Delta Tau Forums
Trajectory Lowpass overshoots Endpoint - Printable Version

+- Delta Tau Forums (
+-- Forum: Delta Tau Data Systems (
+--- Forum: Power PMAC (
+--- Thread: Trajectory Lowpass overshoots Endpoint (/showthread.php?tid=2862)

Trajectory Lowpass overshoots Endpoint - JeffLowe - 03-27-2019

I setup a 2 pole 100 Hz lowpass on a simple point to point trap profile move. The filter had the intended effect of reducing the acceleration profile from steps to something with less frequency content, but when looking at the commanded position I note the endpoint commanded position overshoots before returning to the desired endpoint. IS this normal?

RE: Trajectory Lowpass overshoots Endpoint - curtwilson - 03-27-2019

If you choose the simplest low-pass filter specification in the IDE (click on low-pass, specify a cutoff frequency, click on "Auto-Calculate"), it will produce a 2nd-order Butterworth filter. The Butterworth filter has many desirable characteristics (e.g. very flat frequency response below the cutoff), but it does have a slight overshoot in its step response.

Mathematically, a "continuous" digital trajectory is simply a concatenation of many tiny individual steps. Because the individual steps are tiny, the resulting overshoot is tiny. Most people don't care. But if you have a level of precision that means you care about this, you will want to modify your filter.

The key parameter is the damping ratio of the filter. For a Butterworth, it is 0.707 (shown on the setup screen). If the damping ratio is less than 1.0, the step response will have overshoot -- the lower the damping ratio, the greater the overshoot. You may want to increase the damping ratio if you are bothered by the overshoot.