Buy, or make? All options considered- creating an ultra precise rotary table

[Odin]

Before I begin, I'd like to clarify- this is a sort of continuation/addendum thread of Zoidberg's amazing "backlash free rotary table" thread (see http://www.cnczone.com/forums/showthread.php?t=72261). I don't mean to hijack it, and I have more specific questions/breakdowns for my posting, so I decided to try a new thread.
------------------------------------------

I'm about to buy a Taig manual mill, and I want to do a lot of gear hobbing- so I need a rotary table. Carter tools sells Sherline rotary tables for about 250$. Money isn't a total issue, but I'm not rich. I figure, I've got about 1200-1300 dollars to work with for everything.

Thing is, I plan on cutting very small gears, with extremely high tooth counts, and various other items like time equation cams for clocks where standard rotary tables may work, but I want to kick it up a notch. I plan on working with large gear trains (making full orreys, etc), so even the smallest backlash or positioning error of a normal table isn't acceptable, or the cumulative error will add up over time and ruin the orreys. So I'm thinking of building a rotary table. An exceptionally precisce one, with NO backlash. None.

Put aside for a second what I've said, forget that material tribology will change over time and wear the gears anyway, and other concerns. Just work with me here and dream.

I can afford that Sherline rotary table. But it's not as positionally repeatable and as backlash free as I need. Reading Zoidberg's amazing post, and everyone's comments meticulously, I've decided to repost all the positioning technologies discussed therein here, named, numbered, and with relevant links.

So here's what I ask from anyone who knows them: Can you comment on these things:

A. Cost - how much to buy the components/system, typically?
B. Precision - how repeatable is the the positioning tech?
C. (most important) Fabricability - how feasible would it be for someone to create this in the shop with only a 3 way mill, and NO rotary table? (Or with a standard rotary table as an initial investment?)


1.Worm gear (I'm not really considering this- traditional worm gears have too many shortcomings for a single one to work well for my needs)
2.Dual worm gear (http://www.allytech.eu/index_fichier...dwormwheel.htm)
3.Duplex Worm Gear (http://en.wikipedia.org/wiki/Duplex_worm)- this one's interesting.
4.Harmonic Drive- sounds promising, but the cheap ones are HUGE. I need something I can fit on a Taig mill!
5.Pneumatic Drive (air bearing, ala Onvio’s Dojen Cycloidal Speed Reducers) (http://www.onviollc.com/dotnetnuke/C...Default.aspx)- these look especially promising.
6.Hydrolic Drive (oil)
7.Rack & DUAL Pinion (http://www.globalspec.com/FeaturedPr...lDrive/74196/0)
8.Howimatt Recirculating Ball Drive (Epicyclic Gear)(http://www.detlevhofmann.de/english_.../funktion.html)
9.Hypocycloidal Reducer (http://www.zincland.com/hypocycloid/) Ala user ZincBoy, and the incredible creations of him and others, such as in beautiful metal ala user LaRdArms (

YouTube - hypocycloid reducer

YouTube - hypocycloid reducer

) These look like the most doable on a mill, with great possibility for really high accuracy. These also include versions that incorporate ball bearings.
10. Ball Worm Transmission (http://urobotics.urology.jhu.edu/projects/BW/)
This looks the simplest to me, and also very buildable.
11. Anything I missed not previously discussed, however outlandish or exotic.

Final caveat- I want to do this manually. I enjoy hand machining, and don't see the fun in CNC for what I do. I want to make something that is hand cranked. I can always add a servo later, the idea here is for manual operation, let's concentrate on the bare mechanics of it.

Keep in mind, I need this to fit on a Taig mill. It's not a gigantic rotary table I need. Large diameter is great- excellent in fact, but height not so. I can fit a large diameter of 6-7 or maybe even 8 inches on the Taig table, but nothing very high. Let's try for 3 inches or so tall. It won't be handling large amounts of weight or torque, and I'd like to be able to adjust backlash if possible, to allow quick rotation, but able to lock back in precision rotation as needed.

