openinverter forum

As noted in the firmware rev 5.20r+ be very careful when revving the motor without load or other safeties in place.

I just grenaded my leaf motor/inverter and possibly also the coupling when revving in neutral gear to double check an imbalance I thought i sensed when test driving. I was used to be able to do this with the older firmware so I never thought twice about it even though I had read the warnings. Somewhere past a few thousand RPM the motor sped up on its own, over-revved in an instant and made a horrible sound while doing it. I slammed my "kill" button just as the motor ground to a halt by itself. There was more or less no time to react.

Now the inverter board can not be contacted via wifi any more. Status for the rest of the motor is unkown but it sounded BAD.

Sad day, be careful out there!

Sucks to hear this
The motor is definitely toast. Inverter logic should be fine, not sure about the power stage though.

This has to stop, there must be some rev limiter that always works. Like ramping down field weakening current

Johannes, could something be done like acceleration limiter beyond 12K RPM? I am comfortably running EM61 motor at 12K so there is some leeway here.
I think this could be done totaly out of the loop as additional observer of resolver speed. Its flag would simply trip the 0Nm torque or provide some negative torque as long as the flag is on? Its jolt may not be comfortable but at least motor would not go beyond 15K???

Or would simply using NEUTRAL state on the flag reference be enough?

I've had a couple uncommanded runaways myself, though thankfully I was able to shut things off before I broke anything. So far the only casualty has been the fuse to my DCDC somehow.

What is the mechanism for the runaway? Is it that resolver alignment isn't perfect, so flux weakening d current can turn into q current spinning the motor faster which feeds back into requiring more flux weakening current since you're going faster and repeat?

Maybe could add a fault that trips on a second higher frequency limit than fmax, that when exceeded, all 3 low side IGBTs are turned on to force stop the motor by shorting it. This way you don't overvolt the IGBTs/capacitor like you would if you would just stop switching. And since these runaways can only really happen with a free spinning motor, not a whole lot of energy to dissipate since you are just trying to slow down the rotor instead of a whole vehicle.

I'll run a second rev limiter at 1.1*fmax that derates fwcurmax. This should end in an equilibrium just like the current fmax limiter.

I have now added fwcurmax derating at 110% * fmax
This would need testing with low voltage and a low value for fmax. Anyone volunteer? I don't have the setup right now

https://github.com/jsphuebner/stm32-sin ... 9aca6a2a37

ah that sucks.
did you happen to log this 'destruction'?
I think it would be very interesting to see RPM's - and some real world data on maximum rpm's?

johu wrote: ↑Tue Aug 15, 2023 9:04 am I have now added fwcurmax derating at 110% * fmax
This would need testing with low voltage and a low value for fmax. Anyone volunteer? I don't have the setup right now

I am trying to get hold of a new motor + inverter right now. I have my old 30V test rig complete with relays and a pedal ready to go when I get the new hardware. I am probably going to buy a new gen2 rev3 leaf board as well since the old one doesn't show up on wifi anymore

SuperV8 wrote: ↑Tue Aug 15, 2023 9:27 am ah that sucks.
did you happen to log this 'destruction'?
I think it would be very interesting to see RPM's - and some real world data on maximum rpm's?

Sadly no logs. I had just gotten my DC-DC up and running for the first time, on the first try. Patted myself on the back and thought "hey maybe I should check that potential driveline vibration while I'm at it". Computer was still connected to my EVCU and was not logging the inverter at the time.

Interested? viewtopic.php?t=2769
Will send it for shipping

Thank you for the offer. I am in the process of getting a gen3 160 kW stack with zombieverter instead. Matches well with my other equipment (i3 lim, vw heater, chevy heater, ISA, etc...)

I've now tested above software myself and found it caused Overcurrent. Because I forgot to divide the percentage value by 100. So allowed 100*fwcurmax. Doh.

I'm about to draft a new release that also contains the hardened CAN implementation.

I didn't dare actually testing the overrev protection at full pack voltage, will leave that to somebody with a test bench.

johu wrote: ↑Tue Aug 15, 2023 9:04 am Anyone volunteer? I don't have the setup right now

https://github.com/jsphuebner/stm32-sin ... 9aca6a2a37

Tomorrow, the motor will be driven at low voltage to high speeds due to weakening