Detecting Shorted MOSFETs in Output Stage

[Jdo300]

Hey Everyone,

The title of this thread may be a bit ambiguous but what I am wondering is if there is a way to determine if a MOSFET in an output stage has shorted. The scenario I want to catch is the case where the FET drain is exposed to too much voltage (from a collapsing spike or some other unpredictable transient) and the drain blows and shorts out. At the moment, my protection circuit is simply a small mechanical circuit breaker which cuts the supply (12V in my case) to stop the circuit from roasting, but I'm trying to determine if there is another way to detect that the MOSFET is blown (so the uC will know) other than testing for a lack of supply voltage on the output stage?

Thanks,
Jason O

 

[berkeman]

Is the output stage push-pull, or open-drain?

In either case, can you sense the quiescent output voltage (no output drive), and determine if it is correct? If you have a shorted output FET, the expected output voltage should change...

 

[Jdo300]

Hi berkeman,

In my setup, the FET is essentially being used as a low-side flyback driver (open drain). It is driving a flyback transformer primary with a 12V input but when the switch closes, the flyback voltage can rise as high as 400V. the FETs I'm using will handle this with no problem but I'm wondering what happens in the case where something goes wrong. When the FET shorts, that would simply blow the breaker on the circuit. But even without the input voltage to the output stage, is there any way to safely determine weather the fet is good or bad (to differentiate between a case where there is simple some kind of current fault vs. the fault being caused because the MOSFET failed).

Sorry in advance for the confusing explanation, I'm still trying to get it straight in my mind even now.

- Jason O

 

[vk6kro]

TV sets have a similar arrangement and when the line output transistor breaks down, it also destroys an expensive line output transformer in series with it.

Rather than develop a sensor for detecting a failure, it would be better to improve the circuit so that it is less likely to happen.

If the FET was not normally in saturation (ie with near zero volts on it) you could detect this, but that 400 volts would be a problem with logic circuitry.

If the FET failed by shorting to the Gate from the Drain, then the Gate would be at zero volts, too, so you could detect that. Mosfets normally have a forward bias on them.

If the circuit normally generated high voltage pulses at a constant frequency, you could detect that these were not there any more by using a 567 tone decoder and a small pickup probe near the circuitry.

The transformers normally generate smoke when they fail, so you could use a smoke detector.

 

[waht]

If a transistor begins to fail it will behave more non-linearly and cause a greater harmonic distortion. You could detect the rise of a 2nd or 3rd harmonic with some sort of filter and detector or if the micro-controller had an ADC on board, code an FFT program to monitor the spectrum.

But that's all too much work for a switcher. Good components and protection should ensure it will work fine.

 

[Jdo300]

Yeah, that definitely makes sense. I'm going to rely primarily on the input circuit breaker to stop any major damage from happening to the board. The loss of the supply voltage to the fet would be the best way, in my situation, to determine if something went wrong with the output stage. Thank you everyone for the suggestions. I especially like the "smoke detector" one, that was great, lol. I have another question about temperature sensors related to the same situation but I'll make a new thread about that.

Thanks!
Jason O