A Brief Discussion on Partial Discharge Measurement

[Introduction] Power semiconductors will encounter various problems in actual system applications. Some problems require the accumulation and summary of design experience, some classic problems have been written into textbooks, and some problems are still in the scope of academic research. Readers are welcome to write down and share their experiences in work and study.

Partial discharge phenomenon and hazards

Partial discharges are discharges that occur only in a part of the insulator. These discharges may also occur at electrodes, but they may also occur in the electric field space "without electrodes". Partial discharges can occur in: gases, liquids and solids. When partial discharges occur, not only will there be losses, but the high-energy electrons and UV radiation generated will damage the surrounding insulating materials.

Different insulating media will cause different aging damage due to partial discharge:

■ No damage: moving air, natural substances such as glass mica;

■ Minor damage: sealed gas insulation such as SF6, air;

■ Moderate damage: oil-paper insulation (transformers, DF) casting resin;

■ Severe damage: PE, VPE, almost all plastics.

The measurement of partial discharge becomes particularly important in situations where it causes minor to severe damage to the insulation.

Partial discharge mechanism

■ The local field strength increases above the electrical strength of the insulating medium;

■Locally lower electrical strength (e.g. voids in the casting resin).

Corona discharge

Corona discharges are discharges caused by local excessive electric field strength in gases and liquids. They occur mainly on tips, edges and thin conductors. Since they usually appear on outer contours, they are easy to detect due to their typical glow and cracks.

Surface discharge

It is caused by corona discharge on the electrodes.

Discharge in layered materials

It is another form of surface discharge. Locally excessive voltage will be generated in the boundary layers of various materials, resulting in local discharge.

Air Gap Discharge

Air gap discharge is caused by bubbles in the insulating material or contaminants with different dielectric constants.

Tree channel

Continuous partial discharge in solid or liquid insulating materials will form tree-like conductive paths and cause permanent damage to the insulator.

Air Gap Discharge

The air gap discharge is caused by a fault in the insulating material, such as bubbles in the transformer oil or contaminants with different dielectric constants. The capacitance of the entire insulator to be tested is composed of the bubble cavity capacitance C1 and the capacitance of the remaining insulation distance C2 in series and in parallel with the capacitance of the fault-free insulator C3. Due to the different areas, C3 is much larger than C1, and due to the different dielectric constants, C1 is much larger than C2. When a partial discharge occurs, the voltage drop ΔU1 in the cavity is in the kilovolt level. Since C2 is very small, the voltage drop ΔUt across the insulator is only in the millivolt level, so it is difficult to measure. In the actual measurement process, we use coupling capacitors to accurately measure.

Partial discharge measurements can measure insulation quality without damaging the sample.

As shown in the figure below, consider the sinusoidal terminal voltage U. Once the voltage U1 exceeds the voltage threshold Uz required for the electric field discharge in the bubble cavity, C1 drops to the Ul voltage through spark discharge. The displacement current of C2 may also recharge C1, so multiple partial discharges may occur within a half cycle of the voltage depending on the voltage level.

Test Methods

Ideally, we assume that the coupled capacitance is much larger than the capacitance of the insulator under test. When the insulator under test has partial discharge, a very small voltage drop occurs at the terminals at both ends relative to the power supply, which will be immediately compensated by the coupling capacitance. Therefore, a high-frequency current will flow between the coupling capacitance and the capacitance of the insulator under test. By integrating the current flowing through the coupling capacitance, the amount of charge flowing into the insulator is obtained: the apparent discharge amount. The amount of electricity consumed by partial discharge in the insulator q1 is converted through C1 and C2. It can be noted that C2 is smaller than C1, so the apparent discharge amount is very small compared to q1.

For Ck>>Ct, we have:

When C1>>C2, we have:

The wiring diagram for partial discharge measurement is shown in the figure below. The test current is industrial frequency AC. Ct (Cp) is the equivalent capacitance of the object to be tested, and Cp is the coupling capacitance.

