I will dig out my scope and take some measurements. I only used the Low ESR electrolytics so far. These are very high grade, they came out of a variable speed motor drive, so they were intended to filter high frequency switching DC loads. With only 3,600 uf total, I was not expecting much of a spark, but wow, 50 volts made a nice snap. I will use my pre charge resistor next time I connect them. Glad it didn't pop the 30 amp MRBF fuse feeding it.
I am certain there is nothing wrong with Cell 12. The XW inverter/charger runs up to 80 amps charging or discharging and no cell deviates more than 10 millivolts under those loads, and with the little buck converter off, Cell 12 is smack in the middle of the cell voltage range. And even when the 12 volt buck converter is running and the cell voltage readings are going stupid, my Fluke meter shows all of the cells stable as a rock and within 10 mv from highest to lowest. And the way the cells are in the Chevy Bolt bricks, I can't really swap around cells. Cell 12 is actually 6 cells, consisting of 2 groups of 3, in 2 different sections of the top brick. Cells 1 through 10 are each 3 cells in the bottom brick, paralleled with the matching 3 cells in the middle brick. The Cells 11 through 14 are the top brick. It was originally an 8S3P setup. I cut the middle buss bar, and paralleled the back 4 groups with the front 4 groups to make it 4S6P. The balance leads of the 2 separate 3P strings are all fused at 7 amps and tied between them with #16 wire to the BMS #20 balance leads. The small wire resistance between the strings parallel balances the 2 groups and reports the average between them to the BMS. I was very careful to keep all the resistances matched. When the balancer is moving 2 amps, I see 1 amp to each group on the balance leads. The two main current paths are the factory Chevy Bolt Buss bars, and #2 cables, fused at 125 amps to the main buss bar. So swapping cells around is not really an option.
I guess I could add 14 capacitors across each cell at the balancer end of the leads, but that will be a bit of a pain. About the only way I could see doing it wuld be to cut all the balance wires and splice in th caps about an inch from the connector. The balance wires are about 0.13 ohms from each cell, so even a 1 uf cap should filter a fair bit of RFI. Probable use a good film or ceramic type, but with 14 of them in a row, it will be an interesting experiment. Maybe I can slip a ferrite bead on each lead as well. 15 of those. Since it is measuring the voltage of each cell, I would just use low voltage caps wired the same as each cell at the balance connector.
First I tried an audio range digital scope and it was basically useless. So I pulled out my old BK Precision crt scope. The caps on the input side are certainly doing their job. I can't lock any voltage waveform on the input side of the buck converter, or on the lead to the batteries. I expected to see some 120 HZ as the XW inverter is pumping out 1,600 watts into my house right now. On the output side of the buck converter, WOW, it has all kinds of hash. I could see the 12 volt DC but the screen had a glow above and below it. I put it in AC coupled so I could center the trace without the 12 volt offset. Turned the gain up just a little and it was able to sync a waveform. To see the wave, I had to dial up my sweep rate to the highest setting, on a 15 Mhz scope. 1 us / div. The repetition rate of the main pulse is about 154,000 hz (6.5 us period). Then it has a ringing well over 2 Mhz after each pulse. The peak to peak voltage at the main switching pulse is OVER 12 volts. In DC mod, it went from below 6 volts to over 18 volts. The ring after the pulse quickly drops to about 1 volts, then tapers to just about nothing right as it hits the next switch pulse. Here is the best pic I could get on my phone.
View attachment 25217
The phone image does not really do it justice.
Here are pics of the toroid and the filter installed.
View attachment 25218 View attachment 25219
I am hoping some caps on the output of the buck converter might be enough to stop the balancer from going goofy.