequency for the narrowband pass system is 1 MHz.
If there is high background noise, the center frequency can be varied between 0,85 MHz
and 1,15 MHz.
The first PD measurement (RIV values in V) should be performed at a low test voltage
level (about 10% Ur). This value serves as a reference for the background noise level in
the HV laboratory. According to the Standards, the background noise level must be lower
than half of the required V value for the specific transformer. In a shielded HV
laboratory the background noise level is less than a few V.

The following PD measurements should be performed at each test voltage level indicated in
figure 8.5 (parts
A, C) of section 8 with the exception of the enhancement level (part B). 50 All measured V values at all bushings of
the transformer should be documented, see example Table 2.

During the onehour test (part C in figure 8.5), the PD activity at each bushing should be
checked at least every 5 minutes.

Acceptance criteria for PD test

The PD test is considered successful if no continuous PD activity greater than the
specified RIV level in V is detected at any bushing, and if there is no rising trend of RIV
during the long duration test. According to the
IEEE Standard the PD test was successful if the following conditions were met:
The magnitude of the PD level did not exceed 100 V
The PD level increase during one hour of test did not exceed 30 V
The PD level during the onehour test did not exhibit any steadily rising
trend, and no sudden, sustained increase in levels occurred during the last
20 minutes of the test.

9.7.1

Investigation of external PD sources

The first step is to exclude all possible external PD sources. Typical external PD sources
are, see also clause A 9.4:

Conducting particles on the bushing surface

Nonshielded sharp points on the transformer or in the test circuit
Bad connections on shielding electrodes
Unearthed metallic objects close to the transformer
Noise or internal PD from the voltage source Electric discharges in air (corona)
generated by sharp electrodes (tippoint electrode) are easy to detect using a
portable ultrasonic detector (corona gun), see figure 9.7.

A PD problem in the voltage source can easily be checked by separately measuring the
voltage source.

9.7.2 Investigation of the type

of PD source
A PD source type is defined by its specific statistical behavior (PD pattern). The statistical
behavior of the PD source is mainly influenced by the availability of starting electrons,
which trigger an electric discharge in the weak region in the transformer insulating
system, see physics of discharge processes in clause A 9.1. The availability of starting
electrons is strongly dependent on the PD source itself (conducting or nonconducting
material) and on the position of the PD source with respect to the metallic electrode. As a
result, five typical PD patterns representing the physical processes of the PD sources
exist (for a visual interpretation, see Table 3).
These typical PD patterns are based on physical electric discharge processes in the weak
region, see clause A 9.1, and can theoretically be detected in any insulating system (if
the measuring circuit is sufficiently sensitive). PD patterns (statistical analysis of the PD
signals) are not influenced by the structure of the insulating system [213].

Note:
The structure of extended insulating systems, like those of a transformer, heavily
attenuate the amplitude of the original internal PD current impulses, but do not change
their statistical behavior.
An advanced PD system Phase Resolving Partial Discharge Analyser = PRPDAsystem is
used to record a PD pattern [215]. The advanced PD system, see clause A 9.5, performs
a statistical analysis of the recorded PD data. At the specific test voltage, the PD activity
is saved as a function of the phase position and of the amplitude of apparent charge
during a preset time (two dimensional multichannel analyser). The results are finally
presented as a two dimensional PD pattern, see figure 9.8a. The third dimension (color)
indicates the total number of PD impulses collected during the preset measuring time.
For the statistical analysis of a PD source a minimum of 3000 cycles are needed (for 50
Hz the preset measuring time must be 60 seconds).
Figure 9.8b shows the wellknown presentation of PD impulses during one
cycle, recorded using a conventional PD system.
To investigate the PD source, the first PD pattern should be recorded under the
following test conditions [216]:
Inception voltage of the PD source
Extinction voltage of the PD source
10% above the inception voltage
Further test conditions for the investigation of the PDsource are dependent on the
results of the analysis of the PDpattern.

Figure 9.8a: Registration of PD impulses (bubbles and surface discharge);

advanced PD system (statistical analysis of PD
impulses)
Analysis of the PD pattern is based on comparing the recorded real PD pattern with
the typical PD pattern types presented in Table 3:
In reality, the five typical PD patterns appear in many variations. Due to the continuous
change of both the surrounding area at the location of the PD source and of the PD
source itself (due to the electric discharge), there are only a few PD patterns that exhibit
constant behavior during the test.

Table 3: Typical PD sources in the transformer insulating system

Basic PD pattern characteristics that should be analyzed are [216]:

Phase position of the PD signals

Symmetry of the PD signals during the positive and negative sine wave
Number of PD signals per cycle
Reproducibility of the PD pattern

Interpretation and screening of the correct type of PD pattern from the real PD pattern
results, requires experience and a strong interpolation capability. If PD defects are
superimposed, a comparison with the typical types of PD patterns and finding the correct
type of PD pattern becomes much more difficult.
An overview of the typical PD sources in the transformer insulating system together with
their typical PD pattern and their typical behavior during the test is presented in Table 4.
If there is a clear indication of internal PD activity in the transformer insulating system,
localization of the PD source must follow, see sections 9.8 and 9.9.

Figure 9.8b: Registration of PD impulses (bubbles and surface discharge);

conventional PD system one cycle only

Table 4:Typical PD sources in the transformer insulating system(Figures 9.10a,b,c)

Figure 9.10a: Typical PD defects in the

transformer insulation; conducting
material = PD pattern type 1

Figure 9.10b: Typical PD defects in the

transformer insulation; conducting
material = PD pattern type 2

Figure 9.10c: Typical PD defects in the transformer insulation;

bubbles = PD pattern type 3 and 4

9.8 Detection of acoustic PD signals

An acoustic PD signal is a mechanical vibration in the elastic medium (acoustic wave). Means
to locate PD sources that are generated by the electric discharge in the weak region using
acoustic wave analysis (acoustic emission) are based on time delay measurements between
the electric PD signal (oscilloscope trigger) and acoustic signals detected by a minimum of
three acoustic sensors positioned along the transformer tank wall .
Piezoelectric transducers (crystal) with a resonance frequency between 60 150 kHz are
normally used as acoustic sensors. A fourchannel digital oscilloscope is needed to
analyze acoustic PD signals. A typical result of detecting acoustic PD signals correlated
with an electric PD signal is shown in figure 9.11.
The location of the PD source in the insulating system is calculated from the time delay
between the electric and acoustic PD signals using the triangulation method, see figure
9.12 [217]. The velocity of acoustic waves in oil is around 1400 m/s. This method is
theoretically only applicable for direct waves i.e. acoustic waves propagating through
the oil only.

