What's new
What's new

DIY Roller Rack & Pinion - Need solution from experienced machinists

denhen89

Plastic
Joined
Feb 1, 2024
Hello Guys,
i am not sure if i have ever posted something here, so i will introduce myself as briefly as possible: I am just a hobby machinist who has build a couple of machines like for example a 4x8 CNC Router (linear rails and ballscrews on all axis) and also i converted a old Polish heavy industrial Lathe to CNC with Linear rails, ball screws and servo motors. In near future i would like to use the machines for a side business.

I am currently thinking to machine my own Roller rack and pinion system. (like the NEXXEN RPS)
When it comes to the pinion, i will soon have what i need to machine it very precisly on my CNC Router. I have ordered a 80:1 harmonic drive that i will set up like a vertical lathe machine, but i will not loose time to explain how i exactly will do it, because i think its not that difficult.
But, when it comes to the rack i need some help. The rollers that will run on the rack are from bearing steel (HRC ~60) , and therefor the rack should be be hardened as well and here is the problem:
After hardening, the rack would need to be ground on a grinding machine. Is there any way to avoid the grinding process?

I thought about this options:
- Laser hardening the rack: (i have read that after laser hardening, the grinding process can be avoided. I have contact one company here in Poland, but i am waiting for a reply)
- Brass Rollers or other alternative material: Instead of standard bearing-steel rollers, i could for example use ground brass rollers (havent found such brass rollers to buy) which would run on a e.g. hot-rolled steel rack. I am not sure how fast the rollers would worn out, compared to bearing steel rollers on a hardened steel rack, but the Rollers could be replaced very quickly, without dissambly of the pinion or the rack.

Ps: I thought to machine the rack on my CNC router, but not the whole rack with CNC, but only one "cutout" at once (for the roller-tooth of the pinion). Then move the stock-material of the rack by the specific amount, and then machine the next cut-out. I would use a 1um linear encode to move the rack precisely by the specific distance for every cutout. This would minimize the error - i think and i hope :D.
But, what about the surface finish? Do you think milling can achieve a good enough surface finish for such roller pinion system?

Of course, i could get it done by grinding, but this would be done my lathe machine that i would need to customize a bit, so that i can also create a profile of the grinding wheel, but the problem is that i could also only grind one cut-out at once, then move to the next cut-out of the rack. Due to the fact that i have no experience with grinding, i cannot tell how fast the grinding wheel will worn out on hardened steel, therefor i dont know if i would keep the accuracy from the first cut-out of the rack to the last one.

I hope you can give me some tipps. Thanks in advance and i wish you all a nice weekend.
 

Attachments

  • roller-pinion-racks-1024x682.jpg
    roller-pinion-racks-1024x682.jpg
    49.8 KB · Views: 34
also i converted a old Polish heavy industrial Lathe to CNC with Linear rails, ball screws and servo motors.

That’s pretty neat, photos please!

I am currently thinking to machine my own Roller rack and pinion system. (like the NEXXEN RPS)

Why? You cannot purchase the roller rack and have it shipped to Poland? You’ll not make this with a router with any degree of accuracy.

After hardening, the rack would need to be ground on a grinding machine. Is there any way to avoid the grinding process?

Ok we’re going down this road then. Flame harden the engagement surfaces. Polish by hand with an abrasive pad (3M red) to not mess with the geometry, not that the geometry will be correct to begin with.

I thought about this options:
- Laser hardening the rack: (i have read that after laser hardening, the grinding process can be avoided. I have contact one company here in Poland, but i am waiting for a reply)
- Brass Rollers or other alternative material: Instead of standard bearing-steel rollers, i could for example use ground brass rollers (havent found such brass rollers to buy) which would run on a e.g. hot-rolled steel rack. I am not sure how fast the rollers would worn out, compared to bearing steel rollers on a hardened steel rack, but the Rollers could be replaced very quickly, without dissambly of the pinion or the rack.

Ps: I thought to machine the rack on my CNC router, but not the whole rack with CNC, but only one "cutout" at once (for the roller-tooth of the pinion). Then move the stock-material of the rack by the specific amount, and then machine the next cut-out. I would use a 1um linear encode to move the rack precisely by the specific distance for every cutout. This would minimize the error - i think and i hope :D.

That’s optimistic. You’d be better off to keep the cutting action preloaded in one direction and make successive profile cuts from one end of the rack to the other (as in laying the rack down and profiling out of a flat) But again, the accuracy of this will depend on your router set up.


But, what about the surface finish? Do you think milling can achieve a good enough surface finish for such roller pinion system?

