Running an APM mini 3.1 with the stable release 3.2 firmware. 
I had a series of events leading to this investigation. I performed an auto land with a 550 class drone. 15" CF props and 30a generic SimonK ESC's. The auto land went pretty well and settled into the grass but the motors did not disarm as the system had done on 2-3 other auto land tests (new build, still tuning) I was not paying really close attention as I was demonstrating the function to a friend. About 10-15 seconds after touch down the quad started to increase rpm on one side and it cause the machine to lean over and one of the blades struck the grass before I could disarm via the TX. That immediately smoked that ESC.....dang. I brought it home for repairs and replace the fried ESC with another generic 30a SimonK from a second set of four I had from different manufacturer. Upon executing a rotation test the replacement ESC blew. Thinking I may have shorted something as I always directly solder to cut to length wires to save weight. Installed the next ESC from that second set and it blew also. Pissed now, install third ESC. This one I test run up by connecting a spare RX directly to this ESC and it runs the prop up fine. Hook up to the APM and POOF, There went number three. Now on my fourth, props off all motors.

Findings.
A. upon arming the APM all motors begin to spin at an idle. This build has always done this and I did not mind as it gave a good visual that the quad was armed. 
B. when the throttle is moved just slightly (stabilize mode), there is a large jump in RPM. Not enough for lift off but close. Without the props this was not a problem for this fourth ESC but the sudden increase was enough of a jolt as to make that motor emit a bang type of noise. The three original ESC's do this as well but not as noticeable.
C. Pulse widths sent from the APM: Disarmed - .940 ms, Armed - .870 ms and then the jump to .520 ms. It is not ramped up at all, it just jumps.
D. I checked my radio on the calibration screen. It begins at 1099 and SMOOTHLY transitions to 1900

My guess is the rawness of the SimonK software was attempting to increase the speed of that 15" prop at the same rate the APM did, i.e. IMMEDIATELY! and popped a FET.

Is there some setting or parameter that controls how fast the signal to the ESC's is allowed to change? Immediately seems to be too fast for this build. I am not sure why the APM signal changes like that. Can it be adjusted to ramp up rather than jump?

Keshka