Вы находитесь на странице: 1из 18

Partial Discharge Testing

Decreasing the Field Failures of High Voltage Components

Test engineers have an understanding of high voltage isolation leakage testing, but Partial
Discharge testing is much less understood. We first summarize the more traditional Isolation
Leakage testing, and then follow with a comprehensive account of Partial Discharge testing.

Isolation Voltage
The initial design of a device with an insulation barrier includes a choice of materials and
dimensions to achieve an isolation voltage rating.

The isolation rating is the voltage level under specified conditions that the device will withstand
without breakdown.

Isolation Voltage Failure Modes

Isolation breakdown in devices can occur in several ways.
In an insulator electrons are tightly bound to atoms and molecules. With moderate gradient
potential some electrons are pulled free of their bonds to be later recaptured in collisions with
neighboring atoms or molecules. As the gradient is increased beyond the Intrinsic Dielectric
Strength of the material, collisions occur with sufficient strength to free more electrons than are
captured, resulting in a disruptive breakdown called Intrinsic Strength Breakdown.

In another type of breakdown, a path across the surface of a device can lead to carbonization and
possibly even Surface Flashover.

Another more complex cause of degradation failure inside the volume of the insulation material
is Erosion Breakdown which will be later presented.

In High Voltage Leakage testing, the tester applies line frequency AC voltage and monitors the
resulting device current to not exceed a certain limit. A good device will have both resistive and
capacitive leakage components that are proportional to the instantaneous applied voltage. The
resistive component of device impedance is typically in the order of Gigohms. A Leakage failure
occurs when this resistance is degraded. Excessive device capacitance could also cause leakage
failure, though a physical mechanism to cause this seems improbable.

HTPaper: Partial Discharge Testing 1 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Partial Discharge
Partial discharges are the source of Erosion Breakdown which affects the long term life of an

Partial discharges are discharges that do not completely bridge the insulation between the
terminals. These discharges are termed “partial” because they occur in areas that occupy a small
portion of the electrical path length and are limited in magnitude because they are in series with
mostly good insulation (which may eventually degrade). These discharges can occur in insulators
that contain gaseous inclusions, cavities, or voids.

To understand how and why this occurs the following sections will include Electric Field

Electric Field Basics--Electric Flux Density “D”

Gauss’s Theorem states that the Electric Flux Density at a distance r1 from a concentrated charge
Q on a small conductive sphere may be determined by enlarging the sphere until its radius is
equal to r2 (as the charge is distributed in equilibrium on the surface). See Figure 1. The
magnitude of the Electric Flux Density at r2 is the charge Q divided by the surface area of the
enlarged sphere.
 The Electric Flux Density “D” at the surface of a sphere of radius r1 is charge/area
(Coulombs per square meter).
 The radius of the sphere may be enlarged to find D at another distance r2 from the center of
the sphere.
 Note that D is not dependent on the medium.

Charge +Q
D D2=Q/4πrr22


Figure 1. Gauss’s Theorem. Electric Flux density D at a distance r2 from Charge

HTPaper: Partial Discharge Testing 2 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Electric Field Intensity “ε”

The strength and direction of the Electric Field Intensity “ε” Any point in an electric field may be
measured by the force and direction upon a test charge as shown in Figure 2. Electric Field Intensity
“ε” measured in Newtons1 per Coulomb (dimensionally equivalent to Volts per meter) may be
thought of as the force effectiveness of the electric field.

ε ƒt
ε = ƒt/Qt
Figure 2. Electric Field Intensity at a Test Charge
Unlike Electric Flux Density “D”, the magnitude of the Electric Field Intensity “ε” is affected by the

medium. ε is proportional to the Electric Flux Density but inversely proportional to permittivity of

the medium as follows: ε= D/e , where is e the permittivity of the medium. The permittivity of a

vacuum is the basic electrical constant eo which equals 8.85x10-12 F/m, (Farads per meter). The
permittivity e of all other materials is expressed as a dimensionless ratio er times the permittivity eo
of a vacuum (e = er eo). Table 1 below indicates the relative permittivity values for some selected

Medium Relative
Permittivity (er)
Vacuum 1
Air 1.006
Styrofoam 1.03
Polystyrene 2.7
Plexiglass 3.4
Amber 3
Rubber 3

Newton is the force required to accelerate a mass of one kilogram at the rate of one meter per second squared

HTPaper: Partial Discharge Testing 3 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Quartz 5
Formica 6
Ammonia (liquid) 22
Glycerine 50
Distilled water 81
Barium Titanate 1,200
Barium Titanate 20,000

Table 1. Relative Permittivity Values.