a)
= electric PD signal [0,50 V/div]
b)= acoustic PD signal [20 mV/div], sensor 1
c)
= acoustic PD signal [10 mV/div], sensor 2
d)= acoustic PD signal [50 mV/div], sensor 3
Figure 9.11: Detection of acoustic PD signals
Acoustic wave propagation in the transformer, see clause A 9.6, is heavily influenced by
the complicated structure of the insulating system (winding barriers, core, tank walls). In
a complicated structure the acoustic signal emitted by the PD source changes along its
propagation path. Both the amplitude (attenuation) and the signal shape (absorption,
dispersion) are influenced. It must therefore be possible to distinguish between direct
propagated and wallpropagated waves when analyzing the time difference between
electric and acoustic PD signals for the purpose of localization, see figure 9.12. This
information is theoretically hidden in the wavefront of the acoustic signal detected by a
sensor on the tank wall, see figure 9.13.

9.8.1 Sensitivity of the measurement

Localizing PD defects that emit the acoustic wave directly into oil, similar to a metallic
particle lying on the surface, can easily be detected (amplitude of apparent charge >
100 pC).

PD defects hidden in the solid insulation, similar to a metallic particle in the insulation, are quite
difficult to detect (amplitude of apparent charge >1000 pC), due to the different propagation
velocities in different materials and due to the reflection phenomena of acoustic waves in the
extended insulating system of the transformer, see clause A 9.6.
PD defects in the main insulation of the transformer are the most difficult to detect due
to the transformer board barriers and outer winding, see figure 9.4.
Acoustic waves caused by PD defects in the core are very difficult to analyze and may
give very misleading results.

C = velocity of acoustic wave

t1, t2, t3 = time of signal arrival at the sensor

Note:

Figure 9.12: Location of PD source using triangulation method

Detection of acoustic PD signals should be performed at a test voltage level close to the
inception voltage of the PD source to achieve a reliable correlation between electric and
acoustic PD signals. At higher test voltages the number of PD impulses per cycle usually
increases and is likely to initiate additional PD sources. In case of unidentified PD activity,
a detailed and in depth investigation of the PD source should follow. A minimum of one
day is required for the PD measurements and an additional day to analyze the results,
see section 9.9.

Figure 9.13: Analysis of acoustic waves of PD sources

a)Detection of direct acoustic wave (through the oil)
b)
Detection of reflected acoustic wave or wave propagated through
different media.

9.9 Detailed investigation of the PD source

Detailed investigation of the PD source goes beyond the requirements specified in the
IEC and IEEE standards. This procedure is adapted to the behavior of the PD source. The
goal of the investigation is to find the PD source as quickly as possible [216], [219].
Detailed investigation of the PD source requires the following commercially available
equipment:
High frequency current transformer: High frequency current transformers (HF CT,
100 kHz 30 MHz) must be used as a measuring impedance Zm to detect the real PD
current impulses. HF CTs are connected to each bushing of the transformer
(multiterminal measurement).

Spectrum analyzer: A spectrum analyzer is used both to analyze PD current impulses in the
frequency domain
for localization of the PD sources and to detect PD signals as a variable bandpass filter (quasiintegration of 61 PD
current impulses) as the frontend of the advanced PDsystem.

Advanced PDsystem: For example ICM system or ICMsys8 is used to record and
analyze the statistical behaviour of the PD signals for investigation of the type of the PD
source (PDpattern).
Digital oscilloscope: A digital oscilloscope is used as a control device for the digitized
signals and as an analyzing device for time resolved signals when localizing the PD
activity.

9.9.1 Investigation and localization of the PDsource

The investigation of the PD source should be performed in the following sequence:
Analysis of the frequency spectrum of the PD current impulses
Analysis of the PD pattern (statistical analysis of the PD signals)
Efforts to locate the PD source using analysis of electric PD signals in the frequency
domain
Efforts to locate the PD source using analysis of electric PD signals in the time domain
These steps above should be repeated at the following test voltage levels:
At the inception voltage of the PD source
At the extinction voltage of the PD source
At different test voltage levels up to the required test voltage level
As a function of the time of the applied voltage
The analysis of the results is mainly based on comparing the real PD signal behavior with
the characteristic behavior of the specific insulating system. The characteristic behavior
of the insulating system of the transformer under test must be defined by a special
calibrating procedure.
An example of the characteristics of a transformer using a conventional calibrator for PD
measurements is shown in figure 9.14:
Frequency spectra at each bushing (spectrum analyzer works in full span, see figure
9.14a).
Sensitivity (in pC) of the measuring circuit at each bushing (spectrum analyzer
works in span zero), see figure 9.14b (band pass filter).
Registration of the sensitivity in pC at each bushing with ICMsystem (video
out signal from the spectrum analyzer), see figure 9.14c (PDpattern).
Crosscoupling of calibrating signals in the frequency domain, see figure 9.14d
(localization of PD source).
Crosscoupling of calibrating signals in the time domain, see figure 9.14e (localization of
PD source).
For the crosscoupling characteristics of the insulating system, a calibrating signal of
1000 pC is usually injected at one specific bushing and the response is measured at all
other bushings (multiterminal method). This procedure is repeated for each bushing
(time needed for a threephase transformer is about 6 hours).

Figure 9.14a: Characterization of the transformer; calibration in the frequency

domain.

Figure 9.14b: Characteristics of the transformer, calibration in pC (spectrum analyzer as a

band pass filter)

Figure 9.14 c: Characteristics of the transformer, calibration in pC (ICM record of calibrating

impulse)

Figure 9.14d: Characteristics of the transformer; cross coupling in the frequency

domain

Localization
Localization of the PD sources is based on the following theory:
Electric PD signals (PD current impulses) propagate from the PD source through the
RLCM network of the transformer, see clause A 9.7. The response of this network to
excitation by a PD current impulse at any location in the insulating system can only be
detected at the bushings. The measured real PD current impulses at the bushings are
compared with the characteristic values obtained during the calibrating procedure
described above, both in the frequency domain and the time domain, see figure 9.14.
The theory of propagation of electric PD signals through the transformer RLCM network is
discussed in clause A 9.7.

