[lerman]

As I understand it, the major downside to using rolled ballscrews is that variations in pitch means that you can't eliminate a lot of the backlash by preloading a pair of ballnuts.

Has anyone tried to correct the variation by making a nut to fit out of some soft metal, say brass, and lapping it against the ballscrew with abrasive? It seems to me that we should be able to make the threads uniform to within a fraction of a tenth or so. Of course, that would just fix the variation, but software could correct for any constant error in pitch.

I was thinking that I could build a little fixture with a small motor that would drive the brass nut back and forth over the ballscrew.

Ken

[HuFlungDung]

A good lap needs to be a perfect mate for the surface it is supposed to correct. You would probably have more error in your home made nut, than exists in the ballscrew to start with

You'd need to cast the lap "in place" in order to get a good replica. The lap should also be quite long, in order that it can average out surface errors by many points of contact.

Circulation of the abrasive is also important: a lap is a cutting tool and needs to have a place for the swarf to go. Otherwise, it plugs up in about 10 seconds

Ideally, the lap should run right off both ends of the screw, to insure that you don't end up with a loose fit in the middle, and tight at the ends.

Also, if you can reverse the lap (I'm imagining it in half-shell shape), you will improve the symmetry of the groove.

There is no guarantee that you won't spoil the shape of the groove somewhat. In the first moments of use, a lap becomes impregnated with abrasive (because its soft) and it ceases to wear as quickly as the screw. Grinding wheels have the same property: they are bonded together with a relatively soft matrix, and its the particles that make them seem hard.

If you cast the lap, you should use some kind of a "disposable plastic film" to space it away from the screw by a few thousandths. This is because the abrasive mixture should be of uniform thickness, and the plastic film would help allow for this to occur. In reality, you want the lap to be offset slightly from the ball track: if the ball track has a.0625" radius, then, you would allow .002 for abrasive, and want the lap to have .0605" radius. This will help ensure even wear of the screw. Any high pressure points will be the first to go, and if your lap does not fit properly, you'll just be ruining the shape of the groove.

Perhaps it would be possible to rig up a grease gun (or something) to inject the unhardened casting material into the mold, forcing the plastic film tightly against the screw while the lap hardens. Then, peel the plastic film off and start lapping.

[lerman]

A good lap needs to be a perfect mate for the surface it is supposed to correct. You would probably have more error in your home made nut, than exists in the ballscrew to start with

You'd need to cast the lap "in place" in order to get a good replica. The lap should also be quite long, in order that it can average out surface errors by many points of contact.

How long is quite long? I was thinking something like 15 times the pitch (3 inches for a 5 pitch screw). -- I'm assuming a single start screw.

Circulation of the abrasive is also important: a lap is a cutting tool and needs to have a place for the swarf to go. Otherwise, it plugs up in about 10 seconds

Ideally, the lap should run right off both ends of the screw, to insure that you don't end up with a loose fit in the middle, and tight at the ends.

Also, if you can reverse the lap (I'm imagining it in half-shell shape), you will improve the symmetry of the groove.

There is no guarantee that you won't spoil the shape of the groove somewhat. In the first moments of use, a lap becomes impregnated with abrasive (because its soft) and it ceases to wear as quickly as the screw. Grinding wheels have the same property: they are bonded together with a relatively soft matrix, and its the particles that make them seem hard.

If you cast the lap, you should use some kind of a "disposable plastic film" to space it away from the screw by a few thousandths. This is because the abrasive mixture should be of uniform thickness, and the plastic film would help allow for this to occur. In reality, you want the lap to be offset slightly from the ball track: if the ball track has a.0625" radius, then, you would allow .002 for abrasive, and want the lap to have .0605" radius. This will help ensure even wear of the screw. Any high pressure points will be the first to go, and if your lap does not fit properly, you'll just be ruining the shape of the groove.

I assume you mean .0623 -- or did you really mean to leave .020 for abrasive?

Perhaps it would be possible to rig up a grease gun (or something) to inject the unhardened casting material into the mold, forcing the plastic film tightly against the screw while the lap hardens. Then, peel the plastic film off and start lapping.

I was thinking of perhaps casting a nut out of brass or aluminum (or possibly lead). Slit the length of it and open it up to be of the proper "looseness". Drill some holes to provide space to "inject" the abrasive.

I figured I could start with a screw that is over length and discard the ends. They'd probably be turned for bearings, anyway.

Ken

[HuFlungDung]

Check my figures: .0625 -.002 = .0605"

Casting a metal lap is going to give you the problem with excessive heat shrinkage. Actually, autobody filler makes a pretty decent casting compound. It seems to be fairly stable in reproducing a surface quite accurately (otherwise, it would peel off when it hardened, I guess). You could even premix the abrasive with the filler compound before you cast it.

I'd recommend a half-shell lap, which you can apply by hand, whilst powering the screw with some other means. This way, you can easily lift the lap, wash it, turn it end for end, apply new abrasive, and continue.

[lerman]

Check my figures: .0625 -.002 = .0605"

Casting a metal lap is going to give you the problem with excessive heat shrinkage. Actually, autobody filler makes a pretty decent casting compound. It seems to be fairly stable in reproducing a surface quite accurately (otherwise, it would peel off when it hardened, I guess). You could even premix the abrasive with the filler compound before you cast it.

Sorry about that -- I was thinking .625 radius (as in for an inch an a quarter leader screw).

You are correct.

Ken

[HuFlungDung]

I was assuming a 1/8" ball was being used. Hence, the radius of the groove has nothing to do with the radius of the screw.

[sendkeys]

You could even premix the abrasive with the filler compound before you cast it.
.

very cool idea

[lerman]

So, that sounds like a plan.

Some followup questions:

What grade(s) abrasive should I use? I assume that if change grades, I would have to make a new lap. I assume I would look for an even abrasive pattern on the screw in between grades. Would it make sense to put some Dykem on the screw and look at that?

Then, when it is done, how can I check the precision of the screw without having a more precise long machine to check it against? Of course, I could build a laser interferometer, but I'm sure that would introduce a set of problems that would make building a ballscrew look like child's play.

Ken

[balsaman]

I doubt you could lap the screw to make it better than the .004 per foot that it comes out of the factory as....

Eric

[HuFlungDung]

That's true, Eric. I've never tried to re-pack one of these rolled ballscrews to see how irregular they might be. However, if someone wants to try to set preload with opposing ball nuts, it might work well enough (given enough patience ) to make a screw that won't have to run so loose in order to not be too tight in the bad spots.

[Torsten]

I would not advise this your chances of making things worse are greater then what you may gain.
A Ballscrew rated at 0.004" per foot accuracy has a 99.999% chance of been better
then this, because the manufacturer states that this is the accuracy he has
set for himself to beat consistently.
What you actualy end up with on your machine will depend on many variables, some
of them do cancel themself out.
It also will depend on the specific cut of screw you happen to end up with.
I would use it as is then calibrate your machine with some known samples and
decide from there if and what needs attention.
Good Luck