Very clever design Mike!

With grooved rollers slippage causes no harm.
How is the tap butchered anyway?

By "butchered", I mean the tap is relieved by grinding so that you can use it as a perfect grooving tool that does all the grooves at once in the bushings. They are circular grooves on this one, not helical.
If you think about the tooth profile of the groove, it has to be altered when one diameter thread must lay in another diameter with the same pitch. The pitch angles are different.
Using the tap as a groove cutter automatically makes this adjustment for you, which is way cool!

I just made steel threaded bushings (1/4-32) and pressed them into the ID of 608ZZ bearings (skateboard bearings). I've got drawings somewhere of heavier, more practically applicable ones, but haven't found it yet.


Too bad !!! you can believe that many guy are looking for something like that.
to be honest i just cannot make it working in my head and this is a good reason to make a prototype because your video show it: IT WORK !!!
One question the ratio diameter of the axe/diameter of the nuts is always the same ? because for a small axe it make a big nuts and i fear the external size of the nuts for a 20mm screw (0.8 inches) this might be a severe limitation

Not sure I understand your question well...
The design I posted below has a .75" screw, and the nut assembly is 50mm square.

For the differential screw, the closer the nut diameter is to the screw diameter, the finer the effective pitch gets.

(chair)Very sorry i just look at the post 298 and i find out that it was exactly what i was lookimg for
so focus to understand how it work i did not look at the dimensions:withstupi

Brilliant Mike, so bloody simple.
I like the method you advocate for cutting the grooves, beats single point and pitching with a boring bar.

This has got to be so doable, I reckon it will suffice for all the transmission jobs where pitch accuracy is important and outside dimensions are not.

I'm going to have a look at making it for the cross feed bronze nut replacement on my lathe which is just about worn out now, and I reckon the outside dimensions would probably fit in if I do a bit of surgery.

However due to the bearing sizes required, around a 12mm crossfeed screw, I might try and adapt your design by using needle bearings with the threaded rollers being the inner races but case hardened on the outside.

Looking at your drawing I see the two end bearings with the internal threads would have to have the grooves located precisely to ensure they are both in the thread and bearing on both flanks simultaneously, while the middle bearing "could" have some endways float to position itself in the thread to eliminate any backlash.

I like the fact that the female thread on the male thread naturally centralises the three bearings round the screw when depthed.

I'm going back to have a look at the drawing to see how the assembly is made to get the three bearings to depth correctly on the screw.

There is a way that I'm thinking of to use this for the half nut on a lathe leadscrew, but it entails having the centre bearing a sliding action to get the three to part for disengagement, exciting prospect.
Ian.

Thanks, Ian!
With regard to needle bearings, when you actually get the ones that can take some axial load (which they do not without thrust bearing elements), you end up with a more expensive and just as large OD. Actually, not recent information of mine, really, I've just been there.

As with so many things, small od will be a matter of a little more money on thin section bearings. If I were producing these, which I'm contemplating, I'd use dual row bearings. First out, I'd use deep groove radial single bearings as shown and see if single is adequate.
As to preloading, I've been thinking of adding features so that Belleville washers to provide the spring force, center radially against ends. Currently, it's a three finger squeeze and tighten the bolts that hold the whole stack together. You and I will quickly find how consistent the screw thread is! Last night I thought on this, and figured I'd cycle the snot out of it full travel with jewelers rouge all over the screw to lap it in, then clean and replace that $20 set of three bearings. Don't know if that will work.

I think you needn't get too worried about the pitch tolerance on the leading and trailing grooves. The worst case is that they ride up a bit on the outer or inner two thread faces, and the center one seats fully. That's actually a better condition (nice that it's almost unavoidable simply!).
As to half nut use, yeah! I've been looking at my brand new lathe and pondering it, but won't. ha I did notice that my half nut operates like a clamshell. You're right, though, perfect for that. That's the fun part, making the simplest handle actuator that centers the bushings, disengaging the nut.
Faaaaascinating.

hello everybody,
Just visit the Mike Evermann homepage (I am sorry for making professionnal advertissement ) BUT this guy is a genius !!

Thank you, Rokag! You're giving me a large head. ;-)
I do prefer to make products that are not available anywhere else!

Great work Mike, that's really clever!
What do you think about making the "ring grooves" threaded, but in reverse? It would make the contact area between the rings and the screw bigger.

Great work Mike, that's really clever!
What do you think about making the "ring grooves" threaded, but in reverse? It would make the contact area between the rings and the screw bigger.
Thanks!
Do you mean Left Hand thread? If so, there can be no thread engagement, at least if the roller and screw centerlines are parallel..

Thanks!
Do you mean Left Hand thread? If so, there can be no thread engagement, at least if the roller and screw centerlines are parallel..

Ah well, of course. The ring is not outside threaded as a planetary roller. But how about making it threaded as usual then, as a right hand thread?

Ah well, of course. The ring is not outside threaded as a planetary roller. But how about making it threaded as usual then, as a right hand thread?
Sven,
If you look down a few posts, there is discussion and a video of me showing that very thing.
What you get is a differential speed reduction that is approximately:
(Ring thread pitch dia - Screw thread pitch dia) / Screw thread pitch dia
In the video, that one had a 1:3, so was effectively 100 threads per inch, running on 32!
So in short, yes, it works and is a good thing (with some details to be sorted out!)

Great work Mike, that's really clever!
What do you think about making the "ring grooves" threaded, but in reverse? It would make the contact area between the rings and the screw bigger.

I think position repeatability would be again lost due to slippage.

By "butchered", I mean the tap is relieved by grinding so that you can use it as a perfect grooving tool that does all the grooves at once in the bushings. They are circular grooves on this one, not helical.
If you think about the tooth profile of the groove, it has to be altered when one diameter thread must lay in another diameter with the same pitch. The pitch angles are different.
Using the tap as a groove cutter automatically makes this adjustment for you, which is way cool!

can someone translate it in digest english because it's not very clear to me this is what i understand
you take the red part (the tap)in the lathe
make a hole in the center
install the tools to make the filet with 10 teeth inside the hole
do not move on x axis but only in y axes
the misuse of the tool will produce the perfect profile needed
this primitive syntax express a corect semantic ?

For the differential screw, the closer the nut diameter is to the screw diameter, the finer the effective pitch gets.

True, the Pitch circle (invisible circle inbetween the mayor and minor diameters) of the smaller screw relates to the Pitch circle of the larger screw.
One revolution of the small screw would advance 1/tpi.
At this point the pitch circle of the larger screw has only partially rolled off, so it has advanced somewhat less then 1/tpi.
The difference is the effective tpi of the combination of the two.
If both Pitch diameters are known this could be calculated I think.

Hi Rokag, As I see it you require a pitched set of grooves in the bore of the bearing, the same as the screw, but NOT A THREAD.

The bore in the bearings requires grooves at the same pitch as the thread going through it, and the easiest way to get the pitch right is to use a tap of the same pitch as a form tool, but smaller in diameter, and plunge the teeth into the inside face of the bore, this gives you grooves but not a thread, because the tap can't pull itself forward.

If you don't use this method you will have to use a single point tool in a boring bar and move the tool point forward at the pitch rate for each groove, much easier to use a form tool.

The tap usually has three cutting faces, (sometimes four), and you will have to grind the teeth off of two of the cutting faces to leave an oval "D" shaped tap with one cutting face.

Then you just bring the cutting face with teeth up to centre height and plunge it into the bore of the bearing insert to get the grooves at the right pitch, same as a boring bar, but now you have multiple teeth at the right pitch.

Actually, you might not have to grind the teeth off one side of the tap, if you can find a tap the same pitch as the screw but smaller, which is very unlikely.

The requirement is to have a screw thread of the required pitch for your machine, and then make a grooved bore with pitched grooves that will mesh with it, hence the use of the same tap but with most teeth removed.

Mike,
As a spin off of this method, I think it would be very practical for the drive mechanism of a CD player, as at present all the CD player mechanisms I have taken apart have a rack and pinion drive which is not the most ideal accurate drive you can get, but I believe it's a bit late to redesign that mechanism as now the CD is likely to be replaced by the memory chip which is solid state, oh well, maybe a linear record player arm.
Ian.

Hi Mike, I think the differential type, with the bore of the bearing insert having a pitch instead of grooves, would be ideal for a reduction drive where accuracy is not so important, due to some rolling slip occuring as in all friction drives, similar to Vee belt reduction pulleys, where creep on the pullys is not a problem in this type of drive.

The reduction in friction would be very desirable, and positional losses are acceptable.

I think this type of drive can be very advantageous especially in the very small diam size screw drives that ball screws cannot match, (maybe I'm wrong and they already have them)?
Who actually Invented this design, and is it patented?
Ian.

you take the red part (the tap)in the lathe
make a hole in the center
install the tools to make the filet with 10 teeth inside the hole
do not move on x axis but only in y axes
the misuse of the tool will produce the perfect profile needed
this primitive syntax express a corect semantic ?

Yes, I think you have it!