Figure 9.14e: Characteristics of the transformer; crosscoupling in the time

domain.

Analysis of PD current impulses in the frequency domain [216], [219]

In order to define the background noise condition of the entire test circuit (background
frequency spectrum) the first data registration when taking the PD measurement is
performed using a spectrum analyzer at about 10% Ur at all bushings.
The background frequency spectra serve as the basis for unambiguous identification of
repeated PD activity detected at the specific bushing.
For each subsequent test voltage level, frequency spectra are checked and compared
with the background frequency spectrum at each bushing. Any PD activity in the test
circuit generates a change in the background frequency spectrum (visual interpretation).
PD signals close to the bushing generate a frequency spectrum similar to the calibrating
signal, see figure 9.15 (1U).

PD signals transferred to the bushing via the RLCM network of the transformer insulating
system generate a
spectrum with defined resonances, see figure 9.15 (1V).

1U = PD source close to the bushing

1V = PD signal coupled from the phase U
1W = background noise
Figure 9.15: Real PD signal in the frequency domain detected at
different bushings. The basic frequency spectra characteristics that should be
analyzed are:
Amplitude of the power spectrum in dBm
Frequency range of the power spectrum
Typical resonances
Reproducibility of power spectra
Comparison of the PD signal frequency spectra with the results of the characterization of
the transformer gives the first indication of the location of the PD source.

Note:
Analysis of the PD signals in frequency domain can only be performed for repetitive PD
signals. Sporadic PD signals can only be registered with a peak detector (conventional or
advanced PD measuring systems)

Analysis of PD currents in the time domain [214]

Besides analyzing the PD signals in the frequency domain (spectrum analyzer), the PD
current impulses are also analyzed in the time domain (oscilloscope). The highest PD
current amplitude is used as a trigger signal for the oscilloscope (channel, HV) and the
response of the RLCM network at all other bushings is usually systematically analyzed at
the second channel (NT), see figure 9.16. Additional simultaneous recording of four
specific channels may be used to confirm previous results.
Recorded PD current signals in the time domain are again compared with the calibrating
signals (visual interpretation).
PD signals close to the bushing generate a timeresolved signal similar to the calibrating
signal, see figure 9.16 (HV).
PD signals transferred to the bushing via the RLCM network of the transformer insulating
system generate a timeresolved signal that is comparable with the response of a RLC
filter, see figure 9.16 (NT).

HV = PD source close to the HV bushing

NT = PD signal coupled to the neutral terminal
Figure 9.16: Real PD signal in the time domain.
Basic timeresolved PD signal characteristics that should be analyzed are:

Maximum amplitude of the PD current signal in mV

Risetime of the PD current signal
PD current signal oscillations
Reproducibility of the PD current signals

A comparison of the PD signals recorded in the time domain with the results of the
characterization of the transformer gives the second indication of the location of the PD
source.

Note:
In the event of superimposed PD sources, it is possible to distinguish between two
different PD current signals by varying the trigger level. While attempting to localize the
PD source, the type of the PD activity (PD pattern) is continuously analyzed in the same
way as described in section 9.7

9.9.2 Final analysis of the results

All results recorded during the detailed investigation of the PD source under different test
conditions must be analyzed (an experienced person takes at least 4 hours) before
making any decisions regarding the next steps in the investigation procedure. In depth
analysis of all results will reliably identify the type of PD source (PD pattern), give
information about the location of the PD defect and provide a basic idea of how
dangerous the PD source is for the transformer insulating system.
The behavior of the PD source during the investigation procedure allows us to distinguish
between dangerous PD sources and nondangerous PD sources. The amplitude of the
apparent charge is not always a meaningful criterion for this decision, see clause A 9.3.
Dangerous PD sources for the transformer insulating system are:
PD source with inception voltage below 100% Ur

 PD source with extinction voltage below 100% Ur

PD source hidden in the solid insulation
PD source with continuous change in PD pattern
PD source where the amplitude of the apparent charge increases with the time of
applied voltage
PD source where the number of PD signals per cycle increases with the time of
applied voltage.
Less dangerous PD sources for the transformer insulating system are:
PD source with inception voltage above the protection level (external surge arresters) in
the supply
PD source with extinction voltage above 100% Ur
Gas bubbles in the oil
PD source with a constant PD pattern
PD source where the number of PD signals per cycle decreases with the time of applied
voltage
Successful resolution of the PD problem can be finally reached by discussing the results
with the design engineers to mutually find the real cause of the PD source. Depending on
the results of the discussion, the next steps in the procedure may be:
Additional calibration
Additional investigation of the PD source in HV laboratory
Conditioning the PD source
Redrying the transformer insulation
Modifying the identified weak region (cause of PD defect)
Disassembling the transformer

Setup for threephase induced test including partial discharge measurement.

9.10 Measuring uncertainty

The calibrated values for measurement of apparent charge in pC and of RIVvalues in V
are only valid for PD defects close to the bushings. For all PD defects far from the
bushing, the uncertainty may be more than 50%, see clause A 9.3.
Since the resonance phenomena of the PD detection circuit (bushing, measuring
impedance, measuring cable and band pass filter of the PD measuring system) are not
known, the RIV system may deliver confusing results if the center frequency of the
narrow bandpass filter (usually 1 MHz) is the same as the resonance frequency of the
specific PD detection circuit, see clause A 9.2.
Exact conversion of the measured interference voltage in V to a corresponding
apparent charge is only possible in certain cases, see clause A 9.2.

Appendix A 9
Partial Discharge Measurement

A 9.1 Physics of partial discharge

Partial discharge (PD) is an electric breakdown in the weak region of an extended
insulating system, see figure 9.17.

