Electrical Insulation/Hi-pot Test

[mazzola]

I've been reading UL 2271's standards for battery safety but am having trouble understanding how the "Dielectric Voltage Withstand Test"/Hipot would work. The test is meant to ensure the battery has sufficient insulation between its enclosure and the high voltage circuits within. But I don't quite understand the test procedure and how it wouldn't fry the battery under test, even a well-designed one.

The procedure calls for applying 2.8 times the rated voltage of a battery (e.g. 134.4v for a 48v battery) between its high voltage circuit and the enclosure for 1 minute and measuring if there is any arcing.

But if you apply 2.8x the rated voltage wouldn't you fry the BMS? The test procedure allows you to disconnect/bypass "semiconductors and similar electronics liable to be damaged", so you could remove the BMS but then what is preventing the cells from getting over-charged and go into thermal runaway?

The procedure also allows you to disconnect the cells during application of the voltage but I don't see how this is really feasible if the cells are welded or fused to the busbars (i.e. the high voltage circuit).

 

[amberwolf]

You don't apply the voltage across the BMS or battery terminals, never to both at the same time.

You apply across one of the terminals and the enclosure, then remove it from the one terminal and apply to the other and the enclosure, so it is never connected in a way that can cause current flow thru the battery itself (unless there is a fault in the enclosure insulation), and nothing can charge, etc.

If there is a fault that allows current to pass from enclosure to any part of the battery, then it can certainly damage whatever part of the battery is sensitive to that. If the enclosure is designed correctly it won't fail.

These kinds of tests can be destructive, so they are intended to be done on prototypes for mass production units (or the MPUs themselves); if you have a one-off DIY pack then you need to be willing to sacrifice it to the testing gods, or be able to repair whatever is damaged in such tests.

 

[mazzola]

Ah yeah, as long as there's not a fault in the enclosure, the circuit between one test probe and the other remains open and no current flows - so no damage and no over-charging of the cells.

Is there somehow a way to test a one-off DIY pack in a non-destructive way in case it fails? Or at least, not damage the cells of the pack? I'm okay with frying and repairing/replacing a $100-ish BMS but I'd like to save my more expensive cells

I've seen hipot test equipment that has a cut-off/trip current feature that turns off the voltage if it detects a breakdown, but those cost thousands of dollars. I was hoping to get away with a cheaper insulation test tool like the Klein ET600. Megohmmeters like the Klein have a specified test current that they use, usually in the mA range. Could I replicate the cut-off current feature by just putting a cheap, fast-blow 1mA fuse in front of the battery pack? Or would I still irreparably damage the cells even if they see high voltage for a split second?

 

[999zip999]

To let me understand things better what's the voltage and the AH of your pack and what cells are you using. And on what amperage are you going to be using this battery pack at. Or are you just worried about it getting damaged being dropped or in a crash ?

 

[amberwolf]

Is there somehow a way to test a one-off DIY pack in a non-destructive way in case it fails? Or at least, not damage the cells of the pack? I'm okay with frying and repairing/replacing a $100-ish BMS but I'd like to save my more expensive cells

Test it with dead cells. To validate the design as completely as possible short of using the real cells, you'd still need to build them into the same pack layout / interconnects / etc, built and wired up exactly like the real pack would be.

It wouldn't tell you if your actual pack is actually isolated, just that the design itself probably is (or isn't).

I've seen hipot test equipment that has a cut-off/trip current feature that turns off the voltage if it detects a breakdown, but those cost thousands of dollars. I was hoping to get away with a cheaper insulation test tool like the Klein ET600. Megohmmeters like the Klein have a specified test current that they use, usually in the mA range.

I don't remember if the one I have here has that function.

Could I replicate the cut-off current feature by just putting a cheap, fast-blow 1mA fuse in front of the battery pack? Or would I still irreparably damage the cells even if they see high voltage for a split second?

You'd have to check the cell spec sheet for that, or check with the cell manufacturer. I suspect it isn't a problem, but I don't know what might happen with any internal cell protections that might be present.

 

[mazzola]

what's the voltage and the AH of your pack and what cells are you using

I'm thinking of building a 20s7p configuration using LG INR21700-M50L 4800mAh 21700 li-ion cells - so 84v max and 33.6Ah running at 100A peak. I'm also going to be looking at impact and vibration tests that are a part of the UL spec to inform my battery pack design, but my main concern for this thread is some sort of electrical failure (which could be caused by a crash) that would lead to a voltage spike and shock hazard.

You'd have to check the cell spec sheet for that, or check with the cell manufacturer.

The cells I'm considering don't mention any over-voltage protections on the spec sheet. They do undergo over-charge tests but at 3x the spec'd current and only at 4.2V. Maybe I'll try reaching out to LG somehow?

 

[amberwolf]

It's unlikely to be an overvoltage protection, or anything listed as overcharge/etc testing, because this type of issue would not really ever be expected to happen on a charging setup (which should be designed to prevent that).

But they may specify ohter testing that is done on them, that could include hipot / isolation testing. It's possible this info (if available at all) is only available directly from the manufacturer by request...or they may never have done such testing.

 

[greenno]

HI-pot really?
I used to be a HV splicer back in the the day for the local utility companies and we only HP terms starting at 4160v ac.
Anything less than that nada.
Never had any thing blow or track.
300% of the rated voltage is standard for testing as it is DC voltage that is being applied so it's less destructive than AC.
For what's it's worth there would not be any arcing only tracking at that low of voltage.
Keep it all CLEAN and your golden.