Probably not, but if all you have is a hammer…

Of course, i could get it done by grinding, but this would be done my lathe machine that i would need to customize a bit, so that i can also create a profile of the grinding wheel, but the problem is that i could also only grind one cut-out at once, then move to the next cut-out of the rack. Due to the fact that i have no experience with grinding, i cannot tell how fast the grinding wheel will worn out on hardened steel, therefor i dont know if i would keep the accuracy from the first cut-out of the rack to the last one.

You won’t be able to manually. Both the rad and the tangents are critical surfaces for motion control, an inexperienced grinder will make a dogs breakfast out of this. I believe you’d be setting yourself up for a frustrating time and an inferior system. Like if I was designing/building a CNC whatever for my own purposes, the last thing I’d make myself are the ball screws. Right?
 
Poland - huh? Can't help with your questions, but I struggled to find quality machining capability in Poland - I know it exists as Poland makes some good stuff, but outside of that the pickins were slim or we didn't know where to look. I've been to Poland 40 or 50 times..............
 
You won’t be able to manually.
Finally a suitable job for #3 Moore CNC jig grinder ! (honest !)

Okay, I thought of a way but it would be ridiculous.

This is essentially a cycloidal rack.

So rough it in, get it within say ten thou ?

Now you get a borazon wheel of the diameter of your roller. Goes on an electric or air spindle, which you mount on a disk at the diameter of your base circle.

Now you devise some sort of a driving arrangement - maybe a band wrapped around a disk the size of the base circle ? Then you translate your disk along the rack as it rotates from the bands. Then index the grinding wheel to the next roller location and do it again, until you've run the wheel past every tooth.

Did that make sense ? It's ridiculous but I think would work.
 
@RC Mech
That’s pretty neat, photos please!
Pictures are attached :)
I love my Lathe. I mainly used it as a "manual lathe" controlled with the 2 MPG handwheels on the panel. The rails are SKF 45mm on the Z axis and 25mm on the X-axis. 4kw motor controlled with a Siemens VFD + spindle encoder. Cutting through steel like through butter. I bought it on a scrapyard and gave it a new life:)
Why? You cannot purchase the roller rack and have it shipped to Poland? You’ll not make this with a router with any degree of accuracy.
Yes, of course, but firstly that system is very expensive and secondly i thought to build such system and try to make a small business out of it. But thats just a idea, i like to try new things out and it could be useful for a project i am thinking about that i might build in future.
Ok we’re going down this road then. Flame harden the engagement surfaces. Polish by hand with an abrasive pad (3M red) to not mess with the geometry, not that the geometry will be correct to begin with.
No. If i would build such systems and like to make a business out of it, then it has to be close to tolerances as for example the Nexxen RPS system has. I might be too optimistic, but i created this post to get a more realistic view of it.
Actually, i milled a test rack for 4mm rollers. Also i created a pinion, just for fun, without needle bearings. So the 4mm rollers are just inserted in into a almost perfect 4.00 diamer hole. They are not rolling inside the hole, so the movement on the rack is not very smooth, but 0 backlash. (at least i cannot feel any).
I attach pictures in the next post, because i can only upload 10 images per answer.

Please keep in mind, i milled it on a day where i did have not a lot of time, so i mounted the 6mm plate with 2 sided tape on my cnc router table. Also, the pinion were milled with 0.11 mm runout with a 3mm tool. I knew something was not right, but i did not care about the surface finish of the outside surface of the pinion plates. The rack has been milled with a 4mm tool and about 0.01 runout - without any finishing toolpathes.
That’s optimistic. You’d be better off to keep the cutting action preloaded in one direction and make successive profile cuts from one end of the rack to the other (as in laying the rack down and profiling out of a flat) But again, the accuracy of this will depend on your router set up.
Maybe i dont understand how you mean it exactly, but also its possible that you dont know how i would exactly do it. I would need to make couple of drawings, step by step to show how i would do it, but for sure its not as acurrate as a perfectly profiled grinding wheel that will finish the rack.
You won’t be able to manually. Both the rad and the tangents are critical surfaces for motion control, an inexperienced grinder will make a dogs breakfast out of this. I believe you’d be setting yourself up for a frustrating time and an inferior system. Like if I was designing/building a CNC whatever for my own purposes, the last thing I’d make myself are the ball screws. Right?
Its possible you might be right here, but i will do some tests in the next week. The harmonic drive is on the way, as well as the (cheap)-needle bearings for the rollers and then i will machine a new rack. I have a precisly machinened axis with a HGR15 rail, so i can build a 500mm long test axis with the roller pinion system and see how it works.