PR = press ring
BI = barrier insulation
LV = low voltage
HV = high voltage
RW = regulation winding
Figure 9.17: PD source in insulating system of the transformer.
An electric breakdown of the insulating material between two electrodes means
generally that the distance between them has been bridged by electric charges (high
ohm resistance has changed to low ohm resistance). Electric breakdown occurs if the
following conditions are fulfilled:

 Local electric field E in kV/mm is greater than the breakdown field strength Eb in kV/mm
of
the specific insulating material (different values for different insulating materials)
Starting electrons are available.
For a homogeneous electric field (main insulation between the windings; see figure 9.17)
the breakdown field strength Eb is defined as:

where:
Eb = breakdown field strength
in kV/mm Vb = breakdown
voltage in kV
d = distance between two electrodes in mm

Note:
There are many places in the transformer insulating system, where the electric field is
nonhomogeneous, see figure 9.17. At these locations, the calculated homogeneous
electric field must be multiplied by the electrode shape factor to estimate the maximum
electric field. When designing a transformer, the maximum electric field strength of
critical locations must be calculated using a field program.
The following physical mechanisms describe the electric breakdown between two electrodes
[111] and [220]:
Generation of primary electrons to start the electric breakdown
Charge multiplication and transport phenomena to bridge the distance between
electrodes
Charge storage phenomena to support the next electric breakdown
(memory effect of the PD source)

Generation of primary electrons

Generation of primary electrons depends on:

Maximum electric field Emax.

Electrode material (conducting, nonconducting)
Electrode shape (homogeneous or nonhomogeneous electric field)
Electrode surface condition (uncovered, or covered with insulating material)
Smoothness of the surface (microtip electrodes)

The best condition for generating starting electrons occurs for an uncovered metallic tip
electrode as shown in figure 9.18. Starting electrons are generated by field emission
from the surface if the magnitude of the local electric field E exceeds the field emission
values of the specific material. PD activity appears as soon as the local inception electric
field is reached.

E = electrical field [kV/cm]

x = distance between the electrodes [cm].
Figure 9.18: Charge distribution (upper graph) and electric field distribution (lower
graph) for negative tip electrode (upper figure, a) and positive tip
electrode (lower figure b).
The most difficult location for generating a starting electron is a weak region without
contact to the metallic electrodes, as shown in figure 9.19. In this situation the starting
electrons are generated by radioactive ionization (cosmic, Xray, etc.). Due to the lack of
starting electrons, there is a delay before the PDactivity starts (up to several minutes),
even if the local electric field exceeds the inception value.

E= local electric
field E0 =
electric field
= dielectric permitivity of the material
Figure 9.19: PD defect without metallic electrodes

Note:

For PD sources in the transformer insulating system the electrode material can be either an interface
between 70 different insulating materials (for example, solid insulation with a gas bubble) or a conducting
material (for example copper covered with paper insulation).

Charge multiplication and transportation

Charge multiplication and transport processes, see figure 9.20, are based on an
avalanche mechanism described by the equation:

where:
n = number of electrons at
distance d d = distance
between electrodes
n0 = number of available starting
electrons ed = electron
avalanche
An electron avalanche can only cause an electric breakdown (bridging of the distance
between electrodes) if the following condition is fulfilled:

where:
= factor which is a function of the local
electric field E d = distance between the
electrodes.

d = distance between electrodes

Figure 9.20: Development of electric breakdown
Both physical processes described above take time to develop (breakdown delay), see
figure 9.21. During this time the applied voltage (i.e. the local electric field) must be
constant. The limiting parameters for PD activity in the weak region of the insulating
system are:
Local electric field E exceeds the design rating ( factor)
Size of the weak region (bubble) is sufficient (distance between electrodes in the weak
region)
Duration of the applied voltage (local electric field) is long enough to develop the
discharge processes

T = total time
delay tS =
starting electrons
tA = avalanches
Ud = statistical voltage breakdown

Figure 9.21: Time delay of electric breakdown.

Charge storage
Charge storage mechanisms are important for the weak regions with no contact to
conducting electrodes (bubbles in insulating system), see figure 9.22. Charge storage
and detrapping mechanisms strongly influence successive electric breakdowns in the
weak region. The repetition rate of the PD impulses and the type of PDpattern (PDtype
pattern type 5 in Table 3) are permanently changing. The physical phenomena caused by
the charge storage mechanism can be observed by the recorded PD patterns during a
long duration test (several hours at a constant value of the test voltage). An increasing
repetition rate in the PD impulses indicates continuous damage to the insulating system
in the vicinity of the PD source.

q = charge
E = local electric
field PD = partial
discharge E0 =
electric field
1,2 = dielectric permittivity of the molecules
Figure 9.22: Schematic presentation of charge storage

Note:

This short introduction into the physics of partial discharge has shown that electric breakdown in
insulating materials is strongly influenced by the statistical behavior of the discharge
mechanisms. To interpret PD results, it is necessary not only to consider the amplitude of the
apparent charge, but also to analyze the statistical behavior of the PD source. Statistical analysis
of PD activity is performed using an advanced PD system (Phase Resolving Partial Discharge
Analyser, see clause A 9.5).

A 9.2 Principle of quasiintegration

A PD signal is always detected as a current impulse i(t) via a measuring impedence Zm [212],
see figure 9.23.

Ct = test object
capacitance CK =
coupling capacitance
G = voltage source
i (t) = PD current
impulses i~k,~t =
displacement currents Z
= voltage source
connection q =
transferred charge
Ut = voltage at parallelconnected capacitor
Figure 9.23: Equivalent circuit for PD measurement

