Diagonal travel error

So I have an issue with my setup that I am struggling to find a root cause.
My mill is an old Prototrak that I had to replace the controller
It now runs on a Mesa 5i24 and 2 7i29s.

I am having an issue where I have a random drift in my -X direction. The y-Axis is rock solid.
I have a couple test programs I run in air (spindle off). The Spiral that comes with Axis runs
perfectly (final position error <.001 compared to the linear encoders on the table). However
I made a stepped pyramid that increases in dimension by .2" (.1" in every direction) per step.
This program fluctuates in final position error from -.03" to -.4" each time I run it.
Since the error is always in the -x I dont think it can be backlash. I have tried multiple different
methods of calibrating the Axis but like I mentioned the Y axis which has more weight is rock solid
(<.001 every time) and it has the same calibration.

I thought I had it where originally the spiral wasn't cutting well either and I found that the x-axis handle
dial was tight and loosening it completely solved the problem for that program.

I stepped through my test pyramid program line by line and noticed that the error seemed to accumulate on short diagonal moves
and the system just never caught back up.

Ive checked the pulley on the X-Axis servo and noted no slip (.4" of slip would be more than a full rotation)
I watched the lead screw move I cant see anything binding.
I've tried manually randomly moving the table with the keyboard as fast as I can (rapids, slower speeds and diagonals)
and I cant repeat it manually.

Im out of ideas of what to check.

If you have cumulative errors on a servo system, its often caused by losing or gaining
encoder counts.

Things to try:

1. Use differential encoder signaling (assuming you have differential encoders make sure the 7I29 is jumpered to use the differential signals)

2. Enable encoder filtering in the hal file:

setp hm2_5i24.0.encoder.00.filter true (for encoder 00)
This is probably already set

3. Lower the encoder sample rate:

setp hm2_5i24.0.encoder.sample-frequency 2000000

That (along with the filter bit being set true) would limit the count rate to about 266 KHz and have a
7.5 usec time constant digital filter on the encoder inputs

That and double check all the couplers between encoder, motor, shafts, and pulleys.

The sample frequency change seems to have handled it.
Thank you very much