Ian,
I had a patent search done about 4 years ago, and the differential screw brought up a sizable stack of patents with these very attributes, though again, I do not remember one this simple. Gonna have to drag that search out and look at it again.
I have learned though, not to simply dismiss someones need for added parts and complexity. These guys definitely made these, and learned some things that may not be obvious in our mental exercise. ha

You have a great idea! I was just going to dismiss this today as having far too small a market, and was stupidly focused on high accuracy for open loop type positioning. Same thing made me abandon a different patent of mine that ended up with this same set of pros and cons, but was also noisy! (http://www.freepatentsonline.com/4895046.pdf)

Ha ha, how's this for a device that needs no accuracy, low friction, high mechanical advantage and simplicity, relative immunity to debris, and controllable by a rocker switch?: Log splitter! Gasp!

I have one of the patents here, and they go on quite a bit about tooth geometry, convex on convex, convex on concave. I imagine they wanted a reliable contact radius, and that is by definition the pitch radius.

Hi Mike, I suppose if you really wanted to make a "long lasting" but Rolls Royce item you'd make an insert from nickel chrome steel and cut the grooves in the bore, then have it case hardened and finally finish by grinding the grooves and O.D., but this would also require a hardened screw and that's becoming close to a regular ball screw set up, more like a commercial item, whereas the beauty of your design is the ability to make a practical drive using pretty basic materials and in the average home workshop, well within the realms and capabilities of the average worker.

I think it beats Acme thread and bronze nut, even with compensating nuts, for simplicity, especially as it's capable of practically eliminating backlash and friction completely by simple adjustment.

I'll second your log splitter and add a scissor car jack with 12volt cordless battery pack to drive it.

I am going to redesign the grinding head spindle lift screw on my Macson surface grinder to use this design.

At the moment the grinding head has an Acme screw, and to maintain positive up pressure against the screw when putting on the cut, there is a pulley and wire rope with springs that pulls the head up tp prevent it from dropping when the cut is put on.

You see a lot of things get patented but often they never make it to the market.
Have you seen this type of design on the market yet? Maybe they've engineered too much perfection into it and are up against the ball screw for cost.

I think this design is better than the roller nut type drive with the three rollers around the outside of the screw, and using deep groove bearings to combat end thrust will be the way to go.

I'm still looking at the needle bearing due to it's small OD in relation to length ratio and have seen some in the SKF.com online bearing catalogue that have an angular contact component built into it without increasing the OD of the bearing shell.

If this becomes practical, it will enable a hardened grooved insert to give a very compact housing, it just depends on how small the needle bearings are made.
My main interest would be for the crossfeed screw on my lathe as this is where a lot of action takes place.

On the subject of marketability, there is a huge market for aquarium air pumps, and I can see this type of drive being utilised as a reciprocating drive instead of the bellows type diaphragms powered by vibrating magnets, so bloody noisy and inneficient.

I can visualise the drive being in the middle of a piston that is double acting, just needs a seal on the centre, endless possibilities.

In the meantime my home built power hacksaw using hand hacksaw blades, (will cut a 100mm X 100mm, 4" X 4" solid block of steel in half), is looking ripe for a make over with this design in mind.
Ian.

I'll second your log splitter and add a scissor car jack with 12volt cordless battery pack to drive it.


Ian.
Ha ha. Niiice. Where in Au are you?

Anyway, thanks. I wish I hadn't found so much prior art, but oh well. Might as well have some fun with it.
I'm seriously considering using it on the X on my lathe.

Speaking of materials, harder on the roller side is OK, in fact preferred. When you get serious, then you'd have the bearings come in with ID grooves ground in.

With regard to a reciprocating pump, that's an idea I seem to come around to occasionally as well, but I've though that would be better with a ballscrew. I guess this would be good for high pressure.

I've changed my mind. I think there's nothing good to come from hard roller on soft screw. I'm on the same page now!

Hi Mike, I'm in melbourne in Victoria, the South Eastern part of OZ,

Ive found that soft on soft rubbing surfaces always wears big time due to surface welding whereas hard rubbing surfaces ground and suitably lubed maintain an oil cushion and last longer, but not in all cases.

The rolling and a bit of face rubbing action is less frictional than a straight face rubbing situation as you have in the Acme screw and bronze nut, but with harder grooves and a standard thread I think the form of the thread won't suffer as much as if you had a soft grooved insert being deformed under load against the screw flanks due to the much smaller face area which comes into contact once for every turn of the insert.

I like the idea of the grooved inner bearing, and even if it ran on a soft screw would be a better drive than a soft inner insert, which under load would deform and tend to pick up.

It all comes down to the load, which for our purposes would be relatively on the light side, but to go to heavier applications I think it would be counterproductive to expect a design that lends itself to DIY practice to be able to withstand forces that more expensive components are able to handle.

I'm all for KISS, Keep It Stupid Simple, and this is where this design excells itself, it's so DOABLE, even by the newest nubie wanting to have a go at CNC drive without the backlash problems of Acme screws and bronze nut.

On a long screw there's a lube problem with Acme/bronze nut, but the rolling drive only requires a bit of lube to keep it slightly wet.

Due to the fact that it has low friction, and if you are using a stepper motor drive which requires it to be held on to hold position, would there be a need for a spindle brake that enabled the stepper motor to shut down when stationary, but still have the screw braked to hold position?

I wondered about this as somone had stepper motors burning out after a short time.
Ian.

I won't use steppers, so I won't concern myself with that. ha

But, at 200 cardinal steps per rev, and a 12 TPI screw, even turning the motor off means with the stepper detent torque, you'll stay within .0004". No problem!

This is most likely to be totally impractical, but then I do have my reputation to live up to.:)

It comes from looking at the possibility of replacing the grooves on the insert by a series of retaining rings that constrain the path of a row of balls around the threaded central rod.
I suppose I'm trying to get towards a diy ballnut using your idea as a starting point, and with the same two identical sets at 180 degrees offset, and a third one, identical to the first.

I realise that there will be all sorts of loading issues, but I wonder if the geometry of the idea holds up. I've only shown the bare bones for clarity.

John

Well, keep going!
The problem I see there is that it will be quite bumpy as balls go in and out of the load zone. If you lay it out and zoom in, you'll see that there is a different "happy place" for the ring when you are directly on a ball, vs. on two on either side of a (vertical) line on your drawing.
This is one reason sloppy radial bearings under radial load die without some axial preload.

If you can figure it out, you may be able to curve it around and make a ball worm gear; the "holy grail" !

Who actually Invented this design, and is it patented?
Ian.

Gartner US3789678 pretty much nails it back in '74, so well and truly public domain now. Use freepatentsonline.com It truly is free, you just have to register to get full access.

I think I know now how to do the differential version simply, but I think it's an exceedingly tiny market for it...

Hi Mike, thanks form the patent link.

Hi John,
Been looking at your idea of the "ball" screw type idea, it has distinct possibilities but in single row form might be bumpy as Mike said, however it is crucial to have the screw hardened or you'll get a bumpy screw from surface indentation, also the ball track would want to be angled to the helix angle of the screw to allow the balls to recirculate, and just using one ball track might not be enough for wearability, probably 2 side by side,( three sets required), like the teeth of a threaded nut, which it resembles, but differs by having "rotating threads" instead of fixed threads.

I think that if you were to use an Acme type thread form, (or a Vee thread) for the screw, ( but hardened and ground), and only one ball track, with the balls of the right diameter to press down into the Vee of the thread sides but not to bottom, then you only require three ball tracks as per Mike's design, but I think the ball sets will have to be angled at the helix angle, which would be accomplished by making one end face of the housing angled to the helix angle of the screw, and the others screwed to it to get preload.

This would give you three sets of balls in contact with the screw and might be sufficient for wearability purposes.

The problem would lie in getting Acme (or Vee) threaded screw lengths with a hardened thread face and of sufficient length for your needs.

I suppose you could have a thread form that was more like a ball screw (spherical) and apply this principle, but it would need to be hardened and ground no matter what form was used, and then you are competing with ball screw manufacture which is well established and cost effective for the production rates now marketed througout the industry.

Mike, there was just one aspect of your design that needs adressing, and that is when the inside grooved inserts are hardened they will have a tendency to dig into the screw flanks and shave metal particles off under load, due to the difference between the helix angle of the screw and the none helix angle of the insert grooves, probably reduced by carefull deburring and rounding of the groove flank top corners.

On a soft screw and insert there will be no wear apparent unless they under load, as in an actual drive situation under load.

I'm looking at an Acme screw set-up, (soft), with a diam of about 19mm (3/4") having a double row ball bearing and the insert with just two radial threads (not helix) at the same pitch as the thread, with a form more like a "U" shape so that it forms a type of two point contact with the screw, three sets required as per your design.

The reason for this is that commercially, this size would be suitable for those big router tables and the choice of the thread form is, they're off the shelf items in the soft state, so the grooves in the insert consists of two non helix threads with the U shaped rounded sides that would not wear too much with the soft screw and require little lube.

On that size the inserts would probably have a 1" bore so single point grooving would be easy, more like square thread form but with rounded corners.

The alternative for big table router makers, cost wise, is compensated Acme screws that always have some backlash no matter how tight the nuts are adjusted, or ball screws for the Rolls Royce mob, but at huge cost.

I've seen it said a couple of times that there is a need for a cheap ready made package backlashness drive nut to suit the Acme screw, and I think this would probably fill a niche market, probably start an industry, especially if it's public domain now,
"Everybody wants one, how low can you go".

PS, I'm going to refer to this design as "The tricycle roller nut"
Ian.

