[Ether]

@BornaE:

The BackEMF term seems to react in the wrong direction to disturbances or changes in setpoint.

For example, if a change in load causes the motor speed to decrease, this causes the BackEMF term to decrease as well, instead of holding fast (as might be desired).

Or suppose there is a step decrease in setpoint. Instead of immediately decreasing (as might be desired), the BackEMF term will only decrease as the PID overcomes it and slows down the motor.

Also, at steady state, the BackEMF term provides only the voltage necessary to hold speed for an unloaded system. So the PID must contain an integrator to make up the difference.

Why not just replace the BackEMF term with the expected voltage* necessary to achieve the desired speed ?

Read the links provided in post #4. They provide further detail.


* As might be the case with a spinning wheel shooter for example where the load is predictable and the expected volts vs speed curve can be known. You can determine this voltage empirically by just reading the PWM from the dashboard.