PI controller of speedcontroller drive

[aand74]

What are the effects of the cycle time the PI control loop of the speedcontroller from a drive is executed?
If I go to faster cycle times with the same PI settings, what will be the behavior of the load? If I go to faster cycle times, is it advisible to change the PI-settings of the speed controller, and in what sense?

 

[Peter Nachtwey]

If the PI controller was written correctly there should be no effect of increasing the scan rate. The Ki gain should be multiplied by the scan time so as the scan time decreases the effect of the integrator gain is less for each scan but since there are more scans per second it works out the same.

This worry about scan time and the integrator is due to poorly written PID controllers.

 

[aand74]

I have done some tests on a drive, and indeed as it should be the scan rate does not influence the reaction of integrator.
But what could be said about the proportional part?
If I have shorter scan times (f.i. 500µs to 125µs), with same proportional settings, will the dynamic be the same? Do shorter scan times allow higher proportional values while overshoot at a step response is not worse then in the original setting with longer scan times?
All these questions relate to an upgrade of drives where the old drives had 400µs cycle of the speed controller and the new ones 125µs.
We know we should tune the new drives to the best possible values, but we should also want to have some comparison with the old drives (which are not present anymore, and of which we have no traces, only parameter settings).

 

[Peter Nachtwey]

But what could be said about the proportional part?

It doesn't change the proportional gain.

If I have shorter scan times (f.i. 500µs to 125µs), with same proportional settings, will the dynamic be the same?

The scan time should be much faster than the frequency of motion. Once the scan time is much faster than the frequency of motion then scanning faster doesn't do much. The frequency of motion can be found by doing a FFT on the motion profile. A simple way is to add the time it takes to accelerate and decelerate together and divide this by 20.

Do shorter scan times allow higher proportional values while overshoot at a step response is not worse then in the original setting with longer scan times?

Yes, but there is a point of diminishing returns and the difference in the P and I gains will be small.

All these questions relate to an upgrade of drives where the old drives had 400µs cycle of the speed controller and the new ones 125µs.

You should be able to copy the gains from one to another if the two drives use the same units for gains. 125 micro seconds is more than fast enough conveyor work.

We know we should tune the new drives to the best possible values, but we should also want to have some comparison with the old drives (which are not present anymore, and of which we have no traces, only parameter settings).

You are in luck. I just made a YouTube video about PI control for a velocity system. I show how to CALCULATE gains to get the response you want but this assumes you know the open loop gain and time constant or bandwidth.
https://www.youtube.com/watch?v=E6tH01GGxDg