Chasing Backlash

[MontanaAardvark]

How do you get a shim between the two ballnuts? In mine that dark piece in the middle wobbled around a little but I was trying my best not to let them get apart.


Bob

 

[cs900]

How do you get a shim between the two ballnuts? In mine that dark piece in the middle wobbled around a little but I was trying my best not to let them get apart.


Bob

well that's not right, the second ballnut should be tight up against the shim.

I also think a better option would be to put a belville washer stack in there so that you get preload that will compensate for wear on the ballscrew.

 

[MontanaAardvark]

well that's not right, the second ballnut should be tight up against the shim.

I also think a better option would be to put a belville washer stack in there so that you get preload that will compensate for wear on the ballscrew.

I should clarify that this is when I had the nut removed from the screw with the ballnut removal tool. Still, how do you do that? How do you take them apart to get anything in there and rebuilt it so that it's tight?

 

[TomS]

How do you get a shim between the two ballnuts? In mine that dark piece in the middle wobbled around a little but I was trying my best not to let them get apart.


Bob

I believe the shim is installed as part of the nut assembly procedure at the factory. Thickness is calculated prior to assembly. The shim on mine seems to be snug which is what I would expect. I tried to measure screw axial movement with a dial indicator by pushing and pulling on the screw. The readings were not conclusive but were significantly less than the .007" I measured with it assembled. What's aggravating is I haven't been able to identify the source, or sources, of the backlash. If I knew what was broke I could fix it.

Tom S.

 

[TomS]

Today was a good and enlightening day. First, I will say my backlash is now about .0008". Yahoo!! I'll explain how I got there. To address MontanaAardvark"s questions about the double ball nut and shim, my shim was snug but not tight. I couldn't get a .002" feeler gauge between the nut and shim. It looks like the shim is a two piece arrangement and is retained by the anti-rotation keys. I tried taking mine apart but the key started to bend so I stopped before I broke it. How the two nuts are held together I don't know.

After reinstalling the ball nut and screw and the AC bearings I checked nut backlash by mounting a DTI in the spindle with the needle against my vise. Pushing and pulling the table in both directions I got a reading of about .002" +/-.001". Not what I wanted to see but better than expected. I adjusted the X axis gib using the Bridgeport method and settled on a reading of .002". Ran a backlash test and got .007". What a disappointment after having .002" of ball nut clearance. I backed off the gib and the best backlash I could get was .0035". Better but not good enough.

I wrote a warm up gcode program several months ago and decided to run it to see if it made any difference. Ran it for about twenty minutes and tested backlash. I got .0004" repeatable readings. Things are looking up! I also found that the factory gib adjustment screws were pushing the gib to the side. I cut 1/4" off the right end of the gib making it flat and square and made a bushing that slips over a stud instead of using a bolt, with an OD that just fits into the gib screw counterbore. This works much better. I was able further tighten the gib and maintain .0004" backlash. After a bit more gib adjusting I settled on .0008" of backlash.

So after about a week of testing, measuring, cussing, retesting and remeasuring the source of my excessive backlash appears to be bad gib adjustment screws and not warming up the machine prior to taking measurements. Lesson learned. I've still got some fine tuning to do.

Thanks to everyone that contributed their experiences. Without this forum I'd still be chasing backlash.

Tom S.

 

[JimDawson]

 

[MontanaAardvark]

What Jim said.

Before I took my mill apart, I don't recall actually measuring backlash, but I didn't feel any slop in the handwheels. I figured taking it apart would destroy that anyway. On the other hand, I think the gibs were farther in than they are now. I need to take a look at them in the morning. I think they may be loose.

I tested my mill on rapids and I can get 280 IPM on all three axes. Coming from a system where Z was 15 ipm and the fastest axis was about 50, it's amazing. If I gave up a little speed and improved backlash, that would be an acceptable trade.


Bob

 

[TomS]

What Jim said.

Before I took my mill apart, I don't recall actually measuring backlash, but I didn't feel any slop in the handwheels. I figured taking it apart would destroy that anyway. On the other hand, I think the gibs were farther in than they are now. I need to take a look at them in the morning. I think they may be loose.

I tested my mill on rapids and I can get 280 IPM on all three axes. Coming from a system where Z was 15 ipm and the fastest axis was about 50, it's amazing. If I gave up a little speed and improved backlash, that would be an acceptable trade.


Bob

I don't have handwheels on my mill so wasn't able to detect backlash that way. It showed up when I did a circular pocket. The hole was out-of-round about .006". I learned a valuable lesson that proper gib adjustment and warm up routine go hand-in-hand. You can't do just one and expect to get good results, at least on my machine.

I have my rapids set at 100 IPM. Tried it at 200 IPM and it looked like the machine was beating itself to death on short moves. It's what works for me.

Tom S.

 

[MontanaAardvark]

I don't have handwheels on my mill so wasn't able to detect backlash that way. It showed up when I did a circular pocket. The hole was out-of-round about .006". I learned a valuable lesson that proper gib adjustment and warm up routine go hand-in-hand. You can't do just one and expect to get good results, at least on my machine.

I have my rapids set at 100 IPM. Tried it at 200 IPM and it looked like the machine was beating itself to death on short moves. It's what works for me.

Tom S.

I agree that rapids are pretty meaningless because you can never cut at those speeds. They really only matter if you have to do something with a lot of movements above the work and your time is money. I did an array of holes on the front and read panels of my motor controller box. Drill, retract bit, move to next hole, repeat, a large number of holes, followed by a "carriage return, line feed" to the next row of holes. Rapids sped that job up nicely. (Rapids on my Sherline were 15 IPM, a far cry from these speeds).

Pardon if this is like a rant, but I'm an engineer and I always see things this way. I just don't see why cutting a circular cutout should be something backlash compensation can't fix. In fact, it should be a really easy case. Backlash only shows up when direction is changed, and in a circle, each axis changes direction twice at the most. The CNC controller should know everything it needs to know. It doesn't show up when the axis stops and resumes the same motion, or slows down and resumes. Torque in a stepper is maximum as lowest speed so if the motor just slows down or stops without reversing direction, backlash doesn't apply. It's not going to slip. If it does, that's a different problem.

I guess I just don't get it.

 

[JimDawson]

Computer generated backlash compensation is only partly effective and normally only works well on operations like hole patterns. Cutting a circle is where it doesn't work very well at all. The only true backlash comp is when you have a closed loop servo system, and the encoders are on the load (table). Even that is not 100% effective, because the table can float as the comp is applied. The only real path to no backlash is to have a tight mechanical system that is heavy enough not to have any measurable elasticity in normal operation.

Take a look at this thread on the Machsupport forum.
https://www.machsupport.com/forum/index.php?topic=12143.0