Piezoelectric damage from MOSFET removal?

[Winston]

I was attempting to remove and replace the four low quality (I assume at least two had unusually high Rds(on)) SOT-23 MOSFETs on a brand new, inexpensive "toy grade" miniquadcopter. All MOSFETs worked, but two motors were considerably slower than the others requiring trim adjustments on the TX. The problem remained when other motors were connected to those channels. I didn't want to use my hot air rework station because the board was cheap and had SMD components on both sides, so I was using the desolder-one-pin-at-a-time-and-lift method. The package of one MOSFET just would not lift, so I used my lead cutter to cut it and then removed each lead individually. No pads were damaged in any way. ESD precautions were followed. Or, perhaps, unintentionally, they weren't...

When I tested the flight controller board after replacing all four MOSFETS, any motor I connected to the channel with the MOSFET I had to cut to remove did not work. I didn't scope it out because this thing was already barely worth the effort to mess with in the first place, but I wonder if a piezoelectrically generated voltage pulse from cutting the MOSFET's silicon die might have fried the port on the microcontroller attached to the gate of that MOSFET. If so, this is a failure mode I'd never considered until now.

[MadTux]

No way. The MOSFET/motor control was probably  bad from the beginning. China quality and quality control. Silicon is to my knowledge a bad piezoelectric material and piezoelectric materials usually quite need a quick strong shock to get reasonable voltages.

(Played with some natural quartz crystals a while ago, after watching this video:

Aluminum foil on each side of the crystal connected to scope probe and Tek 7A22, for high pass 50Hz rejection. Used a polyamide rod and a hammer to compress the quartz crystal. Got voltages in the 10V range when hammering hard, so you really need a strong blow to get a bit voltage, so no way with a cutter knife)

[Winston]

Yeah, just looked it up in depth and found that Si is NOT piezoelectric. A summary result from a Google search seemed to indicate it was. Glad it wasn't my stupid fault then.

Agree on the Chinese cra*p comment. Weird that just that channel failed, though. All of the rest worked great, including the one that had originally had a slow spinning motor. The only components between a typical microcontroller port and a medium/low-current MOSFET's gate are typically just a low value series resistor and a pull-up or down resistor, so what else could have failed and why beats me. I guess just the port itself finally died... for no good reason other than it was in a 32-bit ARM from a no-name Chinese manufacturer whose web site is now dead.

[Alex Eisenhut]

What part number? How do you know you even have the same pinout?

[Winston]

What part number? How do you know you even have the same pinout?

Three out of four channels worked after MOSFET replacement and the one with the slow motor worked great. Pinout issues were thus N/A from my description above. The original parts were all marked C009T which is a SI2300 and, based upon that, a compatible but superior MOSFET was chosen, a long part number which I don't recall off the top of my head.

[Rasz]

you killed pcb mechanically by using force instead of hotair, most likely perfectly salvageable with continuity tester (and maybe replacing cracked smd capacitors/resistors)

[Alex Eisenhut]

What part number? How do you know you even have the same pinout?

Three out of four channels worked after MOSFET replacement and the one with the slow motor worked great. Pinout issues were thus N/A from my description above. The original parts were all marked C009T which is a SI2300 and, based upon that, a compatible but superior MOSFET was chosen, a long part number which I don't recall off the top of my head.

Ah yes... perhaps the tx trim adjustments need to be put back to what they were before. I know that my MQX gets cranky when confronted with wrong "throttle" setting on power up. It does things like what you say, one motor doesn't spin. I don't know why.

[iamdarkyoshi]

one motor doesn't spin. I don't know why.

Could the gyro be trying to compensate anything? My X4 quads will run the motors at ramdom speeds until the gyro notices something change, and then the microcontroller can get its bearings. But if the quad cant move, it cant calibrate anything.

[Winston]

you killed pcb mechanically by using force instead of hotair, most likely perfectly salvageable with continuity tester (and maybe replacing cracked smd capacitors/resistors)

No, absolutely no damage to the PCB anywhere under magnification. Not worth the effort to proceed further. Parts (lipo, motors, body, props, etc.) will just be used as spares for an identical quad.

[Winston]

What part number? How do you know you even have the same pinout?

Three out of four channels worked after MOSFET replacement and the one with the slow motor worked great. Pinout issues were thus N/A from my description above. The original parts were all marked C009T which is a SI2300 and, based upon that, a compatible but superior MOSFET was chosen, a long part number which I don't recall off the top of my head.

Ah yes... perhaps the tx trim adjustments need to be put back to what they were before. I know that my MQX gets cranky when confronted with wrong "throttle" setting on power up. It does things like what you say, one motor doesn't spin. I don't know why.

No, trim was tried. Motor channel was dead.

[Winston]

one motor doesn't spin. I don't know why.

Could the gyro be trying to compensate anything? My X4 quads will run the motors at ramdom speeds until the gyro notices something change, and then the microcontroller can get its bearings. But if the quad cant move, it cant calibrate anything.

With all of the "toy grade" quads I know of, calibration occurs shortly after power-up after the quad is placed in its take-off position (level ground). There are also specific stick movements that will reset the controller calibration when stationary on level ground. Neither were productive in this case.