[louieatienza]

Harmonic Drives

As explained towards the end of the last chapter, I was not entirely happy with the simple and obvious design for a Rotary Table based on a spindle and two bearings. So far I had been trying to minimise the cost of the bearings, but then I began to wonder what might come from removing this restriction? Could any increased cost for the bearings be recouped elsewhere? It would not be the first time that has happened to me.

Attachment 271698

Well, Deep Groove Bearings (DBGs) are the cheap end of the market, while Crossed Roller Bearings (CRBs) are undoubtedly the expensive end of the market, at least for my application. So I collected some catalogues for CRBs. (Brief note: Crossed Roller Bearings should not be confused with Cylindrical Roller Bearings. The latter are different. Yeah - I had never heard of the latter either.)

Attachment 271700

Can one use CRBs for a Rotary Table? Clearly the answer is not only yes, but it would seem that CRBs are ideally suited to making Rotary Tables. What's more, they apparently have a very high load-handling capacity, due to all those rollers with their large contact area. Interesting, but how do you mount them? (Nice pictures of CRBs come from a Schaeffler KG catalog on CRBs. Pity they got the force arrows in the left hand pic back to front though.)

Attachment 271702

So I read on through the tech manual. That's where it became clear that handling CRBs was a shade more complex than DBGs. Clamping forces figured high in the discussions about these, along with the need for rigid support. You see, while CRBs are large in diameter, the races are not especially heavy in themselves. They often rely on the surrounding metal for real support. This could be complex - but based on recent experiences, I thought maybe my CNC could handle these tolerances. Somewhat more problematic was whether my wallet could handle them - and anyhow, while very nice, they don't solve the reduction gears problem.

Somewhere along the way I had come across a comment that Harmonic Drives use CRBs - and have a large reduction ratio, and do not have much backlash. There are even a few threads here on CNCZone about using a Harmonic Drive in a Rotary Table. Mind you, the first impression I got was that they also had serious wallet problems, but looking is cheap, and the write-ups were encouraging. So off to the web for more information.



How do these things really work? Well, my first impression was ... confused. I am sure their engineering drawing is useful and accurate, but understanding it was another matter. That required some sort of slow interaction.



So I sat down and modelled the same unit in a simple 2D drawing package (AutoSketch) from the data in the previous drawing. This was not hard - AutoSketch is a dream to use. (No, AutoDesk did NOT create it: they bought it from some Russians.) Anyhow, the Harmonic Drive gear reducton spline (the core of the unit) is the curved red thing with a pink interior, while the rollers for the CRB are the olive green squares. Obviously the races are part of the structure. The black lines with pink interior are the input shaft, and the green bits are the normal output stage. The blue bit seemed to be the 'base' for the whole thing, although there is a gotcha here which I will cover later. Yes, I could have used a 3D modelling package for this - provided I understood what it was I was modelling. First walk, then fly.

However, Harmonic Drives are expensive things. I wondered: do the Chinese make a clone? Off to eBay and AliBaba to search. The answer was no, but there are one or two Asian dealers who have a supply of ssecond-hand Harmonic Drives stripped out of something or other - decommissioned Japanese robots maybe? Being secondhand they are of course much cheaper, and since they are coming from Asia you can always haggle a bit. To my surprise, I found I could get an SHG-25-50-2UH for just a bit over US$300, including FedEx freight.

The 'SHF' means it is a 'silk hat type' (what the h*ll is that? Do they mean 'top hat'?), the unit is 'size 25', whatever that means (the dimensions are in the book), with a 50:1 reduction. The 2UH bit says it has a hole through the middle and it has integrated output bearings. Anyhow, it has the dimensions as shown in the first drawing. And that size was so close to the 100 mm chuck I had in mind too. How incredibly tempting.

Yes, that's a whole lot more expensive than the bar of Fortal and the two bearings I showed before, and you could buy an Asian rotary table complete with stepper motor for that much money - but two things motivated me. First, the very suitable gearing and the excellent bearings were ALL included in the HD, and second, the precision I could get from an HD made the Asian rotaries look very inferior. If I am going to do this, I thought, let's do it properly.

So, I bought one. It turned up quickly, it was as described, and it was in good condition. The gearing was very smooth. All of a sudden, I could see this project becoming a whole lot simpler. The consequent redesign is in the next chapter.

Roger...

I also am using a Harmonic Drive gearset in my design. The integrated bearings are pretty heavy duty and designed to directly carry a load (they are used in robotic arms for example.) I have a 100:1 unit that I plan to pair with a 200W servomotor I have laying around. It really is amazing that the flex spline has only two less gear teeth than the outer ring gear. I also have a hollow shaft version. Looking to see how you implement yours in your design.