[outsider787]

In the process of sourcing parts for changing to ball screws on my BF20 machine.
I would like to go with a double nut setup for all the axis, in order to eliminate/minimize backlash on the system.
As many have experienced, Chai (linearmotionbearings2008 on ebay) has pretty good prices on C7 ball screws and ball nuts and I was going to try and use the parts he has available for my conversion.

I would use two ball nuts, and attach them together (at the flange) between some belleville washers. As the flanges get tightened together, I would get a preload between the nuts.
For the Y and Z axis this is relatively simple, as there's lots of room to put 2 nuts (flange to flange).
The X axis seems a little more problematic due to clearances and the screws that attach the nut to the machine.

Has anyone attempted putting 2 flanged ball nuts on a G0704 type machine? Would like to see how others have mounted two nuts on the x axis.

[Fastest1]

I haven't seen anyone's pictures of the X double nutted. I would think only one of the flanges would be stabilized and the other held in place by the bellevilles. It would seem you could use the original pinch bolts except thread them into one of the flanges. If the slot in the carriage was widened to accommodate both flanges and the bellevilles, it might work.

[outsider787]

I haven't seen anyone's pictures of the X double nutted. I would think only one of the flanges would be stabilized and the other held in place by the bellevilles. It would seem you could use the original pinch bolts except thread them into one of the flanges. If the slot in the carriage was widened to accommodate both flanges and the bellevilles, it might work.

In order to get these nuts on the machine, I need to pocket the saddle towards the center of the saddle. If I put 2 nuts flange to flange (and assuming I widen the slot in the saddle/carriage to fit both flanges) that means I need to pocket towards the pinch bolts.
The problem I foresee is that this extra pocketing for the second nut would likely wipe out/interfere with the threads of the pinch bolts as the bolts are not that far beneath the surface.

[rrobb]

Check out Ryan's G0704 build.

I think he used a double nut from linearmotionbearings.

[outsider787]

I just checked Ryan's thread. He did use double ball nuts everywhere.
Though to fit the X axis nuts under the table, he machined the outside of the table. Unfortunately I don't have that luxury.

Will have to think of something else.

adrian.

[Fastest1]

In actuality I probably did see Ryan's use of double nuts. Just thinking about the dilemma. What about if you made adapter plates for each side of the carriage on the X axis? The flanges mounted outboard pointed toward the center. One side fixed and the other floating on bellevilles. You wouldn't lose much if any travel. The lube ports of the ball nut would be accessible. Damn it is starting to sound better the more I think of it. Though I am totally impressed with the accuracy of the single nuts. Maybe as I get better extreme accuracy might be important. Till then .003-.005 won't make any difference in the outcome. Most likely if those tolerances are that critical, there is a design flaw LOL.

[outsider787]

Fastest,
I think I understand your idea. With the ball screw above the carriage, one nut is fixed and the other (on the other side) is used to adjust the preload.
So the preload will be applying a bending force to the shaft since the nuts will only be restrained on one side (the side below the screw) , right?

adrian.

[Fastest1]

Fastest,
I think I understand your idea. With the ball screw above the carriage, one nut is fixed and the other (on the other side) is used to adjust the preload.
So the preload will be applying a bending force to the shaft since the nuts will only be restrained on one side (the side below the screw) , right?

adrian.

Not really. I was thinking the adapter plates would mount to the carriage similar to the way the X axis end plates mount. By mounting the plates on each side of the carriage you could pocket for the flange in one side, bolt it in firmly. The opposite side would have small bellevilles (or maybe some large od ones around the flanges smaller tubular part (where the balls actually recirculate) and would preload by pulling or pushing the screw in a linear plane parallel with the length of the screw. Whether pushing or pulling would be dependent on which side of the adapter plate the belleville/s were on. My opinion would be that pulling the screw under tension would be better than compressing it. Not that I am an engineer or anything. This would require a small trench or relief go across the carriage at least as long as the tubes of the flanges are. My imagination says the orginal pinch bolts would be discarded or unused. I will have to play with the idea in my head. Usually the first solution isnt the easiest LOL. I will have to reapply the KISS principle. In fact the more I think about it, if you pocketed the floating side to accomdate the flats of the flange it would eliminate the possibility of rotation and only allow the flange to move in or out on the bellevilles. This would have to work better than just depending on the strength of 6 bolts resisting rotation with bellevilles spacing it out.

[Fastest1]

Just imagine it like this but below the table obviously. It would require some machining of the carriage, from what I can see, mostly just squaring up the sides to allow the plates to mount. The plates would also be a great place for a limit switch, under the table and out of the way.

