Flexible / deformable bearing

[erosnicolau]

Hi guys,
I'm new here, so please forgive me if my question is not right for this forum - but I hope it is.
For an intro to my problem, please watch this presentation of the HarmonicDrive principle.
It contains a reference to an elastic/deformable bearing. I know, from their sources, they're using 440C SS (pretty much the same material used for regular bearings) to manufacture the inner and outer races of these bearings. The elasticity must be coming from the thinness of the races.
I need to come up, for a personal project, with a similar, symmetric cam, able to rotate with minimum friction while maintaining a very rigorous big diameter.
My approach, for now, is to use a bronze cam interfacing with a spring steel sheet - but the frictions (and wear) are still pretty big, especially because the stronger surface contact forces needed for an improved function translate to stronger friction forces.
I am tempted to try to build such an elastic bearing myself:

1. The first idea would be to try and weld some spring steel sheet into the required diameter rings.
The questions raised by this idea would be:
- Is there a realistic way of welding spring steel without significantly losing its structural qualities (mainly, fatigue resistance)?
- Or: is there a heat-treatment for re-hardening the steel after welding?
In a nutshell, would this approach make any sense at all?

2. The second idea would be to find some tubing close to the required dimensions, to pre-machine it to bring it as close as possible to the required dimensions, to harden it and then to surface-grind it to micron tolerances. From a metallurgy point of view, this makes much more sense than the first idea, however sourcing any grade of steel (let alone 440C) in tube form, at largely usable diameters, is pretty much impossible - at least for small, prototype or small run quantities.
The question raised by this idea thus relates to material sourcing:
- Are any of you guys aware of a stockist of 440C willing to source small quantities? What should I be googling for?

Thank you so much,
Eros

 

[TVP]

440C is readily available in tube form. This is just one provider:

http://www.mcmaster.com/#standard-stainless-steel-precision-bushing-stock/=szoovl

 

[erosnicolau]

Hi, TVP, and thanks for the tip.
However, mcmaster is limiting in both that it only stocks sizes smaller than what I need, and that it has prices 3-4 times higher than the same lengths of full round bar available on other sites, which makes no commercial sense whatsoever to me, even after calculating the time-saving of not having to hollow out the parts.
Eros

 

[tbuelna]

erosnicolau-

By "spring steel" if you mean something like high carbon steel (1075 or similar), then that would not be a good choice for a bearing race. A case hardened or through hardened alloy steel would be best. If it requires corrosion resistance, you can coat the race with thin dense chrome.

You did not define what you meant by a "big" race diameter. The most common method of producing large diameter thin section race blanks is ring rolling. The process is fairly inexpensive since it does not require any tooling. With a large diameter thin section bearing race the most difficult problems you will face are minimizing heat treat distortions and accurately finish grinding the ID, OD and race surfaces. It can be done but it requires very specialized equipment. There are bearing companies like Kaydon that specialize in manufacturing these types of bearing races. You might take a look at their off-the-shelf bearings to see if there is one that you can adapt for your application.

Hope that helps.
Terry

 

[erosnicolau]

Hi, Terry,

The "steel spring" idea is already being used here (I'm using a sheet of 0.1mm spring steel with the ends cut at 45*, held together externally with an elastic garter), as a sliding interface, and it works. Well sort of: if I am using too thin of a sheet, then it is subject to micro-buckles at the contact points; and if I'm using too thick/strong of a sheet, then I can't find a garter strong enough to force it to assume the cam shape correctly. Since this is not being used as a bearing race but simply as a "bronze-steel" sliding interface, I excluded the bearing-specific hardening considerations for now.

Of course, this is the less optimal and thus less desirable option.

Moving on to the real bearing solution: To answer your question, by "big diameter" I meant the longer of the two diameters of the oval cam, and also the external surface of the outer race. Ring rolling sounds like a very promising idea just by the name of it: I will be googling this one! If you do have any advice on that, please share!
About hardening and subsequent finish grinding: I am lucky enough to have access to some pretty good hardening guys around here, and also to an excellent finish grinding expert - so I'm pretty much set up in this compartment. My idea is to start, by finish-grinding (2-3 micron tolerances) from the calculated outer diameter of the hardened outer race (which is the only truly fixed, un-negotiable dimension), to finish-grind the interior surface of the same race until we find that it is elastic enough to bend to our needs, and then, from this newly found diameter, to subtract the diameter of the rolling elements to obtain the outer diameter of the inner race. After that, it's pretty straight-forward: I can calculate the inner diameter of the inner race and, based on that, its circumference. From that I can then calculate the amount of offset to be taken from the cam, to accommodate the whole thing.

Kaydon has long been contacted before...
1. all their thin section bearings are pretty much square in the side section, while my need is for a very rectangular section (3-4mm wide x 11mm high)
2. those are not bearings meant to be bent (they are hardened for toughness not for elasticity)
3. they do have some prohibitive (to say the least) prices - how about 5-600$ for a piece?

What's odd is that the guys at HarmonicDrive rely on this type of bearings completely, yet there's not a single independent manufacturer of such bearings out there. My guess is they have their own bearing factory. In which case, I don't get it: why do they use cross roller bearings made elsewhere? 'Tis a mixed up world we live in...

So right now, I am looking for a stockist who will not try to sell 3m long bars when I need 40mm for a bearing... For now, the only one that fits the bill somewhat is SpeedyMetals - although I do have to mention that for any small quantity they are charging an arm and a leg for shipping (imagine around 400$ for shipping a 20$ part), unless you contact them directly to ask for a better shipping price.