Q+ Q-

Figure 3. Plots of Electric Field Intensity and Flux Density vs Distance between Two
Oppositely Charged Plates in Uniform Medium (Air)

In Figure 3. the two charges are assumed to be evenly distributed on the inside surfaces, and end
effects are ignored. Because of opposite charges and symmetry, cancelation of the electric field
within the volume of each plate will occur such that the fields will be zero across the thickness
of the two plates. This causes the charges to be present only on the inside surfaces of the two
plates. With the permittivity (e) of air a constant, and since ε=D/e, the ε and D fields will
therefore have similar shapes across the gap.

HTPaper: Partial Discharge Testing 4 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

This is not the case in Figure 4., since there are two different dielectrics across the gap. As
previously stated, Flux Density is not affected by the medium, and will therefore be the same in
both mediums as shown in Figure 4. Notice that the Field Intensity will be twice as great in the
air where er = 1 as in the dielectric where er = 2. It is important to note that the smaller
permittivity dielectric has the higher Electric Field Intensity.

Q+ Q-


er=1 er=2

Figure 4. Electric Flux Density and Field Intensity ε between Two Oppositely Charged
Parallel Plates Separated by Two Types of Dielectrics

Figure 5. below illustrates a model for a capacitor (or the passive elements of an optocoupler). A
quartz insulator 1.0 mm thick is shown between two conducting plates. The insulator has defects
in the form of air voids. If 1.0 kV is applied across the insulator, there would be an Electric Field
Intensity of 1.0kV per mm (ε=volts/distance) across the dielectric. However, the air voids have a
relative permittivity (∈r = 1) equal to 1/5 that of the quartz, which would cause a localized
Electric Field Intensity of 5 kV/ mm in the air voids. This causes unequal charge distributions in
the insulator. Table 2. shows that the dielectric strength of quartz is 30 kV/mm. The air voids,

HTPaper: Partial Discharge Testing 5 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

however, have a dielectric strength of only 3 kV/mm. This produces sufficient ionization in the
voids to cause an arc that shorts out the charges that accumulate on opposite sides of the void.
This discharge path is in series with good insulation which limits the bandwidth of the
picocoulomb discharge. The effect is similar to discharging a capacitor through a high resistance.

er = 5
Air er = 1

Figure 5. Insulation Barrier with Voids in an Electrostatic Field

When the void discharges, the arcing stops and a slow recharging of the void through good
insulation begins again until the critical breakdown voltage of the air in the void is
again reached, repeating the process of slow charging followed by fast discharging. Thus the
void essentially becomes a relaxation oscillator. This effect is known as Partial Discharge.

Table 2. below illustrates the relative dielectric strength of some sample materials.

Material Dielectric Strength

(kV per mm)
Air (atmospheric pressure) 3
Oil (Mineral) 15
Impregnated Paper 15
Polystyrene 20
Rubber (hard) 21
Bakelite 25
Glass (plate) 30
Paraffin 30
Quartz (fused) 30
Mica 200

Table 2. Material Dielectric Strength

HTPaper: Partial Discharge Testing 6 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

When high voltage is applied to a device that produces Partial Discharge it may be observed that
the effect starts at a certain voltage level, and once started, the voltage must be reduced to a
lower voltage before it ceases. These two voltages are called the Inception and Extinction
voltages. This effect is illustrated in Figure 6. In typical AC voltage testing Partial Discharge
cycles may occur many times during the positive and negative peaks. In applications, if this
happens with sufficient magnitude over time, arcing in the voids will degrade the insulation,
even producing tree-like patterns in the dielectric that lead to failure. This effect is called Erosion

Figure 6. below illustrates the repetitive nature of Partial Discharge. Once Inception Voltage is
reached the frequency of partial the discharge will increase as the AC voltage is approaches its
peak value (frequency change not illustrated).


Typical Partial Discharge

cycles in a void

Figure 6. Inception and Extinction of Partial Discharge with AC Voltage Testing

HTPaper: Partial Discharge Testing 7 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

There are a number of Regulatory Agencies governing standards throughout the world:
 USA UL 1577
 Germany VDE 0884
 Canada CSA comp. acceptance notice 5

The standard test adopted by the European Community (CE) for semiconductor manufacturers is
VDE 0884. This has also become recognized as an international standard. There are three classes
of tests that exist for VDE optocoupler test specifications: Type, Random, and Routine.
1. Type Testing is performed on a specified number of devices that are sent to the testing
agency. It includes both destructive and non-destructive tests as well as
environmental test and is the most all-encompassing, determining whether products
meet design requirements.
2. Random Testing, which is performed on random samples from normal production
batches, also does both destructive and non-destructive testing.
3. Routine Testing is non-destructive testing that is performed on all production units.
Routing Testing (as well as Type and Random) includes partial discharge testing to
verify isolation barrier integrity.

