steppers or servo drives

[mmjpotter]

Arvid

Different terminology, what you know as feed-forward I call look-ahead, it’s the same thing and I use it to induce an ERROR into the system. I agree we MUST have an error but this does not mean a loss of accuracy. In my implementation of the loop I have considered 2 types of error. The first is a POSITIONAL error. This can occur if we physically try to move the motor, the PID loop will respond automatically to pull the motor back to position. The second is the FOLLOWING error, this error is induced when the control requests a new position and again the PID loop responds by pulling the motor to that position. My system looks ahead along the move and is always 1 millisecond ahead of the motors, this produces a continual FOLLOWING error while the move is in progress. Any POSITIONAL errors during the move alter the error value to maintain accuracy.

Am I making myself clear? I find it easier to do it, than to explain what I’m doing. I must stress that this is only my implementation of a PID loop and it may not conform to text book designs, but it does work and it does not overshoot.

Regards
Mike

[arvidb]

Mike,

I'm not sure this is the same thing (but it might work equally well ).

Feed-forward by-passes the PID controller and introduces a move signal after the PID, while your look-ahead inserts an error signal before the PID, if I understood you correctly.

What (kind of machine) are you running with your controller?

Arvid

[mmjpotter]

Arvid

I agree my look-ahead function is not the same as feed-forward. If I understand correctly this is a predictive function based on the historical performance of the system. I suspect the benefit of such a function depends on the type of application. My controller is dedicated to the positioning and coordination of 3 axis mechanical machines. At this time I can’t see what would be achieved by introducing such a function, probably because I don’t fully understand the concept. At the moment I am trying to simplify my control algorithms, this would introduce a complication I can well do without. Perhaps when I’ve plotted a few graphs I’ll be in a better position to predict what is going on!!


Kind regards
Mike