why interpolation in control loops?

[BobM3]

The manuals for the robot I just bought state that the controller for the AC servo motors interpolates the command. I can see doing this for coordinated motion between 2 axes but why would this need to be done if moving just one axis? Don't you just give it the command (pulses to get to) and the control loop looks at the feedback and generates an error signal?

 

[itsmoked]

I agree! Are you sure you don't mean interpret ? Or the the manual-writer meant interpret? Or they say interpolate because normally most moves are coordinated and so they just consider them all interpolated.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[sreid]

Interpolation in motion control means the calculation of positions between "Way Points." Usually there are far fewer Way Points than are required for every servo update time so some means of generating intermediate points is required.

 

[cswilson]

If you don't interpolate the command, you might as well hit the axis with a giant hammer to get it going. Without interpolation, you are giving the control loop a giant command step and then wait for the response to settle out.

For diagnostic purposes, you sometimes give an axis a small command step and plot/analyze the response. But the reason you do this is that the step contains a broad frequency content (i.e. a lot of high-frequency energy) so you can see how the physical system responds to a broad spectrum of frequency. But this is the last thing you want to do in the actual application, especially with large steps.

With a large command step, you are basically telling the servo to go instantly from zero torque to maximum. Aside from saturating the feedback loop, with all of the problems that this introduces (such as integrator windup), it is incredibly hard on the mechanics. Systems that do not control the third derivative of position (jerk) well tend to wear out their mechanics faster than those that do. Think of the implications of not controlling the first (velocity) and second (acceleration) derivatives.

Also, think about how the system would settle at the destination point if given a giant step, versus a smooth interpolated trajectory with limitations on at least two derivatives.

Curt Wilson
Delta Tau Data Systems

 

[BobM3]

Getting further along in the manual, I see that controlled acceleration profiles are offered for the beginning and end of each segment of motion. I guess I'm used to older analog hydraulic PID loops where you'd just take the gain down and set the rate up a bit to control the overshoot and ringing on step commands.