Partial discharge measurement per VDE 0884 (June 1992) is used to evaluate the insulation
integrity of optocouplers. Partial discharge testing replaces the common dielectric withstand
voltage test, because dielectric test voltages may pre-damage the insulation of an optocoupler.
Partial discharge testing qualifies an optocoupler for operation at voltage levels that are below
the inception level so that no harmful breakdown of the insulation occurs during its normal
lifetime. The illustrations below describe the partial discharge test for Type and Sampling
(Procedure A—Figure 7.) and for 100% production (Procedure B—Figure 8.) testing in
accordance with VDE 0884.

HTPaper: Partial Discharge Testing 8 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

AC Voltage

initial (Maximum
V initial (MaximumTransient)

V pr (1. 5 Viorm)



tini tm ( 60sec)

Figure 7. Procedure A: (for Type and Random Tests)

tini = (measuring time for device leakage) = 10 seconds

tm (measuring time for partial discharge and leakage) = 60 seconds
td = (partial discharge detection delay, for PD extinction) = Adjustable 0.0 to 9.9 seconds

HTPaper: Partial Discharge Testing 9 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Vpr =1. 875 (Viorm)



Figure 8. Procedure B (For 100% Production Testing)

tm (measuring time for partial discharge and leakage) = 1.0 second

td = (partial discharge detection delay, may be kept at 0.0 sec.). Adjustable 0.0 to 9.9 seconds

HTPaper: Partial Discharge Testing 10 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Term Definition
Vinitial Maximum test voltage for the Partial discharge test. It is also the maximum
transient over voltage occurring in a rated mains service class. At this initial
voltage, partial discharge (but no breakdown) may occur. Vinitial also equals
Viotm (transient over voltage) which is listed in the applicable VDE
insulated related characteristics section. Preferred values for Vinitial are
shown in Table 3 below. This is taken from table 2 of the VDE 0884, June
1992 revision.
Vpr Partial Discharge test voltage applied to an optocoupler and maintained for
a specific time period, tst. During this time, Partial Discharge is measured at
a specific time interval, tm. Vpr=1.5 X Viorm for procedure A and
Vpr=1.875 X Viorm for Procedure B.
Viorm Working voltage (maximum service insulation voltage); this is the
maximum continuous permitted voltage that may be applied to an
optocoupler. This value is specified by VDE to each insulator.
tm Test time for Partial Discharge equal to 60 seconds for Procedure 1 and
1 second for procedure B.
tini Time beginning at Vinitial test voltage, which equals 10 seconds.
td Test voltage initialization time.
Pass/Fail No leakage failures and no optocoupler to have more than 5 pC Partial
Criteria Discharge during Partial discharge test time, tm.

Table 3. Definitions of Terms used in VDE Partial Discharge Testing.

HTPaper: Partial Discharge Testing 11 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Rated Mains 1 2 3 4
voltage up to Peak Peak Peak Peak
and including V
rms. & V dc
50 330 500 800 1500
100 500 800 1500 2500
150 800 1500 2500 4000
300 1500 2500 4000 6000
600 2500 4000 6000 8000
1200 4000 6000 8000 12000

Table 4. Preferred Insulation Test Voltages for Service Class (Vinitial)2

Comparative tracking Index (CTI)

CTI is a measure of the optocoupler mold material and its relative insulating capability. The
surface of the mold material is subjected to an alternating low voltage stress, which produces a
small current flow. When the current reaches a predetermined value, the corresponding
numerical value of the applied voltage is the CTI value. CTI impacts both external creepage and
maximum allowable working voltage for the same value of external creepage distance.

Material Group
Because the behavior of insulating materials is very complex under various contaminants and
voltages, direct correlation between deterioration of the insulating material and formation of
conductive paths on the insulation surface is not practical. Correlation between the Comparative
Tracking Index (CTI) and ranking performance of insulating materials has been found by
empirical observation. Consequently, CTI values can be used to categorize insulation materials:
Material Group 1 600 < CTI
Material Group 2 400 < CTI < 600
Material Group 3a 175 < CTI < 400
Material Group 3b 100 < CTI < 175
In some equipment specifications, material Group is used in conjunction with Pollution Degree,
Creepage distance and the Working Voltage Table.