...Mike, there was just one aspect of your design that needs adressing, and that is when the inside grooved inserts are hardened they will have a tendency to dig into the screw flanks and shave metal particles off under load, due to the difference between the helix angle of the screw and the none helix angle of the insert grooves, probably reduced by carefull deburring and rounding of the groove flank top corners.

...or run it in for some period at high preload to just remove the offending material, clean and re-grease everything and off you go. I did agree with eliminating the hardening in post #323..

As to the rest, go man, go. Though if you're going to run soft on soft, you'll want as many contact points as possible, though, so not sure I agree with the two groove approach, nor the point contacts.

I think I'm changing my mind about the "grooves cut with a mutilated tap". It's close, as you suggest, when the bearings are canted at the helix angle. I realize now that it is not close to the right tooth form for bearings parallel to the screw as I have shown. the higher the helix angle, the pointier and shallower the ridges get. I can generate that exact profile in solidworks and take a look.

Are you using a single start screw? Is it 3/4"-what tpr? I'd like to model something real and we can trade it off against the other ways you are suggesting. I'm still going down this path, as I can't think of a way for the bearings to follow the helix angle without making the bearing holder parts much more difficult to make.

Hi Mike, just for availability the 3/4" single start Acme screw seems like a good start in the bigger size, not sure of the pitch, probably something like 4 tpi.

If you went for a two start Acme screw you'd get a much bigger helix slant that would push the grooved threads out of the screw thread leading to point contact only.

In some ways I think the finer pitch Vee form thread might be a better way to go, as the helix angle gets less for the bigger diam and finer pitch, but on the down size the Vee form threads are usually Allthread, unless there is a supply of precision Vee thread available.

Just to theorise further, the only way possibly to get a better contact drive face would be if you had thin housings with just one groove in the bore and stacked the housings together in batches of 4 like saucers, so that each housing took up a slant to the screw and as such were all bolted together, (12 in all), to form the three components required to depth the teeth.

I think this would overcomplicate the set-up and even if extra thin ball races were available would probably lead to instability due to the races developing side slop from end thrust wear.

You might be able to have the two groove set up for each bearing housing using a twin ballrace and this might be able to slew over to run parallel to the Acme thread, but anything longer than that just would climb up the thread faces at the ends if slewed over.

For simplicity I think your original method with multiple grooves in the bore with twin row bearings scaled to suit the screw size used is the way to go, even if it has a not too efficient contact profile.

The beauty of it is that, if it did wear in and develop a bit of backlash, it can be adjusted to depth better by loosening the fixing screws and redepthing the set up by any DIY'er making or being supplied with one.

I would make one of the bearing housings with a large flange for attaching to the machine frame, giving a natural thrust resistant mounting face, and have the other two housings attached to it but clear of the frame.
An alternative fixing could be an L shaped end housing for mounting under the machine table if required etc.

At any rate using a 19mm or 3/4" diam Acme or Vee thread will still give a relatively large housing, but I've always advocated that you can work better with a machine part when you can see it and get your hands round it, and it's easier to make.

Just a bit more food for thought, using Acme thread, if you had the centre housing only with grooves and the two outer housings with PLAIN bores to ride on the top of the Acme form, but on the opposite side, this would have the same effect, as the plain bores would be able to ride on the top of the flat on the Acme form, won't work for Vee thread though.

Now with this thought in mind it might be a way to have the half nut for a lathe, as you only have to move the centre grooved housing out of contact with the screw, while the two outer housings remain on top of the thread and are able to slide along it without making flank contact.

To use this method would require a single point cut grooved bore of sufficient number of grooves to give a good wear rate without damaging the screw.

The main thing is, you've proved the concept to such an extent that it is entirely feasible to make it as you originally showed.
Ian.

Ha ha, Ian, as with most simple solutions, there's so much to talk about!
4 tpi is coarse!
I like and have adopted the idea of using commonly available unplated all-thread, at least for the DIY version.

I laid out the 3/4-12 and the perfect tooth form for the roller grooves is stubbier as expected, but only .55 degrees wider than the standard 60 degree tooth form, so my concept of the "tap as grooving tool" is wrong, but not far off and probably close enough.

Just a bit more food for thought, using Acme thread, if you had the centre housing only with grooves and the two outer housings with PLAIN bores to ride on the top of the Acme form, but on the opposite side, this would have the same effect, as the plain bores would be able to ride on the top of the flat on the Acme form, won't work for Vee thread though.................

.......................To use this method would require a single point cut grooved bore of sufficient number of grooves to give a good wear rate without damaging the screw....................
Ian.

Ian, what if your central grooved insert was slightly deformable, within the appropriate accuracy of a particular set up, but gave more contact surface.

Like wood ?
OK OK (chair)

Or epoxy, suitably loaded ? It could even be molded then as a replaceable part.

John

Hi John, nothing wrong with wood, the only problem might be from a deformation standpoint that led to wear and short life, bearing in mind that the application is for a transmission screw so end loads against the screw flanks are equal to the force required to move the table against the cutter, maybe nylon grooved inserts on a steel thread might be longer lasting as it's a rolling force as opposed to a rubbing force like in bronze nuts.

I might venture to add that if using soft grooved inserts the centre housing could be spring loaded against the other two housings at 90 deg to the thread axis,to maintain a constant thread flank thrust as long as the inserts didn't bottom on the thread due to flank wear, which would lead to backlash.

Hi Mike,
the 4 TPI was probably true for an Acme thread which is pretty coarse in the 19mm-3/4" diameter size, so 3/4" X 12 Althread would have a better helix angle for this application.

The other aspect is the method of production used for Allthread, which I think is by a die head type machine tool and is subject to a pitch error as the length of the screw progresses, but if pitching accurately is not a problem, such as required for doing definite co-ordinate locations in hole positioning, then Allthread would do the trick.

You might only be out a couple of thou' per inch and for wood routers this is nothing.
At least you get to go there and back and be on the same spot when you go forward again, even if it is out of position by a few thou'.

I would think even if you were cutting metal, the ability to go back and forward without backlash is far better than losing the plot when backlash gets in the way.
You can always program in a progressive allowance for pitch error, which would be constant, but backlash is a pain in the rump.

I bet somone will ask for a design for a revolving nut based on this design, so you might get your patent yet.
Ian.

Ian,
Agreed, I think standard thread form is better than Acme for this, and that die cut threads are accurate enough for most.

I'd want to do six rollers in a ServoNut, and yeah, I'm thinking of jumping straight to that. I've got a potential customer for it; a semiconductor vacuum elevator thing that I'd like to not put a power off brake on to hold position when the servo is off.

My current thoughts on a design for a roller nut for a wood router machine with light loads.
Based on the ideas and comments by Mike and Ian.

A filled epoxy internally grooved roller ( blue in the diagram), to match the threaded shaft, with a pair of ball races mounted on eccentrics( or sprung) acting as rollers maintaining the pressure of the thread into contact with the resin roller.
The epoxy insert would be molded, and replaced as needed.

Comments please.
John

I think that is perfectly workable, John. I don't much like rollers made of epoxy, as Hertzian stresses can be high. It's not much good for any cyclic contact stress of any type, really.
Delrin might not be too bad, thermoformed on the proper shafting.

Mike,
Great work mate. Thanks for the calc and drawing in post 332. I have always wanted to know how a 60 Deg thread angle would look to a roller screw roller. So is 60.45Deg calculated with a roller axis parallel to the thread axis?

Chich

So is 60.45Deg calculated with a roller axis parallel to the thread axis?

Chich

Thanks, and, yes it is. I need to do one more bit on that and see what's the smallest diameter of roller grooves before the helix interferes (axially).

Hi Mike, I'm in melbourne in Victoria, the South Eastern part of OZ,


Ian.
Hi Ian, I just ask blokes in OZ where they're from because I have another trip in the works out there next year. I have family in Newcastle (go Knights!) and friends I do pulse-jet engines with in the Blue Mountains, Brisbane, Bribie Island and Melbourne. I may put together a jet meet somewhere central to those locations, burn fuel, make noise and drink beer, so I'm making a list of folks that might just be interested.

I own www.pulse-jets.com and there is a pretty sizeable Aussie contingent in our forum.

HI mike, if you're in my area I'll give it a whirl.

The thread form of 60.45 deg is practically a 60 deg NC, NF or metric form so it would probably be available off the shelf for most people, if they make NC or NF allthread commercially that is, I know they make metric.

Hi John, I think you will have problems with the ball bearings bearing on top of the pointed Vee's, and the epoxy for the internal thread doesn't appear to be a good material anyway, more likely Delrin or even plain old Nylon, but whichever way you go the number of grooves required in the single threaded centre insert will make manufacture difficult unless you make it in two pieces and pitch the two halves carefully.

Alternatively you could try making 6 or 8 washers, each with a single thread (not helix) bore, so that when they are stacked together in a housing they will be the correct pitch.

They will need to be contained in the housing to prevent turning and end movement, maybe with a peg and end circlips, or swaging the ends of the housing to hold the washers in.

If the thread tops were lightly filed or turned to form a flat it might give a better surface to run the ball bearings on, but possibly a long plain roller parallel to the grooved bore, eccentrically adjustable would spread the load better.

Having the bearings or roller, bearing on the OD of the thread, is getting close to the same design approach as the three roller nut method.

I personally like the original design Mike showed as it's so simple and easy to make.

The whole heart of the set-up will lie in the accuracy of the screw used, more from a clean cut thread form point of view than from a pitch error which can be programmed out once known.