International standards (IEC 60270) [26] require that apparent charge should be measured:

where:
q = apparent charge
in pC i(t) = PD
current signal
Integration of the PD current impulses can be performed either in the time domain
(digital oscilloscope) or in the frequency domain (bandpass filter). Most PD systems
available on the market, perform a quasi integration of the PD current impulses in the
frequency domain using a wideband or narrowband filter. The following assumption
is made when nonperiodic PD current impulses in the frequency domain are integrated
using a bandpass filter, see figure 9.24 [109]:

where:
q = apparent charge F(0) = amplitude frequency spectrum at frequency f = 0 Hz
F(f) = amplitude frequency spectrum at frequency f Hz
The limiting frequency for constant spectral amplitude density, i.e. for a correct
integration of the PD impulses in the frequency domain, depends on the pulse shape of
the PD impulse as shown in figure 9.24 [109].
For a PD impulse with risetime T1 = 1 s and halfvalue time T2 = 5 s, the limiting
frequency for F(f) = F(0) is about 5 kHz. For PD systems that are based on quasi
integration of the PD impulses, the lower cutoff frequency f1 should therefore be
sufficiently low (kHz range), otherwise correct measurements are not possible.
A wideband PD measuring system consists of a bandpass filter with lower and upper
cutoff frequencies f1 and f2, see figure 9.25 [212]. Recommended values in IEC 60270
[26] are for f1 = 50 kHz and for f2 = 150 400 kHz.
The amplitude smax of the wideband filter response to excitation by a PD current
impulse is proportional to the apparent charge q if the center frequency f0 of the filter
corresponds to the frequency range of the current impulse where F(f) = F(0), see figure
9.24.
Typical wide band filter response is presented in figure 9.26 [109]. For the amplitude is valid:

where:
smax = amplitude of filter
response q = apparent
charge
F(0) = amplitude frequency spectrum at the frequency f = 0 Hz

a) Idealized PD current impulse

i (t)= PD pulses
T1 = time to maximum
Imax T2 = halfvalue
time

b) Normalized spectral amplitude density

1 = T1/T2 = 1s / 5 s 2 = T1/5T2 = 5 ns / 50 ns
3 = T1/5T2 = 5 ns / 15 ns
F(0)= amplitude frequency spectrum at the
frequency f = 0 Hz F(f) = amplitude frequency
spectrum at frequency f Hz
Figure 9.24: Quasiintegration of PD impulses
For the center frequency is valid:

where:
f0 = center frequency of the bandpass filter
f1 = lower cutoff frequency of the band
pass filter f2 = upper cutoff frequency of
the bandpass filter
Advantages of a wideband PD system:
distinguish positive and negative polarity of PD current impulses
high resolution capacity for repetitive PD impulses, with typical filter response
duration between 2 s and 5 s a PD impulse repetition frequency of 100 kHz
can still be resolved
Disadvantage of the wideband PD system:
Sensitive to external noise (not suitable for unshielded HV laboratories)

Z = voltage source connections Figure 9.25: Wideband

pass filter Ck = coupling capacitor
Ct = test objekt capacitance
Zm = measuring impedance (R, L,
Ctype) Cc = capacitance of the
measuring cable i(t) = PDcurrent
impulse
u1(t) = response of the measuring
impedance Zm u2(t) = response of the
band pass filter
f1 = lower cut off frequency of the band
pass filter f2 = upper cut off frequency
of the band pass filter f0 = center
frequency of the band pass filter

smax = maximum amplitude ~q

= impulse duration (response of the filter)
Figure 9.26: Typical response of the wideband pass filter
A narrowband PD measuring system, see figure 9.27 [212], consists of a bandpass
filter with variable center frequency f0 and narrow bandwidth f (between 3 kHz and 30
kHz). Values accepted by IEEE C57.12.90.1999 [51] are:
Center frequency f0 between 0,85 MHz and 1,15 MHz (recommended 1 MHz)
Bandwidth f between 3 kHz and 10 kHz (9 kHz recommended)
The amplitude smax, see figure 9.28, of the narrowband filter response to excitation by a
PD current impulse is proportional to the apparent charge q if the variable center
frequency f0 of the filter corresponds to the frequency range of the current impulse
where F(f) = F(0), see figure 9.24.
Typical narrowband filter response is presented in figure 9.28 [109].
Advantage of the narrowband PD system:
Less sensitive to external PD sources (successfully used in unshielded HV laboratories)
Disadvantages
Not possible to
frequency f0)
Low resolution
duration ~
resolved.

of the narrowband PD system:

distinguish the polarity of the PD impulses (signal oscillation with the center
capacity for repetitive PD impulses, with typical filter response
200s a PD impulse repetition frequency of 4 kHz can only be

The response duration of a narrowband filter to the PD current impulse input can be
estimated as:

where:
= impulse duration (response of
the filter) f = bandwidth of the
filter.

Z = voltage source
connections Ck =
coupling capacitor
Ct = test object capacitance
Zm = measuring impedance (R,
Ltype) Cc = capacitance of the
measuring cable i(t) = PD
current impulse
u1(t) = response of the measuring
impedance Zm u2(t) = response of the
band pass filter
f0 = center frequency of the band pass filter = fm (variable)
Figure : Narrowband pass filter

The RIVPD system (radio noise or field strength meter = RIVmeter) recommended by
IEEE Standard C57.12.90.1999 [51] is a narrowband pass filter, which includes the CISPR
weighting circuit (nonlinear psophometric curve, see figure 9.29) to quantify the
magnitudes of the repeated impulses. The readings on this type of instrument depend
not only on the amplitude of apparent charge but also on the repetition rate of the PD
impulses.
For a given constant PD impulse amplitude with a variable repetition rate, the reading on
the RIVmeter with the CISPRweighting circuit will increase linearly with an increasing
repetition rate up to n = 100 impulses per second, see figure 9.29 [109]. For a pulse
repetition rate higher than 4000 per second, which corresponds to a narrowband filter
pulse resolution time of = 220 s with f = 10 kHz, the reading becomes useless.
Because of the CISPR weighting circuit, there is no relationship between the pC and the
Vreading. Only if equally large impulses occur on the positive and the negative half
cycles a conversion can be made. With a repetition rate of twice a cycle (n = 100 in a 50
Hzsystem) and a measuring impedance of 60 the relationship 1V = 2.6 PC holds.

smax = maximum amplitude ~q

= impulse duration (response of the filter) Figure: Typical response of the narrowband
pass filter.

Note:
For PD sources in the transformer insulating system far from the bushings (for example
in the main insulation), the PD current impulses are heavily attenuated by the RLCM
network of the transformer. Such attenuated PD impulses may be outside the range of
the PD system if the lower cutoff frequency f1 of the wideband filter (recommended 50
kHz) or the center frequency of the narrowband filter f0 (recommended 1 MHz) are
higher than the limiting frequency of the attenuated PDimpulses F(f) = F(0), see figure
9.24. In this case the quasi integration is not correct and the amplitude of apparent
charge no longer corresponds to smax.

f(N) = variation of meter

reading [V] N = pulse
repetition rate
Figure 9.29: CISPR weighting characteristic for periodic pulse sequence with constant
amplitude.