[Fastest1]

If you pointed the tubular part of the flange outwards you could get away with less carriage machining. Then for max travel, the bearings on the outer adapter plates would need to be moved outward on a tubular standoff much like steppers are mounted. This would create a pocket for the flange to recede into. It would require a longer ball screw. I believe Y could be done in a similar if not easier way.

[outsider787]

I understand your idea now and it may work.
Just wondering how difficult it may be to apply the preload once the table is on top of the saddle.

adrian.

[Fastest1]

I was thinking the same thing. You would want the belleville on the outside of the plate so as it is preloaded the screw is being pulled. You could install the plate and flange on 1 side (without the table on) and unwind or wind the opposite plate to preload the belleviles, then attach it. The table is then slid on and the end plates are swiveled around and mounted. Btw I use BD Tools mounts, I simply remove the 2 bolts at each end of the X axis (and dowel pins) let the plates rotate around 180 degrees to their normal mounted position and slide the table off without ever disconnecting the ballscrews or steppers. I do keep the stepper closest to the carriage when doing this as to minimize the leverage on the ball screw. After that inner preload was adjusted, I would remount just like that. The more I think about it, there are ways to make it adjustable from the accessible side after being mounted.

[691175002]

Just some thoughts regarding backlash elimination.

In my experience, effectively running double nuts requires some pretty hefty preload.

If you have a vice in the middle of the table and you are running at the very ends of travel, I wouldn't be suprised if the screw needs 100lb+ of force to move the table. Throw in cutting forces and add a safety margin and you need 200+Lb of belleviles on the nuts. Make that 400lb+ for the Z axis.

Might just have been my screws, but under that much load they start to feel "grainy". I'd imagine that a hobby machine will never wear out the screws, but it is something to consider.


Reducing backlash is a never ending game. Each time you fix a source of backlash, another one pops up.

Other big sources of backlash include the motor couplers and the fixed bearings that hold the end of the ballscrew. Even the ballscrew itself acts a bit like a spring, my X axis has a measurable increase in backlash when on the side away from the fixed end because the extra screw length has windup.

When you are measuring 0.001" or 0.0001" increments, literally everything bends under load.

[109jb]

I didn't use Chinese ball screws, but I did use double nuts on all axes. I used Thompson ball screws and nuts from McMaster Carr. I have run the machine enough to know it works very well. Like Ryan said though, you need to run the preload up there. I am running about 130 pounds or so. You can see how I did it in my build thread at the link below

http://www.cnczone.com/forums/bencht...lans_soon.html

[outsider787]

I realize that the machine will bend and flex under load and/or accelerations with some mass on the table and I realize that .001" is quite a small distance.
My goal with doing the double nut is to eliminate as much play as I can out of this one (the screw/nut) component.
I gotta start somewhere...
Besides, the cost is pretty minimal for a second nut per axis, and outside of the x axis, implementation of the second nut seems trivial in the y and z direction.

adrian.

[outsider787]

Fastest,
I was throwing some ideas around on how to mount the 2 nuts.
This is similar to your concept (each nut on the outside).
Each nut would only be supported by 2 bolts (10-32 bolts), but that should't be a problem.
One nut would be hard mounted, while the other would get the the belleville washers to adjust the preload.
I'd make the support bar out of mild steel, as the threads in aluminum may not hold both the preload and usage load..

It should be straight forward to adjust the preload with the table off, then slide the table on the saddle, and connect the ball screw to the end plates of the table.

[Fastest1]

That would work. The only difference in what I was thinking is that the mounting plates would have been perpendicular to the screw. Yours is parallel. In fact yours would require much less fabrication than my thought. You should be able to adjust the preload with the table on with just a little planning.

[zamazz]

I may just borrow this idea on my G0704, with a couple of tweaks. I'd like to machine both rails in a single piece and do as little machining to the saddle as possible. In your design, is the ballscrew pretty much resting on the surface of the saddle?

[outsider787]

Zamazz,
Yes, the ball screw is just above the surface of the saddle. There is very little room between the bottom of the saddle and the underside of the table.

Pretty much all ball screw/ball nut conversions involve machining the surface of the saddle in some way.

Mind you this "machining" of the saddle doesn't have to be pretty. Nothing actually rests on that surface. It's only there for clearance purposes.

adrian.

[outsider787]

As for making the support rails out of one piece; I had thought of that, however with the ball screw running between the rails you're left with very little material holding the rails together.

It would have been nice to have one piece support both ball nuts, since alignment would be dictated by the machining, but I think this will actually be easier.
I'll let you know how I make out once I put this in practice.

adrian.