1.7*0.74*0.4m Mill / Router building...

[routercnc]

Hi Jonathan,

Interesting idea, but it's got me a bit puzzled. If the thread rotates, and the bearing rotates with it, how does it progress along the thread? For a nut in a thread to progress, one part has to rotate whilst the other stays still. Yet in the video the bearing inner and the thread are both rotating. Is it perhaps vibrating it's way along? If so I'm not sure this will give reliable position control.

The only way I can see this working is if you removed the inner race (somehow) to create a pseudo ballnut. Or maybe I'm missing something here so please enlighten me!

[Jonathan]

]
Interesting idea, but it's got me a bit puzzled. If the thread rotates, and the bearing rotates with it, how does it progress along the thread?

Er...it just does!
The bearing is rolling on the thread properly, not vibrating. It's difficult to explain and visualise. Try just getting a random bearing and screw and fiddle with it, you'll soon see how it works!
If you imagine the inner ring of the bearing was stationary then clearly it would still work...just like a normal nut. The inner ring rotating makes no difference to this, except of course giving rolling friction.

Sorry I can't really explain it any better!

One slight issue I can see with this is the bearing applies torque to the screw, trying to bend it. This could be solved with having two bearings angled opposite to each other but I think M16 is strong enough for this not to be an issue.

I'm in the workshop at the moment..lots of tapping, joy!

[Wobblybootie]

Ummm like this???

[routercnc]

Mulled it over a bit more and managing to visualise this now. Yes, it does work. Nice idea, and worth a look compared to the cost of ballscrews.

[Jonathan]

Ummm like this???

Yes, exactly. I'm sticking with one bearing for now though as I think it will be ok.

[Jonathan]

Mulled it over a bit more and managing to visualise this now. Yes, it does work. Nice idea, and worth a look compared to the cost of ballscrews.

Certainly worth a try. £8 for a meter of M16 threaded rod, plus a few quid for the bearing compares very nicely to the price of a ballscrew! Even a *standard* ballnut has some backlash whereas I can't really see how this could have any? Granted the pitch of the thread being slightly uneven with threaded rod isn't ideal ' but that could be solved by using ACME.

[routercnc]

I think there will be a torque limit (from the stepper motor) which can be applied to the leadscrew, otherwise the bearing will want to jump over the teeth. Depends on the spring preload holding it against the teeth - higher is better for this, but then more bending load on the leadscrew which is less good. This effect will probably be the determining factor in the success of this idea, but good thinking!

Better tooth engagement from a brass insert, as you suggest, might help although there will still be a limit - you'll have to see how it goes. You might be able to arrange a locking up nut so that when the spring tension is developed you can lock it off at that position. Of course this won't compensate for small variations along the thread, but might be a fallback position if it keeps jumping.

[Jonathan]

I can see bearing jumping could be a big issue with a long screw since for the bearing to jump the screw would need to bend a bit...

Locking nut on the spring is a good plan. Just thought instead of using a standard spring I could use a couple of belleville washers since the travel required is tiny. Can also adjust the spring constant by stacking the washers different ways etc...

Back to the main build.

I went into school today. Their laser cutter had arrived over the summer...it's huge! Only 80W but still useful I guess.

I milled the 600x160x20mm Z axis plate square and drilled and tapped a couple of holes to bolt the rotary table to it for when I want a 4th/5th axis, not that that's very likely since I can't afford the software.

I also cut and milled the steel gantry sides square which took much longer than anticipated. It's not a nice grade of steel to mill.... Started milling the surface flat on either end to make a better contact between the gantry cross piece and this. I'm putting a piece of 20mm aluminium plate in both ends of the 80x80 box section to make a good strong joint.

Photos coming soon!

[routercnc]

Hi Jonathan,

I'd need to draw it out and look at the angles and forces etc. , but I suspect that the bearing jumping would occur on any length of leadscrew. If the bearing is mounted in a sort of gimble as you have shown and held against the screw thread via a spring, then if you push the leadscrew it will try to rotate (not spin) the bearing about the gimble pivot axis against the spring force, and ratchet the bearing over the screw threads. A bit like a handbrake ratchet on a car where the pawl runs over the teeth against the pressure of a spring.
If you zoom in close to the exact point where the bearing contacts the thread you'd see an inclined plane sliding under the angled face of the bearing. When the inclined plane of the thread moves forward (either you pushing it or the action of the stepper rotating the leadscrew) it will push the bearing upward and away. If it pushes the bearing far enough away it will jump a thread.

Anyway, I was thinking of a way around this and was thinking about a setscrew with a small rubber buffer on the end, but your belleville washer is similar and neater. Hope it all works out because it's a nice idea and will save some cash. I went with 3 ballscrews for the fit-and-forget method, but it was not cheap.

Look forward to your photos . . .

[irving2008]

There's a big thread regarding this idea on CNCZone... seems some have used it successfully and theres a commercial offering based on the idea.