If you have a lathe and screwcutting experience then it is a piece of cake.
Ian.

HI mike, if you're in my area I'll give it a whirl.

The thread form of 60.45 deg is practically a 60 deg NC, NF or metric form so it would probably be available off the shelf for most people, if they make NC or NF allthread commercially that is, I know they make metric.

Cool, I'll give you a shout.

Ha ha, I like the washer idea.

I'm sure you realize that the thread form of 60.45 on my layout is how the walls of a 60 deg standard thread form appear to a straight grooved roller. It occurs to me that this is true only at the tangent point, and as your groove roller approaches the diameter of the screw, a different kind of interference will crop up, that of the helix angle vs. the "zero angle" of the roller. That's a more difficult thing to lay out.
I think I can do it in Solidworks as an interference detection routine. I'll try it and post the result.
I'll have to switch the geometry around so I can iterate the roller pitch diameter down until I see a non-working condition...

Hi Mike,
So, the smaller the bore of the insert to the thread diam, the more interference you get, IE. the thread will contact groove flank in the insert bore at two points , being the entry on one side and the exit on the other.

That means if you have a bigger bore in the insert you get a larger internal radius presented to the helix of the screw, which if the screw is a fine thread will be better due to the helix angle being less.

I can see that there would be no advantage in having the housing as small as possible when you really need to maximise the bore of the insert, unless you get some bearings with a large bore and small outside diam.

If you had a 12mm diam thread and an insert of 25mm bore size, by my guesstimation and a quick sketch you would need a ballrace with about 35mm bore and 60mm outside diam and about 20mm wide, very special, also you would need them double row at least or two 10mm wide bearings, with about 4 or 5 grooves in each insert, giving a 2mm pitch on the screw.
Ian.

agreed on all points!
I think I'm going to build one up with a 5/8-11 thread, because I can throw it on a ServoNut drive without too much trouble. (Current ballscrew size on that is 16mm)

Hi Mike, been doing some research on the SKF.com bearing site to see what bearing sizes are available.

I looked at the needle bearing layouts and one interesting set came to light, the NKIB 59 series, which is a combined needle bearing and double angular contact bearing in one package.

The size I'm mostly interested in is the NKIB 5907 which is a needle bearing with inner race sleeve and a double thrust angular contact bearing built into the end, so it will cater for thrust both ways as well as radial loads and is rated to 9500 rpm.

The dimensions are 55mm OD, 35mm bore and 30mm wide.

Depending on the cost, I reckon this bearing would lend itself to this project 100%, especially as it combines both radial and axial thrust in both directions.

All you need is a housing with a straight through bore, no shoulders or grooves to lock it in place, as it can be Loctited into the housing.

The grooved insert would be a sliding fit into the middle of the bearing bore, and only needs to have a small shoulder at one end, and either a circlip retainer at the other end to hold the bearing against a preload Bellville washer or a threaded collar to preload it.

The internally grooved sleeve, with an internal diameter of about 25mm, gives bags of room for a screw size of 20mm diam.

Lubrication is by a groove on the outside of the bearing that leads to a hole through the bearing wall so this would be one way to keep sh1t out of the bearing by regular positive oil injection.

This bearing layout is so neat and compact that I think for the smaller sizes it would be the way to go too, especially as it cuts out a lot of fussy machining in the housing and inserts.

Tomorrow I'll contact my local bearing supplier and see if they can give me a price, or supply something similar.
Ian.

hello,

(price *3) I am affraid that you may have a bad(very)surprise but if someone can find some chinese or russian alternative (for a diy we do not need SKF quality and logical price that goes with it).
Eventually, if this type of bearing is used in car industry then there is opportunity in scrapyard

Hi Rocky, How much is too much?

When you have a solution to a problem that will break new ground and give you a whole new approach then sometimes the initial cost is justified.

There are of course many ways to solve a problem, but it's up to the individual whether or not it's worth it.

I knew from some time back that this type of bearing was available, but didn't have too many details on its virtues or the layout of the components.

When I get the price I'll post some details, probably other bearing makers (Chinese, Taiwanese, Indian etc) will have something like it.

Which ever way I find, there must be alternatives to the design that haven't been thought of yet.
Ian.

price i have is around 60 us$

AAAAhhhhhhhh Rocky! now that's a lot of dough, $60 bucks, I wonder if the Chinese make one?

Still, $180 for a backlash free practically frictionless drive on allthread, bet it cost a lot more for the same ballscrew set-up.
Ian.

Hi Rocky, just got a quote from our local bearing supplier.......approximately A$120.

That's a lot of dough.

At the same time I priced up a double row deep groove bearing with 55mm outside diameter, 35mm bore and 10mm width, bearing number (SKF) 6907 at A$15 each, A$45 total, big difference, but would work just as well, and could be loctited into the housing with the grooved insert loctited into the bearing bore too, no shoulders to worry about for location.

I think having two deep groove double row bearings side by side in each housing, (six in all), would allow a longer nut and so give more stability and wear life under the push and pull conditions of the screw drive, that would cost about A$90 all up for the bearings.
Ian.

...At the same time I priced up a double row deep groove bearing with 55mm outside diameter, 35mm bore and 10mm width, bearing number (SKF) 6907 at A$15 each, A$45 total
Ian.
Those 6907's are single row...and it's $9.95 out of VXB.com

ha ha, I was just getting ready to extol the virtues of deep groove radial bearings for this, and there you are! I am going for a bit smaller, so 20-25 bore is better for me.
The nice thing is just a little more money gets you actual two row angular contact units for about $15, the 520X series.
6006ZZ is just $3.99

For the dual row radial bearings, the 620X series:
6204RS Bearing 20x47x14 Sealed $4.95

What we shouldn't lose sight of here is that load capacity is not the criteria for selecting the bearing. the sizes we're talking even for a single row radial bearing, and accounting for many less balls in the load path will still be in the hundreds of pounds per bearing dynamic load capacity.

I prefer the angular contact series because it will be slightly stiffer axially, but a single deep groove bearing per roller has the advantage of being slightly self-aligning for the roller grooves, which could be really good for a DIY'er.

I already use the dual angular as track rollers on my ServoBelts. They're the best deal going!

Here's a couple helical interference plots. It's not much, and can't really be addressed other than making the roller bigger in diameter.
Screw on this is 8mm x 1.0 lead.

Hello,
So why not use
2 x AXK3552 : thrust needle bearing 35x52
1 x RNA4907 : needle roller bearing 42x55x20
the tap will seat directly on the needle all the load are taken both way by specialised bearing and its cheap !!!

Hi Mike, been lookin at the SKF catalogue, seems the bearing dealer gave me some prices that didn't add up, for double row angular contacts to suit a 20mm, (just a ballpark size for starters), Allthread screw size and the 3207 bearing with 35mm bore and seals seems like a good package, only problem is it's 72mm OD which is pretty big.

So I looked at the double row deep groove bearing, but that comes in 72mm OD with 35mm bore and 23mm thick, still too big, but when you go to the single row deep groove with seals, the 4207, you get 47mm OD, 35mm bore and 7mm thick, which if you put two in each housing, (total six) makes a very small housing size for a big screw.

I think I've got some single row bearings to play with in my scrap box, that I aquired years ago when some cigarette machines were being rebuilt and I collected all the used bearings that went for scrap, so that will give me a reference point to plan on without actually buying something (tight arrse inc.).

I see what you mean by the area of contact, and making the bore bigger is logical as well as if the pitch is fine and the screw OD is big.

Conclusion, I think I'd go with two deep grooved sealed bearings in each housing for somplicity and cost effectiveness.

The other popular screw size would be the 12mm or 1/2" size, so I reckon the bearing package on that size would be very cost effective.

It only leaves the Allthread to be run down with a die nut to clean the burrs etc, or to get one precision cut if pure accuracy is required.

I can't see Acme being of use in this set-up due to the narrow thread profile, but I might give it a serious looking at to see if it can be used.
Ian.

Mike-
In post 330 I believe you stated that you are wondering how to align the bearings to keep the screw in contact with one side of the roller without making it difficult to machine. If you are planning on rotating the nut, I think the answer is to have wheels (bearings) that are perpindicular to the screw. If the threaded shaft isn't turning, as in your servo nut, all you need is linear support. The first thought is that the wheels would bump along the crests of the threads - but if the screw isn't turning, what good are the threads on the top of the screw if the roller only contacts the bottom section of thread? Machine the top of the threads flat so the wheels roll along a flat surface. If you want to get fancier (crazier) you could machine two flats at say a 60° angle and have a sort of V-Rail arrangement right on the screw. Sort of a helical rack with a V-bearing on the back. If the threaded rack was made large enough in diamater, it could even function as the guide rail with a type of linear ball spline built into the back. This V-Block on the back would serve as the anti-rotation device between the rotating nut and the fixed frame of the motor.

In theory, you could have a belt driven rotating nut that is mounted in a fixed bearing block. The shaft could then extend and retract without any other external guidance needed.

If this isn't communicated clearly, I can try to create a solidworks model and post it later this week.

