I am having a hard time cutting accurate circles. when I cut them they are slightly elliptical it seems that there the motors are not moving at the correct rates. At the points where the x axis is acceleration that quadrant of the circle is skewed away from the x axis. As x is decelerating it flattens out that quadrant of the circle. this problem is repeatable each and every time. If I reverse the cutting path the problem is exactly the same.

I have a taig mill with a xylotex board. If I run a program the spindle will return to the home position with absolutely no error on the dial indicator, so it is safe to say that I am not losing steps.

I have tried different bit sizes and speeds to eliminate the possibility of bit flex. All gibs are very tight with no play. I have the axis weighted via a pulley system so there is no backlash. 

I have run g code generated from mastercam as well as the wizards included with mach2, both sadly produce the same results. I have tried both constant velocity and exact stop with no change in the outcome.

There is one question that may answer my dilema. If I use exact stop when cutting a circle does the stop occur at the points indicated in the g code or does the mach software compute each individual stop point?

The only way you can get what you are seeing is if you have backlash... put and indicator on your axis and set the axis to zero. now jog away from the indicator and type G0 X0 into the MDI window. did the indicator go back to zero? If it did go back to zero jog the axis into the indicator and do the same G0 X0 in the MDI window. Did that give you zero?

That is just the first test...

I think you should try MAch3 with Sherline mode. this will tell us if your drivers are having trouble


There are a few more things that we can test after that

Crazy idea. How about swapping stepper motors? X-Y or X-Z, Y-Z to test your steppers. How elliptical are the circles? .001 in 3" or what?