Source: VDE 0884, June 1992 revision, Table 2

HTPaper: Partial Discharge Testing 12 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

HT 9464 High Voltage Isolation Tester

Figure 9. shows the High Voltage signal paths that are used by the HT 9464 Tester to obtain
device leakage and partial discharge measurements. The primary of the H.V. transformer
receives regulated constant voltage AC power from the settings of the front panel variacs V1 and
V2 for a duration determined by front panel T1 and T2 test time settings. The secondary of the
H. V. transformer applies the test voltage to the H. V. test terminal contacts in a handler. Not
shown is the safety test site cover that is interlocked to the main controller to prevent operators
from touching the contacts when high voltage is present. Notice that the H. V. path includes Z1
and Z2 in series with the contacts. Both of these consist of a 50Mohm resistor shunted by a 22pF
capacitor. These limit the energy to the device contacts for the following reasons:
1. Safety. The maximum 50/60 Hz AC current will be about 100uA. By comparison,
perception of alternating current through the human body requires at least 0.5 mA.
2. Less degradation to failed devices that may pass when retested at reduced voltage settings.
3. Less high frequency radiation (that can disturb digital logic in both handler and tester) when
shorted or arcing devices occur during testing.

HT 9464 Leakage Measurement

When a device is placed in the contacts the 50/60 Hz current from the transformer high voltage
secondary will pass through Z1, Z2 and Zin (less than one ohm) and through the SIGNAL coax
to the leakage current measurement circuit in the Main Controller. The current will return in the
coax shield to chassis ground and the H. V. secondary. The measured device peak current will be
compared with the Leakage Threshold setting and the tester will produce a Fail Leakage if the
measurement is excessive.

HTPaper: Partial Discharge Testing 13 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

High Voltage Box

H.V. Box Transformer
50Mohm Partial
.01uF H. V. Test
1nF Terminals
5Kohm Calibrator,
0 - 9 Pico-
Zin Z2 Coulombs
Amplifier &
Comparator DC
P. D. Power
Threshold Cable
+/- 15
Coax Coax Volts

Front Panel Variacs Volt Meter Circuitry Leakage Current Partial Discharge
V1 and V2 Peak Detector Measurement Signal Processing
0 to 120 VAC 0 to 7.50 KVAC uApeak
(0 to 10.6 KVpeak)
Main Controller

Figure 9. HT 9464 High Voltage Diagram Showing Signals to Main Controller and Partial
Discharge Calibrator

HT 9464 Partial Discharge Measurement

Because of the high frequency nature of partial discharges that may occur in the device under
test, the associated current will take a different path than that of 50/60Hz leakage current. The 1
nanoFarad capacitor now becomes the low impedance high voltage source for partial discharge
transient current that travels through the two 22pF shunt capacitors in Z1 and Z2 (and the device
under test) to Zin (the first stage of the narrowband amplifier). Passing through Zin, transient
current then goes directly through the 0.01 microfarad capacitor (bypassing the leakage
measurement circuit) and returns to the bottom of the 1 nanoFarad capacitor. The narrowband
amplifier will respond to the high frequency components that get amplified and sent to the
comparator for amplitude discrimination. The comparator is referenced to a three digit partial
discharge threshold setting that has been previously set with the Partial Discharge Calibrator
(usually at 5 picoCoulombs). If a partial discharge is detected by the comparator, the red LED on
the side of the HV Box will blink and the signal will be sent to the main controller for

HTPaper: Partial Discharge Testing 14 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

HT 9464 Partial Discharge Threshold Setting Using the Calibrator

The HT PDC700V Partial Discharge Calibrator is capable of producing accurate 0 to 9
picoCoulomb pulses that are used to set the three digit Partial Discharge Threshold adjustment.
Normally the single digit thumbswitch on the calibrator is set to 5 picoCoulombs. The calibrator
spring contacts (pogo pins) are then placed against the handler contacts and the three digit setting
on the side of the H.V. Box is increased until the red LED stops blinking. This usually occurs
between “050” and “100” when calibrating the detector to 5 picoCoulombs.
Other comments:
1. This procedure requires that the Calibrator has been integrated into the handler so that it
can make contact with the H. V. device contacts (while the DC power cable is connected
between the Calibrator and the Main controller). The handler contacts should be in the
closed position when calibration is performed to simulate normal contact capacitance. A
device may or may not be present in the contacts.
2. This procedure is done when no High Voltage is present. This is usually guaranteed by
the necessity to raise the interlocked Safety Cover over the handler test site before
placing the calibrator in contact with the H.V. contacts. The calibrator will be damaged if
it experiences High Voltage.
3. It is also assumed that the handler H. V. Test Site has been characterized to see how high
in voltage the empty contacts can go (in the closed position as when testing) before
failing partial discharge. The test site must be able to go higher than the voltage at which
devices will be expected to pass partial discharge.

