Here's the latest video of babysitting the mosfets, to discover what caused previous failures. It still could possibly be noise on the probe, being clamped by the 15v TVS diode:
I did some tests with the transient voltage suppresion diodes, and they get pretty warm at even 20-30mA continuous, when they are clamping to 18v. I just tested that on my little variable power supply. So, I glued on a thermistor to one of the TVS diodes on the driver board, and am now waiting for the glue to dry. Then I'm going to see under what conditions the tvs diode gets hot. If it does get hot, then the spikes are real, and I might need to move up to a higher powered TVS diode. I'm using a 600w peak, 1w continuous at the moment, so I figured a 6000w peak, 8w continuous would do it. haha. Those things are fast! It takes them less than 1 picosecond to start clamping the voltage. To give an idea how small that is, each horizontal division on the oscilloscope in the video is 20000 picoseconds.