Position drift with closed loop?

Hi guys, new on the forum here

I've converted a BF30 mill with DMM DYN4 motors and Mesa 5i25 & 7i77-kit but experiencing some issues when running gcode. I've managed to tune the motors so that ferror is <5 microns after a 50mm cruise (f1000mm/min)

But when I run gcode, the x-axis drifts up to 50mm (!) when i try to mill a lego brick (5min job). The DRO show no drift and the ferror and min_ferror is set to 0,2 and 0,02. If I decrease these, it throws a following error.
X-axis also makes random harsh accelerations in the middle of the code, and it sounds like light taps with a hammer

Could it have something to do with the encoder signal? Could the encoder be throwing out more pulses than my computer can handle?
I've tried to change the default servo period to 150 000 ns without any noticable differance.
Tried to set the encoder output to the minimum; 500 lines/revolution in the drive, and encoder scale to 400pulses/mm in linux (5mm pitch screw), but it did not work

Anyone know how I can fix this? Thanks

From your description it looks like that you might have a mechanical problem rather a encoder one. Encoder problems normally trough following errors as the error created by the encoder failure is detected as a "large movement".
Maybe a broken keyway? a slipping shaft?

I would recommend a repetition test where you move slowly your axis in one direction (as long as possible) then return to the original position, make a pause and have a gauge right were it stops. Confirm that it stops always on the same position. If true make the test faster, then faster. You can also try with different accelerations.

Hey thanks for your reply! We've been troubleshooting for a few days now and we (think) we found the problem:
Probably bad cable between 7i77 and the drive. When we switched the X&Y cable, the drift started to appear on the Y-axis

Switched the cable for a new one, and it hasn't drifted since!

Watch out for crappy DB9 cables folks!