Killed a MOSFET, how to prevent killing another?

[lurkingdevil]

I am driving a coil with pwm from a microcontroller via P50NE1 N-Channel MOSFET. The coil has a resistance of about 4 ohms and driving voltage is 15V. I was using 1N4007 diode across the coil.

Last night I was fiddling with the code in the micocontroller for hours and the coil and mosfet were working fine. This morning I try to run the coil but the coil won't turn off.
Upon inspection, the mosfet is hot to touch and the mosfet will not turn off when I connect gate to gnd. Looks like its dead.

I connected the multimeter inline with the diode and it measured 0.8A current through the diode.

I have another of the same MOSFET and also a 1N5819 diode. How can I prevent killing the second mosfet?

 

[Reloadron]

If you are driving the MOSFET directly from the uC I would venture a guess your gate voltage is about 5 volts. The MOSFET you are using (VGS Gate-source Voltage ± 20 V) taken from the data sheet, is not a logic level MOSFET. Just based on the information you provided I would guess you are not driving the MOSFET into saturation, so it gets hot and cooked. I would suggest using a logic level MOSFET better suited for use with a uC. You also don't mention if you have a gate resistor between the MOSFET and uC output.

If you want to use the remaining MOSFET you might use the uC to turn on a transistor and have the transistor drive the MOSFET with a much higher gate voltage, like around 12 volts. The simple solution would be to use a logic level mosfet.

Those are my thoughts.

Ron

 

[lurkingdevil]

No resistor between uC and MOSFET. Yeah it was being driven by 5v. When I first tried the circuit it worked fine and the mosfet was cool so I thought it was ok.

It appears that the transistor would invert the signal. Using a pnp in common emitter would give positive to mosfet when I give negative to base. A npn would give negative to mosfet when I give positive to base.

Should I use common collector?

 

[Reloadron]

Personally I would start thinking logic level mosfet like the ones in the link. You will not turn on fully (drive into saturation) the mosfet you have with a 5 volt output from a uC (assuming the uC has a 5 volt output and adequate current. Since the circuit is PWM I am not surprised the mosfet worked but as I mentioned, the mosfet is not being driven into saturation. It will get hot and eventually self destruct. On a side note I would likely use a diode more like a 1N4002 across the motor.

Ron

 

[lurkingdevil]

Personally I would start thinking logic level mosfet like the ones in the link. You will not turn on fully (drive into saturation) the mosfet you have with a 5 volt output from a uC (assuming the uC has a 5 volt output and adequate current. Since the circuit is PWM I am not surprised the mosfet worked but as I mentioned, the mosfet is not being driven into saturation. It will get hot and eventually self destruct. On a side note I would likely use a diode more like a 1N4002 across the motor.

Ron

I am using a transistor to drive the second mosfet like you suggested. The transistor is connected to 15V. I would use a logic level mosfet if I had one, but I'd have to wait a day or two to get one. And my circuit isn't fully functional so I would like to continue right now and later I'll swap the mosfets.

1N4007 appears to be superior to 1N4002, why would you suggest that?

 

[alec_t]

I connected the multimeter inline with the diode and it measured 0.8A current through the diode

That suggests either the diode is dead or it's connected the wrong way round.

 

[Reloadron]

The 1N4000 series of diodes 1N4001 through 1N4007 all have an average rectified forward current of 1 amp. The main difference between them is their PRV (Peak Reverse Voltage). The diode I chose has a PRV of about 100 volts. The 1N4007 has a PRV of 1 KV. Your motor is running off 15 volts I believe. Actually thinking about this since you have a motor and not a solenoid coil out there I would likely place a .47 uF cap across the motor terminals and a .1 uF cap between each motor terminal and the motor case. When a device like a mosfet is driving a coil the diode acts as a snubber and goes into conduction when the solenoid coil collapses. If your motor is a brush type DC motor the issue is more noise than an inductive kick so there is a need to filter the noise, this is especially true when we have uC circuits out there. Thus capacitors are the choice for noise filtering. Since the circuit works I wouldn't worry about it. When a diode is used across a motor the popular choice is the use of a Schottky diode having a voltage rating about 4 times the supply voltage to the motor.

Yes, if you have transistors lying around then use a transistor and get the gate voltage of the mosfet up. Something else you may want to consider is placing about a 10K resistor between mosfet gate and ground to make sure that the mosfet does switch completely off. See how things work first.

