I'm trying the backlash function and it seems odd behavior to me.

Jogging the axis back and forth results inthe backlash being taken up at a very slow movement regardless of the speed setting in the backlash config box, or the feed rate commanded.

Am I missing a setting somewhere or is this the normal behavior?

I cut aluminum at 6,500 RPM .002 per rev feed on a 5/16" end mill. The extreme slowdown during backlash takeup is going to be a problem methinks.

Hi Simpson,
Servos or Steppers ?
What is your "Shuttle Wheel Setting" set at ?
You might try reducing it.
RC
 
 

RC, thanks!

Steppers. Two 495oz-in and a 900+ on the Z

20,000 steps X and Y

13,333 steps on Z

I didn't know the shuttle had anything to do with it. I'll check that out.

What is the speed setting in the backlash dialogue box for? It doesn't seem to have any effect.

The backlash speed is % of max in motor tunung but it appears to ignore the accel setting so it needs to be set accordingly to prevent missed steps.
RC

My observation is that these setting in fact have no effect (using newest .20 version). I've played with it a bit more now and no matter what shuttle or backlash sleed is set to, there is still a large and very noticeable delay at each change of direction, both for jogging and during program execution.

The actual backlash takeup speed seems to be stuck at 20%. It will go slower if set lower, but it will not go any faster no matter what number is put in the box, including 100% with shuttle set to 0.

I noted that using backlash defeates the CV, which would not be a good tradoff, methinks.

My machine now has .005" backlash on each axis using rolled Thompson ball screws. AM I being too anal wanting to use the backlash function? What do most people do? What is a reasonable amout of backlash to accept?

Is this an undocumented unsupported feature in the demo verion?

Sorry for all the questions. Newebee here.

Did you ever get this sorted out?
I recently had the same issues with Mach v3.042, whenever I'd move in any direction the machine would stop, wind out the backclash, stop again, and then resume movement.
I reinstalled the older version of Mach, and then the newest again, and the problem went away.

The backlash thing and some CV issues and threading problems are things I am watching to see if they get sorted out before I take 'the plunge' and commit to Mach 3. I already have a fairly big investment in time into Mach 3 and so I would want to use the same control software for both mill and lathe.

The backlash, as noted in the documentation, is really only good for things like increasing the precision of drilled holes or things like that. The implementation currently in mach is not useful in my opinion for general machining because of the pauses when the slack is taken up. I simply cannot use it for milling.

I watched a youtube video from Hossmachine where he has to use backlash compensation just to get a round hole. I'm not nearly that bad off, but I did have .005 on both x and y and a good deal more on Z.

My solution has been to buy oversized balls for the leadscrews. I located .0009" oversize balls on ebay and they have taken the lash to less than .0005" on the X a little under .002" on Y (my problem axis) and I designed and built an entirely new mechanism for the Z and it now has only .005 and I am waiting on the new oversize ballscrew bearings for that axis to get that number down . . I am hoping for .002" or better.

For my purposes, this will be accurate enough so long as it can hold or a while.

I am no expert in CNC control programming by any means, but it does seem to me that adding the backlash amount to the distance of first movement after a change of direction (per axis) would not be a difficult thing to accomplish. While that would not be as accurate as stopping and taking out the lash and then proceeding (with jerky, tool burning pauses), it certainly would increase the accuracy of the machine while maintaingin a smooth CV. That would not be a perfect solutio, but in my opinion it would a logical and valuable compromise between the current 'useful for drilling, useless for milling' gap.

You mentioned that there is a change in the backlash comp behavior  in your software that is version dependent. Can you elaborate on that please?