[de_]

Thanks Everyone.

I sounds like FRC should just make it a rule that the compressor has to use a breaker and that 12 gauge, not 18 guage wire should be used.

I'm still somewhat concerned that in one case the case of the spike melted where the output terminal goes into it and the output terminal was desoldered from the spike pcb. One possible explaination is that high (excessive) currents led to heating in the push-on terminal that led to the metal softing reducing the contact pressure leading to increased resistance leading to more heat...
But the plastic insulation showed no heat damage so it sounds more like a weak solder joint on the pcb.

Out of spec excessively tight clearance between the compressor piston and cylinder wall could increase current draw but since nobody else has seen this and since the compressor seems to be starting and running fine, that doesn't look likely.
I will measure the current the next time I see the robot. I just ran out of time. Thanks again everyone.