Let the brainstorming begin!

[ZincBoy]

Obviously I am partial to the hypocycloidal reducers and, interestingly enough, I made my reducers on a Taig. Trying to do this without a CNC would be "interesting". The cam profile is not simple.

I did see some descriptions of how the hypocycloids can be ground using what amounts to a heavy duty spirograph. I think you would need a rotary table as a starting point on a manual mill. As the table is rotated, there is a offset shaft turning against a drive ring that creates the eccentric action that forms the cam profile. If I can find the paper that I saw this in, I will post it.

The tool I think you will find it difficult to do without is a lathe. Cutting the shafts for which ever table will be difficult without one.

One thing to consider is do you really need zero backlash? I would think that for gears, you could always turn the table in one direction only and lock it when milling. The angular position error would be much more important for your application and by all accounts the Sherline table is very good in this regard. My reading indicates that this does tend to be a weak point of the hypocycloidal type reducers. Errors in the cam profile turn into angular errors. A very good reducer might have zero backlash (preloaded bearings) but would not have as good an angular error rating as a worm gear with backlash.

[Odin]

Zincboy, I consider you the foremost expert on here for Hypocycloid reducers, so thanks very much for your take on it. I agree with your commentary, but a couple questions:

Assuming I always rotate in one direction, how could I lock the rotary table in place for milling? Machine a locking tab into the table crank?

I was actually planning on making a lathe with the mill at some point soon after getting it.

So hypocycloidals look amazing, but even assuming they never wear (they would), they wouldn't be ideal for this application. What do you think would work best? A ball worm? A duplex worm? Easiest to make manually? Let me know what you think.

And on that spirograph idea- I just had a great idea of my own. If one had to reproduce very accurately a complex cam profile on a non CNC mill, why not turn the mill head into a sort of moving pantograph, with the profile following portion much larger- make a very accurate profile large, say 3 feet, and the pantograph head adapted mill would hold to this at a much smaller scale.

Come to think of it, if I figure out a way to make that, creating a pantograph mill would be an incredible thing all by itself! Midevil CNC, as it were.

[RICHARD ZASTROW]

Define "ultra precise". To me that is 1 arc second territory and $$,$$$.

Dick Z

[Odin]

Originally Posted by RICHARD ZASTROW Define "ultra precise". To me that is 1 arc second territory and $$,$$$. Dick Z

I know, I know. I knew that would come up. "ultra precise" would indeed mean 1 arc second.

I mean, in practical terms, as damned accurate as possible for several hundred dollars/and or massive time investments dedicated towards accurate machining, bordering on zealousness.

Does that help?

[BillTodd]

As Zathras pointed out here: http://www.cnczone.com/forums/showpo...&postcount=199

If you can get away with less than (or just over) 360' of motion, then the most accurate would be a preloaded shim-steel strip, wrapped around an drum. Then you can use a an accurate lead-screw (or ball screw if you require speed) to pull on the tape.

The mechanism is simple, easily built by hand and should be easy to calibrate (simple maths).

If you require continuous accurate rotation, then the problem is the same as stated Zoidberg's thread with the added problem of measuring the error.

[edit] I'm trying to think where I have seen a tape drive with over 360' of movement - maybe in a dream ? Until it comes to me, it's best to assume a tape drive will only do less than 360'.

[Mike Everman]

Hi Bill,
Since you won't be doing NC paths, do you also not need to crank the table for radius cuts?
Point being, if you just want a great indexer, then spend the money on a good direct reading encoder and digital readout, and then just spend time on a decent bearing set and locking method.
I'm sure you want to crank it, but wanted to ask.
Speaking of cranking it, use the approach above and add a long handle, then watch the numbers. For thin brass and such, you might not need much in the way of mechanical advantage, and all the issues it brings up.

Nice to hear you're into clocks. I just joined the NAWCC here in the states, and the HS chapter 161. I think I've licked the isochronous pendulum problem! (but I won't take it prime-time until I build it, of course.) Apologies to those who do not understand. There are very few of us nerds that care about such a thing.

