Hi, apologies if I’m missing this but I don’t seem to understand the starting condition for the manual tuning procedure.
When Step 1 in the tuning procedure says: " Set vel_integrator_gain gain to 0" and * Make sure you have a stable system. If it is not, decrease all gains until you have one." It does not explicitly state which system state we should be in during tuning.
What axis state, command mode and/or pos/vel setpoint should we be operating while conducting the tuning. Should the motors be installed in the target application or standalone on the bench? If tuning should be done in the application, for example I have a two wheeled differential drive robot, should tuning be done with the robot on the target terrain?
Thanks for all your great work and any guidance with this issue.
Ideally, you should start in Current control mode, using current_setpoint step commands, and looking at Iq_measured vs your current_setpoint, to make sure it’s clean and stable. Then once it looks good, go into velocity mode, tune the P gain, then tune the I gain until you have nice velocity steps. Then do position control P gain.
1 Like
Thanks for this method, I’m finishing up testing various tuning procedures specific to velocity control and I’m finding the velocity integral gain tuning values are significantly different for the desired torque response vs position control mode. I want to post findings here once complete but I’m fairly convinced that tuning procedure should be done A) in the application robot and B) in the target control mode to be effective and C) Then tested against a disturbance if possible, but given those three variables tuning will differ, at least in velocity control mode. Does that square with your understanding?