NEATman
Keith

Mike-
In post 330 I believe you stated that you are wondering how to align the bearings to keep the screw in contact with one side of the roller without making it difficult to machine. If you are planning on rotating the nut, I think the answer is to have wheels (bearings) that are perpindicular to the screw. If the threaded shaft isn't turning, as in your servo nut, all you need is linear support. The first thought is that the wheels would bump along the crests of the threads - but if the screw isn't turning, what good are the threads on the top of the screw if the roller only contacts the bottom section of thread? Machine the top of the threads flat so the wheels roll along a flat surface. If you want to get fancier (crazier) you could machine two flats at say a 60° angle and have a sort of V-Rail arrangement right on the screw. Sort of a helical rack with a V-bearing on the back. If the threaded rack was made large enough in diamater, it could even function as the guide rail with a type of linear ball spline built into the back. This V-Block on the back would serve as the anti-rotation device between the rotating nut and the fixed frame of the motor.

In theory, you could have a belt driven rotating nut that is mounted in a fixed bearing block. The shaft could then extend and retract without any other external guidance needed.
Keith

As to flats on a static screw: brilliant! For light duty bearing, that's the ticket. That's basically a "Robotte" (sp?), ever seen one?
I've been designing this roller screw into an "economy" ServoNut, with NEMA23 side mounted and belt, so I was half way there with you. That's a beauty of an idea! Makes for the same kind of simplicity to become a Z on a ServoBelt. Heck, you can put some pretty beefy screw stock on there cheaply, now that it's also a linear bearing.

I'd use V's or radiused grooves ground into the screw, and now I'd need more rollers, cause they will go in and out of contact, but that's no biggie. Each roller probably costs $25, so why not 6 or 8?

I knew there was a reason I grabbed a 25mm spline and the Robotte out of the toybox the other day, but I hadn't made the connection. You definitely beat me to that, dammit. ha. Nice show. I've seen a couple examples on this forum of neat grinding jigs that would work for the DIY'er.

Now, for the OTHER way. mwahaha
I've got a neat way to couple two spinning nuts on two parallel screws, constant velocity (a belt or gears would do, but), let's just say it's a zero backlash hard parallel coupler. So I could put nuts on two screws and either rotate both screws or both nuts, and the nuts are the linear bearings. As I write this, I realize I'd heard about this from Alex Slocum, so hat off to him! He was working on a method of making lathes in let's say, "primitive settings".
Of course it had nothing to do with roller screws. It was all bronze nut stuff.

Ha ha, if you've ever gone through that 4 volume set of orange mechanism books; there's damned little new under the sun, but there's still some things to discover. Still, a lot of prior art in this area, seemingly by people that couldn't exploit it, or it proved impractical for some reason we will soon discover. My cup is half empty tonight. ha.

Mike - for the diy'er, how about dropping your large diameter screwthread into a small angle, then epoxying them together, so no grinding ;)

John

Edit
I can now see this is impossible for the idea you mentioned - the nut has to be able to rotate around the thread (chair)(chair)(chair)

Hi John, keep at it mate, some of the best ideas occur when you bounce ideas off of someone else.

Now all it needs is to grind the grooves into the bore of the bearing as per Mikes original design and Bob's yer uncle, a smaller package again.

Maybe SKF or someone will consider it.
Ian.

I never did look for a bearing house to do that. Maybe someday... Might be a good bearing to have a couple of sizes for this very thing. hmmm..

Hi Mike, I think it would take a bit of doing grinding 60 deg thread form grooves in a bearing bore, not impossible, but then EDM can do this and also give a very clean form profile, better than grinding.

The principle is so practical and user friendly as far as making it with the inserts method you advocated originally, but it needs a source of reliable screwed rod to compliment it.

A firm I used to work for in UK in the 60's made their own Acme thread transmission screws, about 3 meters long, and the method was by thread milling, and I reckon it could also be used for making precision vee thread using a milling head instead of a single point tool.

The machine resembled a long bed lathe with a traveling steady that supported the bar length as the threading progressed.
The Acme thread was milled in one hit with a cutter that resembled a gear tooth cutter as used on a horizontal mill.

It's possible, but I don't know if and where, that precision vee threaded lengths are available on the market, if so, Bob's ya' uncle, problem solved.

The ideal material would be Nickel Chrome steel, nitride hardened to avoid distortion, and possibly even hard chrome plated and finish ground.

Whichever method is used it would pay dividends to have a decent bit of material for the screw to avoid any pitch error due to wear, but on the other hand if Allthread was found to be acceptable it might be the way to go, and build in a wear rate factor that just lasted so many hours and got replaced say once a month, cheap enough if accurate enough.

I would prefer the screwed rod to be replaced rather than having to rebuild the nut assembly every time, so that's why I would recommend a hardened thread insert in the bearing bore.
Ian.

Mike-
I have the same set of four ingenious mechanism books. It's amazing what people have come up with over the years - it almost seems we have it too easy with CNC these days...

I had posted a few links in another thread back in December about a "threadless screw"
http://www.cnczone.com/forums/showthread.php?t=47782

The thought about grinding v-grooves inside bearings is interesting, it would almost be a threaded version of this:
http://amacoil.com/html/page3c.htm

This mechanism has special bearings with a non-cylindrical shape through the bore to allow them to be angled to the OD of the Threadless rotating leadscrew. I think that the variable speed & reversible model is especially slick - considering that the rotational speed of the "screw" stays constant.

A couple of these if they were enlarged could be used for both driving a gantry and as the linear bearings.

BTW, I think I have an idea of how you coupled the two threaded rods. I was just thinking of a way to couple two parallel threadless rods.

NEATman
Keith

Now, how about combining this:
http://amacoil.com/html/page3a.htm

Over a hardened hollow tube that is one of these encoders:
http://www.newall.com/DROs/advantage.htm

It could be the linear bearing, the drive "screw" and the encoder feed back to see if it slipped position.

Or if you want rollers on the outside, this is a brand new idea from 1937:
http://patimg1.uspto.gov/.piw?Docid=02204638&homeurl=http%3A%2F%2Fpatft.uspto.gov%2Fnetacgi%2Fnph-Parser%3FSect2%3DPTO1%2526Sect2%3DHITOFF%2526p%3D1%2526u%3D%25252Fnetahtml%25252FPTO%25252Fsearch-bool.html%2526r%3D1%2526f%3DG%2526l%3D50%2526d%3DPALL%2526S1%3D2204638.PN.%2526OS%3DPN%2F2204638%2526RS%3DPN%2F2204638&PageNum=&Rtype=&SectionNum=&idkey=NONE&Input=View+first+page

NEATman
Keith

snip... this is a brand new idea from 1937:
http://patimg1.uspto.gov/.piw?Docid=02204638&homeurl=http%3A%2F%2Fpatft.uspto.gov%2Fnetacgi%2Fnph-Parser%3FSect2%3DPTO1%2526Sect2%3DHITOFF%2526p%3D1%2526u%3D%25252Fnetahtml%25252FPTO%25252Fsearch-bool.html%2526r%3D1%2526f%3DG%2526l%3D50%2526d%3DPALL%2526S1%3D2204638.PN.%2526OS%3DPN%2F2204638%2526RS%3DPN%2F2204638&PageNum=&Rtype=&SectionNum=&idkey=NONE&Input=View+first+page
NEATman
Keith
There is truly not much new under the sun. Did you see the little one with finer pitch on that patent that just does the garage door limit switches? ha
I've always liked the Newall encoder idea. so simple.
This type of threadless design was on our truly huge CMM machine at my last company. Doing two in parallel would be neverending grief, though, since the ratios will never be the same between two, or at least you probably couldn't get them "same enough" ha

Mike-
I overlooked the finer pitch limit switch actuator on the same rod no less! To make the door travel less or more distance do you move the limit switch or change the angle of the rollers to effectively change the lead? I would guess that whip may be a problem on a contraption like that though.

As far as the different "pitches" between more than one threadless nut, that's a good point. One would certainly fight the other, and that fighting could cause something like backlash to occur, especially if the axial overload forces were almost the same.

I'm still thinking along the lines of the internal v groove on a bearing. You could have several of them arranged just like the rolling ring concept, and preload them radially to take up the axial backlash - thanks to the angle of the thread.

Keith

Hi Mike, here's a few sketches of a housing arrangement I think might solve the problem of the straight groove mating with the helix angle of the threaded rod.

In the arrangement at (A) & (B) I've shown a single groove (straight thread?) mating with the Acme thread but having a self alligning bearing in the housing that allows the thread insert to tilt freely at the helix angle, so maximising the thread contact line.

The self alligning bearing has a certain amount of end thrust capability, and so will handle reasonable end loads.

The 3 housings should be made wide enough to cater for the extra width required when the bearings are tilted over a bit.

The same arrangement at (F) shows a double thread insert on Acme thread for more contact, giving a total of six threads in contact in the three housings.

I favour Acme if possible because it is more accurate than Allthread.

A different approach is used at (C) (D) & (E), where double row deep groove bearings are used, but the housings are bored at the helix angle of the thread, to maximise thread contact, and the bearings remain fixed at the helix angle.

The limiting factor is the length of the thread insert in contact with the screw, but with a vee groove insert on Allthread and tilted at the helix angle, should give at least a number of threads in each housing in full line contact.

This should prevent the scuffing of the threaded rod, from flank engagement, that would occur if the straight grooved insert is squarely mated with the helix angle of thread.

I think it can then be considered truly frictionless, as the Vee of the grooved insert will engage 100% with the flanks of the threaded rod, be it Acme or Althread.

The important factor to be considered is the accurate pitching that will be required when the three housings are at the helix angle and are mated together, but this can be done simply by increasing or decreasing the distance between each housing face with shims or a single spacer on assembly, or facing the housing to suit.

