My water-flosser died yesterday, and of course I couldn't dump it without attempting to fix it.
It responded to the controls, at least with the LEDs, but the main motor wouldn't run.
The motor is controlled by two paralleled MOSFETs (SMT code A27K, thus should be AO3402).
They act as PWM controllers and power switch (the motor/MOSFETs are directly connected to the Li cell).
Shorting D-S made the motor run, and when in the ON state, a PWM waveform was present on the gate, but the amplitude was ~1Vbe, too small to turn the FETs on.
Upstream of a 330R gate stopper, the signal had a normal amplitude, ~full battery voltage.
I immediately suspected that one of the gates was damaged, and had a parasitic junction behaviour.
To ascertain that hypothesis, I connected a component tester in low-voltage, high-current mode between the G and D, and this made the motor run immediately.
The tester only showed a linear resistance between the terminals, probably the 330R and subsequent logic output in a low state.
When I attempted to start the flosser normally, with the control button, it turned on as it should: connecting the tester cleared the defect in some way.
Apparently, there was something like a parasitic SCR behaviour active, and since the power was always on (via the drain), it couldn't clear by itself.
Connecting the component tester forced a negative/positive gate excursion of several volts, and unlatched whatever parasitic device was present.
Not a completely outlandish behaviour for some ESD protection junctions, but the mystery is why an unprotected device, as the AO3402 is supposed to be, shows such a weird behaviour?
It responded to the controls, at least with the LEDs, but the main motor wouldn't run.
The motor is controlled by two paralleled MOSFETs (SMT code A27K, thus should be AO3402).
They act as PWM controllers and power switch (the motor/MOSFETs are directly connected to the Li cell).
Shorting D-S made the motor run, and when in the ON state, a PWM waveform was present on the gate, but the amplitude was ~1Vbe, too small to turn the FETs on.
Upstream of a 330R gate stopper, the signal had a normal amplitude, ~full battery voltage.
I immediately suspected that one of the gates was damaged, and had a parasitic junction behaviour.
To ascertain that hypothesis, I connected a component tester in low-voltage, high-current mode between the G and D, and this made the motor run immediately.
The tester only showed a linear resistance between the terminals, probably the 330R and subsequent logic output in a low state.
When I attempted to start the flosser normally, with the control button, it turned on as it should: connecting the tester cleared the defect in some way.
Apparently, there was something like a parasitic SCR behaviour active, and since the power was always on (via the drain), it couldn't clear by itself.
Connecting the component tester forced a negative/positive gate excursion of several volts, and unlatched whatever parasitic device was present.
Not a completely outlandish behaviour for some ESD protection junctions, but the mystery is why an unprotected device, as the AO3402 is supposed to be, shows such a weird behaviour?
The parasitic NPN between drain and source has snap back behaviour, but that makes the MOSFET conduct rather than not conduct. Maybe there was some microscopic gate-source short that evaporated when you connected the component tester?
Yes, that is one of the hypothesis I considered, but the resistance of the "short" would have needed to be ~80R? Is it realistic? It looks a bit large
Do those MOSFETs have a common gate stopper or did they both behave weirdly?
The resistance doesn't sound unreasonable to me if the damaged spot is small. Maybe there was a weak spot in the gate oxide that suffered from time-dependent dielectric breakdown, or some local ESD damage. It would be very weird if that happened simultaneously with two MOSFETs, though.
The resistance doesn't sound unreasonable to me if the damaged spot is small. Maybe there was a weak spot in the gate oxide that suffered from time-dependent dielectric breakdown, or some local ESD damage. It would be very weird if that happened simultaneously with two MOSFETs, though.
A dry joint on the gate would leave a MOSFET on for many minutes.
I did it with an analogue resistance meter and assumed that the FET was dead short many years ago.
The gates hold charge.
It took me weeks to work out what was happening.
I did it with an analogue resistance meter and assumed that the FET was dead short many years ago.
The gates hold charge.
It took me weeks to work out what was happening.
I placed the O-scope probe directly on the transistor terminal, so that' excluded
They had a common stopper
The same problem reappeared 24h later. The same manipulation cured it again, but it came back after a few minutes, and then more and more frequently until it remained dead for good.
I unsoldered the most accessible transistor, and for once I was lucky: it had a low G-S resistance. The other appeared intact.
I replaced it, and everything worked like charm.
Thus, the gate oxide layer had a tiny defect which could momentarily be cleared, but came back each time with a vengeance, until it grew so large as to practically short the gate.
I unsoldered the most accessible transistor, and for once I was lucky: it had a low G-S resistance. The other appeared intact.
I replaced it, and everything worked like charm.
Thus, the gate oxide layer had a tiny defect which could momentarily be cleared, but came back each time with a vengeance, until it grew so large as to practically short the gate.