Ball Nut Holder - Manual Machining

[Bob La Londe]

Ball Nut Holder - I am sure somebody can tell me why this is a stupid idea, but I needed to get this ball nut off the ball screw without dumping the balls all over the floor. I don't know if the screw is salvageable or not. Its got a bend in it somewhere. If its just in the machining in the end I might be able to turn it down and machine some bushings. If the screw is bent in the main span its done. I might be able to straighten that, but it wouldn't be very accurate anymore. Normally guys screw the nut off onto a piece of cardboard tube, but I have bad luck with that. This hand made stub of aluminum "ballscrew" mates up with the end of the ballscrew I need to check, and I was able to just screw the nut off onto the stub. I had to hand grind a radius tool out of HSS in order to thread the holder. It took me a couple hours, but it worked out perfectly. Of course that time is wasted except for the learning experience if I can't save the screw.

Anyway, now I can roll the screw on my granite surface plate and determine where its bent.

[Planeman41]

Looks like you are getting no suggestions, so I will throw one in here. See if you can somehow inject some very heavy grease into the ball bearing with the hope that it will hold the balls in place when you disassemble it. Should this work, you can reinstall the bearing and wash out the grease with a solvent like acetone.

[danlb]

Looks good to me.

[Sparky_NY]

I have had many apart and just do it over a towel and let the balls fall out. The advantage is that it gives you a chance to clean out the internals of the nut real good as well as the balls. You wouldn't believe the junk that accumulates in there. They run much smoother clean, the smallest spec will be felt. Be advised, it is very common for ballnuts to have 2 size balls, about a thou different. That is so the nut spins easier. You have to sort the balls but it goes quick. Reloading the nut is pretty easy, again do it over a towel, the balls alternate small/large. There are youtube videos on the process.

It may be of interest that old screws that get excessive backlash is usually due to wear on the balls, they wear first. Reballing greatly reduces the backlash when the balls are worn.

Your mandrel should work, provided you align the treads perfectly with no gap, otherwise the balls will dump anyways. It may want to push your mandrel when transitioning off the screw onto the mandrel.

[Bob La Londe]

Your mandrel should work, provided you align the treads perfectly with no gap, otherwise the balls will dump anyways. It may want to push your mandrel when transitioning off the screw onto the mandrel.

Its already on the mandrel in the picture. The screw is in the background. Came off nice and easy. I did machine the holder with a little extra clearance on purpose for that reason. The screw just happens to have a 3/8-16 threaded hole in the end, and I just happened to drill a 3/8 hole all the way through my holder. That should make re-installation pretty easy too.

I was thinking about making one or two of these for the other size ball screws I have on machines in my shop too. 1605 long enough for a double nut certainly. 1204 maybe. Sometimes I get new screws and the nut is on backwards from the way I want to use it. A tool similar to this with a screw on instead of fixed end cap would make swapping them around dead easy. This is the only machine i have with 1-1/4 inch ball screws, so the fixed end is fine, but I could easily change it to a removable cap. The biggest headache is making the turning tool to cut the groove. I hand ground it and matched it up to a radius gage I happen to have a set of. Finding lathe radius inserts the right size would make the task go much faster.

[Sparky_NY]

Nice work.

After my first cnc retrofit I stopped using rolled screws and all the other retrofits used ebay used ground screws, some new, some used. I found I get much better performance that way, no backlash at all and no double nuts to fuss with not to mention very accurate pitch. The bridgeport I have now is a old boss 3 with servos and linuxcnc control. I reballed them, new support bearings and the backlash is difficult to measure, maybe a tenth or two. Had a lot of fun over the years doing retrofits.

Again, nice work on the threaded stubs.

[dian]

can you please explain why you were not able to screw the nut onto the appropriate tube? i mean, they come like that and you screw them on, right? did you check the balls. were they two sizes?

[J Tiers]

Nice work.

After my first cnc retrofit I stopped using rolled screws and all the other retrofits used ebay used ground screws, some new, some used. I found I get much better performance that way, no backlash at all and no double nuts to fuss with not to mention very accurate pitch. .....

What is the difference in spec and / or performance? I see mostly reasonably large errors per unit distance on most available screws that are not in the really high cost area. Presumably that is the difference between ground and rolled.

[dian]

your on the right track. ground usually is c0-c5 while rolled would be c7-c10. i also see a t and a p. probabl depends on manuf.

[Bob La Londe]

can you please explain why you were not able to screw the nut onto the appropriate tube? i mean, they come like that and you screw them on, right? did you check the balls. were they two sizes?

Well number one and I mentioned this although not so strongly. I hate tubes. Some guys use them and they have great luck. I have not. Ever. Second, I would have had to make a tube, and a tube would have needed to have ateast a flange anyway, because it won't be sitting there for ten seconds. It might be sitting there for a few days... and lastly I wanted to do it just to annoy you. LOL.

[Bob La Londe]

The screw doesn't seem to be significantly bent. I got maybe 0.00075 variance from one end to the other by resting the screw on a granite surface plate and passing an indicator over the top of the thread. I checked on multiple lines along the screw. Then to double check myself I grabbed a piece of .002 thin shim stock and tried jamming it under the screw all along the screw at 4 different rotational positions of the screw. I know the indicator is better, but I can get a feel with the shim stock. A sort of confirmation.