At assembly the thread would be engaged fully and the tops and bottoms of the respective Vees given clearance to prevent bottoming of the threads, so that backlash was non existent.

End preloading is not advisable due to the vees requiring full flank engagement on both flanks while rotating, so preload would be accomplished by the depthing as you originally did.
Ian.

Nice work, Ian. I've been studying your sketches for a bit.

Some observations:
If you kick the bushing to follow the helix angle, only the leading and trailing grooves will engage when the goal is several (though if you're going to do only one or two grooves, never mind). Should be a slight interference, though, and I think will wear in until all engage. Maybe run it with over-high preload for a bit, and make sure the nut is softer than the screw...
Nothing wrong with any of this, except for the self aligning bearing thing. I think if you built that, it would give you a few WTF moments. ha (this is getting too long, but there's a story about self aligning gone very badly for me!)

I was trying to avoid wedgie plates, and so was trying to find the silver lining in doing it straight across, beside ease of manufacture. One neat thing about it is the nature of the interference, leading and trailing. The closer the groove and screw diameters get, the larger an angle between contacts. So, we can get the same two contacts we had theoretically, but they are spread apart radially about the screw, and we still get a set for every groove.

Faaaaascinating! (to a diminishingly small percentage of humanity!)

How about just making a form tool from the same lead screw thread stock to finish cut the rollers? I would think then you would have a good match of the helix angle of the rollers to lead screw using the form tool.

I made this form tool out of an old acme lead screw I replaced with ball screws in my mill. Pretty simple to do.
As a test I plunged the entire 20 groove tool into some 6061 without first single point cutting the grooves just to see how it would cut.
A thin cutoff tool to first cut the grooves and then final finish the grooves with the form tool should result in a very good mating roller to lead screw.
Thoughts?

http://i109.photobucket.com/albums/n48/S_J_H/misc/formtool002.jpg
http://i109.photobucket.com/albums/n48/S_J_H/misc/formtool003.jpg
http://i109.photobucket.com/albums/n48/S_J_H/misc/formtool004.jpg

Very niiiice!
If you do an internal, though, you'll have to relieve the backside of the form tool tips, as the helix will interfere, either burnishing or worse, galling up.

I'm talking to a company that makes threading dies. I figure they have the internal grinding machines for high production, and all the knowhow. Just have to talk them into it!
It starts with deep groove Conrad style bearings, take them apart and send the inner race for grinding.

:cheers:
How about just making a form tool from the same lead screw thread stock to finish cut the rollers? I would think then you would have a good match of the helix angle of the rollers to lead screw using the form tool.

I made this form tool out of an old acme lead screw I replaced with ball screws in my mill. Pretty simple to do.
As a test I plunged the entire 20 groove tool into some 6061 without first single point cutting the grooves just to see how it would cut.
A thin cutoff tool to first cut the grooves and then final finish the grooves with the form tool should result in a very good mating roller to lead screw.
Thoughts?


at least a splendid technique to make the //roller screw

#369 Today, 05:23 AM
Mike Everman Join Date: May 2008
Location: USA
Posts: 84


roller screw

--------------------------------------------------------------------------------

I'm talking to a company that makes threading dies. I figure they have the internal grinding machines for high production, and all the knowhow. Just have to talk them into it!
It starts with deep groove Conrad style bearings, take them apart and send the inner race for grinding.


definitely the ultimate solution, but i am wondering; how much ?
because' this type of machine do not come cheap, and, if there is not the sufficient number it will be as expensive as it could be cheap
Meanwhile i am ready to buy 3 of them if the price of the bearing is only multiplied by two and do not go more than 60$(180$)

ha ha, Rokag, I'll put you down for 3. :-)
I'm starting with 10mm x 1.5 thread. It's a size I need to make the low cost Z axis for my ServoBelt stages. Well, maybe I'll go up a size to 12mm. If I were to really do this thing, I'd get the grinding done in China. I source thousands of castings from there; shouldn't be hard to find a tap and die manufacturer to do it for practically nothing.

Greetings
I have read this thread at least twice, enjoyed it both times. If i missed this or if i am all wet just smack me down i can take it.

if you use a 1/4-20 threaded rod and mate it to a 1/2-10 double start nut
it seems the helix angles would be the same, so the bearings would all sit square.Any interger multiple of the rods pich diameter with that integer multiple of thread starts for the nut should work.
edit
of course its a 1:1 drive then

Hi CXEVALO, the internal part must be grooved not threaded with a helix, otherwise you'll get a variation in the drive resolution due to slip between the threaded rod and the internal threaded insert.

All this has been brought to the stage where the bearing is going to be internally grooved with the pitch of the screw it will mate with, best regards to Mike who is pioneering in this project.

Mike, I think you're right about the non validity of the self centering bearings, especially if they get a bit slack and develop a slight wobble.

For simplicity the housings could be bored straight through as you originally showed, and the helix angle/goove contact lived with, but I think it would run better and smoother if the housing was tilted to the helix angle.

This could be accomplished simply by mounting the first housing against an angled mounting bracket and allow the other housings to butt up against it but with the necessary requirement of maintaining pitch distance by shimming etc.

I reread with interest your previous posts some time back where you mentioned the use of the roller nut as a linear slide and drive all in one, using two nuts on two seperate screws spaced apart and coupled together to synchronise, very ingenious.
Ian.

http://www.youtube.com/watch?v=PuBtaV_L0fk


video and pictures of my proto installed on my lathe
the comparator is installed to measure the variation of the chariot .

the chariot is very tight so there is a very small backslash (<1/100mm)but due to the cantilever as you can see the screw is very far from the gravity center, the 1/100 is the bending of the flange of the roller, my next one will be design to fit mach closer anyway i wish you have the same performance on your x axes.

Also no backdrive so you can use your lathe manually !!!

So do not hesitated it work and it's rather cheap(bearing is the problem here 12 bearings so look on ebay ) the screw is the original one

Beautifulllllll!
Well done.

Nice work!! Well done!

hello guy,
this forum is a little bit dead show a little bit what you have done because it work !!!

Opps :)

Oldmanandhistory,

Yes This video has already been displayed on this thread by Mike Everman in post #280
Thanks for sharing it again any way!

Chich

Oldmanandhistory,

Yes This video has already been displayed on this thread by Mike Everman in post #280
Thanks for sharing it again any way!

Chich

OPPS sorry about that will go see :)

Yes This video has already been displayed on this thread by Mike Everman in post #280
Thanks for sharing it again any way!



Mike Everman as shown something VERY interisting on post 280 but it is completely different

rokag3
I'll explain. There WAS a video in post #379 but it has now been removed. It was the same one as #280. It's all fixed now so there's nothing to worry about.

Have a Happy Day,
Chich

thank Chich for the explanation
Lucien

Hi everybody !

Hi Lucien,

I saw your proto and found it awesome, would you please post some simple but commented schematics of your proto, especially the side plates, even if it's a hand draft as long as it contains the numbers.

And please the formula to determine the diameter of the rollers from the diameter of the screw.

Thanks !
cnc2.

hello,
I saw your proto and found it awesome, would you please post some simple but commented schematics of your proto, especially the side plates, even if it's a hand draft as long as it contains the numbers.

And please the formula to determine the diameter of the rollers from the diameter of the screw.



basically the diameter of the rollers where determined with the size of the bearings i could get for little money.
I will say that biger rollers mean cheaper bearing less roller and good rigidity also you can use longer roller so everything positif the drawback is the size of the screw
for X axis it's not a problem
for y axis there is much less rooms and the use of miniature bearing cannot be avoided
you must remember that the push is made to the extremity of the roller
so the numbers of rolls will be determined by:
1)how much rooms do i have?
2)what are the bearings I can find at a reasonable price (for wood machine delrin can be used but for metal forget it) 4 bearing for 1 roll
3)If I can use hardening technique then i can have a smaller roll
but my mother screw must be of good quality too (in my case i keep the original screw and bronze is the only possible metal)

There is one point very important and probably the most difficult
the determination of the corect profile !
this can be done with solidwork and several trials

for the excentric this is very easy to make and help a lot to get a good ajustment

If i was not so leasy ,i will design a system with a spring (to keep the pessure even on the screw) and through elbow attach to a bar a bit longer than the chariot and paralele to the axe, in case of crash the bar will lift the roller from the screw absorbing crash energy through the spring and saving the machine it could also be use as the traditional (but lousy)half nuts

on post 267 you have a good picture of the flange
total thickness 15 mm
4.9mm depth here for the thrust bearing 4.9 in the roller
the depth is dependant of the thrust bearing
the diameter will be external diameter of the thrust bearing +excentricity
5.1 mm for the wall the diameter = diameter of the axe+excentricity here 8mm+3mm
5mm depth for the excentric
the diameter is = to diameter of the axe+excentricity+what it need to make it strong enough with the guide line that all the diameters should never get in conflict with the axe :nono:
the center diameter is diameter of the screw+(as less as possible so it will be more easy to adjust here 21mm for 20 mm screw)
the excentric is your choice I choose complete disengagement
for the roller do not forget to calculate the position of the rings
in the case of 3 rolls 2 rolls with a ring on the left and one with the ring on the middle one of the 2 identical roll will be put in opposition with the other so left midle right

hello,


basically the diameter of the rollers where determined with the size of the bearings i could get for little money.
I will say that biger rollers mean cheaper bearing less roller and good rigidity also you can use longer roller so everything positif the drawback is the size of the screw
for X axis it's not a problem
for y axis there is much less rooms and the use of miniature bearing cannot be avoided
you must remember that the push is made to the extremity of the roller
so the numbers of rolls will be determined by:
1)how much rooms do i have?
2)what are the bearings I can find at a reasonable price (for wood machine delrin can be used but for metal forget it) 4 bearing for 1 roll
3)If I can use hardening technique then i can have a smaller roll
but my mother screw must be of good quality too (in my case i keep the original screw and bronze is the only possible metal)

There is one point very important and probably the most difficult
the determination of the corect profile !
this can be done with solidwork and several trials

for the excentric this is very easy to make and help a lot to get a good ajustment


Thanks ! the size seems to be the only real problem here.
What do you think about these (http://www.surpluscenter.com/item.asp?UID=2008092610304296&item=1-LM11949&catname=) ?
it might be really difficult to use'em but who knows...