@markz528
Poland - huh? Can't help with your questions, but I struggled to find quality machining capability in Poland - I know it exists as Poland makes some good stuff, but outside of that the pickins were slim or we didn't know where to look. I've been to Poland 40 or 50 times..............
Where are you from if i may ask? I am asking, because you were so many times in Poland. I am actually born in Germany, but i am living here in Poland since about 10 years. I think here are enough companies with quality machining capality, but it depends what you think about exactly. For example, my Lathe surfaces for the rails has been machined by a company that renovates machines and they do a very good job. Also my tailstock is made by them and they did a excellent job. The machine that has milled the surfaces of the rails is big as 2 tanks :D. Old machine, but holds 0.01/m. It was a old shop, but experienced machinist. I dont know if they had CNC machines.
When it comes to newer machines and complex parts, i think of a company that is about 5 minutes drive from my place. https://ragor.com.pl/
You can check it. I was once there to ask if they can machine the ends of the 3 meter ball screws for my cnc router, but they had no time for my job:(.
I could name a couple companies, but i think generally here are a lot of machine shops. Maybe not as high end as in Germany, but for me it seems that maching is quiet a typical thing here in Poland. I have at least 2 guys in my family who are machinists.

@EmGo
Finally a suitable job for #3 Moore CNC jig grinder ! (honest !)
I have never seen a jig grinder :D. Interesting, but i have to spend a bit of time to understand how it works. From a couple of short videos it seems that it would be really a good machine for that job, but i dont have it :D.
Okay, I thought of a way but it would be ridiculous.

This is essentially a cycloidal rack.

So rough it in, get it within say ten thou ?

Now you get a borazon wheel of the diameter of your roller. Goes on an electric or air spindle, which you mount on a disk at the diameter of your base circle.

Now you devise some sort of a driving arrangement - maybe a band wrapped around a disk the size of the base circle ? Then you translate your disk along the rack as it rotates from the bands. Then index the grinding wheel to the next roller location and do it again, until you've run the wheel past every tooth.

Did that make sense ? It's ridiculous but I think would work.
I am interested, but i dont know if i understand it correctly. Do you mean something close to this here:
 

Attachments

  • TUD-50__SKF 3.png
    TUD-50__SKF 3.png
    666.3 KB · Views: 31
  • TUD-50__SKF 2.png
    TUD-50__SKF 2.png
    603.3 KB · Views: 29
  • TUD-50__SKF 1.png
    TUD-50__SKF 1.png
    642.5 KB · Views: 35
  • 7.jpg
    7.jpg
    2.5 MB · Views: 38
  • 6.jpg
    6.jpg
    1.1 MB · Views: 38
  • 4.jpg
    4.jpg
    1 MB · Views: 36
  • 3.jpg
    3.jpg
    34 KB · Views: 36
  • 2.jpg
    2.jpg
    2.1 MB · Views: 37
  • 1.jpg
    1.jpg
    2.3 MB · Views: 40
*
Please keep in mind, i milled it on a day where i did have not a lot of time, so i mounted the 6mm plate with 2 sided tape on my cnc router table. Also, the pinion were milled with 0.11 mm runout with a 3mm tool. I knew something was not right, but i did not care about the surface finish of the outside surface of the pinion plates. The rack has been milled with a 4mm tool and about 0.01 runout - without any finishing toolpathes.
 

Attachments

  • test_roller_pinion_rack___3.jpg
    test_roller_pinion_rack___3.jpg
    90.9 KB · Views: 19
  • test_roller_pinion_rack___2.jpg
    test_roller_pinion_rack___2.jpg
    80.5 KB · Views: 20
  • test_roller_pinion_rack___1.jpg
    test_roller_pinion_rack___1.jpg
    70.3 KB · Views: 18
  • test_roller_pinion_rack___4.jpg
    test_roller_pinion_rack___4.jpg
    92.4 KB · Views: 16
  • test_roller_pinion_rack___5.jpg
    test_roller_pinion_rack___5.jpg
    54.7 KB · Views: 18
  • test_roller_pinion_rack___6.jpg
    test_roller_pinion_rack___6.jpg
    51.7 KB · Views: 20
I have never seen a jig grinder :D.
The #3 moore is a bit of a joke around here now, sorry :) But yes it would do your job.

Do you mean something close to this here:
Yes I mean that method of rotating the part while simultaneously translating. That's an old method but it's accurate.

No, what you would have to do is different because you want to make the rack, not the pinion and also because it is a cycloid, not an involute. But the bands rotating the device during translation, yes.

I was a little shocked - he went to all that trouble to make a nice setup, then mis-indexed and instead of making a new one that was right, just jammed the part in with two bad teeth. Sheesh ! To get so close then just hackjob butcher the finish, sad :(

Personally, a section of ground rack with a ground pinion would probably do what you want faster and cheaper but sometimes, we gotta do things the hard way :)

IMG_7202.jpg
 
I do not see indexing to cut single teeth being accurate. Anything off at all adds up fast, fewer indexes, longer paths in the router/mill.

