Appologies that I haven't been as active as I would like on the forum, not enough hours in the day! Some good questions have been asked, so hopefully some answers.
I'm not sure I understand the logic of the charger coming back on if the cell voltage dropped to 3.6V?
The bms boards turned on a 250ma balancing load at 3.6 volts and switched it off at 3.55 volts. Since I didn't have any way to control the charger I simply switched it off when a cell got to 3.75 volts. If I had left it off my pack would not have been balanced since some of the cells would have been at lower voltage. The BMS board continued to load the cell with a 250ma load and when that cell got back down to 3.6 volts the BMS would have allowed the charger to switch back on. This allowed the lower cells to catch up since their BMS boards would not be loading their cells.
If it was a cell going short circuit, from what I understand of your BMS you would have noticed, the problem develop over time, a cell taking a lot longer to charge than the others, for some days or even weeks before
My simple analogue setup provided no individual data for the cells. Had planned on moving to the full digital version as that offers individual cell monitoring and also any BMS failures are immediately detected and the system can just shut down. I had been using a Paktrakr that gave me a reading of each "12 volt" battery but I had stopped using it because it powers itself from the first 12 volt battery and was increasing my balancing time. I was also happy that all the cells were behaving normally and had no reason to suspect that a problem was brewing.
In normal auto fires, 99% of the time its wiring catching fire from lose terminals getting hot and the insulation or sound deadening material catching light.
That was exactly the cause of this PHEV Prius fire
. BTW, its a 2mb pdf. In this case the heat caused from the poor connection caused the cells to overheat and vent, but in a much more violent manner than shown in the youtube video above. In my case their was no high currents to generate large amounts of heat. One possibilty is that the relay controlling my charger simply stuck closed in the night. The charger would have stayed on, overcharging a cell, possibly causeing it to swell. Since they were not restrained against swelling, they could have burst out of their boxes and that could have possibly allowed them to short and start the fire. This is of course all hypothetical, but a plausible explanation, so should be kept in mind for future projects.
One thing that the Prius fire and the you tube video show is that these cells can generate tremendous heat when something goes wrong and that heat can easily start a fire even if the cells themselves don't catch fire. Temperature monitoring is essential and doesn't need to be complex. Even simple temp snap switches that open at say 40 Celsius would be a step in the right direction. And give careful thought about what happens if any one component in your system fails, what is the backup?