1 ACCEPTED SOLUTION
Accepted Solutions
> After a bunch of testing I found that with the encoder enabled I could successfully do position commands with MCI_ExecPositionCommand(). It's weird the Motor Pilot doesn't work as expected for speed or torque control though. Any idea how I might be able to run the velocity and torque control with the encoder enabled?
Motor Pilot does actually nothing else than sending commands to the firmware. From what you describe I suspect a hardware issue around your encoder signals. It would be interesting to do a live plot with the Pilot (By clicking on the highlighted button) of STO-PLL angle and Encoder angle with STO-PLL as primary sensor as you did before but for a longer period and increasing the speed target, I suspect that at some speed your encoder signal is corrupted, and the encoder angle will be distorted. That could explain why for position control the angle is properly tracked.
Plugin an oscilloscope at encoder signals could also be very helpful if you can.
Regards
Cedric
I have not been able to get the Motor Pilot to work for plotting signals. I can get the Motor Profiler to work but but not the other part of the tool that connects to the motor and you can give commands. V1.2.9 of Motor Pilot.
I tried reversing the motor phases, and then reversing the encoder phases, which resulted in the same behavior as above: one connections of encoder wires makes the motor vibrate in place and the other connection can sometimes make the motor rotate for a bit.
But when the motor rotates it does so with inconsistent speed and the torque is very weak. 1/3 times the motor will make it a full rotation when using the encoder mode. Then where it stops I cannot get the motor to spin again until I move the motor manually by a few degrees, then the motor will spin again weakly as described above. All of this functional testing is by pressing the SW1 button on the dev kit.
Can you point me towards some variables to check? It looks like STO_PLL_M1 is the struct to monitor the PLL and ENCODER_M1 is the struct to monitor the encoder. Is there a way to switch the firmware from using the encoder to the PLL while the program is running?
Is there anything obvious in the encoder struct below that might need to be changed?
For the PLL struct I see the "is speed reliable" bit being flipped when the motor stops rotating
It's weird the motor will spin a little bit sometimes and weakly. I would expect no rotation and more consistent torque.
I was able to get the Motor Pilot working by loading the MC_FOC_SDK.qml file, instead of leaving the Pilot in default mode. Idk if this is in the documentation somewhere but it's confusing I need to do this step when the Motor Profiler will work ok on the next tab.
I recorded some data from encoder and FOC control. Here is a plot of the encoder angle vs foc when the firmware is using encoder for control. You can see how the FOC angle is pretty far off from the encoder angle. Also interestingly, the FOC angle (purple) keeps increasing after the motor stops spinning. In this mode, the motor sometimes spins one direction for a rotation, then the other direction, then after that it has a weak torque pushing in one direction which I can get the motor to spin a bit by hand by pushing it.
Start-up and first ramp:
Change of directions and second ramp:
End of run where FOC angle seems to roll-over repeatedly. The motor was not moving during this period:
Here is a screenshot of the encoder vs FOC when the firmware is generated with FOC control. In this mode the motor spins smoothly at a certain frequency.
Interestingly, the encoder seems to track FOC estimated angle closely for a few seconds, then it seems to stop updating properly:
Here is the FOC angle from that entire run, during which the motor was spinning smoothly at a certain velocity:
Let me know if there's any other tests I can run or example code using a quadrature encoder. When I debug the firmware in FOC mode I see the encoder values change and then eventually they stop changing, but the motor is still spinning in FOC mode.
Thank you for the tip on the Profiler, I found that out the hard way after struggling for 2 days and another ST employee on the forum told me this non-obvious step.
The plots are generated from saving Motor Pilot data and then plotting it with Plotly Dash python scripts. I couldn't figure out how to see the whole plots in Motor Pilot so I had to export the data.
If it helps I'm using CubeMX 6.12.1 and Firmware 1.6.0 for generating the project and get the same results. Could there be some gains I need to tune? Here's the working Motor Pilot with sensorless FOC control:
It seems like the encoder is kinda working: in sensorless mode it tracks position for a few seconds before stopping for some reason. And in sensored mode it can spin for a rotation or 2.
After a bunch of testing I found that with the encoder enabled I could successfully do position commands with MCI_ExecPositionCommand(). It's weird the Motor Pilot doesn't work as expected for speed or torque control though. Any idea how I might be able to run the velocity and torque control with the encoder enabled?
> After a bunch of testing I found that with the encoder enabled I could successfully do position commands with MCI_ExecPositionCommand(). It's weird the Motor Pilot doesn't work as expected for speed or torque control though. Any idea how I might be able to run the velocity and torque control with the encoder enabled?
Motor Pilot does actually nothing else than sending commands to the firmware. From what you describe I suspect a hardware issue around your encoder signals. It would be interesting to do a live plot with the Pilot (By clicking on the highlighted button) of STO-PLL angle and Encoder angle with STO-PLL as primary sensor as you did before but for a longer period and increasing the speed target, I suspect that at some speed your encoder signal is corrupted, and the encoder angle will be distorted. That could explain why for position control the angle is properly tracked.
Plugin an oscilloscope at encoder signals could also be very helpful if you can.
Regards
Cedric