A 9.3 True charge, apparent charge and measurable charge

A partial discharge can be interpreted as a rapid movement of an electric charge from
one position to another. For very fast changes, or during the first instant after charge
movement, the individual insulation links in a series of connected links between two line
terminals can be regarded as a number of series connected capacitors. If the two line
terminals are connected together via an external capacitance Ck the charge movements
within the series connected insulation links (capacitances) will also be reflected in the
charge of external capacitance Ck.
The charge movements can be detected as circulating current impulses in the equivalent
circuit, via measuring impedance Zm, see figure 9.23.
In figure 9.30 a schematic drawing of the PD source (capacitance C1) in the insulating
system (capacitances C2 and C3) and the corresponding equivalent circuit are
presented [212].
If the PD source is assumed to be a small cavity in the solid insulation, the following
ratio exists between the capacitances

The electric breakdown in the cavity (capacitance C1) is represented by the spark gap F and
resistance R1.

The electric breakdown in the cavity generates a PD current i1(t), which is a local current and
cannot be measured at the bushings.

Voltage drop U1 across the cavity is caused by discharge current i1(t) and releases a
charge q1 = true charge.

q1 = U1 C1
The discharge of C1 causes a rapid charge transfer in capacitances C2 and C3. This
charge transfer causes a measurable voltage drop Ut at the test object capacitance Ct
(at the winding bushing connection).

Assuming:

The theoretically measurable charge qt is:

q t q3
q 3 = U3
Ct Ut
U3

BU = bushing
HV = high
voltage NT =
neutral terminal
C1 = weak
region
Ct = test object capacitances
(C2 and C3) Ut = test voltage
F = simulation of the PD
source R1 = damping
resistance
U1 = internal voltage drop caused by PD
Ut = voltage drop measurable at
the bushing i1(t) = local PDcurrent
impulse
Figure 9.30: Detectable PD signals
a)schematic representation of a part of the transformer insulation
b)equivalent circuit for PD defect in insulating system
In an extended insulating system the Ut values are in the millivoltrange, while the
magnitude of U1 (at the PD source) may be in the kilovoltrange. The theoretically
measurable charge qt is linked to true charge q1 via capacitances C1 and C2 and is
defined as apparent charge. Because the location of the PD source is not known,
capacitances C1 and C2 cannot be estimated. The apparent charge (theoretically
measurable charge qt at the bushing) can therefore not be derived from the true charge
q1 at the location of the PD source.

Sensitivity of the measurement:

In any PD test circuit, the real measurable charge qm (integral of the circulating PD current i(t)), is dependent 80 on
the ratio of coupling capacitance Ck to test object capacitance Ct, see figure 9.31.

The charge transfer processes between capacitances Ct and Ck cause a residual voltage
drop Ures. If there is PD activity in the test object (Ct ), the real measurable charge qm
released by coupling capacitance Ck is estimated as:
Apparent charge qt is:

qm = Ck Ures

qt = Ct Ut = (Ct+Ck)
Ures
The ratio of the real measurable charge qm and apparent charge qt (theoretically
measurable charge) is defined as:

To improve PD measurement sensitivity, a sufficiently large coupling capacitance Ck

should be installed, see figure 9.31 [212]. In power transformer PD circuits the coupling
capacitance value is determined by the type of bushing (from 200 to 600 pF). The
sensitivity of each PD circuit is defined by the calibration procedure.

q = apparent charge
qm = measurable charge

Figure 9.31: Influence of coupling capacitor Ck on the measuring sensitivity

qm/q
The maximum sensitivity for detection of a compensating PD currentimpulse i(t) is
reached for Ck >> Ct (Ck = 100 Ct, see figure 9.31). As the value of coupling
capacitance Ck decreases, the sensitivity of the PD circuit is reduced due to the lower
compensating current i(t).
The minimum coupling capacitance of any PD test circuit is the stray capacitance Cs of
the HV potential electrodes to ground. A PD circuit using Cs as a coupling capacitance
has a very low
sensitivity because Cs << Ct.
Theoretically a RIVsystem without CISPR weighting circuit delivers the same amplitude
of measurable charge qm as a narrowband filter (quasiintegration of circulating PD
current i(t)), except that the reading is calibrated in microvolt.
For any RIVsystem with a weighting circuit, no conversion is possible between pC and V.

A 9.4 Typical external noise sources

A sensitive PD system connected to an extended test circuit for PD measurements on HV
transformers is able to detect all high frequency impulses in the test circuit. The PD
system cannot distinguish between the real PD impulse and external noise. If PD activity
is detected, the possibility of an external source must be investigated. An overview of
external PD sources is given in figure 9.32 [212].

1, 2 = low voltage power

supply 3 = high voltage
source
4 = HVfilter
5 = connections and electrodes
6 = Ck = coupling capacitor
7 = conducting object
8 = pulse shaped interferences
9 = harmonic interferences
10 = interference currents in the earthing system
11 = low voltage power supply
Figure 9.32: Typical noise sources (see text for explanation)

Low voltage power supply (1, 2, 11)

Noise such as thyristor pulses or harmonics from the low voltage power supply, may
especially influence a sensitive PD system which is directly connected to the power line.
If these are present, a lowpass filter or insulating transformer should be used. Due to
the filtering effect of the stepup transformer (3) and of the HV filter (4) in the connection
to the test object, the noise from the lowvoltage power line is usually sufficiently
suppressed. If there is a noise problem, a second stepup transformer may be used as an
additional filter or a PD system with a narrowband filter could be used (f0 > 1 MHz).

High voltage source (3)

An HV source must generally be PDfree. If there is a problem, the coupling capacitor Ck
can be connected directly to the source (without the test object) to easily check the HV
source.

HV filter (4)
In difficult cases, a PDfree HV filter (low pass) must sometimes be used.

Connections in the test circuit and electrodes (5)

All bushing tops (even earthed bushings) and sharp metallic parts on top of the transformer
(especially close

to the bushings) should be shielded; see clause A 9.8. All connections should be PDfree (sufficient
radius). All impedances Zm must have a good connection to earth. If there is a problem an ultrasonic
detector (corona gun) may be used to detect an external PD source. The PD type can be determined
from the statistical analysis of the PD signals (typical PDpattern, see Table 3).

