Axis Positioning Problem

[joeby]

Hello,

I am trying to sort out a problem with an Okada Graphite Master with a Fanuc 160 control.

I have run a ballbar test on the x-y axes and corrected the backlash parameters, which currently show on the ballbar plot at .0001" or so. The problem now is the plot shows a reversal spike of over .0015" but only on the negative x axis, x positive shows very little spike. The circularity is at .0018" and has been at that value even before adjusting the backlash.

I would think a mechanical problem would show the reversal spike on both sides of the plot, correct?

Could this be a bad pitch comp. value?

I need to get this problem sorted before I can fix the real issue on this machine. The machine is equipped with glass scales which have never been enabled because of "jerky" movement. I have seen machines do this because of a faulty scale; but supposedly this happens on all three axes. I will have to wait 'till I can get an acceptable ballbar run before I enable the scales again.

Sorry I cannot post the ballbar test results, the files are on a computer which I don't have access to at the present.

Thanks for any suggestions!

Kevin

[tc429]

I'd look for 'kick' in the x slide- put a indicator from the saddle to side of the slide at each end of the slide, with the axis in approximately the same point as where youre seeing the spike- put a little .100" incremental back and forth program in there and watch for any kick at either end...I'll often push/pull a little while its moving back and forth too- once its sliding, a horizontal axis can easily be deflected if its got a loose gib. a tight gib acts like a brakeshoe, and as the guideway has a fair distance from the ballscrew thrustline, a tight gib will cause a torque load kicking the slide sideways a bit at the opposite side- have seen some ballbar plots like youre describing from this...you dont always see the reverse kick as it might be in another(looser) gib contact point, so the return kick is more a gradual distortion...

please be sure to let us know what you find
Tim

[joeby]

Tim,
Thanks for the suggestions!

I had to think about it a minute, as the machine has linear guide ways, but it could be a similar issue with binding. I still need to go back through the paperwork on the machine to see what had been changed. I think this issue is rather recent, possibly only since a recent relocation of the machine, which I don't think went as smoothly as planned. I need to check into that as well (possible machine damage).

This machine also shows an out-of-square condition on the ballbar plot. Another point worth mentioning is that I was asked to look into a chatter problem that occurred in the upper left quadrant of a circle cut on this machine.

Looks like it's all pointing to a mechanical problem.

Thanks again,

Kevin

[geeserteg]

Did you ever find out anything with this?? We have 2 GM656's and 1 has this issue in the Y. We've run a ball bar and checked just about every part mechanically.

[memoryman]

a quicky: did you look into reversal error compensation amount and direction for each axis?

[Dan Fritz]

Sometimes backlash compensation is set using an improper procedure. If you put an indicator on the slide and try to just move back & forth a tiny amount, you may "see" backlash where there isn't any and over-compensate. Any axis with a little bit of "slip-stick" or a bit of ballscrew "windup" will show no slide motion when the servo actually moved the correct amount. Try making a short "back & forth" program at a normal feedrate with some G04 dwells between motions and see if the axis settles in the same postion from both directions. Also be sure that the lube system is working well before testing.

[tc429]

Dan, thats almost exactly what I do- but I dont put g04...instead at very low feedrate (1 IPM or so) I just watch for a jump or pause at the reversal...the program has to have several 'back and forth' sequences before the m99, as the m99 results in a in-position check...anyway, if the indicator jumps at all, too much comp, any 'pause' at all means not enough... we do aluminum, and biggest 'issue' is cosmetic rather than dimensional, and slow reversal with no pause or jump is all we need to get the best finishes...tried a lot of different ways, but this was always easiest way to get the best surface finish.

I agree theres a LOT of folks that set comp the wrong way...our most common is MPG at .010 a click, and just like you said, stickslip lies to you- it always makes for steps if set that way...
Tim