dual loop with 1 micron linear scales?

A little background first:

This adventure to turn my manual knee mill into a Trak style cnc mill has worked out great! Best of both worlds, I can still use it as a manual mill but when needed I can go CNC. Its working so good I decided to upgrade to ballscrews. They have adjustable pre-load and I’ve been able to get the backlash out of the X-axis such that with a good push it only deflects no more than .0003” but then springs back to 0. The Y- axis has been more of a challenge I’m on the third adjustment I think I’m at the end of the mfgrs recommended adjustment range so hopefully that will do it, if not I think it will be a result of the Yoke flexing but I would still expect it to spring back to 0. I have angular contact bearings and have checked for any movement there, I get no movement. I’ll assume I can get both x-y to be super close to zero backlash.

Currently, I’m running closed loop but not a dual loop. I’m finding it difficult to reconcile the encoder scaling with the Lead error in the ballscrews and the 5 micron resolution of the linear scales in order to get the external DRO to agree with Linuxcnc.

I’m thinking if I can get near zero backlash that I can use a dual loop setup, then dealing with lead error basically goes away.

I do a lot of tooling work. Fixtures locating on dowel pins etc. it would be really nice to have the DRO agree with Linuxcnc to four decimal places so I’m looking into 1 micron resolution scales.

Questions:
1. Has anyone else achieved this and have it be stable, that is not constantly tweaking to keep it in?

2. I want to keep using the external DRO, its capable of the resolution. I’m thinking of daisy chaining the encoder cables from the DRO to the Mesa 7I77, but am also thinking I could invite all kinds of unknown noise issues. Has anyone done this?


Thanks for all past and present help.

-Doug

Why not just rely on one measuring system ? Maybe the DRO is out of scale ?