Hello all, and thanks ahead of time.

So this is a simple power mosfet circuit to drive a 36v 300w dc motor. I'm having an issue with the IRF 640 mosfet popping when given ~ about 20% signal at 1khz, and I can't really explain why. Everything works fine below that load.

Assuming stall current is approximated close enough via 0.8 ohms (measured the motor at different turns of the shaft and 0.8 was the lowest) internal times the 12v power... this gives ~ 15A. So even under a open load it should still not pop a mosfet rated at 18A continuous. That being said, it does not pop until I set my drive signal around or above 20%, roughly 2.5v averaged at the motor.. and only 3A. Speaking of which, the IRF640 is rated at 72A pulsed, thus leading me to believe this has nothing to do with the mosfet's rating. Also when it popped, the mosfet was warm, but not hot to the touch.

The IRF640 is thresholded around 2-4v, but the 18A rating is at 10v so running at 5v shouldn't allow that much current anyway. I have already rewritten the drive circuit into a pull-up 10v, however I wanted to know what was inherently wrong with the current circuit before more mosfets die. Yes the circuit in the diagram is high by default, however I set it low before connecting the 12v +.

As far as flyback is concerned, that shouldn't affect the circuit given a gradual increase / decrease in drive signal, so I've only put a 60v 5A schottky on there.

Would anyone point out what I've missed / not accounted for?

 

Horrible gate drive circuit. Make the 1MegΩ resistor 4.7KΩ. Get rid of the 16KΩ resistor.

How is it heatsinked?

 

Are we decreasing the resistance to increase the switching speed? (What makes it so horrible?)

No heatsink on it at the moment, but it was barely warm to touch when it popped. Wouldn't it stay on if it fused?

 

Look at this:

Note the gate drive to M1. It cascades to U1 which is barely on and not switching properly. Note the power dissipation in U2 (yellow trace). How long do you think a non-heatsinked TO220 can take 28W?

 

Here is what it would take to make this work:


Note that to get sufficient gate drive to U1, I had to return the top of R2 to 12V. If you increase that to 36V later, you will have to limit the gate voltage by using a ~12V Zener off the 36V.

Note that the IRF640 has a really high Rds, so it is dissipating ~10W when trying to start the motor, even at 12V. It will be far worse at 36V. It should be on a huge heatsink!

ps: I'm looking at it again, and am wondering what R4 is doing there???

 

Well I was thinking of flipping the gate states anyway to something like this:

 

R4 keeps it from shorting to ground, but is the 100k specific? Just as well to use as little current as needed if it's just keeping it saturated

And as far as the heatsinking, I see your point. I'll mount it on something tomorrow first thing. If i later run the motor at 36v @ 15A (or parallel and bump the full 24A it's rated for), will passive cooling be enough? Sounds like the standard aluminum to220 1"x 0.59" x 0.4" sinks I ordered won't cut it alone.

 

How is your 2N7000 even turning on. With 3.3V on the gate it's barely over threshold voltage, so it's acting as a resistor rather than a switch.

 

You seem to not be clear about MOSFET operation.
MOSFETs have a large gate capacitance that must be rapidly charged and discharged for rapid switching, thus you need a low impedance source for that purpose.
If the gate is not rapidly switched on and off then the transistor will dissipate high switching power during the long switching process. For high currents/voltages this can cause enough instantaneous heating to zap the transistor, even if it's on a heat-sink.

Normally you do not put any resistance in series with the gate (or only a small value of a few tens of ohms to minimize any gate ringing). Any larger resistor value slows down the switching time.
I don't understand the purpose of most of the other resistors in your circuit but they are not needed (or even wanted).

Standard MOSFETs require a Vgs of 10V for full turn-on (look in the data sheet where they specify the ON resistance). This is not the Vgs threshold, which is the point where the transistor just starts to turn on (again look at how Vgs(th) is tested). Any voltage less can cause higher ON voltage and power dissipation.
Logic-level type MOSFETs can fully turn-on with a lower Vgs, typically 5V or 3V.

 

@shortbus yeah I see that, but it's acting to drop the volts across the gate of the IRF640 rather than saturating it. Given the datasheet, looks to me like the 2N7000 only generates a few ohms resistance at the given 3.3v. Even if it was higher, it would still drop the volts approaching 0 with the 1M there. Valid point though, of course I don't know how necessary that is to change?

@crutschow yes thanks, there's a completely useless branch in that circuit. Question though, what does gate ringing affect?

 

In most cases gate ringing probably does not cause a problem.
It's just something you don't like to see in a circuit, in case it does.

 

Horrible gate drive circuit. Make the 1MegΩ resistor 4.7KΩ. Get rid of the 16KΩ resistor.

?

That's pretty much what I was going to say - but I'd add that the TS probably needs a bit more than 5V to guarantee the MOSFET switches fully on when required to do so.

Without bothering to search for datasheets and doing the TS's maths - my gut says a single IRF640 isn't enough muscle, luckily MOSFETs parallel easily.

 

@MikeML was right on with the heatsink. Very happy, thanks guys!