If i was not so leasy ,i will design a system with a spring (to keep the pessure even on the screw) and through elbow attach to a bar a bit longer than the chariot and paralele to the axe, in case of crash the bar will lift the roller from the screw absorbing crash energy through the spring and saving the machine it could also be use as the traditional (but lousy)half nuts

This system is amasing, but, to me it won't be easy to figure out the springs stiffness and find a supply for this size/stifness spring.

Thanks !
cnc2.

on post 267 you have a good picture of the flange
total thickness 15 mm
4.9mm depth here for the thrust bearing 4.9 in the roller
the depth is dependant of the thrust bearing
the diameter will be external diameter of the thrust bearing +excentricity
5.1 mm for the wall the diameter = diameter of the axe+excentricity here 8mm+3mm
5mm depth for the excentric
the diameter is = to diameter of the axe+excentricity+what it need to make it strong enough with the guide line that all the diameters should never get in conflict with the axe :nono:
the center diameter is diameter of the screw+(as less as possible so it will be more easy to adjust here 21mm for 20 mm screw)

Thanks Lucien !

I'm sorry to tell you this but, i didn't really understand what you wanted to say, so, make an UML diagram ! :p
Man, a picture with comments on it or a hand draft even with MS Paint, is worth a thousand words.

Especially here

the excentric is your choice I choose complete disengagement
for the roller do not forget to calculate the position of the rings
in the case of 3 rolls 2 rolls with a ring on the left and one with the ring on the middle one of the 2 identical roll will be put in opposition with the other so left midle right

Regards,
cnc2.

Thanks ! the size seems to be the only real problem here.
What do you think about these ?
it might be really difficult to use'em but who knows...


i can tell you that on skf for 15mm axe you have 42mm outside on the tapered bearing you show me you do not have external cage so it mean that you have very high precision machine and a good knowledge in metalurgy to harden your roller .
If this is not a problem then probably 2 tapered bearing will make the job, one at each extremity |>=====<|but if you look for bearing watch very carefully all the dimensions and remember that the external dimension must be smaller than the roller minus the height of the rings




http://www.skf.com/skf/productcatalogue/jsp/viewers/productTableViewer.jsp?&lang=en&tableName=1_14_1&perfid=131002,131003,131005,130002,130005,131000,130000,&presentationType=3&startnum=1

i can tell you that on skf for 15mm axe you have 42mm outside on the tapered bearing you show me you do not have external cage so it mean that you have very high precision machine and a good knowledge in metalurgy to harden your roller .
If this is not a problem then probably 2 tapered bearing will make the job, one at each extremity |>=====<|but if you look for bearing watch very carefully all the dimensions and remember that the external dimension must be smaller than the roller minus the height of the rings




http://www.skf.com/skf/productcatalogue/jsp/viewers/productTableViewer.jsp?&lang=en&tableName=1_14_1&perfid=131002,131003,131005,130002,130005,131000,130000,&presentationType=3&startnum=1

Great info, thanks a lot man ! (i especially like the |>=====<|, very neat !)

Yes, the needed precision will make it hard for DIY production, but i saw someone on the forum that said he could make any needed parts in China & get them for cheap...don't remember who.

I don't think hardening to be a real issue for rollers of that size, a small furnace could do the job.

Regards,
cnc2.

ok a bit more neat
but if you can harden
make economy and just buy the neddles this will save a lot of rooms

That's tight man ! (what software did you use ?)

So, what would be the roller's final diameter ? is it smaller than your proto's diameter ?

Any idea on how to calculate the conicity of the roller sides ?

***EDIT***
I've found these (http://www.kml-bearing.com/products/bearings/needle_roller_bearing.jsp?bearing_no=AXK0414TN&cat_id=061201), machining the rollers could become a lot easier !

Thanks !
cnc2.

hello,

this is made with solidwork.
Well if you want to do it with a cone since i am not mechanical engineer the only thing i can tell you is greater the angle better is the x force and smaller the y here the cone is 18 deg .
thrust needle is a very good solution for the x force 90 deg but you need to put a needle bearing inside 0 deg so you solve the y problems and then you come back to my solution which is the most economic if technically primitive
note:i show you on the fig that for 3 roller (here 20mm diam 50 mm long) the ring do not start at the same place (3 start) this is VERY important
it's easy to do
in the case of 3 roller
if you have 4 mm pinch
the first roller start at 0 mm of the begining of the ring
the second at 4/3 mm of the begining of the ring
the third at (4/3)*2 mm of the begining of the ring
you cut at the position you calculate and then 50 mm after

concerning the diameter smaller is nicer but more expensive
if you can find for a screw of 20mm a thrust needle bearing that can take an axe of 8mm with an external diameter of 17mm and if your teeth are not more high than 1.5mm you can use a roller of 20.1 mm
if your axe is very hard steel you can use 6mm and may be external diameter 13mm then you can use a roller of 16.1 mm

hello,

this is made with solidwork.
Well if you want to do it with a cone since i am not mechanical engineer the only thing i can tell you is greater the angle better is the x force and smaller the y here the cone is 18 deg .
thrust needle is a very good solution for the x force 90 deg but you need to put a needle bearing inside 0 deg so you solve the y problems and then you come back to my solution which is the most economic if technically primitive


concerning the diameter smaller is nicer but more expensive
if you can find for a screw of 20mm a thrust needle bearing that can take an axe of 8mm with an external diameter of 17mm and if your teeth are not more high than 1.5mm you can use a roller of 20.1 mm
if your axe is very hard steel you can use 6mm and may be external diameter 13mm then you can use a roller of 16.1 mm


I have no idea of the thrust needle's price, but for the needle bearing, they can be found at shops that sell spare parts for 2 strok engines(for scooters & mopeds), they are used for the piston pin wich is 10mm dia. on a 50cc engine, some of them have an outer housing so, no need to harden the center of the roller(so it won't "eat" your screw)... and they are cheap 2.5eur aprox.
This is what they look like:
http://www.vxb.com/Merchant2/graphics/00000001/kit786-1.jpg

The piston pin is 30mm long on a 40mm piston. maybe too short but no hardening needed and can find a used one for nothing they usualy remaine "brand new" because the piston is dead before the pin starts to wear out.

I agree that simpler is better as long as you don't get tied by the size issue.

note:i show you on the fig that for 3 roller (here 20mm diam 50 mm long) the ring do not start at the same place (3 start) this is VERY important
it's easy to do
in the case of 3 roller
if you have 4 mm pinch
the first roller start at 0 mm of the begining of the ring
the second at 4/3 mm of the begining of the ring
the third at (4/3)*2 mm of the begining of the ring
you cut at the position you calculate and then 50 mm after


Thanks for the note, i didn't understand that detail on the fig & thought that it was a "software default" ruler.

I've made a fig for your exemple (yes with paint) please correct it if it's wrong.

Dreaming WARNING: if you find a way to put a joint between the flange and the screw, then you could make an epoxy resin housing for your roller nut(between the two flanges), fill that housing with engine oil ... yea some call it the nirvana :)

Thanks !

cnc2.

hello,
have you seen my post 267 ?
because this is what i use
due to the very slow rotation grease can be use
regarding your calculation this is corect in the case of a disposition in star 120 deg

here an exemple of flange

here an exemple of flange
Thanks a lot Lucien for the amazing figs !

Now i understand what excentric you were talking about... but wait....what is it used for ?
Do we need three excentrics on each flange ? and how are they oriented from each other ?

For your post 267, i saw it, but the needle bearing was "hidden to the naked eye", did it make the outer diameter of your roller bigger ? if no, then your design only needs small thrust needle bearings, to become smaller.

if i wanted to make a housing filled with oil it's to be able to forget about the roller nut because no contaminents would get inside... no cleaning or maintainece needed.

Thanks !
cnc2.

[QUOTEFor your post 267, i saw it, but the needle bearing was "hidden to the naked eye", did it make the outer diameter of your roller bigger ? if no, then your design only needs small thrust needle bearings, to become smaller.

