Replacing burned-out TIP-102s and the damage done around them makes me want to focus more on prevention, which made me realize I'm not 100% sure I know all the reasons those suckers blow up. Here are my theories:
1) Shorted coils (or a shorted flyback / "catch" diode across the coil). If the coil is shorted, the transistor will try to short the 70V supply to GND, and it's a race between the fuse and the transistor as to which one pops first. I didn't really think shorted coils was a thing (figured they'd fail open), but the "ramp down" coil on my Getaway schooled me on that. A shorted diode would cause the same thing.
2) Open diode. Every coil needs a diode across it to limit the inductive kickback when the transistor turns off. Sometimes the diode is on the coil (it's most effective when mounted right at the coil), sometimes it's on a PCB. When the diode is working, it limits the max voltage the transistor sees to (coil power supply voltage + 0.7V). If the diode is open or missing, the transistor will see a kickback spike equal to (coil power supply voltage + 100s of volts) (basically the thing that will limit the voltage is the transistor's VCE breakdown voltage, and if it keeps seeing that spike, the transistor's gonna fail).
3) A logic or software failure causing the transistor to be stuck on, so the coil is always energized (or a broken EOS switch/circuit that keeps the coil in the high power mode while someone's trapping a ball). This could generate a lot of heat (explaining some of the heat damage you see on PCBs) and eventually blow a fuse/coil/transistor/trace.
4) Old age. Eventually the transistors just wear out and fail open (we hope) or closed (hello burned-up coil).
So it seems like replacing your diodes every 5 years or so would be an inexpensive way to decrease your chances of driver board disaster.
Thoughts? Am I missing any other causes of transistor death?