Ballscrews & Backlash

[rustyolddo]

I'm putting the finishing touches on my Mini-Mill conversion and I added 5/8 ballscrews to the x,y,z. It took some shoe horning to make the X ball screw fit. Anyway, there was some backlash associated with the ballnuts so I sent them back to Roton to be fitted, they did a fantastic job at no charge.

Anyway, I don't know about other software out there because I've been using TurboCNC to test with, but TurboCNC has backlash compensation built into the software. If the backlash can be compensated for in software, why is there so much fuss over compensating for it mechanically?

Is it due to the need for manual control? Granted it's a PITA to work manually in two directions +/- when there is backlash present. Just wondering.

[ChrisJ]

I'll take a stab at this. I could be wrong, so wait for the experience users to validate or shoot this theory down.

Unless your backlash compensation that you enter into the software matches exactly what the actual backlash is on the movement of the screw, the difference/error will accumulate. The more back and forth moves you have on that axis in cutting a part the more error builds up.

But then again, I am thinking this error may cancel out. Maybe the experienced will set us straight.

Is this correct?

Chris

[FLUTE HEAD]

I would like to know also. I have heard people having trouble with circular cuts. It would seem if the backlash is measured accurately, then computer could compensate. Maybe its dependant on quality of program used. Any comments?

[JFettig]

The problem with backlash is even though you can compensate for it in software, the machine can still move on its own within the backlash. Soon enough, I plan on sending my screws and nuts into homeshopcnc to preload them.

Jon

[RotarySMP]

Try doing climb milling with mechanical backlash and electronic compensation.

[fpworks]

The backlash will vary over the length of the ballscrew, so you have a "tolerance" on the backlash measurement. This tolerance will be greater with lower grade ballscrews. Each time you use backlash compensation (by switching directions of travel) you will accumulate error.

Depending on the work you're doing and the class of ballscrews, you may or may not find this error accumulation acceptable.

Also, with reversing curved cuts (over 90 degrees), sometimes the software compensation cannot make your servos or steppers "jump" (backlash compensate) quick enough if cutting at a high rate, and you'll get "flats" where the curved cut switches direction.

Justin

[Rob T]

I dont agree that errors in backlash compensation accumulate (but I do agree that minimizing mechanical backlash is better than compensating it in software).

Suppose backlash compensation is set to some value "e" for a particulat axis. If you tell the machine to go to position "P" on that axis, then the software actually moves the screws to a position "P+e" or "P-e". Which of the 2 depends on if the current position is numerically above or below P ("P-e" if (current position) > P ). You'll never get "P+2e" or "P-2e" etc, no matter how many times you change direction.

The real problem with backlash is that it's not a single constant that can be perfectly compensated in software. It varies with load on the cutter (as Mark said above, climb milling will show up the problem with sofware compensation by making the backlash behave differently then normal milling).

[John Labutsk]

I have found that backlash on single ballnuts can be significantly reduced by repacking the ballnuts with new balls. It's 4 bucks for a package of a hundred SS .125 balls from enco. Just mic them for uniform size and save a few undersize ones for final packing. Be careful not to overpack or you will bend the pointed tabs on the 2 piece ball return tubes. This was recommended to me by Rick of Rick-0-Matic fame. He has pic's on this site. It sure makes a difference. You can plug in .001 backlash in your software and this helps compensate for flexing of small milling cutters.

[ViperTX]

and where is Rick's site....he has promised to update us on his Rick-O-Matic spindle.

[ESjaavik]

As Mark wrote, you cannot climb mill with backlash, compensated or not. There's several reasons you'll miss that. It makes a cleaner cut with less marks from chomping old chips. The wear on your cutter is lower. And while hog milling you use less G0 moves just cutting air.

Also you don't say if you use servos. If you do the backlash will make it impossible to tune them properly because there is virtually no resistance to movement in the backlash region.

[fpworks]

Rob,
The error will accumulate because of the fluctuation in the pitch. Lower grade (even ground Class 7) ballscrews have quite a bit of pitch fluctuation, so the mechanical backlash measurement will be different at along the length of the ballscrew. Software compensation (that I know of) will only allow a single value for backlash compensation.

Take a look at how much backlash is in higher grade [non-preloaded] ballscrews. Higher grade ballscrews have less backlash because there is less fluctuation in the pitch of the ballscrew.

[Rob T]

fpworks, I think you're talking about different forms of error. Backlash is just one form of error and describes the positioning error when the tool is moved to the same point from different directions. True the amount of backlash can vary at different points along the screw, as might happen if the thread profile was not perfectly uniform. By measuring the backlash at various point along the screw and using an average value you can get a backlash error that is reduced from the uncompensated value, provided the forces acting on the tool continue in the same direction (that's the point about climb milling- they dont!).

Software backlash compensation alone will never, ever, lead to an error that accumulates over multiple moves of the machine or that increases with distance travelled.

That's not to say that other errors dont accumulate. Errors in the pitch of the screw would lead to a positioning error that may well increase with distance travelled. That's a positioning error, but it isn't backlash. It would be a mistake to compensate for an error in screw pitch by adjusting the backlash compensation, just like it would be a mistake to change the configuration for the distance travelled per step to try to compensate for backlash!

I agree that a better screw / nut has less of all forms of error, backlash and otherwise. So I'm not arguing against the use of good screws and nuts. I just dont want to see people scared off using software compensation to improve on whatever screw they are using for fear of things that dont actually happen!