XHC-Motion-Card-MK3 positioning error

Hi
I have a setup that is running on a cheap usb mach3 controller on 3 axis for a tablesaw, using this kind of controller:

Redirecting to https://redirect.viglink.com?key=a7e37b5f6ff1de9cb410158b1 013e54a&u=https%3A%2F%2Fwww.ebay.com%2Fitm%2FCNC-USB-MACH3-100Khz-Breakout-Board-4-Axis-Interface-Driver-Motion-Controller-r%2F233445300969%3F_trkparms%3Dispr%253D1%26hash%3 Ditem

I have an issue with on of the axis, the more moves it makes, the more it drifts off positioning. so after approx 20 moves, with same start and end pos hitting a dial indicator, it has drifted off approx 0.7mm
This is a card labeled 100khz so in theory it should be able to pump out enough pulses.
But I thought I would try another board, and got the ethernet version of the XHC-Motion cars MK3:
Redirecting to https://cnc-controller.com/product/xhc-ethernet-mach3-breakout-board-3-4-6-axis-usb-motion-control-card-resume-2mhz-support-for-cnc-lathe-engraver-newcarve/

This is a 2Mhz card, so it should be able to shuffle out enough pulses.
However, this seems to be even worse.
I cannot get consistent positioning trying to run the same axis as the one I drove on the usb version.
Does anyone have any experience with this controller?
It comes with a designated driver for Mach3, but this mainly has settings for I/O´s, and no motor tuning settings.

best reg
Cory

I guess that it has to be with the polarity of step pulses.
If your software generates pulses the way that drivers should trigger on rising edge and you have the polarity wrong, it will trigger on the falling edge of the step pulse, but it will not be synchronized with direction pulse so you will, efectively, loose step pulses in one direction.

Thanks for the feedback. I am really poor at explaining, sorry about that. For instance I start with the dial indicator at 0.00mm. I then tell it to move 10.0mm. It moves to 10.0mm. When I say move to 0.00, it moves to ie. -0.09 mm, when I say go to 10.0 mm its moved to 9.89mm. So it seem to move the correct amount, but just shifts off in one direction. I run it with mach 3. I setup another linear actuator and tried, and it gave me the same problem, so I guess it is not in the Parke servo drives there is an issue. I am going to try another computer to see if it has any effect