Sad story, from oDrive developer:
“ODrive does not support running both motor channels in parallel. The output FETs are PWM’ed and the channels are not synchronized. It’s likely that, even if both channels were “50%”, they would not be in phase. That will definitely blow up the output stages!”
I suppose it’s not trivial to put the output stages in phase…
"ODrive v4 is in development and it will be single axis at a similar power level to each ODrive v3 axis. It is feasible to do some firmware modifications to get v3 to run both axes in parallel, but it would be kind of crappy on our part to get this up and running for you only to have it be discontinued (I don’t really have a timeline for the v3->v4 switchover, but eventually v3 will be discontinued). In other words, it would be an unintentional bait-and-switch which is not something we are willing to do. Bad for us and bad for ODrive users. I have made a note with the team that a higher power level ODrive v4 is desired.
It is, however, feasible to run two motors in parallel mechanically, each one going to a different channel. This would be forwards compatible with the single axis ODrive v4. Torques on a shaft sum through the shaft, so there is no synchronization issue. ODrive typically requires an encoder, but there is a sensorless velocity control mode. I imagine this would be better for something in use underwater."
But wait, I’ll make the efforts to put the channels in phase…
"Our lead firmware dev estimates that it would take 2 days to make it work, so I’d estimate 3-5 days of work for a developer not familiar with ODrive.
On the hardware side, there are some other considerations. ODrive v3 supports a maximum modulation magnitude of 70% as a consequence of how we do current sensing (v3 only measures the currents of two of the motor phases. This is valid mathematically due to KCL, but also means that we require a certain amount of off time to measure both shunts in a single cycle). There’s also a limit on commutation speed of around 600Hz electrical due to the control loop rate and PWM frequency. For that motor, that works out to a 6000RPM limit from commutation ((600Hz / ) * 60) and a 4704RPM limit from modulation magnitude (56V * 0.7 * )
By my math, that works out to 6kW still being possible with both channels in parallel, but at a max speed of 4704RPM. It’ll be a bit slower due to battery voltage sag at high current draw."