Will multi-time flashovers constitute a failure of Hi-Pot test?

[Roland chung]

In 3rd edition, single momentary flashover is not regarded as insulation breakdown during carrying out Hi-Pot test. How about multi-time flashovers?

During the test, breakdown constitutes a failure. Insulation breakdown is considered to have occurred when the current which flows as a result of the application of the test voltage rapidly increases in an uncontrolled manner, that is, the insulation does not restrict the flow of the current. Corona discharge or a single momentary flashover is not regarded as insulation breakdown.​

​

​

​

​

​

​

 

[Peter Selvey]

The "Hi-Pot" test is intended to verify solid insulation. If the corona discharge has nothing to do with the insulation being tested, you can ignore it.

At voltages above 3000Vpeak, corona is common and may occur away from the actual solid insulation under test, especially if care is not taken with the set up.

However, there is a possibility that corona / flashover is related to the insulation partially breaking down and then recovering. For this it is important to make sure to use the 100mA supply as required by CB scheme/CTL, to avoid that the recovery is limited current capacity in the hi-pot supply.

In general though, it is normally possible to design solid insulation that is >> requirement, so corona/flashover is a test set up issue.

 

[Roland chung]

Thanks, peter.

The continuous flashover occurred in transformer's core when hi-pot test was carried out between input and output of switching power supply. The applied voltage was 4 kV, the test instrument didn't fail the test since the current was only 2 mA (tripping current was set to 10 mA). Did you mean that the test result should be judged based on the current only?

 

[Peter Selvey]

Almost certainly a set up issue, you rarely see 2mA at 4kV if the set up is correct (but often see if the set up is wrong).

The high leakage indicates a side path is getting picked up, probably via EMC capacitors, from secondary to earth which is the typical case.

This usually means one or both of the following:
1) areas not designed or required for 4kV get overstressed (hence the flashovers)
2) the area designed for 4kV is not being stressed/tested

 

[Roland chung]

In case of the setup is correct (just in case), this test fails or not?

 

[Peter Selvey]

I should distinguish between corona and flashover, I assumed at the start you were talking about corona.

Corona can occur around a surface near an electrode without actually bridging to the other electrode. It's just the point where the local field strength is higher than the molecular O2 bonds. Near an electrode the field strength (V/m) can be high, but it usually drops off quickly as you move away. As such steady corona (light buzzing sound, smell of ozone) is normal for the 4kV test, and can be ignored.

A big flashover is different and needs investigation. Flashover is a full path from two sides. Because of the high cr/cl limits in IEC 60601-1 (e.g 8.0/5.0mm for 2MOPP) ), flashover is not really possible in normal testing.

So, if it occurs it is likely to be a set up issue.

If one were to insist that the set up correct and flashover still occurs I would fail it as the standard says only "single" flashover is allowed.

 

[Marcelo]

AS defined in the standards, "breakdown is considered to have occurred when the current which flows as a result of the application of the test voltage rapidly increases in an uncontrolled manner, that is, the insulation does not restrict the flow of the current".

So, you have to verify the current and, if it flows with our restriction, it's a breakdown.

Just take care with incremental flow due to capacitance, if your hipot does not have enough power and have a limit to current (most have to avoid hipot damage), it may trip due to the capacitance (this is easily checked, if you increase the voltage and there's a related increase in current, and when your step a step of the voltage and the current also stops, this is a capacitance problem, not a breakdown - in this case you probably forgot to remove the capacitance from the tests).

 

[somashekar]

With a valid setup, can the flashover be the extension of the corona effect perhaps due to faster rate of voltage ramp up or high humidity in the test place ....

 

[Peter Selvey]

You would need to be pretty quick with the dial to have any concern about the ramp up effects. The only risk would be if the voltage was already set up and then applied by a switch, that could cause an issue.

The breakdown voltage in air is around 3kV/mm, this means 5mm clearance requires about 15kVpeak. The 4kVrms test is about 5.6kVpeak, well below this level. Influence of humidity is around 2% from 40-80% and actually higher humidity means higher breakdown voltage (it's counter-intuitive, but fairly well documented). Different geometry such as sharp points can influence, but sharp points are rare and probably not enough.

If the dielectric strength test is not done in the RH chamber, condensation can cause breakdown along any surfaces. But this is not a failure as it is unrelated to the solid insulation under test. It is just lazy testing .

As already stated, far more likely is a set up issue where the volts are being transferred to a location not required or designed for the 4kV.

If you read the standard carefully, solid insulation is best removed from the equipment to avoid stray path issue and to ensure the targeted insulation is actually being tested. For example, I would normally remove a switching power supply PCB from the equipment and any metal frame and then solder in test wires to the primary and secondary, inspect to make sure all parallel paths are removed, and then put that into the chamber. If you do that, you will never get any flashover.

I very (very) rarely do 4kV test on a fully assembled device. It's almost impossible to eliminate the stray paths and make sure the target insulation is is tested. Rare cases are Class II devices where the lack of earthing eliminates many stray path, but even then not always.

 

[Marcelo]

The only risk would be if the voltage was already set up and then applied by a switch, that could cause an issue.

Yeah, that's the case I was thinking, for automated testers. But in this case there would not a corona in fact, only a tripping.

As already stated, far more likely is a set up issue where the volts are being transferred to a location not required or designed for the 4kV.

If you read the standard carefully, solid insulation is best removed from the equipment to avoid stray path issue and to ensure the targeted insulation is actually being tested. For example, I would normally remove a switching power supply PCB from the equipment and any metal frame and then solder in test wires to the primary and secondary, inspect to make sure all parallel paths are removed, and then put that into the chamber. If you do that, you will never get any flashover.

I very (very) rarely do 4kV test on a fully assembled device. It's almost impossible to eliminate the stray paths and make sure the target insulation is is tested. Rare cases are Class II devices where the lack of earthing eliminates many stray path, but even then not always.

This is a common problem in some places (it is in Brazil, at least), when manufactures do not clearly state insulations and then declare to put the "4k from main to AP", but in fact there is a double insulation, and the tester simply put 4k from one to the other. This goes back to the comment in insulation diagrams not bring clear enough.