Undersize teeth then diamond paste disposable rollers and mesh the two together?

the roller rack is neat, even with high precision on all parts accuracy is not going to be as good as rack/pinion with same build precision. Every moving part introduces play. Now if you are going for speed with ok accuracy then the roller rack becomes viable.

Linear motor.
 
I do not see indexing to cut single teeth being accurate.
That's how it's done tho :)

I don't see this thing as being much more than a gimmick either, but you can index between teeth pretty accurately.

Of bigger concern is, any inaccuracy in mounting or the flatness of the rack or the parallelism of movement or pretty much anything, will have a much bigger impact on this system than involutes with a rack. That's the reason involutes are so common and cycloids not - cycloids are really really sensitive to any mounting being imperfect.

Other than that they are better, there's a lot less sliding, but .... real is the sworn enemy of perfect, guess which one usually wins :)
 
Where are you from if i may ask? I am asking, because you were so many times in Poland. I am actually born in Germany, but i am living here in Poland since about 10 years. I think here are enough companies with quality machining capality, but it depends what you think about exactly.
I'm from Connecticut originally. Company I work for owned a factory in Poland for around 30 years. I was travelling to the plant over a 25 year period. We sold it a couple years ago.

We were looking for capability in the electric motor repair business. It was poor. We looked all over the country. I have no doubt there is strong capability in the pure machining world, just wasn't where we were looking.
 
I have no experience with this sort of drive, but it sure reminds me of a bicycle chain and sprocket, except reversed. Have you thought about inverting your system: make a "rack" of dowels in a channel and a conventional-looking sprocket that engages them? A toothed pinion would be easier to make, easier to harden, and the dowel-rack could be made in segments to get you the required length. Dowels are cheap, and as you point out, you want the pinion to wear before the rack.
 
I have no experience with this sort of drive, but it sure reminds me of a bicycle chain and sprocket, except reversed.
It looks similar but it's a little different.

A chain sprocket is not truly a circle - it's a polygon. The arc of a sprocket tooth is described by swinging a point from the next pin, one link over. Chain sprockets don't rotate at a constant speed, they actually speed up and slow down as the links go into and out of the mesh. You can see this on high-speed video, that's why motorcycle drive chains vibrate.

Cycloidal teeth is, if you take a big circle, say 12" diameter. That's your base circle. Now take a small circle, say 2" and mark a point on it, maybe at a radius of 3/4". Roll that smaller circle around the big circle while you let the pin trace a line on paper. Those arcs are cycloids. (Calling David_M ! Calling David_M !).

If you run that in reverse, with the pin on the small circle driving the big circle, if the small circle rotates at a constant speed then so will the big circle. This system just spreads the large circle out into a flat shape. A 'rack' will have straight sides on the teeth, this will not. But the tooth shape is not just simple arcs from link to link, either, like a chain sprocket would be.
 
Chain sprockets don't rotate at a constant speed, they actually speed up and slow down as the links go into and out of the mesh. You can see this on high-speed video, that's why motorcycle drive chains vibrate.
Hence the recommended minimum sprocket size of 19 teeth.
 
I ran a rambaudi 5 axis that used rack and pinion drive. I'm pretty sure they were not roller pinion. I would use a standard off the shelf rack and pinion and sort out the inaccuracies using pitch error compensation. During covid I built a automated press for large PCB's using rack and pinion, it was surprisingly accurate. There are some nifty ideas around backlash control on rack and pinion drives.
 
It looks similar but it's a little different.
Well, that sent me down an internet rathole. But thank you, it was interesting.

It looks like the Nexen stuff achieves "zero backlash" or "near-zero backlash" (depending on what page you read) by always having at least 2 rollers in contact with the rack at all times, where one of said rollers contacts the left side of a tooth and one contacts the right side of a tooth. So, I guess the tooth profile is pretty special. Screen Shot 2024-03-11 at 10.26.39 AM.png

And I did not appreciate that they are rollers, not just dowels. Because each roller requires 2 bearings, it might be expensive to reverse the system and have a sprocket-type pinion with a bazillion rollers forming the rack. The bearings in the pinion appear to be plain sleeves (like oilite?), "no lubrication is required".
 
Well, that sent me down an internet rathole.
:D Here's someone who went to the trouble of doing animations. It's a lot easier to figure out when you can see it. Whether it's worth doing or not, that's a different question ...

 
roller pin rack and gear pinion, this sounds better. Grit will not stick in the rack teeth. More parts, bad- better life is good.
 








 
Back
Top