Missing steps on just a certain part of the part

[ZipSnipe]

I have a Lightmachine 0200(Sherline) with Lightmachine 100oz steppers and controller. I was making a sign for a friend and there are 4 identical borders(an L shape design in each corner). When the machine gets to the 3rd border it screws up by 1/8 of an inch and then does the next one fine as is everything else is fine. I,ve checked the code with Mastercams compare tool, and re examined the code several times to find no fault. The verifications show no probs. I also slowed the feed down to 30%(turtle) and still it skips steps ever so slightly that I can't here it skip. My question is what are some of the things that can cause this type of skipping?

[SORCHEROR]

if it missed steps on the third border how could the fourth one be correct?that makes no sense to me,usually once it misses steps everything after is wrong,it is possible the machine has a tight spot in it to cause the missed steps,but doesnt explain why the fourth border is good

[neilw20]

Motor Tuning.
When you move to a corner one axis has to stop, and another has to start.
Due to inertia, nothing stops instantly.
It has to decelerate to a stop, before accelerating in the new direction.
I am not familiar with your system.
Is it Mach3 based?
More info please.

[kreutz]

It could be Backlash. We CNCd one Sieg X1 using the original nuts and screws, and had a similar problem while cutting a circle, 180 degrees were perfect, the other 180 degrees were a disaster, no lost steps, we returned to the same position after finishing without losing it. Using the same motors and controllers on another machine with grounded ball-screws the circles are perfect.

[ZipSnipe]

Kruetz I think ya nailed it. That makes sense to me as the culprit. Yeah the screws look like allthread to me. Yeah Neil I originally thought that was the case but I have the feed running painfully slow. By the way the machine controller is Fanuc based. Where can I get 1/4 ballscrews at?

[ZipSnipe]

Uumm upon further inspection of the gcode I found something weird. When I designed this design(MastercamX) all I did was make a curvy L shape and then xform mirror(which copies a mirror image opposite) and then a second xform mirror which should have created 4 L exactly the same size and depth. But when I checked the gcode I noticed the 3rd L is several thou shallower and about .050 off on its return path(it makes 2 passes on each L shape). Thats one part of the weird, the second part is when I check the home position the x is out! This led me to believe it was missing steps. Here,s the code http://www.cnczone.com/forums/attach...1&d=1199302874
Todd and Angie.txt Look at line #114 ,318 and 726. Compare that with line#522(3rd L)
Any ideas?