Coupling capacitor (6)

The coupling capacitor must be PDfree. If there is a problem, the coupling capacitor
must be measured separately.

Conductive objects close to the transformer under test (7)

Unearthed conductive objects close to the transformer under test are charged to a high
potential due to the electric field. If the breakdown field value is reached, a pulselike
discharge occurs. These PD impulses are coupled to the PD test circuit and detected at
the measuring impedances, and normally exhibit a very high apparent charge amplitude.
This PD source can be recognized by comparing it with typical PD patterns, by visual
observation in the laboratory, or by using an ultrasonic detector.

Interference in nonshielded laboratories (8,9)

Pulseshaped interferences (switching phenomena) or harmonic interferences (radio
transmitters) heavily influence the sensitivity of the PD system. In these cases a narrow
band filter with variable center frequency should be used, see clause A 9.2, to suppress
the ambient interference. The best way to suppress the effect of such external
interferences is to apply an advanced PD system, which uses a spectrum analyzer as a
frontend, see clause 9.9. Repetitive external sources like thyristor pulses can be gated
out. A gating function is available on several PD systems. The final possibility for
suppressing these electromagnetic waves is the construction of a shielded HV laboratory
(Faraday cage).

Note:
Measuring impedance Zm (lower cutoff frequency at 10 kHz) normally suppresses the
power frequency displacement currents at capacitances Ck and Ct . If the power
frequency current limit for the specific impedance Zm is exceeded, suppression of power
frequency current is lost. Before applying a test voltage, the maximum power frequency
current in the test circuit, which is dependent on capacitances Ck and Ct in the test
circuit, should be calculated. The maximum power frequency current in the capacitance
is defined by:

A 9.5 Advanced PD system

The ICMsys8, (manufactured by Power Diagnostix Systems GmbH, Germany) is a
specially designed modern PD system to meet the requirements of partial discharge
measurements on power transformers. The ICMsys8 uses wideband filters for both
digital data acquisition and further data processing of conventionally detected
PD signals, see clause A 9.2.
True parallel acquisition of PD impulse currents on eight channels is achieved by using
eight individual amplifiers (wideband filters) connected to eight measuring impedances
at the bushings via eight pre amplifiers. The PD activity is detected simultaneously on
all eight channels and processed in the controller unit, see figure 9.33a. PD readings can

be weighted according to IEC in pC or according to IEEE in V (analog interface for

RIVmeter). Typical result of the ICMsys8 is shown in figure 9.33b.

Figure 9.34: Principle PhaseResolving Partial Discharge Analyser (PRPDAsystem)

Figure 9.35: Registration of PD impulses (bubbles and surface discharge);

advanced PD system (statistical analysis of PD
impulses)
A phaseresolving partial discharge analysis is important for identification of the type of
PD source for the following reasons:
PD patterns identify a specific type of PD source (image of the physical
process; see Tables 3 and 4 and clause A 9.1)
PD patterns are not influenced by the signal transfer function of the
extended insulating system (statistical behavior does not change)
PD patterns can be used to distinguish between superimposed PD
defects on the basis of different statistical behavior

A 9.6 Detection of acoustic PD signals

An acoustic PD signal is a mechanical vibration characterized by its frequency f.
Theoretically, the PD source acts as a point source of acoustic waves. The intensity of the
emitted acoustic waves is proportional to the energy released during the discharge.
where:
W1 = locally released
energy q1 = local
charge
U1 = local voltage drop

W1 = q1

U1

Acoustic wave propagation occurs only if the wavelength is small compared with the length of
the propagation path.
In a specific medium the wavelength is given by:

where:
= wavelength
v = sound velocity in a specific medium
(1400 m/s in oil) f = frequency of
mechanical vibration (acoustic wave)
Oil would be a perfect medium for the propagation of acoustic waves, with no
attenuation or dispersion occurring. In the transformers, acoustic waves propagation is
heavily influenced by the complicated structure of the insulating system (winding
barriers, core, and tank walls). In a complicated structure, both the amplitude
(attenuation) and the signalshape (absorption, dispersion) of the acoustic signal emitted
by a PD source change along its propagation path [217].
Besides the absorption and dispersion phenomena, the multitude of wave types
complicate the analysis of acoustic PD signals detected in a multimaterial structure. Two
types of waves must generally be considered for the analysis of acoustic signals, see
figure 9.36.

Figure 9.36: Schematic presentation of two types of acoustic waves.

Transversal waves; attenuation is dependent on wall thickness
Longitudinal waves; higher velocity than transversal waves (approximately a factor of two)
Possible propagation paths of acoustic waves from a PD source to the transformer tank
wall are shown schematically in figure 9.37. The two waves (superimposed) are created
at an interface, either by incidence or reflection, see figure 9.37. An acoustic sensor
positioned at a defined location on the tank wall detects both, directly propagated waves
and wallpropagated waves. The two wave types have different propagation velocities.

v1 = velocity of direct
wave v2 = velocity of
wall wave =
incident angle
Z = distance of the PDsource to
the wall h = thickness of the
wall
Figure 9.37: Possible acoustic waves at the sensor
In order to localize the PD sources by analyzing the time difference between the electric
and acoustic PD signals, there must be a possibility to distinguish between directly
propagated and wallpropagated waves. This information is theoretically hidden in the
wave front of the acoustic signal that is detected at the sensor on the tank wall, see
figure 9.38.

Note:
The fastest propagation path is not the direct path, but the path with an incident angle , see
figure 9.37:

where:
v1 = velocity in medium 1 (for example in oil)
v2 = velocity in medium 2 (for example in the tank wall)

Figure 9.38: Analysis of acoustic waves of PD sources

a)
Detection of direct (through the oil) acoustic wave
b)Detection of reflected acoustic wave or wave propagated through different
media

Advanced detection system:

ThreeTransducer Detector (TTD)
A minimum of three acoustic sensors (piezoelectric transducers) and a fourchannel
digital oscilloscope are theoretically required to localize PD sources using the time
difference between electric and acoustic signals. An advanced detection system, Three
Transducer Detector (TTD) is used.

