Threadless screw LISA

[see3d]

I like the LISA concept. It can be REPRAP or REPSTRAP, but it is really the small number of moving parts and linear print area in the Z like other popular deltas that give it extra appeal. The one thing that seems to be a drawback is the high cost of the quality threaded rods and nuts. There is also that thing about sliding motions.

The threadless screw idea previously put forth would seem to be an answer to the cost hit, and no sliding motions. I am sure there are some downsides to the practical implementation, and many questions, like:
1) How large a thread "pitch" is practical on an 8mm rod?
2) Can 3 ball bearings be the whole bearing solution for each rod?
3) How much variation would there be in "thread pitch" with practical tolerances?
4) Could each of 3 rods be custom calibrated to its natural "thread pitch"?
5) How much cost would it really save?

~Dennis

[Nicholas Seward]

1) I think a practical limit on bearing tilt would be 30-45 degrees. When you work out the pitch it comes out to be 15-25mm.
2) You need two sets of 3 per rod separated by a bit. You can use 608's. That $1.50 per rod.
3) I would imagine that the pitch would be fairly constant along the whole rod. I would like to test this but I don't know why this wouldn't be the case.
4) No sure what you mean but each rod will have a slightly different pitch. Calibration is key.
5) I am about to buy 7/16" ROTON Torqsplines and nuts for a total of $180. (Note: GUS's complete mechanicals are under $40.) With these shafts the cost will only be $50.

I am going to go with the ROTON screws for now to verify the design with something I am 99% sure of. I will then probably build a threadless one just to play around.

My only concern is the repeatability/drift. Several people have assured me that there is no drift so I am getting less and less concerned. However, this just isn't in my toolbox yet. I have seen people use them for a z axis on a Mendel style bot. I haven't seen a whole bot converted over to them. That is understandable but there is a lack of hard data. When I build one the first thing I will do is drive it up and down 1000 times and then measure the drift.

[see3d]

4) No sure what you mean but each rod will have a slightly different pitch. Calibration is key.

What I mean is that the amount of bearing skew which defines the pitch will have a build tolerance. Each rod will have a slightly different pitch associated with it. It will be a constant pitch for the whole length of the rod. I would expect that each rod would have to be independently calibrated for its unique pitch (number of steps per total travel), in addition to all the other calibrations that would have to be done on a threaded LISA.

My concern is if two sets of 3 bearings are required, tolerance variations between them might cause a thread pitch fight. Perhaps just a spring load between bearing pairs could take up any slight pitch variation.

[Nicholas Seward]

How about we make six sets and then measure all the pitches. We then pair them up with the closest one.

[owens-bill]

1) I have seen people use them for a z axis on a Mendel style bot. I haven't seen a whole bot converted over to them. That is understandable but there is a lack of hard data.

I'm aware of one such project: http://www.youtube.com/watch?v=l25R_nUwyJM RifRaf discussed it on the #reprap IRC channel when he was working on the project, and it appeared to be successful, but then seemed to indicate that he was moving on to other things and wasn't developing it further.

[Ralphxyz]

And for those of us that wouldn't recognize a threadless screw if we saw one on the street.
http://www.amacoil.com/rs-features.html

Very impressed, what's the cost difference?

Ralph

[Nicholas Seward]

From their literature...

The maximum difference in pitch for the two directions of travel is 2 %. We therefore recommend the use of positional sensors for positioning applications.

This is why I hesitate.

[Jason Fisher]

Could you maintain alignment by taking advantage of the slipping nature of threadless? Spacers at the top of each rod (for set clearance) that you purposefully bump the arms into, and enough torque to drive an arm to top and break friction to spin the rod freely? Then you could just drive all three arms a bit beyond top, and the arms that hit early will spin freely until the others catch up?

[Nicholas Seward]

@Jason Fisher It depends on how good the tracking is with the screws. I imagine if we had to repeat this procedure every couple of layers that it would be bad on the machine.

[Jason Fisher]

A magnetic levitating drive is probably the holy grail for this arrangement. Stepperless.

http://www.youtube.com/watch?v=45-9al2hig0
http://www.youtube.com/watch?v=WRFUkME0Pz8