MCSDK 6.3.1, FOC with STOPLL, EVSPIN32G4, this motor.
We're using a custom board based on EVSPIN32G4 as a drone ESC in a traditional sensorless configuration. Naturally, in our application the motor starts up to a low RPM under idle load and then after FOC takes over it spins up and down within the whole range as commanded by the autopilot. The MCSDK is in the speed control mode.
We have found an interesting fault condition (also recreatable with the original EVSPIN32G4): if the motor is held in place somewhat elastically, e. g. by hand, during startup, the following things might happen:
- Speed feedback fault flag is set ~45% of the time -- expected
- Startup fault flag is set ~45% of the time -- expected
- No fault flag is set, the motor is not spinning (sometimes making noises), the indicated RPM jumps to very high negative numbers, followed by speed feedback fault after a few seconds -- unexpected, but easy to catch and handle, RPM flips are easy to see.
- No fault flag is set, the motor is not spinning (sometimes making noises), the current stays very low, the indicated RPM actually follows the setpoint (no changes to the motor behavior, not spinning at all), until the motor is commanded to stop, then it continues with scenario (3).
Naturally (4) produces a disaster because not only there's no fault, but even the RPM follows the setpoint, so there is no way to detect that but to look at the current, which is not a reliable indicator because it's only apparent at higher RPM.
Feels like the observer converges onto some sort of a false equvilibrium in which the stator field is rotating without affecting the motor at all. I guess this might be unavoidable without external sensors but it's a bit too easy to induce in our case.
Everything runs fine and as expected otherwise.
Perhaps this is a known condition and there're known strategies to deal with it?