Note:

This system is not manufactured any more, but is still in use. The three transducers
(piezoelectric crystals) are placed close to one another (equilateral triangle with L =
0,15m) so that the propagation path of the acoustic waves from the PD source is nearly
the same. TTD defines its own x, y and z coordinate system (normalized vectors). The
normalized coordinates (x, y, z) give the direction to the source and depend only on time
differences in the TTD system, see figure 9.39c. It is assumed that the distance between
the transducers L is considerably less than the distance d of the TTDsensors to the PD
source when deriving the localization formulas, see figure 9.39c.

Figure 9.39a: Main components of the TTDsystem

A typical result obtained by the TTDsystem is shown in figure 9.40.
The procedure for localizing the PD source using the TTD results is given in the following
and will be explained further using an example:
A decision is made whether a direct wave or a wall wave was
detected by analyzing the shape of the detected signal
The apparent velocity, vapp is calculated by approximate expression to
distinguish between direct wave and wall wave

where:
L = distance between transducers
t1, t2, t3 = time difference between electric and acoustic signals for the three
transducers
the apparent velocity vapp is compared with sound velocities in
materials relevant for the transformer insulating system, see Table 5.
a decision is made about the type of wave (direct or wall wave)
the coordinates x, y, z of the PD source are calculated using TTD software
for the specific wave (direct or wall wave)

Figure 9.39 c: Normalized co

ordinates There are three different conditions for the calculated
apparent velocity vapp.
vapp > vtransversal > direct incident wave
with shallow angle vapp vtransversal >
transversal wave
vapp < vtransversal > should not happen (possibly in thin walls)

Note:
Apparent velocity vapp is always higher than voil (1400 m/s).
Table 5: Example of sound velocities

Example
Calculation of the PD source location using the results presented in figure 9.40.
The shape of the detected acoustic signal indicates a wall wave
(superimposed longitudinal and transversal waves)
Apparent velocity vapp = 3417 m/s for a longitudinal wave, calculated values for a
longitudinal wave (t1L ~ 336 s, t2L ~ 363 s, t3L~ 320 s) do not clearly indicate a
longitudinal wave

Figure 9.40: Typical TTD system result

apparent velocity vapp = 3236 m/s for a transversal wave, calculated values for a
transversal wave (t1T ~ 382 s, t2T ~ 418 s, t3T ~ 375 s) indicate a wall wave (3200 m/s
= velocity of transversal wave for steel plate)
calculated position of the PD source for wall wave (TTD software)
x= +0.19, y = 0.52, z =+0.34 m

Note:
Due to the different paths of electric and acoustic PD signals through the insulating
system of the transformer, there is no simple relationship between the amplitude of
apparent charge and the amplitude of acoustic waves.
Knowledge of the transformer insulating system and experience in the analysis of the
results are needed to localize PD sources by analyzing acoustic PD signals.

Localization of the PD source using analysis of the electric signals

Electric PD signals (PD current impulses) propagate from the PD source through the
RLCMnetwork of the transformer, see figure 9.42. The response of this network to
excitation by a PD current impulse at any location in the insulating system can only be
detected at the bushings. All information available in the detected signal must therefore
be analyzed. [214, 219]
All signals in the time domain (recorded using an oscilloscope) and in the frequency
domain (recorded using a spectrum analyzer) are generally linked via Fourier
transformation, see figure 9.41. [121]
For a given nonperiodic PD current impulse i(t), the complex frequency spectrum I(j) is
obtained from the Fourier integral:

Allthough the PD current signal is the same, the data recorded in the time domain and in
the frequency
domain contain different information, which is important for localization of the PD source.
Theoretically a specific response of the RLCM network exists for any type of PD pulse and
for any location of
the PD source in the insulating system, and may be detected at the transformer
bushings.
A time consuming characterization of the transformer RLCM network is generally
necessary to be able to
analyze real PD signals in the time and frequency domains. A calibrating signal (usually
1000 pC) is injected at
one specific bushing and crosscouplings are successively recorded at all other bushings.
This procedure is
repeated for each bushing, see clause 9.9.

A
=
amplitude f
=
frequency t
= time
Figure Signal presentation in time domain and frequency domain

Analysis of PD signals in the time domain [214]

The PD current impulse in the time domain detected at the bushing can be
characterized by three components, see figure 9.43:
Capacitive component caused by transmission through the capacitive ladder network
Traveling wave component caused by electromagnetic wave transmission
Oscillating component determined by the resonance frequency
of the LCcircuit Basic timeresolved PD signal characteristics that
should be analyzed are:

Maximum amplitude of the PD current signal (mV)

Risetime of the PD current signal
Oscillations of the PD current signal
Reproducibility of PD current signals

Figure 9.43: PD signal analysis in time domain

An indication about the location of the PD source is given by comparing the recorded PD
signals in the time domain with the results of the characterization of the transformer.

Analysis of PD signals in the frequency domain

The PD current impulse in frequency domain detected at the bushing can be
characterized by two typical frequency spectra, see figure 9.44:
Typical frequency spectrum generated by very fast PD signals (close to the measuring tap)
Typical frequency spectrum generated by attenuated PD signals (inside the insulating
system)

Y1 = close to the measuring top

X1 = coupling via RLCM network
Figure 9.44: PD signal analysis in frequency domain
Basic frequency spectra features that should be analyzed are:
Amplitude of the power spectrum (dBm)
Frequency range of the power spectrum
Typical resonances
Reproducibility of power spectra
The indication about the location of the PD source is found by comparing the PD signal
frequency spectra with the results of the characterization of the transformer.

Note:
All the theoretical components of the signals in the frequency domain as well as in the
time domain are strongly dependent on the design of the transformer (size, type of
windings), on the true location of the PD defect with respect to the terminal where the
PD signal was detected and on the original shape of the PD signal (discharge in gas, oil,
or solid material). Knowledge of the transformer insulating system and experience in the
analysis of the results are required to localize PD sources using the analysis of the
electric PD signals.

Corona shielding

External discharge (corona) can be prevented by shielding. A guideline for shielding is given
in figure9.45.Thegraphshows permissible voltage on the shield without giving rise to corona.

Figure 9.45: Shielding dimensions to eliminate corona