Hello Guest it is May 25, 2022, 05:36:56 AM

Author Topic: Backlash Comp doing weird things.  (Read 2980 times)

0 Members and 1 Guest are viewing this topic.

Backlash Comp doing weird things.
« on: April 01, 2009, 08:14:01 PM »
I've got my BP clone CNC conversion up and running and I'm getting about .0005 backlash and want to use the Mach Backlash Comp function but when I turn it on and then move let's say Y5 in MDI, it spins fast for a second, then spins slow then spins at normal speed until it reaches the coordinates. What's going on here? Thanks.

Offline Graham Waterworth

  • *
  •  2,456 2,456
  • Yorkshire Dales, England
    • View Profile
Re: Backlash Comp doing weird things.
« Reply #1 on: April 02, 2009, 04:04:26 AM »
What % speed do you have set for the backlash, is you have very fast rapids you may need to reduce this %

Without engineers the world stops
Re: Backlash Comp doing weird things.
« Reply #2 on: April 02, 2009, 01:06:49 PM »
Thanks for the reply. I found some info on it. I didn't realize that I had to restart the program, then I turned shuttle to .005, then I set speed to 100%, but someone else just told me I shouldn't have it at 100%? Any suggestions.

Also anyone know how to ACTUALLY do ballscrew mapping. I've looked and looked and seems no one actually knows how to do it?

Offline RICH

  • *
  •  7,428 7,428
    • View Profile
Re: Backlash Comp doing weird things.
« Reply #3 on: April 03, 2009, 08:52:36 AM »
I have seen very few posts on the ballscrew mapping and not sure it even works ( would assume it does).
Have never tested, tried it out, or implemented it in MACH for an axis.
I have done / profiled a screw and graphed it just for kicks using optical allignment euipment.
Basicaly you give a distance and move over that distance and adjust the total travel or incremental
distance, Mach then would adjust your axis travel based on the mapping.

Ok, here's an example. A lot of rolled ball screws have leads that may gain or loose over some distance.
They are not perfect and have specs on them. My screws will gain .003"/ foot, but negelecting any backlash, they will return loosing .003" / foot. Now in 1" of movement that's a gain of 0.00025" ( a swipe of a piece of sandpaper on a bore ) being practical. So they are repeatable. So if i mapped the screw for those numbers, then Mach should adjust the travel to get rid of the screws gain. Now if the screw varied you can adjust the travel based on different points along the way. What is important though is what the actual disatnce is compared to when doing the mapping. The distance needs to be done relating to a known standard. And it looks like there are input boxes to provide that info or you can use the electronics of the program.

If you were to do a profile manually the graph ( plot of movement to actual movement ) would be done along many incremental points. Each point would be plotted against a lne running on a 45 deg angle.
Any points above or below the line shows deviations from perfect. So maybe you would use a mean of the points, or select a spot you work most often at along the screw and that could be the an adjusted value.

Just one little note that comes to mind. If you don't use limit switches / home your machine then Mach would not know where you are and thus all of the above would not be of much value for a really screwy

Just some additoinal thoughts on mapping. Hope this is was of value. Now i should see if it's in the manual!  ;)


Here is what i would recommend for initial settings when using a machine with more than one axis.

"My"  50-50-50 rule of thumb.

Motor velocity setting: 50% of max velocity ( down from where you would start to miss steps post motor tuning )
Backlash Speed:          50%
Shuttle acceleration: .050 ( .5, or .050, or .0050 just an easly way to remember the setting )   

Then you can play with individual settings to get the system just so, such that you have no problems
during a two axis move. The motor acceleration can also be payed with.