Killed fan MOSFET on a Marlin/SKR v1.3 rig

TL;DR - If you’re driving a diode laser by a fan control N-ch MOSFET like on a common 3d printer setup, it may fail in the closed position and keep your laser 100% on when the job ends.

More people these days are reporting failed MOSFETs on SKR v1.3 boards from BIQU/BTT, now me too. I’m posting this here because I see a lot of people running lasers on FDM/FFF machines lately and they’re using very similar configurations to mine with Marlin on a BTT SKR v1.3 using the part cooling fan PWM output for driving the laser by chopping the input power to the laser driver board. Now I feel like we shouldn’t be doing this, we should be driving the PWM line on the external driver with this signal, but the sellers of these things swear you can just run off the fan line, and I see plenty of people doing it, so I am too because this board doesn’t have squat for extra pins if you’ve got it Gucci’d out for printing.

My PWM control went away suddenly, and after a short test burn the fan output MOSFET on my SKR v1.3 board failed into a closed circuit and will not turn off. It is on even before the OS loads, as soon as power is applied. This seems to indicate an overheat melting the low temperature solder in the chip which will short the source and drain. I noticed the failure when the very light gradient I was using to calibrate (burned as a solid rectangle, but light colored which was confusing me) became a 3mm deep trench all of the sudden when my laser power was only set to 10%, and when I hit stop, the steppers quit but the laser stayed on and punched through the plywood and into the aluminum surface of my print bed. I sent a gcode command to kill it, didn’t work, hit the soft e-stop but it stayed on, unplugged the laser driver, reset, output was still on. Removed mains power, reapplied it, still on… A dangerous failure mode if you’re not ready for it.

If you have a similar problem, the method to troubleshoot is to remove power and the load connected to the FET and check for continuity or very low resistance from source to the drain, you shouldn’t see any. If it failed closed, you will. Also check your CPU output under power by turning that output off and on, and probing from ground to the gate. You should see the voltage change normally. If the voltage on the gate isn’t changing, your CPU may be damaged. I’ll replace the FET and investigate further on an external driver before proceeding.

Anyways, I think it serves as an important lesson about why you don’t ever operate these machines unattended. I think very high frequency PWM with such a load did it? I’ve got Marlin set with FAST_PWM_FAN enabled to try to get high frequency output but it looks like the junction temperature may have risen and reflowed the solder in the chip making it fail into a closed state… I’m thinking I should reduce the FAST_PWM_FAN frequency to something lower… According to the Marlin documentation, a 100MHz CPU should default to run at 100MHz / 510 = 192kHz? and I don’t see anywhere that is being scaled either, so that’s a bit high eh? Watch out for your MOSFETs failing closed.

Has anyone else experienced something similar? If so, how did you resolve it?

Wow, Scary stuff. Thanks for the heads-up.

If the voltage drop across the FET (when on) is low enough, you could put two in series so the one that hasn’t failed yet can still open the circuit. A couple of LEDs to prove that both FETs open would inform you about their health without much additional cost.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.