Integral winding up in start-stop-repeat servo application

  • Thread starter Miles Koolmeister
  • Start date

Thread Starter

Miles Koolmeister

We are tuning a velocity driven position PID loop and are observing a strange problem.

Cycle parameters:
Trapezoidal input
Motor is rotated 75° in 15 msec
Dwell (under load) 10 msec
Return 75° in 15 msec
Dwell (no load) 35 msec
Repeat (repeats up to 20 times).

Observation (problem):
The "under load" condition on the first cycle is only capable of sustaining ~50-75% of the load of any of the other subsequent cycles. If the "no load" dwell time is increased significantly (several seconds), the "under load" condition of the first cycle is repeated (ie only ~50-75% capacity) for all subsequent cycles.

In viewing plots of the commanded position vs actual position, it is clear that during the 35 msec "no load" condition, the error reduces to a zero value.

It was stated that the reason the first cycle load capability is significantly lower than the rest is due to integral windup. Since the system has zero error at the start of each cycle (every 35 msec and/or several second), how could this be integral windup?

Please let me know if anymore information is needed.

Thanks for helping,