HTPaper: Partial Discharge Testing 15 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Paralleling Multiple Devices in High Voltage Isolation Tester Contacts

The question is occasionally asked whether more than one device can be tested in the High
Voltage contacts. This is an area that introduces uncertainty into the test due to unknowns at the
test site caused by several test devices in parallel. The particular uncertainties from putting
several devices in parallel depend on whether Leakage Testing or the Partial Discharge Testing is

Leakage Testing Case

When devices are put in parallel the Leakage Current is the total from all devices. This means
that the operator will have to use a larger Fail Leakage threshold, but he must not use the total
number of devices times the max permitted leakage for one device. The problem is that we loose
the ability to know about each device when we put several in parallel. For example: Suppose at
some test AC Voltage we have devices that are typically 1uA with a specification of 5uA max.
Also suppose that three devices are tested in parallel with a resulting leakage of 8uA. The test
must reject all three by assuming (1uA + 1uA + 6uA) even though all three parts may be in spec
at 2.66uA. This uncertainty becomes greater with more devices in parallel. If a high percentage
of the parts are good this may be acceptable. Rejected parts may also be tested one at a time to
retrieve good ones.

Another consideration when using multiple devices is the effect of the resultant higher device
leakage current upon the tester. With higher device leakage currents, the tester Source Voltage
will have to be raised more than normal to obtain the desired Device Voltage. These two
voltages typically have a small difference due to a drop across the tester source impedance.
Typically the Device Voltage is more than 95% of the tester Source Voltage. With multiple
devices and larger leakage currents the actual Device Voltage will be a smaller percentage of the
Source Voltage and more dependent upon the device leakage currents. If device leakage current
is not fairly constant then the Device Voltage will be difficult to keep at the desired setting.

HTPaper: Partial Discharge Testing 16 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

Partial Discharge Testing Case

During testing, partial discharges in a dielectric produce small transient voltage spikes at the
tester contacts that can be detected if they are large enough to exceed the setting of the Partial
Discharge Detector Threshold (on the HV Box). When several devices are placed in parallel the
capacitances of all devices will add. This increased capacitance will reduce (filter) the small
transient voltage spikes from any failing device so that they are more difficult to detect. This
may be compensated for by reducing the Partial Discharge Detector Threshold. This may be
done accurately by performing the 5pC calibration of the PD Detector Threshold (using the PDC
700V calibrator) when the actual number of devices to be tested are in the tester contacts. As
more devices are in parallel the signal produced by the PDC 700V calibrator will be reduced and
the PD Detector Threshold setting for 5pC will be set smaller. This will become impractical
when the threshold setting becomes so small that it reaches the noise detection level of the
system. It should also be noted that the devices that are used for this calibration should be typical
in terms of capacitance.

The Test Equipment required for Partial Discharge Testing is considerably different from the
standard Automatic Test Equipment (ATE) utilized to test analog, digital and mixed signal
circuits. The design must be able to generate high voltages (up to 10 kV peak) and unlike Hi-Pot
Testers, detect leakage in the sub-µA range. Additionally for Partial Discharge measurement the
equipment must have the ability to detect and measure minute charges at the picocoulomb level.
For accurate and repeatable Partial Discharge measurements, the equipment must be calibrated
against a traceable standard. Additionally for use in production interfacing to a handler is
mandatory. Most importantly, the equipment needs VDE certification to allow the user to test to
the International VDE Standard.

HTPaper: Partial Discharge Testing 17 of 18 December 2004

Partial Discharge Testing
Decreasing the Field Failures of High Voltage Components

HT, Inc. is a privately owned company specializing in Semiconductor Equipment. The HT 9460
and 9464 product lines previously owned by Hewlett Packard, were acquired by HT in 1998. HT
is located in the heart of Silicon Valley. For further information:
Email: info@ht-world.com
Phone: 408-980-9738
Fax: 408-727-0344.

HTPaper: Partial Discharge Testing 18 of 18 December 2004