Thanks,
Eros

 

[erosnicolau]

OK, just enlightened my nooby self on ring rolling

I'm noticing it's definitely NOT something I could accomplish in our little shop, but rather an industrial process. And speaking of industrial processes, the engineer from the Chinese factory I contacted for the possibility of manufacturing such bearings told me that they make the bearing race blanks by mold pressing - so probably that's the most efficient method of getting rolls of smaller sizes.

So I'll be counting on some good old lathe work for this, for now

Eros

 

[tbuelna]

erosnicolau-

Seamless ring rolling. This is the most common process for producing bearing race blanks since it is seamless, does not require dedicated tooling, can produce virtually any diameter and cross section from a wide range of wrought bar stock, and produces excellent grain properties in the finished ring.

If you want to make a ring from spring steel, you might be able to roll form a strip of the spring steel into a hoop and use a solid state process like friction welding to join the ends. You would still need to put the hoop through a full heat treat cycle to stress relieve it and give it the mechanical properties you require.

The reason I suggested looking at a Kaydon thin section bearing is because the races are through hardened. So if you were to re-grind the outer race OD, the resulting surface would still have a hardness similar to the original surface.

 

[erosnicolau]

Hi again, and thank you for the link!

It confirms what I found by myself yesterday, too: that ring-rolling is for larger diameters, generally (my bearing would be 2" OD, while ring rolling is for minimal 7.5" IDs)
If you say seamless ring rolling is the most common process for producing any size of bearing races, it means I simply haven't searched hard enough for smaller size capable machines?
About Kaydon - didn't know they are through-hardening their races - thanks for the tip!

Eros

 

[tbuelna]

Mass production of bearing races requires careful cost control. The double vacuum melt steel used for bearing races is very expensive, so it makes economic sense to use a rolled ring for even 2" OD bearings.

I also hope that you fully appreciate just how difficult it is to heat treat and finish grind a thin-section bearing race to micro-inch tolerances. If you don't have experience doing this type of work, you will have a very steep (and expensive) learning curve ahead of you. It might be worthwhile for you to approach one of the companies that manufactures harmonic drives about making this bearing for you. In the end it will likely save you time and money.

 

[erosnicolau]

Hi again, @tbuelna,

We only have a small shop here (a couple of engraving CNCs, lathes, a regular milling machine, drill-presses, regular grinders etc.) So for anything else we must find outside help.
We're currently developing an alternate kind of strain-wave gearing mechanism (similar to the harmonic drives, but simpler in construction), which we're trying to patent. Business size aside (HD = huge, we = small shop with creative ideas), HD is a direct competitor in this regard, so getting any kind of help from them seems light-years away.

Here are some videos with our current prototype, grinding away on a lathe, in my hand, for scale and during a backlash test (which so far yields around 0.75...1 arc-minutes of backlash).

The steel-bronze interface we're currently using is working, as you can see, but it generates quite a bit of friction and hence bigger input torques - that's what I'm thinking about the elastic bearing.

As you can imagine, we're still in a late prototyping stage, so money is being carefully pondered and seeking help from mass-producers has proven quite useless so far. So far, the cheapest way to achieve 2 such prototype bearings would be to buy 100mm length of round 440c stock, to rough machine it ourselves in our shop, to have it properly heat-treated and then to have it precision-ground elsewhere (as I mentioned, we do have access to an external guy who already confirmed he has the means to do this). Sure, a rolled ring would involve less machining and (since I'm a green thinking guy) less wasteful - if only I could find a ring rolling mill able to help with such a small quantity and with such small size rings...

Eros

 

[TVP]

eros,

There are other suppliers that have larger diameters than McMaster. I just included them as an example. ELG Carrs in the UK is one example of a metal service center that actually manufactures hollow bar, not a just distributor like McMaster:

http://www.elgcarrs.co.uk/product.asp?ID=57

Another possibility is hollow bar produced by a casting (foundry) process rather than by seamless tube/pipe process or machining of solid by trepanning:

http://www.timcasthollowbar.com/centri-bar/

 

[erosnicolau]

Thank you for the links!
I will write to them, to see what they say
Eros

 

[tbuelna]

erosnicolau-

Sorry for the tardy response. If you have decided you want to use 440C and you only need enough material to make a couple races, there are many online vendors that will sell you a small quantity of 440C bar in the size you require. Just be aware that they usually charge premium prices for small quantities. A 12" piece of 2" diameter CF 440C bar will cost you $77 from these guys.

For your small quantity of races, the cost of raw material will be insignificant compared to the costs for heat treating, precision grinding, tooling/fixtures, etc. If you'd like to see an example of how the track surface of an outer bearing race is finish ground to microinch precision, do some research into the "shoe grinding" process.

Thanks for posting the videos of your device. Looks like a fun project.

Regards,
Terry

 

[erosnicolau]

Hi, Terry,

No problem, I kept busy asking around - including with the two hollow bar suppliers you kindly suggested. As expected, they completely ignored my emails. I'm being neither sarcastic nor pessimistic when I say it's extremely tough to find people willing to engage in small quantities in this field - I'm simply expressing a judgement based on my previous experience.

I know about "onlinemetals" - as I told you before, the best deal I've got so far, elsewhere, is around the same 75$, but that's including the shipping (which in some cases is 2-3 times more expensive than the metal itself)

About shoe-grinding: not sure, but I think that's what my expert grinder uses (I refer to him as a guy, but he actually works as a specialist in a big company on some quite precise machines, he's not a one man shop - and still he has some hesitations...) Looking for shoe grinding I also stumbled upon RON-centric grinding - also very interesting, at least if you believe the marketing hype around it

Anyway, I took your advise (kind of) and also resumed my search for potential bearing manufacturers let's see what the cat drags in...

Eros