Power cycle for restart?

[jon volk]

Does the board need a power cycle before restarting if an overcurrent limit is hit? I recently set up a function with the intent to auto start/restart the car over CAN once it receives a waitstart status message. I've confirmed functionality as it will kick everything on once I key on and press the brake.

However, if I hit an overcurrent trip, the status will go to waitstart, CAN sends start/brake over canio to restart, but there is no PWM generation unless I cycle power to the board. Looking at the wifi interface, it does receive the CAN message and transitions to a status of "none". There may be something I have overlooked, but I wanted to ask the question before chasing my tail.

[Isaac96]

What do the inputs look like after reaching waitstart? I seem to remember some sort of latching input, maybe with a NAND gate, that would prevent resetting without a power cycle.

[jon volk]

canio will show startbrake and then the status goes to none, which then clears canio. My VCU that drives the gauges will stick the tach around 3k during a shut down event and previously stay there until I cycled the key. It will now stick there for a second then drop back down to 0 showing that CAN traffic has resumed. There's just no PWM generation even after coming to a stop and cycling neutral back to forward to rule out any dirch rpm issues that may arise. All normal operation resumes after a power cycle.

[jon volk]

So, I had a realization while trying to fall asleep last night.

Thinking back to my first LDU inverter with a blown IGBT, a gate driver board fault would show up as an overcurrent error on the software. Im thinking there is probably some sort comparator circuit for overcurrent protection on each inverter phase driver. This would explain why the board responds, but I dont get anything out of the motor without a power cycle.

If that's the case, the positive is that there seems to remain a hardware overcurrent protection. I've probably hit this 100 times in testing so it seems to be an effective means of preventing inverter damage from a spicy tune up. The downside is Ill probably need to continue the occasional key cycle in my quest for maximum usable limits of this motor.

[Isaac96]

You might be hitting desat events -- apparently those can destroy IGBTs after a couple hundred.
Such is the price of high performance.

[jon volk]

I've got a pretty good handle on not making it do that. At this point Im chasing battery temperature compensation to reduce parameter aggressiveness at higher temperatures. From what Ive put this motor through lately, it owes me nothing.

[johu]

You could probably hack into the supply lines of the gate drivers and only run them when DCSW is enabled. That would clear the fault as soon as the DC switch drops out. Not sure if there are any drawbacks - may not be worth it, really

[jon volk]

I’ll probably abandon the effort unless I decide to throw a relay in the 12v feed to the inverter controlled with my VCU. I really only entertained the idea since it was supposed to be all over CAN with some keyboard bashing.