In actual application testing, the coupling capacitance will not be infinite. Usually it is approximately equal to the capacitance of the test insulator, and it will only compensate for part of the voltage drop occurring at the measured end. In this case, the measured charge will be smaller than the apparent charge, so we must calibrate the instrument before the experiment to determine the ratio of the measured charge qm to the apparent charge qs.

Current integration

In order to convert the decoupled current pulses into charge values, they need to be integrated. This can be done in two different ways: time domain integration and frequency domain integration

Frequency domain integration

Generally, we can use an RC low-pass filter to integrate the current. However, it cannot avoid the interference caused by the 50Hz AC current, so we will use a bandpass filter to integrate the area where the current spectrum density is relatively flat. Broadband partial discharge measurement equipment usually measures in the frequency band of 10 to 100kHz. Some narrow bandwidth filters are used for integration in the frequency domain. Some partial discharge measurement equipment can be adjusted to a narrow bandpass filter with a very small bandwidth (IEC60270 standard recommends a center frequency of 50kHz≤fm≤1MHz and a bandwidth of 9kHz≤Δf≤1MHz) so that the detection range can be set to any frequency range to avoid specific interference frequencies.

Partial discharge current pulses in the time and frequency domains

However, the disadvantage of narrow bandwidth measurement is the long pulse blanking time. If two partial discharge pulses follow closely together, the second pulse can only be recorded correctly after the previous pulse has oscillated. If the second pulse arrives while the bandpass is still oscillating, two pulse responses with different phases will be superimposed on the output, which may cause measurement errors in subsequent peaks. Since the decay time of the bandpass filter is roughly inversely proportional to the bandwidth, narrow bandwidth measurements will produce errors when the current signal rises very quickly. Another major limitation of this narrow bandwidth measurement is that it is impossible to distinguish the positive and negative poles of the partial discharge pulses. As shown in the following figure:

The left and right figures are the pulse responses of narrow bandwidth measurement and wide bandwidth measurement respectively.

Time domain integration

By integrating the current pulse in the time domain it is equivalent to measuring with a wide bandwidth. According to IEC 60270:2001, measurements are made with a bandwidth between 100 and 400 kHz. Modern PD measuring devices operate in a bandwidth range of up to MHz. However, if the test environment is not electromagnetically shielded, radio interference may be coupled into the measurement signal and cause errors. Broadband measurements allow more detailed information about the aging of individual parts of the insulation material to be obtained from the PD pulses. Compared to narrowband measuring devices, PD charge measuring devices that integrate in the time domain do not have any special requirements for the frequency spectrum. Due to the large bandwidth, these devices have a very good pulse resolution.

Coupling method

According to IEC 60270, there are three methods for testing partial discharge through coupling capacitors: direct measurement, indirect measurement and differential measurement, which correspond to the following figures:

Indirect and differential measurement

It can be found that in indirect measurement, the coupling device and the coupling capacitor are connected in series, and the coupling device will not be destroyed when the sample is damaged. However, its sensitivity is lower than that of direct measurement (the coupling device and the insulator to be tested are connected in series). However, when the test sample breaks down, direct measurement is more likely to damage the coupling device.

Actual measurement

In our experiment, an indirect measurement method is used, using frequency domain integration. Different center frequencies and bandwidths are adjusted before measurement to keep the basic noise level low. Each time the sample to be tested is changed, the ratio of the measured charge qm and the apparent charge qs needs to be recalibrated. The plate capacitor of the insulating oil contaminated by the sample to be tested mainly shows partial discharge in the air gap caused by impurities or bubbles.

Partial discharge of defective plate capacitor

Basic noise level: QIEC = 4.3pC
 

The effective value of the voltage at the start of partial discharge: UTEE = 8.8 kV

The green line in the figure below is the test voltage waveform. When the voltage RMS reaches 8.8kV, the sample to be tested begins to have partial discharge in the negative half cycle of the voltage. When the voltage RMS reaches 10kV, the number of partial discharges in the negative half cycle of the voltage increases.

Locally placed at 10kV

Partial discharge of IGBT modules

The structure of the IGBT module is shown in the figure. It is an insulating device with voltage levels ranging from 600V to 6500V. For voltage levels above 3300V, partial discharge assessment is required.