TSDU - Strange intermittent OVERCURRENT fault [SOLVED]

[cloudy]

Damien board V4 running on a 55V 60A PSU for HV. Tried some 4.x firmwares and also just tried 5.0 with no difference

I seem to have an intermittent fault with overcurrent being thrown as soon as you try to enter Run mode. Sometimes it's fine and drive unit works fine, other times it instantly kicks out when going from precharge > run.
I've tried inverting the sign of ocurlim (this one is especially weird as I feel i've had it working with both a negative and positive sign without changing firmware level)

I think I'll have to split the inverter, thinking it must be a dry contact - I can't see any way to explain why sometimes the unit sometimes works fine. Is the only way to trigger the OVERCURRENT purely from the hardware overcurrent circuit? This is a current limited supply, so in theory there is no way I can be hitting 1200A.

EDIT: I've switched over to using a bench supply for the HV - anything up to 27V seems to be fine and the unit reliably gets into run mode and works fine. Not seeing any current spikes at all going from precharge > run. anything above 30V seems to intermittently hit ocurlim (but again there is no current spike) - It has to be a board issue I guess, if it spins OK on less presumably the inverter and power stage is fine

[johu]

OVERCURRENT is triggered by 3 items, unfortunately:
- The comparator that compares the current sensor signals to ocurlim
- The 3 fault signals from the gate drivers
- UVLO

One of these may be floating or have a bad solder joint

[cloudy]

Thanks - that will help troubleshoot - agree, it does sound like something is floating...

[cloudy]

Also, just out of interest - is it possible to go from waitstart > run with active overcurrent condition? ie: are inputs evaluated during waitstart, or only once in run?

[johu]

Not all shutdown signals are evaluated, just emcystop and mprot which are always high on Tesla hardware. So yes you can go from Off to run with active overcurrent flags.

[cloudy]

For anyone trying to fix similar - I found a tombstoned capacitor (C50) on the HVIL line into the nand gate - resoldering that appears to have fixed the issue. I guess HVIL was triggering - which it seems would also cause an OVERCURRENT error on the tesla board.

[johu]

Yep, sounds like a root cause, well spotted!
Did you buy that from me? I'm trying to spot these upon arrival. Meh.

[cloudy]

Not one of yours! No need to panic