tom91 wrote: ↑Thu Jan 08, 2026 6:43 pm
Is all the 12V to the inverter via that relay?
If the issues remains like it does need to confirm what inverter Greg has in his rabbit and what firmware was used as a base for testing on it (yes I did write it but its a while ago).
Yes all the 12 volts is via the relay to the inverter.
The two inverters
Greg thinks that it was either version 2.17 or 2.19 with a in between custom release that you sorted out for him.
I sent 2.20A logs and parameters to perplexity.ai to see what advice could be had.
This is what I got…
Is any of it worth trying/following up on?
With a Prius Gen3 inverter and BMW safety box, the fact it ran on 2.15A but not on 2.20A almost certainly points to a configuration / input‑mapping change rather than a hardware fault.
What is already working
The logs show opmode reaches 1, DC bus is ~362 V, and the inverter status is on, so the Prius inverter is actually being started by the VCU.
potnom reaches ~100% and dir flips correctly, so the throttle pedal and direction inputs are being read and scaled.
Motor speed changes slightly, so the inverter is producing some torque, just nowhere near enough to move the car.
This means the basic Prius inverter wiring, BMW safety box, and HV contactor control are OK.
Prius‑specific trouble spots with 2.20A
For a Prius + BMW safety box setup, these are the areas most likely to have changed between 2.15A and 2.20A and silently kill torque:
Drive‑enable / HV request input
If you use a 12 V “drive enable” (from the BMW box or ignition) into one of the generic inputs (e.g. HV request, park/drive, brake), confirm that:
The input is still mapped to the same function (HVReq, DriveEnable, etc.).
It actually shows the expected state in the web UI while your foot is on the pedal.
2.20A has slightly stricter logic in some builds: if that input is not in the “drive” state, it will let the inverter run but clamp torque to almost zero.
BMS / shunt configuration and limits
With a BMW safety box, the pack voltage/current may come either from a shunt, BMS over CAN, or be “fixed”. Check that:
ShuntType and BMS_Mode match what you physically have (for a simple BMW box without CAN current, often this wants to be “no shunt / simple BMS”).
CCS_ILim, CCS_SOCLim, SOCFC, BMS_VminLimit/BMS_VmaxLimit and BMS_TminLimit/BMS_TmaxLimit are reasonable so the VCU does not think the pack is out of range.
If 2.20A now expects current or SOC from a different source than 2.15A, it can decide “no valid current / SOC” and reduce traction torque.
Prius “gear” / direction handling
On the Prius connections, there is usually one “direction” line into the inverter and an optional “gear” input or logic inside the VCU.
Check that your dirmode and Gear / GearLvr settings are still consistent with your wiring (for example: forward when your BMW shifter is in D, reverse in R).
If the firmware believes it is in neutral or in a disabled direction state, you will again see potnom move but almost no torque.
Nope none of this has impact the code reports a torque so that will be sent to the prius inverter.
The issue possibly is in the start up method for the prius Gen 3 inverter. I would not expect this to be released without testing though so no idea why its an issue for you and not others. I believe my changes that Greg tested stayed into V2.20 and beyond.
tom91 wrote: ↑Fri Jan 09, 2026 8:14 pm
Nope none of this has impact the code reports a torque so that will be sent to the prius inverter.
The issue possibly is in the start up method for the prius Gen 3 inverter. I would not expect this to be released without testing though so no idea why its an issue for you and not others. I believe my changes that Greg tested stayed into V2.20 and beyond.
Thanks, Greg suggested that I check if my HSDN wire from the inverter is connected to, he thinks it goes to the Park signal from the transmission safety neutral.
Okay let me know how you get on. If it drives as it should I will revert the Prius specific changes in a test branch of the V2.30A release for you to test.
Went for a short drive just using MG2 “doing a Greg” and the acceleration seemed similar, less regen didn’t have enough available road to see how fast it would go.