Cause For Losing Steps?

I was cutting a part yesterday and failed to doublecheck as I progressed. It turns out that my part is off in every direction by 30 to 60 thous. Could someone list causes for losing steps? My distance per step running 13/64 lead ballscrews is figured like: 13/64 / 1600 steps per revolution(default 1/8 step with xylotec) = .000126953125. start speed=500 accell=1000 max speed=5000. I kept these numbers low, not sure if this is good for stepping.
step pulse = 0. Any help on what could be the cause? specs. turbo cnc with xylotec. motors don't seem to struggle and all lock up when powered. Running turbocnc v3.1a

Try .000125 for the step size. Everything I've read says those ballscrews are 5 tpi.

Thanks ger, so instead of calculating screw lead per rev. I should have been going off of thread per inch? Mcmaster specs claim 13/64 screw lead. 13/64 movement per revolution. Is this wrong in thinking? Motors do seem to move machine ok but not sure about my speed settings. somewhere around 600 oz.in I believe 2.6 amp per phase motors, bipolar driven one to one w/coupling. There wired for low speed higher torque since there moving a rong fu size machine. I'm hoping there not losing steps because of torque/ speed issues.

I think Mcmaster -Carr rounds of the 5 tpi to the nearest 64th. I don't think the 13/64 is the actual lead, from what I've heard. I'd check that out first to see if it helps.

The Xylotex is probably a little small for that size machine, but if you're going slow enough you should be OK. What Feed speed we're you running at?

Yeah I was thinkin on similar terms when Iput the system together, low speed. Actually I was mainly jogging the machine to drill some holes. I think it was set at 10 ipm. I'll use your calc. for step. If I calculate the difference in what I had verses yours I come up with .003125 difference per revolution @ 1/8 step. So this may have been it.

Flute Head,
first off, I would upgrade TCNC to V4.01 as there have been a LOT of changes to the engine since your version.
You "may" find this will solve many of your problems!

Also if you are using a Xylotex controller make sure the vref adjustments are correct and that you are cooling the heatsinks actively by blowing air across them when vrefs are near the upper limit. It will drop steps like mad if either the vrefs are just to high or if the chips get to warm. This time of year in it could be that you are overheating the controller.

Thanks guys, fan is cooling the xylotec, also vref is set. I actually have 4.0 I'm just not using it yet. I had planned on it eventually. So the new changes in tcnc may help, great. Last night I checked out the machine with a test indicator. It seemed after bumping the pulsewidth from 0 to 2 the machine responded better. I actually do not know how to calculate the correct pulsewidth. Machine seems to be within a thousandth when jogging back and fourth also running one line codes. OH, most of all, I did input the .000125 for step. I kinda like the layout of tcnc 3.1 better than the new one. Although I only played with 4.01 for a short time.