Then I ran the indicator over to find the high point of each turned step/shoulder. The results were about the same. I double checked myself by using .001 difference gage block stacks as improvised go-nogo gages to slide under the turned surfaces. I confirmed it's within .001 all the way around.

I'm at a loss.

The thing is I could see the screw bind and release before it was removed, and I could feel it bind and release while I was moving the machine back and forth by hand. I'm totally at a loss.

The only thing left I think is to go ahead and dump the balls out of the ball nut and check it out. The thing is it felt perfectly fine once the screw was out of the machine. Maybe it has alternating ball sizes and at some point in the past somebody packed the balls without alternating them creating the same affect as a bent screw. I think that is a shot in the dark, but I don't think it can hurt anything to check.

I'm just guessing now.

[Sparky_NY]

What is the difference in spec and / or performance? I see mostly reasonably large errors per unit distance on most available screws that are not in the really high cost area. Presumably that is the difference between ground and rolled.

Yes, pitch accuracy vs distance is one difference BUT the big real world difference is backlash. Rolled screws, with a single nut, .003 is not unusual at all. Double nuts can reduce it considerably but take up a lot of room and still the backlash can increase under high loads because a bellville washer is typical used to preload the double nuts. If you put too much tension on the double nuts, the screw binds up due to pitch variances. Ground screws, due to the accuracy are typically preloaded. Backlash is for all purposes non existent in ground screws (fraction of a tenth). Backash creates a lot of problems on cnc machines.

Another way of looking at it is comparing it to general purpose acme screw stock vs precision grade intended for machine tools. Very different beasts. Low end commercial cnc machines often have rolled screws. I have seen lathes, selling for $15K, that spec positioning accuracy at .001 !!! Naturally, that .001 is .002 on the diameter !! Those same class machines NEVER spec backlash at all !

I used rotron rolled screws on my very first retrofit, never again. The latest class of rolled screws, from china, via ebay are not bad for the price. If being used on a router or a tabletop type mill/lathe they are probably plenty good enough due to the machines limitations.

Here is a cut/paste from a THK ballscew spec "Accuracy grades C0 to C5 are defined in the linearity and the directional property, and C7 to C10 in the travel distance error in relation to 300 mm. " SO... rolled grades the error is spec'd over 300mm, ground screws its for the TOTAL length. Here is the doc for some light reading, there is a chart for the accuracy grades. https://tech.thk.com/en/products/pdf/en_a15_011.pdf
Again, backlash is the real issue for DIY users, not accuracy.

[Sparky_NY]

The screw doesn't seem to be significantly bent. I got maybe 0.00075 variance from one end to the other by resting the screw on a granite surface plate and passing an indicator over the top of the thread. I checked on multiple lines along the screw. Then to double check myself I grabbed a piece of .002 thin shim stock and tried jamming it under the screw all along the screw at 4 different rotational positions of the screw. I know the indicator is better, but I can get a feel with the shim stock. A sort of confirmation.

Then I ran the indicator over to find the high point of each turned step/shoulder. The results were about the same. I double checked myself by using .001 difference gage block stacks as improvised go-nogo gages to slide under the turned surfaces. I confirmed it's within .001 all the way around.

I'm at a loss.

The thing is I could see the screw bind and release before it was removed, and I could feel it bind and release while I was moving the machine back and forth by hand. I'm totally at a loss.

The only thing left I think is to go ahead and dump the balls out of the ball nut and check it out. The thing is it felt perfectly fine once the screw was out of the machine. Maybe it has alternating ball sizes and at some point in the past somebody packed the balls without alternating them creating the same affect as a bent screw. I think that is a shot in the dark, but I don't think it can hurt anything to check.

I'm just guessing now.

Alternating ball sizes put in incorrectly will not make the screw bind. The binding you felt could be junk inside the ballnut, due to the close clearances it only takes tiny specs to create binding. Another possibility is your end support bearings. Excessive preload on non precision end bearings will create binding, as will any angular misalignment. REAL ballscrew support bearings are ABEC 7 precision bearings, and heavily preloaded. DIY retrofits often use china standard grade angular contact bearings in the support housings. Those have slop and many cut shims and such to preload them (take out the slop), to much preload and they will bind up.

The concept of alternating ball sizes is this. Only half the balls actually carry any load, the smaller ones function as spacers. Only half load carrying still is plenty to take huge loads for their size as the specs show. IF they used ALL full size balls the screw would be much harder to turn. Ease of turning is the primary reason for the alternating ball sizes. Alternating ball sizes is the norm for ground screws, not often seen in rolled screws. In a alternating size scheme, the smaller balls are typically about ,002 smaller.

[Sparky_NY]

How about some more details on the binding problem? Is this a new build of some sort or a repair job because of the binding? Is it a new screw?

[J Tiers]

LOL.... Roton......

I used to work about a block from them, didn't know it then.

If I recall, it was hard to find any specs, and when I did, they were not great, no better than a McMaster expensive precision Acme, which is about 7 thou error per foot. While that is actually not bad, it sure is not very encouraging on a travel of 2 or 3 feet.