[/QUOTE]
yes !!! small thrust needle bearing
excentric are very easy to make
since your screw is fixed and something as to move to be in contact yes you need excentric
you must put them in a way that the roller are not engaged at 12 o clock
then you turn them on the right or on the left but all in the same direction. the screw should not be bend (of course)the pressure apply should never be such than the roller do not roll
more pressure less backslash
it is less complex to adjust than it seems and the excentric will help you for this operationsince you c an adjust each roller separately.
If you choose soft metal it will give you the oportunity to compensate the effect of the time meanwhile if your roller are long enough since they roll they do not absorb a lot of energy so they will not be eaten very fast (much less than half nuts)
if you make your roller in hard metal the profile must be perfect I suggest that you make your first roller in bronze use them and see with the time what is the good profile since when the profile is perfect they should no be eaten

Many thanks Lucien for this explanation !

I first thought that the excentrics would turn providing a "funny" mouvement of the rollers, but now that you said it is just for adjusting the rollers on the srew it all makes sense, it's a realy smart and easy way of tweaking the nut ... i agree.

Any idea on how to make the profile, or what to ask the machine shop for ?

Thanks !
cnc2.

I first thought that the excentrics would turn providing a "funny" mouvement of the rollers, but now that you said it is just for adjusting the rollers on the srew it all makes sense, it's a realy smart and easy way of tweaking the nut ... i agree.

it also give the necessary pressure on the thrust bearing
about the profile
Is there any mechanical engineer that can give us a trick ?

Hi Lucien, hi everybody,

For making a smaller roller nut, i think the best design is this one(in the video) (http://www.cnczone.com/forums/attachment.php?attachmentid=14078&d=1137973000), look at the rollers they are pretty small, i think 15 mm dia. roller is reachable with that design (excentrics would be hard to "implement" but still possible).

I've attached a fig i've made (with paint).

For the ball bearing you can use those miniature bearing found inside hard drives those of the read/write head they are 12 MM O.Dia the I.Dia may differ, with'em excentrics are still possible to implement but with less adjusting clearence.

For the thrust needle bearing those (http://www.kml-bearing.com/products/bearings/needle_roller_bearing.jsp?bearing_no=AXK0414TN&cat_id=061201) could be used, they are 4 mm I.Dia and 14 mm O.Dia. but i've no idea of their price...

So, what do you think about it ?

Thanks !
cnc2.

For the thrust needle bearing those could be used, they are 4 mm I.Dia and 14 mm O.Dia. but i've no idea of their price...


they are probably not very expensive the dimension is a problem if the axe is only 4mm that is very (too )small
the needle bearing inside the roller do not take much room and it is very easy to insert
about the excentric he too cannot be biger than the roller
in your fig the blue part will conflict anyway the thrust needle bearing is the problem regarding the size
the mention "if possible" for the excentric, is a bit strange. you may use another system to adjust, but, you need one. the precision required to make a roller screw without it is out of reach for most of us and for a unique piece even the "pro" will use an excentric

about the video sorry i cannot read it (use youtube it's easy)

they are probably not very expensive the dimension is a problem if the axe is only 4mm that is very (too )small
the needle bearing inside the roller do not take much room and it is very easy to insert
about the excentric he too cannot be biger than the roller
in your fig the blue part will conflict anyway the thrust needle bearing is the problem regarding the size
the mention "if possible" for the excentric, is a bit strange. you may use another system to adjust, but, you need one. the precision required to make a roller screw without it is out of reach for most of us and for a unique piece even the "pro" will use an excentric
Hi Lucien,

The blue part is just here to show the position and support of the ball bearing and the thrust needle bearing, man it's done with Paint so i didn't take care about the shape of the blue part. around the ball bearing that is 12 or 14 mm Dia you can put an excentric that will provide 2 or 3 mm of adjustment play, and i think the real adjustement play needed is less than the depth of the thread. If you think the excentric system is impossible to implement so we'll have to find an other one.

For the video it's not mine i've found it in the attachements of the Thread and it gave me the idea, it is in a zip file, me too i can't play it but with media player classic you can see some of it... the important in this vid is the shape of the rollers.

Thanks !
cnc2.

Well,
you are not obliged to have an excentric that provide a full disengagement
But i am asking me why to choose this way ?
make a hole is very easy, and the needle bearing is cheap. different size of cages can be found and the size problem do not come from here

Well,

Look at the pic, i'm dreaming of reaching that size of rollers and for this, hard drive bearings are the best + thrust needle bearing, IMHO.

Do you think we can make a flage of the needed precision to be able to implement that design ?

(don't worry about the outter "nut" because it's a more complex system i'm not even dreaming of)

Thanks !
cnc2.

hello cnc2.
Or you are a very high grade pro and in this case i cannot bring you any valuable info
Or you are a dreamer and you never produced anything
by the way read this thread (I know it's not short) you will see that in the case of satellite roller you have no force on the extremity of the roller everythong is on the side of the multistart screw so no need for thrust bearing.
Since such superb piece of mechanic cannot be made by hands I suggest that you get access to a good design software in order to be processed by your <>1 million$ machine !!!
so long

I wait for the pictures of your realisation

hello cnc2.
Or you are a very high grade pro and in this case i cannot bring you any valuable info
Or you are a dreamer and you never produced anything
by the way read this thread (I know it's not short) you will see that in the case of satellite roller you have no force on the extremity of the roller everythong is on the side of the multistart screw so no need for thrust bearing.
Since such superb piece of mechanic cannot be made by hands I suggest that you get access to a good design software in order to be processed by your <>1 million$ machine !!!
so long

I wait for the pictures of your realisation
Man i told you not to look at the whole system, the pic was just there to show the shape of the roller ends, and the size of the rollers compared to the size of the screw, read my previous post, i said i wasn't even dreaming of it ... The design i was talking about is the one of the small rollers with a shaft going out of them. Is it possible to design and make a flage for that size of rollers with the precision of a simple machine shop ? That's what i meant in my previous post.

Sorry for the confusion !

cnc2.

Man i told you not to look at the whole system, the pic was just there to show the shape of the roller ends, and the size of the rollers compared to the size of the screw, read my previous post, i said i wasn't even dreaming of it ... The design i was talking about is the one of the small rollers with a shaft going out of them. Is it possible to design and make a flage for that size of rollers with the precision of a simple machine shop ? That's what i meant in my previous post.

Sorry for the confusion !



no I read well, but evry approach as is drawback by putting the load on the side the planetary does not need any thrust bearing so this technology will not bring you any solutions at all
Many engineers have worked on this subject if the solution i am using is not widely used it's mainly due to his size it can be improved but there is some limits i suggest that you start making one very simple and then build another one more sophisticated

rokag3,

Your work is awesome and as far as I can see no one else has made anything like it.

JerkyBoy

I don't know but all material I've managed to come across lists roller screws as having superior accuracy and repeatability to ball screws and due to the large number of contact points also exhibits better stiffness and load resistance.

But all this is way beside the point. What I wanted this thread to concentrate on is making a DIY version of it as ball screws are nigh on impossible to make at home and these suckers look a lot less complicated.

And here are some examples of what people have allready done, albeit much simpler and cruder versions of what the real deal looks like:

from posix post 13 with beautiful pictures

from posix post 13 with beautiful pictures
Thanks Lucien, i had missed those pics.

So there's definetely a way to make your design smaller, and posix did it, with no bearings at all, but i don't think it will stand heavy efforts without locking or slowing... would be enough for a small machine.(i assume the outter pipe, on posix's roller screw, isn't threaded... that's what it looks like)

And please forget about planetary or satellite roller screws, i just wanted to make your design smaller by keeping the same concept and trying to improuve the roller mounting system to be able to use smaller rollers.

Regards,
cnc2.

So there's definetely a way to make your design smaller, and posix did it, with no bearings at all, but i don't think it will stand heavy efforts without locking or slowing... would be enough for a small machine.(i assume the outter pipe, on posix's roller screw, isn't threaded... that's what it looks like)

on th post 13 it is with bearing
on the post 16 this is a prototype we will call "for educational purpose" which never work read all the post if you want to know the story.
on the hungarian site you can see a very primitive system that can work pretty well the size of the roll is mainly due to the size and type of the bearing used
one of the solution to make smaller roll will be to make your own flanges for the thrust bearing or to find a cheap big inner diam small outer with needles or balls at a good price (maybe in big photocopier)

never saw bearings in photocopiers but will see.

Thanks !
cnc2.

hellon
If you can find big photocopiers, tou have
servos motors with encoder
bearings of all kinds
axes of very good quality that you can use as raw material for lathe
capacitors and power suply
optical
aluminium that you can use for casting
so you should consider phorocopiers as a mine
I am speaking for photocopiers like big xeros for example
if they weight more than 100kg it's probably good

hellon
If you can find big photocopiers, tou have
servos motors with encoder
bearings of all kinds
axes of very good quality that you can use as raw material for lathe
capacitors and power suply
optical
aluminium that you can use for casting
so you should consider phorocopiers as a mine
I am speaking for photocopiers like big xeros for example
if they weight more than 100kg it's probably good
Hi Lucien,

The only copiers i saw, were small sized like the cannon 6030 ...etc, all of them had stepper motors (5 phase 500 steps/rev nema23 0.46Nm) at the exception of one that had a 20 VDC geared motor (http://www.cnczone.com/forums/showthread.php?t=54079)with a tachometer on it, but i'll look for bigger copiers.

When you say "axes" do you mean the linear rails of the copier carriage ? if yes i'm thinking of using them as linear slides for YX table for the mill but have no idea about their rigidity.

I want to add that on some copiers you can salvage timming pulleys and belts.

Okay i know it's the (wrong)

Thanks !
cnc2.