Ballscrew Comp 2nd & 3rd Attempts

[adrewfis1980]

I watched the video on how to do the backlash and the ballscrew compensation. However, I am still off and would like to dial it in a lot closer. What is the process for adjusting the tables to account for the small differences on the 2nd or 3rd go around? I've tried doing a separate run and then adding the differences to the original number in the table and still don't get what I'm desiring for accuracy. I've attached the most recent report for reference.

report_54453800FA74-0602237253_2024-09-13_20-26-10.zip

(914.95 KiB) Downloaded 2 times

[cnckeith]

need more info. photos of your machine for starters. make a photo album and post the link here.
what kind of differences are you seeing in the runs? .00001"? or .01"?

[tblough]

Ballscrew comp is not something you iterate on. You command a position, measure the actual position, and record the difference in the table. Comp tables will have values in tenths, or less than 10 microns. Backlash values will be less than a couple of thousandths (50 microns or less).

If your values are larger than this, then your turns ratio is wrong, or you have a mechanical problem.

[adrewfis1980]

Ballscrew comp is not something you iterate on. You command a position, measure the actual position, and record the difference in the table. Comp tables will have values in tenths, or less than 10 microns. Backlash values will be less than a couple of thousandths (50 microns or less).

If your values are larger than this, then your turns ratio is wrong, or you have a mechanical problem.

I was under the impression you could improve upon the ballscrew numbers based on Alan's video with the comment he made regarding this.

What distance do you set the turns ratio? For some reason I typically always set it to the furthest point and get it within .0001" with several attempts.

Also, do I need to set the backlash when using ballscrew compensation? Thanks for your help!

[adrewfis1980]

need more info. photos of your machine for starters. make a photo album and post the link here.
what kind of differences are you seeing in the runs? .00001"? or .01"?

I have attached a pic of the machine and one of the most recent ballscrew attempts on the Z axis. Some of the numbers are well within my tolerance (.0004" or lower) but as you can see some are out .01 or more which is very odd.

https://www.dropbox.com/scl/fo/95grkb6m ... brc6r&dl=0

[Nigelo]

The lash compensation figures in your report suggests that your basic calibration is incorrect. Lash Values are X=0.0101" Y=0.0507" Z=0.0234" which are enormous at 10, 50 and 23 thousands. This cannot be correct unless you have extremely serious problems on your Mill which, in my opinion, would make the ballscrew mapping a complete waste of time.

However, if you are satisfied the machine is in good mechanical order, I suggest you turn off ballscrew mapping, zero each axis Lash setting, reboot CNC12 and once homed, recalibrate all 3 axis for the machine as accurately as you can. Only then, recalibrate the basic lash settings ideally using Marc Leonard's (cncsnw on here) ready made instruction (and downloadable code) on his site http://cncsnw.com/ then navigate to "Maintenance" section then "How to measure Backlash".

Once the above is completed, you can proceed with the ballscrew mapping.

[cnckeith]

backlash compensation should be set to zero when using ballscrew comp. by definition ballscrew comp is taking care of any lash.
think about it, you don't want backlash compensation adding or subtracting to the screw compensation table as the screw table is what is really happening in space.

[cnckeith]

.62625 turns per inch? .01" lash?

does this machine have ballscrews?

what is the pitch of the ball screw?

[cnckeith]

related info.
viewtopic.php?f=63&t=1801

and
viewtopic.php?f=63&t=8649

[cnckeith]

post a screen shot of your Clearpath MSP software settings.