Tom,
I have the X,Y, and quill working pretty well now, but I came across a very weird problem.
As you can see, the motor is mounted to the end of the table (x axis). The motor drives the ball screw through a toothed belt. The encoder is driven directly by the ball screw.
The system is stable enough when the table is in the position shown, and when it moves almost all the way to the right. However, within 4 inches of the end of travel, the control output starts oscillating at about 127 Hz.
I can tune the system to eliminate this - mostly by adding a 127Hz notch filter, and reducing the gain from 20 to 15, but then, when the table is moved to the left (as shown in the photo), it will start oscillating again. With the notch filter, it oscillates at about 30 hz, though.
So far, I haven't gotten a tuning that works at both ends of the table travel.
If I move the table by hand, I can't feel any binding, nor do I feel any other differences between the two ends.
I have also attached Bode plots from the two ends of the table, with the tuning without the notch filter.
I also attached an unsmoothed Bode plot, and plots of the axis move function.
 
Any thoughts about this?
 
Thanks,
Michael