<EDIT> I just saw alec's post and he brings up a good point I had not caught. Since it worked I had not given much thought to the diode but why are you seeing 800 mA through the diode when the diode should not be conducting? </EDIT>

Ron

 

[hkBattousai]

During the development process, don't connect any actuator to your circuit. Instead, connect a LED for low frequencies, and connect a resistor and observe the voltage on it using an oscillator for high frequencies. You can connect your actuator and do the final tests when you make sure that the system is working OK.

 

[lurkingdevil]

The 1N4000 series of diodes 1N4001 through 1N4007 all have an average rectified forward current of 1 amp. The main difference between them is their PRV (Peak Reverse Voltage). The diode I chose has a PRV of about 100 volts. The 1N4007 has a PRV of 1 KV. Your motor is running off 15 volts I believe. Actually thinking about this since you have a motor and not a solenoid coil out there I would likely place a .47 uF cap across the motor terminals and a .1 uF cap between each motor terminal and the motor case. When a device like a mosfet is driving a coil the diode acts as a snubber and goes into conduction when the solenoid coil collapses. If your motor is a brush type DC motor the issue is more noise than an inductive kick so there is a need to filter the noise, this is especially true when we have uC circuits out there. Thus capacitors are the choice for noise filtering. Since the circuit works I wouldn't worry about it. When a diode is used across a motor the popular choice is the use of a Schottky diode having a voltage rating about 4 times the supply voltage to the motor.

Yes, if you have transistors lying around then use a transistor and get the gate voltage of the mosfet up. Something else you may want to consider is placing about a 10K resistor between mosfet gate and ground to make sure that the mosfet does switch completely off. See how things work first.

<EDIT> I just saw alec's post and he brings up a good point I had not caught. Since it worked I had not given much thought to the diode but why are you seeing 800 mA through the diode when the diode should not be conducting? </EDIT>

Ron

Highier PRV = better, right?
I already have the 1N4007 diodes, so I'm using those, I guess thats okay? I also have a 1N5189 diode, would that be better?

I have a solenoid coil, not a motor.
I am using the diode as a snubber. Remember I'm operating with PWM, everytime the field collapses there is current in the diode. The multimeter basically averages out this current which in my case was 800mA.

Can you help me with the transistor circuit?
I can't seem to be able to make it without inverting my signal..
I have both npn and pnp transistors.

 

[Reloadron]

Beats me where I got "motor" from.

Try configuring the transistor (am guessing NPN switching transistor) in an Emitter follower configuration. Easier than using two transistors. Try about 10 K on the emitter. You will lose the forward voltage drop of the transistor but I doubt 700 mV will matter, you should have plenty to turn on the mosfet fully.

Ron

 

[lurkingdevil]

Is this what you mean :


That will give me less than 5V at the emitter..

 

[Reloadron]

That should give you 15 volts - about .7 volts (the forward drop of the transistor) when the transistor is fully turned on. Where did you get that sim from? Unless the 10 K is way too high. If that is a sim try 1 K then 470 ohms. The problem I think is the current I see is so low with 10K in there. This part sure as hell isn't my forte.

Ron

 

[lurkingdevil]

I can explain.
Consider the loop consisting of base and emitter. 5v(from uC) - Vbe = 4.3V across the resistor (4.44 in the sim).

The value of the resistor does not matter, the emitter voltage will follow the base voltage with a difference of 0.7V. Thats why its called emitter followr

I ended up using a PNP and NPN to get 15V at the MOSFET gate.

Do you know how I can eliminate AC noise/hum. I have a 15V power supply(laptop charger). Usually AC is not much of a problem but in this case, I have

signal --> NPN --> PNP --> MOSFET --> LED

Even if I touch the wire(with insulation) at the base of the NPN the LED lights up faintly.
However, I don't think this would be a problem once I connect it to the uC but if it can be fixed then better.

EDIT: In fact even if I move my hand closer to the wire connecting the base of npn (without actually touching it), the led gets brighter!

 

[Reloadron]

You could try a few filter caps in there off the power supply like around 100 uf and .1 uF to ground. Also, you could try placing some resistance to ground on the NPN like 10 K to make sure it is held off. Damn thing is sensitive.

Ron

 

[lurkingdevil]

A 0.1 uF cap from npn's base to gnd seems to do the job. A 220k resistor from npn base to gnd also seems to work well. Actually I'm using a 10k from uC to npn's base. If I put another 10k from base to gnd, that would divide the voltage in half, I had a 220k lying on the breadboard so I just put that in.

I think the cap might interfere with the pwm from uC, so resistor is a better idea?

 

[Reloadron]

Yeah, I don't think you want a cap in there with the PWM signal. This is getting to be quite the project.

Next time a different MOSFET!

Ron