closed loop using linear scales vs rotary encoders

Somewhere along the line I think the smoothstepper or its config are at fault. There is a setting in one of the smoothsteper specific menus which dictates how frequently the velocity info is updated. It was set to 1k/sec, there were other options for 500hz, 2k and 4k. I am going to try fiddling with those when I do another test cut. If it gets set too fast there is risk of the smooth stepper and Mach losing sync, resulting in the dreaded "lost communication with smoothstepper" error. I've seen that when ethernet powersavings kicks in or windows runs some file I/O heavy cron job in the background. To get any semblance of stability I had to disable any form of windows powersavings and re-time all the crons to run in the mornings instead of mid-night..

As I mentioned, the first half of the part looked fine, the second half which gets rotated 180 and then ran at a faster speed had progressive drift. I looked over older versions of the parts and the associated gcode, everything was setup to run 120 in/min. They were all OK too.

In the end it just strengthens my desire to move the whole control system to linuxcnc and a mesa 7i80