[Odin]

Bill,

Interesting idea. You could make such a metal tape drive rotate through as many degrees as you want- hell, 720 if you want- all that needs changed is the strip needs lengthened. If you have a 4 inch rotary table, and use a strip that is 2*[4(3.14...-pi-)]inches, and it could rotate the table through exactly 720 degrees, or twice around in either direction. Adjust tape accordingly for parts grasped by the moving ballnut/whatever you use to pull on the tape.

The only thing that makes this suck is that you have to have room for all this tape to one side of the table, and it's adjusting mechanism.

I thought of a mobius strip version of this, which would offer endless rotation and indexing, but then we're back to the original problem- the part that the belt wraps around that moves it would ALSO have to rotate as if it were attached to one of these belts to remain that accurate.

Thus, I think unless you make that part of the mobius very large in relation to the rotary table, with some sort of linear indexer that could be bent into a large circle, that would never work.

It might be possible to, instead of wrapping around the outer perimeter of the table, run the belt off a smaller diameter internal hub inside the table itself, say the inside of the through-hole axis, and run the tape and indexer through slots to the outside of the table to take up much less room (less tape needed at that smaller diameter), or house it all inside an elongated table compartment adjoining the rotary portion.

This would be something buildable, and very very accurate indeed. Can anyone speculate how accurate in terms of some sort, like arc-seconds? However, it would need me to buy the custom made steel tape, and some sort of accurate linear indexing part, like a ground ballscrew, or something, to make it work. That link you added had Zathras talking of using a harmonic drive to move his. Don't those work in rotary fashion only? And if so, wouldn't that defeat the purpose of using the tape to move the table- why not use the harmonic drive itself- for any imperfection in movement on the part of the harmonic drive in the first place moving the tape would be redundant if you only used the harmonic drive alone.

So, how about it?

[Odin]

Originally Posted by BillTodd As Zathras pointed out here: http://www.cnczone.com/forums/showpo...&postcount=199 If you can get away with less than (or just over) 360' of motion, then the most accurate would be a preloaded shim-steel strip, wrapped around an drum. Then you can use a an accurate lead-screw (or ball screw if you require speed) to pull on the tape.

I can't imagine how a straight tape could be hooked up to move on a ballscrew. Can anyone do a mock up/model of this?

[BillTodd]

Interesting idea. You could make such a metal tape drive rotate through as many degrees as you want- hell, 720 if you want- all that needs changed is the strip needs lengthened.

I think you have a different idea there. A typical tape drive has two tapes each has one end fixed to the output drum, the other ends to the actuator. There's usually a mechanism at the actuator end to pre-load (stretch) the tapes.

The accuracy of the device comes from the known diameter of the drum. If you wrap a second loop of tape over the first the accuracy is diminished (because of transition between wraps).

It is possible to wrap more than 360' around a drum in a single wrap by splitting the wrap into a 'Y' shape (imagine the top two parts of the Y fixed to the drum at 0' with the intersection say near 360'. The tail can wrap between the upper pair for another near 360' . The whole tape can be in contact with the drum surface, thus maintaining the best accuracy.

(I'll try a make a drawing later)

[mactec54]

Hi Odin

I just sent you a PM I have a Harmonic Drive around the right size to suit your machine
it is new still in it's box the input shaft size is .375 the output mounting is 1.244 ratio I do not know but you could go to there web site to get that

Model # HDC-3C-060-2A (Harmonic Drive Techologies)

[CalG]

To precisely divide angles, Optical methods are best. Obtain a goniometer if you really must

Skip the drive Mechanism! For gear cutting, position is everything. You require a rotary work holding device of some type. That can be as simple as a couple of bearings. Plain bearings at that. Great work can be produced with only a dividing head and an index plate or two. What you ask for is what dividing heads are made to do!

Work on clamping methods for your apparatus that do not upset the desired position.

This has all been done by those who have gone before, and your budget is not a show stopper!.

With your head all wrapped up in stuff not required, nothing truly good will will come of your excitement.





Regards

CalG