Index Homing odd offset

Note that tuning is partially done in your drive, any tuning issues there will become
apparent when you try and close the loop in LinuxCNC.

(If you run the drives open loop you might not ever notice the errors)

Tuning issues in the drive will not be fixable in LinuxCNC and must be fixed by tuning the drive

That is a very good point. The dmm drives do come with a basic tuning program. I adjusted the pid settings there and ended up with something close to the default numbers. Too much P and the motor got noisy so I backed off. I have a very small amount of overshoot on the heavy Z, around .0005" to .001", always returns to 0 on the dial indicator. This is in stepper mode, slow feed or rapids, and I think that can be adjusted out in the drive pid loop or I can live with it.
In stepper mode, it feels responsive and I have not been able to fault out the drive itself.
In closed loop, it feels way off, more like some parameter is missing. Axis either creeps, dithers or is bouncing back and forth with frequency increasing at no external jogging command given. Axis jogs okay at 12-17 ipm or below, above that it joint following errors out. Direction change sometimes triggers error at slow speed.
If the drive remains powered up with no steps being sent, the motor stays at rest with no oscillations as expected.

Andy,

I will try to capture that with halscope.

Also if the drives have any input command filtering options enabled they will likely need to be turned off.
Any significant delay from the step command to motion will make closed loop tuning nearly impossible

I think in ste/dir mode input filtering is always active. I glanced thru their manuals and I didn't see an option on turning it off.

www.dmm-tech.com/Dyn3.html

Maybe I should have purchased an analog Mesa board instead???

I'm starting to like halscope
The encoder signal looks messy
What do you think? The first two are of P0 at 12 ipm, the next are at P50 and the dithering at rest which is around .001" tir.

You need to add ferror as a trace in halscope. You are tuning to minimize it so you need to see it. If ferror is close to zero all the time then the commanded and actual velocities will match.