Is this from servo "droop" and ballscrew backlash?

[Wade C]

Okay, in the process of getting most of my program to work, I stumbled across a little issue, that, though could be overlooked and dealt with in the other ops, Id rather it not be there...

On the bottom of a two quadrant arc,
N8117 G3 X11582 Z30195 I-00399 K-0009
N8118 G2 X11040 Z12150 I27809 K-9443
N8119 G1 X09668 Z10516

say its cutting from 5:00 to 7:00, at 6:00 there is a small ledge. My best attempt at measuring I get .001 - .002" lip (.002 -.004" dia change) and its larger from 6 to 7 than it is 5 to 6. Which makes me think either the machine is incorretly compensating for backlash, the droop is too fast and its over correcting iself to quickly, or, that even though it cuts the arc, the program needs to be broken into quadrants....

Any thoughts or suggestions?
Thanks
Wade

 

[CDiPrecision]

It is possible...can you mill a simple circle and attempt to measure it with an indicator??

 

[Dan Fritz]

Sure sounds like a backlash compensation issue. Lots of people set the backlash comp to be more than necessary, so the control over compensates at each quadrant boundary.

The best solution is to eliminate as much backlash as possible by adjusting or replacing the thrust bearings on the ends of the ballscrews. You should not have much backlash to compensate for if everything is mechanically up to par.

Most CNCs have a parameter for settting the backlash comp in each axis. Use a .0001 linear indicator and move the axis back & forth at a slow feedrate. I like to write a short program to do this, and I move about .100 inch in the plus direction, then back to zero. Then, I move about .100 in the minus direction and then back to zero again. The axis should always return to the same spot regardless of which direction you move back to zero. Adjust the backlash comp parameter until you get the axis to always come back to zero on your indicator in both directions.

Be aware that a worn ballscrew can have more backlash in the center than on the ends, so make your adjustment at the point where you do most of your work.

 

[Wade C]

CDi - sorry should have specified... this is on a lathe. okuma LC 10 with OSP3000L control

Dan,
This is a 1982 machine, and Im sure seen some use... and Im just a little hobby guy pretending to be a big guy with a big machine... I guess you could call it machine envy... Ill check out the backlsh... I dont have .1000" travel on my indicator that reads .0001" resolution, I only have .008" travel on it, so I cant go quite as far, but Ill be sure to check it out.

The part that I find odd, is it actually would be an over correction on the backlash, as in, someone measured it, and set it for more backlash than it really has. Otherwise, I would think it would be a step down rather than the step up in the direction of cut.

Thanks Ill be sure to spec it where it does its most work on the X and Z

Wade

 

[Dave K]

Sounds like over comp for backlash to me too. I've seen this on a 1982 Hitachi that I ran a while back. I would set the backlash so that you have about .0002 backlash. Don't set it so tight that you have 0 backlash because you run the risk of overcomp. .0002 backlash is tight enough that you won't even know it's there.

 

[Wade C]

went in, the x axis backlash was set at .0055 so I backed it down to .0030 and it seems to be smooth as a babies but now.

So I must have some good wear for having that kind of backlash. Didnt even measure it.... just pulled a couple off. Ill measure it for sure this next weekend.

THanks for being a soudingboard and help narrow and confim the issue.

Wade