Pulse Width Modulation (MOSFET Question)

[tarmon.gaidon]

Hello Everyone,

I am working on a system for an undergraduate project that uses pulse width modulation to power two electromagnetic clutches. I have a pic microcontroller that is supplying a PWM signal to a MOSFET that is in turn switching the ground side of my clutches.

Everything seems to be finally working great but my MOSFETs get extremely hot quickly. I am pretty sure if I gave them more than a minute they would be burned out.

So my first idea was that I was switching to fast so I tried to reduce the speed of my PWM and nothing changed.

Could the inductive properties of the clutches be causing me trouble? Is there anything else I could look at?

The circuit is attached.

Thanks,
Rob

EDIT: Add resistors in circuit.

 

[marcusl]

You'll need to supply more information to allow a diagnosis. How much current do the clutches draw? If it's well below the MOSFET I_D rating, and you are using a proper heat sink with thermal grease, then suspect the drive. Big MOSPower FETs have a lot of gate capacitance ("Miller capacitance") that require a stout driver. If you are using TTL drive, for instance, the current drive available from them is small so the gate will charge/discharge slowly. This puts the FET into a resistive state for a considerable time while transitioning between the fully off and on states (and to some extent vice versa), thereby dissipating considerable power.

Look into a FET driver chip like those from TI, Intersil or National.

 

[tarmon.gaidon]

The current is way below what the FET should be capable of. I believe each clutch should pull no more than 2 Amps. On the other hand I do not have any heat sink on the MOSFETs, is that something that will definitely be needed?

The weirdest thing about this is that I had this working once before. It was using a slightly different microcontroller but basically the same set up and it worked until something shorted elsewhere and fried the FETs. I then remade the whole set up using a different pic microcontroller and all the sudden the FETs are overheating. This makes me think that maybe it has something to do with the pic not supplying enough to drive the FET?

 

[Averagesupernova]

I would guess too slow of a transition between high and low on the gate.

 

[tarmon.gaidon]

I would guess too slow of a transition between high and low on the gate.

I am going to get a scope on it tonight. I was reading up on some motor control stuff earlier and it sounded like I might need a diode in reverse parallel to my clutch and one from ground to my gate due to the inductive nature of the motor/clutch. Any thoughts on that?

Would it be worth my time to try it with say an LED first to see if it works and if that works I can say it has something to do with the clutches?

 

[schip666!]

I'd place a small wager on needing a reverse-bias diode across the inductive load. You can try using an equivalent resistor (pay attention to wattage as well as resistance) to see if your drivers still heat up.

 

[mheslep]

I would guess too slow of a transition between high and low on the gate.

2nd, given "MOSFETs get extremely hot quickly"

 

[Topher925]

2nd, given "MOSFETs get extremely hot quickly"

I would like to 3rd that.

How is the PIC connected to the mosfet gates?

 

[tarmon.gaidon]

I would like to 3rd that.

How is the PIC connected to the mosfet gates?

Just as it is shown in the schematic I attached to my first post.

 

[marcusl]

Add a FET driver. They are available in quad packs so one chip will drive both of your FETs.

 

[tarmon.gaidon]

What purpose does a FET driver preform?

 

[marcusl]

It is designed to source / sink a huge current (for a short time) to rapidly charge the MOSFET gate. This turns your FET on / off in nanoseconds. Use good layout practice and solid low-inductance grounds to avoid ringing. Your FET's will run cool.

 

[vk6kro]

You said:

On the other hand I do not have any heat sink on the MOSFETs, is that something that will definitely be needed?

Absolutely, you should have heatsinks.

A transistor without a heatsink may be able to handle 5 watts of heat before it gets too hot to touch. If it was on a heatsink that was suitable, it may handle 100 watts.

 

[Topher925]

Just as it is shown in the schematic I attached to my first post.

Well, obviously, I meant how do you have it connect physically. Are you doing this on a breadboard, PCB, deadbugged, what? Also, what's the model of the MOSFET you are using? Whats Vgs values to turn the thing on?

Generally, driving a MOSFET strait from the logic of a uC is a bad idea. If you don't want a full blown MOSFET driver you can use a separate signal FET or BJT that's connected to your positive power-in source and the gate of your MOSFET and then drive the FET or BJT with your PIC. That will give you the greatest Vgs possible along with more current and therefor less on resistance.

 

[p.Usha]

I am designing a buck regulator using mosfet,UC3845 output pulses are b/w +10v and -10v what amy be the problem? I need pulses b/w 0 and +10v . Can anybody help me?

 

[LotusEffect]

Use a diode to clip off negative voltages.

 

[Xarren]

I didn't read through all of this, but you're missing diodes reverse biased in parallel with the inductive loads. That is because in simple terms, inductive loads do not like being switched off quickly, and will produce a negative voltage spike. This should sort your problem out.