The capacitors I bought for this project are shit.
They are SACON photo-flash electrolytic, rated for 350 uF 360 WV (WV == working voltage, AFAIK). When I charge one up (new capacitor!), it starts to seriously leak above about 250V. Which is a problem, because if some capacitors are leaking and others aren't, substantial charge imbalance in series connection can build up. This can cause low-leakage "cells" to charge up to more than rated voltage. And when the bank is discharged, leaky capacitors will end up with reversed polarity charge, which is very bad for their health.
It seems that if I let the leaky capacitors leak for a while, the leak diminishes. So I've set up a "conditioning" process. I just soldered a few caps in parallel, and connected them via a rectifier to a variac, so that they are held at their rated voltage of 360 V till the leak disappears. The remaining leakage can be easily seen with thermal camera.
This is a thermal camera shot of the capacitors in about 2 minutes after the voltage was applied.
Their temperatures are reasonably even, apart from one cap, which I think I had conditioned previously already, I just forgot which one so I just soldered it into the bank.
About 20 minutes later, I was wasting time in front of computer, when I heard the variac started roaring because of load current increasing. I rushed to switch it off, but before I did that, a bang happened. One capacitor has vented. It turned out, these capacitors contain a cloud worth of magic smoke! It all happened within 2 seconds I think. Luckily I was right next to it all to switch it off promptly, otherwise it might have resulted in a fire.
Continuing the conditioning...
most caps seem to be doing well, apart from three. Strangely they are right next to the one that vented. I think they might have been upset by the heat from that faulty one.