Problem with IRF540N based brushed motor controller

I’m running out of ideas here. This motor controller really is a problem child, and I don’t understand why. I’ve got to be missing some key thing.

Schematic:
https://www.digikey.com/schemeit/project/brushedcontroller-O5RA4U04001G/

The motor works perfectly in the entire process of throttling up and down, but when a load is applied, well…

I’m a little low on the amount of mosfets, but that should cause it to heat up over time, not instantly explode in a glorious manner.

If you have any ideas, please respond. I’m about to tear my hair out here.

Thanks,
Jack

Hello @jack202020
Thank you for posting your question!
I will need to have this inquiry reviewed by some of our engineers.
They will take a look and advise.

Additional details on the physical implementation and the observed behaviors would be helpful (FET waveforms particularly) but two guesses offhand would include inadequate heat sinking or some sort of violation of max gate-source voltage. Planting a zener between gate and source to prevent the latter is good practice, as exceeding Vgs(max) has a way of producing effects like those pictured.

1 Like

Vgs of IRF540N is 20v, 12v is my Vgs.

Heatsinking I do not believe was the problem, there was a fan and heatsinks on all MOSFETs. The heat sinks weren’t even hot, just a sudden explosion. They got warm when running with no load, prior to explosion. I will add some gate diodes.

Thanks,
Jack

I just tried your suggestion of adding zener between gate and source. (1N5245BTR). The motor controller throttled up, suddenly went to full throttle, and blew the mosfets. I was monitoring gate driver during this time, was still outputting PWM, not a solid output. Here are images of the actual board:



Thanks,
Jack

The circuit in front of you is different from the one in the referenced schematic, and may differ also from the concept thereof that exists in your mind–it’s important that the three be consistent. Among other things:

  • There’s 5 FETs on the board, and 3 in the schematic.
  • The anode of what seems likely to be D1 in the schematic appears to connect to circuit common instead of the FET drains. This could lead to an over-voltage condition during FET turn-off due to winding inductance.
  • What’s presumably R1~R3 in your schematic appear to be 100 ohm resistances, rather than zero as described on the schematic. That’s likely excessive, and would contribute to excess switching losses.
  • What’s presumably the added zener appears to have been installed at a common connection of R1~R3 and driver output rather than across gate and source directly, apparently putting 100 ohms between the zener and the gate(s), which would tend to negate it’s potential benefit.

It’s common for FETs to fail short-circuit, to a marginally higher resistance than their normal fully-enhanced Rds(on). This could easily cause a connected motor in a circuit such as this to appear to turn full-on, followed shortly thereafter by the failed FET self-disassembling as a consequence of excess power dissipation due to the fault current. It seems that the only FET in the images showing physical damage is the one nearest to the apparent connection point of the motor return lead. This is, perhaps, not entirely coincidental.

Watching the Vgs and Vds waveforms will go far towards getting a concrete idea of what’s going on.

1 Like

Try above, otherwise may be a problem with reverse voltage spikes and/or D1 protection diode failing first causing Mosfets to fail. D1 diode is for reverse polarity protection but see below:

  1. D1 can only drop less than a volt so the excess voltage (sometimes peaks up to 60V +) would be shorted across the mosfets in a fast switching manner (which can explode mosfets).
  2. D1 may not be big enough, or high enough switching speeds, this diode should never generate heat. If D1 damaged and opened, the reverse voltage spikes will be sent to the mosfets. Check to see D1 is not failing before the mosfets.

There are many protection circuits, but to keep it simple you could add a series resistor with D1 to help absorb excess voltage (5-10 ohm?), this may slow the switching speed down however.

Last two are diodes.

I added the 100 in attempt to fix the issue, hoping it would drop any voltage transients.

The zener is incorrectly configured however, that is correct.

Vgs waveforms were nominal, I can attach some pictures if you wish. What you would normally expect. Vds was the same, until the explosion.

I also just found that the motor was incorrectly manufactured. Cheap Chinese motor surprisingly had one of the brush springs removed, so only three brushes worked. I don’t think that’s necessarily related but it could be.

Yes, it would be helpful if you could post those FET waveforms.

If they don’t show anything amok during the time frame captured, that’d be suggestive of something changing mid-process. Might be design/operating point related, or something more pedestrian like a rogue connection somewhere–the perf board and fly wire construction method lends itself to that sort of thing.

One way or another, the FETs are failing because one or more of their physical limitations are being exceeded–the trick is figuring out which one. If it’s not voltage related, it’s likely current/power related; instantaneous power dissipation during switching events can be intense enough to heat the die to the poof point before the package exterior warms significantly, if the switching business isn’t completed promptly. That could conceivably result from oscillations in the gate drive (with zero ohm gate resistors) in one case, and from excessively long gate charge times in another after switching to 100 ohms. Or a sketchy signal to the gate driver might cause a few rapid transitions that get averaged out to a Vgs of ‘half-on’ for a millisecond or so. Or it could be something else entirely… I can’t 'scope things myself from this side of the screen, so the more info and detail you can provide, the better.

1 Like

I came here while scouting around for ideas working with a brushed motor.

One thing that wasn’t clear from the original problem description is the amount of load placed on the motor. Driving an unloaded motor versus a loaded motor can easily differ in load current by 5x. A stalled motor could draw 50x the current of an unloaded motor. The amounts vary – the motor datasheet will provided the exact values.

The schematic shows 48V 1000W nominal for the motor. That’s 20A. If that is the spec’d running power, then the stall power could be closer to the neighborhood of 100A.

The IRF540N shown on the schematic is rated for 23A continuous.

It might also be instructive to read AN-7514 (formerly AN
9321) as referenced in IRF540N’s figure 6: https://www.onsemi.com/pub/collateral/an-7514.pdf
and also AN=7515 (formerly AN9322): https://www.onsemi.com/download/application-notes/pdf/an-7515.pdf

Apologies for reviving an old thread, but I figured this might be useful information for others…

1 Like

Welcome to Digikey Tech Forum, Thank you for your contribution