Cncfusion Lms Upgrade Kit #5

[grepper]

I sat down to order the CNCFusion ballscrew upgrade kit, but have some questions. I wish the CNCFusion site was more pre-sale informative.

They offer a pre-loaded ballnut service for $20 per ballnut. CNCFusion says,

The stock Nook ballscrews can have anywhere from .004-.014 backlash, it really just depends on the batch of ballnuts what you get.

The nuts will come from the factory, loaded with .125" balls. All kits come with the ballnuts already installed onto the screws, ready to go.

When you order pre-loaded ballnuts, we remove the stock .125" balls, and load the nuts with the largest balls that will fit without binding. This is called pre-loading the nut, so that it fits tighter to the screw, eliminating pretty much all backlash. Ordering pre-loaded ballnuts ensures that you will receive ballscrews with .001 or less backlash.

Does, or how does Mach3 compensate for screw backlash? If Mach is just counting screw rotations, it would seem that backlash could be very problematic. Think I need to pony up and get the pre-loaded nuts? I don't understand the ramifications or exactly what to order.

They also offer Heli-Cal zero backlash couplings. Is that worth it?

Thanks!

 

[JimDawson]

IMHO, spend the $20. Any backlash in the system will cause accuracy problems. Mach3 can compensate for backlash in some instances but can not compensate when cutting a circle or an arc, and in many cases a straight cut. It will compensate if doing a drill hole pattern.

As an example, I have a couple thousandths backlash in my Y axis (I need to fix that). By it's nature, my system automatically compensates for backlash. But when cutting a circle I can count on it to be out of round by about 0.001, in most cases this does not cause problems, but when trying to make a bearing pocket or something like that, I wind up cutting undersize and boring to size.

I think I read somewhere, but now I can't find it, that it comes with the Z-axis pre-loaded, so if it's necessary I would only need to order two. Am I correct in this?

If that is true, then yes. I looked at their website and could not find any information on that. I think an email or a call to them might be in order.

I would go with the Heli-Cal couplings, the Lovejoy couplings are not zero lash.

I'm here to help you spend all of your money

 

[Bob Korves]

Backlash also causes problems with rigidity of the machine, making life much more difficult for cutting tools.

 

[grepper]

I'm here to help you spend all of your money

Well Jim..., Take pride in the fact that you are succeeding! Your wonderful "help" just added another $150 to my already over budget little project.

 

[compsurge]

Did you buy Mach 3 already? If not, can you use LinuxCNC? That will save you some. I also find it much easier to use.

 

[cozmogeek]

You measure the backlash then configure it in your software. It then knows any time it reverses directions on whatever axis it needs to move .001" further (or whatever the backlash is) to take up the slop in the screw. It is kinda weird but it seems to work. I am liking Linuxcnc as well.

 

[Firebrick43]

Most cheaper systems act just as Cosmo said. It's not the machine compensating it's the software using values you provide. More expensive systems use glass or magnetic scales and reference it absolute motion measurement to the commanded motion measured by the encoder. It will self compensate up to a certain value and then alarm out if they are exceeded.

The next step combines the two. Big machines can actually compensate in different areas of the screw/travel to compensate for the inevitable wear, imperfectly ground screws, sag/flex of long tooling and settling of the foundation.

 

[grepper]

You measure the backlash then configure it in your software. It then knows any time it reverses directions on whatever axis it needs to move .001" further (or whatever the backlash is) to take up the slop in the screw. It is kinda weird but it seems to work.

Wow. I'm surprised that works. I would think that if the measured backlash was off by even a little, cumulative error over many operations could be significant.

 

[JimDawson]

Wow. I'm surprised that works. I would think that if the measured backlash was off by even a little, cumulative error over many operations could be significant.

I would think so also!

 

[grepper]

Wow. I'm surprised that works. I would think that if the measured backlash was off by even a little, cumulative error over many operations could be significant.

I would think so also!

Well, then, it begs the question: How well does this method pan out? How do folks deal with cumulative error? I would even think that backlash could vary over the length of the screw due to variation in the accuracy of the screw. How does that pan out?

What does all this mean?