Mosfet bridge safety

[steelgoofy]

Hy,

I'm using a Mosfet bridge to drve a BLDC motor, and I want to add something to it to keep the mosfets safe. Now my circuit looks like this on the schematic. I'm using a heatsink on each mosfet, so it doen't get hot, but it seems that it is not enough. Today I was driving the motor, everything was fine, the mosfets were cool, then suddenly my motor starts shaking, I switched off the power supply. It turned out, that one of the mosfets go wrong (short), but it wasn't even warm. What could be a basic safety function ? (My program allows only the two right mosfets switched on the same time.)

 

[BobK]

One form of protection is a zener diode from source to gate to prevent too high a voltage on the gate. It should have a breakdown voltage a little above the drive voltage.

Also, are the body diodes of you MOSFETs rated to carry the max current you are sending to the motor coils? If not, you should put an external diode in parallel with them. This is because the current will briefly flow through those diodes when you switch the winding off.

Bob

 

[KrisBlueNZ]

What about the other characteristics of the body diode, Bob? (Or anyone?) Are they fast enough? Is it just their current (and voltage, obviously) ratings that are important?

 

[Arouse1973]

What about the other characteristics of the body diode, Bob? (Or anyone?) Are they fast enough? Is it just their current (and voltage, obviously) ratings that are important?

In most standard MOSFETS the body diode is a poor one. It was not intentional to have this diode but happens to be there as an intrinsic component. They tend to be slow and have a long reverse recovery time.

However there are new fast recovery MOSFETS available which have better diode characteristics and are used in resonant convertors simplifying the design a bit by not having separate clamp diodes across each FET.

As far as the current of the diode, some data sheets refer to this as the source drain diode current (Is) and in a couple I have used it's the same as the max drain current.

Adam

 

[BobK]

Yes, I should have mentioned speed as well. I am currently using a MOSFET bridge for motor control and I do not have external diodes, but the ones in my MOSFETs are rated at 1.6A and the motor current is 200mA. I do not see any voltage spikes.

Bob

 

[(*steve*)]

What about the other characteristics of the body diode, Bob? (Or anyone?) Are they fast enough? Is it just their current (and voltage, obviously) ratings that are important?

To expand on Adam's comment. There certainly are potential pitfalls. There are some mosfets that have an additional diode integrated in the package. I seem to recall a Schottky diode in some device.

It's always worth checking the performance of the body diode. Some can be quite slow and have poor recovery characteristics (in which case you should parallel another diode so this one never gets forward biased to the point of significant conduction) while others can be quite high performance.

There's some good information here, here, here, and here. (and here).

I'm not sure how much the OP needs to understand this. Parallel diodes (or Schottky diodes) with a low inductance path will go a long way, and ensuring there is sufficient dead time will prevent shoot-through. The use of a gate driver designed for half-bridges will help by enforcing the dead time

 

[(*steve*)]

(My program allows only the two right mosfets switched on the same time.)

One issue may be dead time.

It takes time for a mosfet to turn on and off. If you try to switch both mosfets simultaneously then this may result in a brief time when both are turned on. This result in shoot-through, where you essentially short the supply briefly with the mosfets.

This current may be very high, but the duration very low, so the average power may be low. But the mosfets may still be damaged.

Try changing the program so that you turn the mosfets off before you turn other mosfets on. A slight delay (perhaps only a microsecond) may make the difference.

Also important is the amount of gate current you can supply as this controls the switching speed. Generally if you switch too slowly the mosfets heat up.

 

[BobK]

All this is great info in general about using MOSFET bridges.

But the OP stated that the MOSFET failed and was not even warm. So I think shoot through or overcurrent are probably not the problem. More likely the gate destroyed by a voltage spike.

Bob

 

[(*steve*)]

If one is shorted and none of the mosfets is getting hot, then that's pretty interesting.

Perhaps knowing the part number of the mosfets, the driver IC, and which mosfet failed, and (importantly) how it failed.

Was the failed mosfet shorted between drain and source? Make sure you connect the gate to the source while you do this measurement!

 

[steelgoofy]

Thanks for the answers. I put a 1 us switch off delay to my program, but I find out, that the problem was that, the inductive voltage spikes from the motor, also the voltage generated by the motor superimposed on the supply voltage and dameged the driver logic, that caused wrong switching, so that the mosfets damaged. I’ve replaced the wrong parts and tried to fix the problem, by adding a low pass RC filter (filter the spikes) and a zener diode to stabilize the voltage for the logic. Now the logic is slightly protected, but the motor doesn’t work as before. No rotation as before, just turning back and forward. (I added a schematic)

 

[KrisBlueNZ]

Are you driving the gates of all six N-chanel MOSFETs from outputs of the LMC6484 that's powered from 11V? If so, that won't work; the high-side MOSFETs need a gate voltage quite a bit higher than the positive supply rail because when they conduct, they pull their sources up to +12V and you need gate-source voltage (around 10V is recommended) to make them conduct. So the gate voltage should be about 22V to turn a high-side N-channel MOSFET ON.

There's also no need to use op-amps to drive the gates. There are proper high-side gate driver ICs available. Some of these, like the IR2110, use a "bootstrap capacitor" from the high-side MOSFET's source, to provide the supply for the gate voltage; these are only suitable if you are switching the MOSFETs regularly. If you need steady state control of the outputs, you need a higher voltage rail, or a high-side MOSFET gate driver that has an oscillator and charge pump built in, to generate that rail.

 

[steelgoofy]

I use the op amps (3 of them) to drive only the high side of the mosfets, so I can apply a voltage level from 0 to 10 volts to the high side, and use only the uC level PWM to drive the low side of the gates form (0 to 3,4 V).

 

[(*steve*)]

Do we know what the midgets ate yet? Without some sort of voltage boost for the gate drive you need N and P channel devices. You shore only a single type of device.

edit:

Do we know what the midgets ate yet?

What was I drinking?

Actually I think it was a stack of auto-corrects from my phone (with the swipe style keyboard in use)

 

[KrisBlueNZ]

I use the op amps (3 of them) to drive only the high side of the mosfets, so I can apply a voltage level from 0 to 10 volts to the high side, and use only the uC level PWM to drive the low side of the gates form (0 to 3,4 V).

That won't work. Read my post #11 until you understand it.

 

[BobK]

Unless he actually has P channel MOSFETS for the high side.

In any case, an opamp is still a silly way to drive them.

Bob

 

[KrisBlueNZ]

Unless he actually has P channel MOSFETS for the high side.

Not according to the schematic.

In any case, an opamp is still a silly way to drive them.

And driving the low-side ones directly from a micro, especially at only 3.4V, is also silly.

steelgoofy, look for high-side and low-side MOSFET driver ICs. Digi-Key (http://www.digikey.com) has a wide selection.

 

[BobK]

Not according to the schematic.

The symbol in the schematic is missing the arrows that tell you whether they are N or P.
Bob

 

[KrisBlueNZ]

The sources both point downwards.

 

[(*steve*)]

It's sad that we need to do forensic analysis of the schematic when the process could be made far easier by the OP telling us what the actual devices are.

Given the conclusions we are reaching from our analysis, I find the assertion that the mists were not hot to be... astonishing.

 

[KrisBlueNZ]

Given the conclusions we are reaching from our analysis, I find the assertion that the mists were not hot to be... astonishing.

Auto-correct acting up again? Midgets in the mist?