Maximum current value to test dielectrics

[50AE]

Hi folks,

I will be building a HV DC power supply with current limitation to test the dielectric breakdown voltage between the primary and secondary of my transformers.

But, what should be, a safe current value in order not to carbonise the breakdown insulation spot?

I guess we need to calculate dissipated power per surface here? But what is the surface of an internal arc?

 

[Vovk Z]

I don't know but I'm shure it definitely depends of your transformers size. They can be from 1 VA to 333 MVA rated .

 

[FauxFrench]

Hi folks,

I will be building a HV DC power supply with current limitation to test the dielectric breakdown voltage between the primary and secondary of my transformers.

But, what should be, a safe current value in order not to carbonise the breakdown insulation spot?

I guess we need to calculate dissipated power per surface here? But what is the surface of an internal arc?

A good question raised by someone who seems to understand the problematic.
What if we turn the question around and decide a maximum leakage current you will allow and still consider the insulation in order?

Example: say 1uA is OK. If you then apply a purely resistive high voltage and test if that current threshold is exceeded.
Telling the energy level where a partial discharge is likely to cause a small carbonisation seems really difficult and depending on the particular test object.

 

[Elvee]

In principle, HiPot testing is seen as destructive: the DUT is subjected to the test potential during a defined period of time, and if a discharge (= current higher than a set value) occurs, the DUT is discarded, because it does not comply with the spec, and because the discharge will have initiated a failure path, even for lower voltages.

If safety insulation is involved, that rule should always be adhered to, but for functional insulation, a more relaxed approach is probably possible: I have been making and testing various types of insulators, and I didn't want to necessarily discard any sample that failed the test, so I developed this instrument:
A Low-Lethality Dielectric Strength Tester

The materials I test are mainly silicone/mineral based, and thus have a low carbon content, which is a valuable help.

In a transformer, the insulators are generally carbon based: either polymer or cellulose, and could be irreversibly damaged by a breakdown (there are alternatives, but not commonly available for ordinary transformers)

 

[trobbins]

50AE, are you aiming to comply to a particular standard, or aiming to exceed a particular level and gauge what the actual withstand level is, or ?

Is this a mains AC frequency withstand voltage test, or a DC withstand test or an insulation resistance type test?

One sequence of tests I have done is an IR test, followed by a mains frequency withstand, followed by another IR test. The commercial voltage withstand tester (mains frequency test voltage) had something like a 200mA trip on it, but that current level was to allow physically large transformers to be tested, which have similarly large capacitance.

Another commercial tester, the BPL RM-215L and later the AVO RM215-L have a breakdown/flashover indicator with variable setting from 100uA to 1mA, and a 1-5ms detection period. They also have an ionisation/leakage indicator (0-100uA and audible) for the on-set of leakage. That tester has a circa 4mA max short circuit current, and can dial up to 6kVac and 12kVdc.

FFIW, you may want carbonisation to occur if you are trying to identify/confirm the location of a leakage path (eg. perhaps when not visible with the lights off).

 

[Mark Tillotson]

One problem may be that the atmospheric and surface leakage currents may mask the onset of avalanche breakdown in the bulk solid insulation. In other words can you be sure what you are actually measuring? There's a risk of actually measuring the relative humidity of the atmosphere.


I think the best test is to apply a test voltage well below the theoretical breakdown point,
but high enough to ensure safe insulation levels, and measure the leakage current accurately to spot issues.


No straining materials into failure, less risk of flash-over, corona discharge and other confounding effects.

 

[50AE]

Okay folks, your valuable answers gave a lot of food for thought.

My main idea is to discover the maximum DC breakdown voltage in a specific AUDIO transformer and publish it in its datasheet, probably /2 the measured value for a safety margin.

If the dielectric will really be destructed, it sounds like a better idea to wind quick DUTs with a single layer of primary and secondary, test them, then ditch them out.

Most of the dielectrics I'm using have datasheets. But we must add into account the enamel of the wire, plus moisture inside the dielectric, etc.

Elvee, as always you bring precious project ideas to this forum!

 

[trobbins]

How are you including creepage and clearance performance with your construction, versus solid insulation breakdown (and possibly multiple layers for safety redundancy)?

 

[50AE]

Can you please clarify? By "creepage", do you mean the wire pressure deformation on the dielectric and the "clearance" by layer ends air space?

 

[trobbins]

Creepage is minimum distance along the surface of an insulator between two conductors (eg. edge of a layer of primary turns, and edge of a layer of secondary turns on next layer.
Clearance is similar but can be via an air gap.
Some standards require certain minimum distances.
These are not distances through solid insulation.

 

[JMFahey]

I think the main point of testing primary to secondary insulation breakdown or failure voltage ifor a consumer product is not "destruction" level but **SAFETY** level.

Who cares your transformer can happily pass 100mA all day long if anything above 5 mA is dangerous and 100 mA through your chest will paralize and most probably KILL you?

Since it seems to be a non destructive and safe test , I would aim at, say, 1mA to provide a safety margin with respect to 5mA.

And really, at 1mA it´s not much of "insulation" anymore.

 

[50AE]

Creepage is minimum distance along the surface of an insulator between two conductors (eg. edge of a layer of primary turns, and edge of a layer of secondary turns on next layer.
Clearance is similar but can be via an air gap.
Some standards require certain minimum distances.
These are not distances through solid insulation.

I got this. Well, yes, different types of transformers will have different allowable creepages. The idea here is not to have the clearances and creepages to be the Achilles' heel, but the solid dielectric itself.

JMFahey, no current should pass, we're talking mostly about output transformers. If the dielectric breaks, the amount of current will depend mostly on the circuit, which for an output stage, will be high enough to kill a person.

My primary idea is to TEST the DUT's (transformer) dielectric to determine the maximum breakdown voltage, preferably averaging the value from testing several DUTs, then finally dividing that value by half and set it as maximum allowable primary to secondary voltage.

It seems destruction is an inevitable part of the testing?

 

[Elvee]

It seems destruction is an inevitable part of the testing?

No, just do as everyone else does: design for 10kV, test (most of the time) non-destructively at 3kV, and specify an abs max voltage rating of 1kV (just examples)

 

[trobbins]

I recall there is a EU or US standard for commercial audio equipment, which could be related to the incentive some decades ago to use Speakon output connectors due to hazardous voltage.
I'd be surprised that any standard would just pass on solid insulation alone, as that is often bundled with compliance for creepage and clearance distances with some pollution degree and for a rated working voltage.
Wrt to testing, you have options such as routine testing on every sample, and for typical working voltages of up to 1.5kV, then a 3kV test level for a non-mains isolating device is likely common, and could be ok to do for 1 sec compared to nominal 60 secs. But standards are a labyrinth of twists and turns, and costly to acquire.

 

[Mark Tillotson]

No, just do as everyone else does: design for 10kV, test (most of the time) non-destructively at 3kV, and specify an abs max voltage rating of 1kV (just examples)

Another approach is test every Nth unit to destruction, and monitor the statistics in order to catch when the manufacturing process is going wrong - you also test every unit's leakage at full rated voltage too, which should correlate with destructive tests (if not, you need to figure out what's happening better!)

 

[50AE]

No, just do as everyone else does: design for 10kV, test (most of the time) non-destructively at 3kV, and specify an abs max voltage rating of 1kV (just examples)

The dielectrics I'm using have datasheets. The most common one I'm applying between primary and secondary layers is Nomex-mylar-nomex. One easy way would be to just use the values in the datasheet.