The generic question I have is what's the "expected" behavior for a damaged BJT. Let me add more color..

I was trying to fix my kid's toy where a motor would run even when it wasn't supposed to (no matter if the switch was on or off, it would always run). I tried to reverse-engineer the schematic driving the motor and here's what I think they had on the PCB:

enter image description here

The motor is represented by an inductor (ignore its value). What I think is damaged is the transistor Q3 because when the motor was running there were ~8V at the base of Q3 (so it should have been off) and yet I could measure ~1V at the gate of Q4, enough to keep it on. Q4 didn't seem to be broken because if I drove 0V to its gate, the motor would stop running.

Is my analysis correct or the problem of them motor always on could be elsewhere? I guess my question is: is this an expected behavior for a damaged PNP? If not, what is? I'm asking because I'd rather expect a burned BJT not to conduct when it's on, not the other way around (i.e. conduct even when it's off). And if that's the case, what might have caused it? Some specific incident (spikes in current or similar) or do transistor wear out over time anyway?



So I opened it up again and there was indeed a diode across the motor as Spehro suggested (just not on the PCB). And as andre314 suggested, I measured the voltage across R4: it was 0V. So it looks like Spehro was right: the transistor that's damaged is Q4, not Q3. Now I need to learn how to replace it...

Thanks everyone for your help and suggestions

  • 1
    \$\begingroup\$ Q4 could still be damaged and cause this. \$\endgroup\$
    – Andy aka
    Apr 2, 2021 at 16:37
  • \$\begingroup\$ The failure mode of most BJTs and MOSFETs is to fail with two or more legs shorted, or nearly shorted. Less common are base/gate open or leaky. \$\endgroup\$
    – rdtsc
    Apr 2, 2021 at 17:21

2 Answers 2


Remove R4 and confirm that the motor remains off. Shorting R5 is not sufficient to be sure, there could be D-G leakage on Q4 for example. I would suspect Q4 first, damage to Q3 would likely be more of a random failure (toys are not known for using the finest space-qualified components).

Semiconductors do tend to fail "on" unless they've been abused to the point where the case is blown apart or wirebonds are vaporized.

It would be better if there was a diode across the motor unless there is some reason that is not shown in this schematic that would make that a terrible idea (possibly damaging other components).

  • \$\begingroup\$ Unfortunately I cannot remove R4 easily because it's SMD and I don't have the tools (nor the skill) to unsolder it. I believe there is a diode across the motor (inside the transistor), I just didn't report it in my schematic. \$\endgroup\$
    – fab
    Apr 2, 2021 at 16:47
  • 2
    \$\begingroup\$ A diode across D-S (body diode) is not the same as a diode across the motor. The problem is that when the MOSFET tries to turn off the current continues flow due to inductance and the drain voltage increases greatly, which can (immediately or not) damage the transistor. The body diode has no effect. In the case of an H-bridge the body diode in another transistor may perform that function. \$\endgroup\$ Apr 2, 2021 at 16:51
  • 1
    \$\begingroup\$ If you can't remove R4, you could measure the voltage across it. If it is ~50mV, then the current throught R5 comes from Q3 (which means Q3 KO), if it is ~0mV, then the current comes from the gate of Q4 (which means Q4 KO). Otherwise : the situation needs further investigations. \$\endgroup\$
    – andre314
    Apr 2, 2021 at 17:33
  • \$\begingroup\$ Or just remove Q3 (since you suspect it anyway) and carefully replace it if it does not confirm your hypothesis. \$\endgroup\$ Apr 2, 2021 at 17:39

The diode inside the FET protects when used as a high side inductive switch (with Vboost on Vgs) so it is unable to protect low side switching that creates a positive spike on Motor V- and drain.

Due to Miller capacitance on the FET the PNP BJT can see a reverse Vce. You can try to verify the PN junctions on the BJT without a battery using a diode test on a DMM, but with 1V on Vgs this FET would have to be a sub-threshold type of Vt~0.6V (if it powered the motor at the same RPM as normal, but certainly <1V.

With Vgs=1V Vce= 8V @ 1V/220 makes the Rce leakage now look like a 1k6 resistor possibly in the BC junction.

So my vote is to replace Q3


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.