[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 ([ame="http://www.youtube.com/watch?v=cnJCWX2nr4M&feature=channel_page"]YouTube - hypocycloid reducer[/ame]) 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]

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]

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

[Odin]

Cal,

Thanks for your advice. I tried very hard to lay out specifics without any emotional terms, just facts. I figured, though, that someone would see that I'm getting my first serious machine, and see what I wanted to do, and give me the whole "you're too serious" routine, thinking I posted this in some sort of excitement.

No offense taken, none intended toward you or anyone else. I know I'm new to all this.

I was hoping for people here to take my question as seriously as I tried to post it and research it. I've had years of planning in my head for what I wanted, without a clear way to get it. I have means now, that's all. I'm not overly excited- I'm methodical about what I want to do.

My eventual goal is to work with tolerances as high as or higher than the horologist Janvier, who really strived to make orreys and sphères mouvante as accurately as the work could be done. He ignored his contemporaries, who settled for less precise models, and pushed the limits of tolerances in his day to model the planets as accurately as he could, simply because that was the point to him.

I won't compare myself to a genius like that. I do, however, aspire to work with things at a similar level of precision or better than he did. I am a beginner, but I am serious about starting with tools as precise as I can because I choose to be, not because I think I need to be. What I wish to do may be accomplished with simpler means, but I prefer to outfit myself for all forseeable and nonforseeable circumstances. And so I asked for honest commentary regarding the options I posted above.

I specifically left out dividing heads/index plates for a reason- I cut tourmalines on an old, hand controlled cutting rig. It uses a dividing head. Even with that head, and it is well made, it wears, and does not always hold the exact angle I need, off by a few thousandths, and so I must manually adjust the cutting table on the head in three places. So, I'd prefer a rotary table for something like this. Maybe later on, I could try dividing heads again, if I found a suitable method for holding them exactly in place.

This is why I want zero backlash. Especially to do work with orreys as Janvier did- he used very large geartrains to model motions that many others used small geartrains for, but his were far more accurate. This kind of work, with even a few thousandths off on one gear, can make seconds or minutes difference a year in showing the motion of one planetary body. I want to avoid that at all costs. And yes, if you think I'm crazy- Janvier often had geartrains made to model accuracy to the seconds level for a planet in a year- it was done 300 years ago! So no, that's not nuts (in case anyone thought it was). It's been done.

No more long posts on reasoning, I promise. I'm not saying you're wrong, but you might not fully get why I need what I asked about. If you have some specific equipment examples you think may help me, please link them, by all means- it is refreshing to see what someone suggests when they really must know what they are doing, as no doubt someone like you does.

[CalG]

I stand with the suggestion that optical methods will be the ready answer to your needs.
.25 arc seconds is possible......perhaps.

References

Hilger Watts
Davidson
Keuffel and Esser
Moore

"Wonder tools" will not build the mechanisms you aspire to. You will need to do that yourself.

Or, You could use the same methods as did Janvier. Are they known?

Cal

[Odin]

Janvier most certainly used division plates with wheel cutting engines, and probably constructed his own to do so, as any serious horologer (watchmakers, clockmakers, those who study time) would have done in those times. New tech in building more accurate wheel cutting engines was like trade secrets in diamond making are now.

So Janvier used your method, most certainly. But even engines to cut gear teeth had error- he most likely hand filed all teeth and polished them by hand- something I'm willing to do. But if Janvier had lived today, he'd be trying to get at what I was with this thread, of that I'm sure. He sought always the most accurate means of acheiving something, even if it was to others, overkill.

I am like Janvier, I feel, in that I search for higher accuracy as an absolute end.

And by the way, I am planning to use the Taig to build my own tools as well, including a wheel cutting engine in the style of the 1800s, a fusee cutting engine, and a geometric lathe (aka rose engine, aka guilloche machine). Until I get that good, I want something I can use to make anything radially precise. Thus the table.

If I asked you, "would the Sherline suffice for my needs?", what would you say? And those names, references, I don't understand them, forgive me. Are they machinists or companies?

[BillTodd]

Odin,

Your rotary table problem can be broken down into two parts: Moving the part and knowing that it's in the right position.

Having a backlash-free precision mechanism will help immensely when it comes to positioning the work piece, but little to help you know it is in the right place.

What Cal (and Mike previously) is suggesting is that you concentrate your time and money on the measurement of position rather than the positioning mechanism. (amazing what can be achieved with a small hammer )

I applaud your search for accuracy and your wish to become a first class Horologer, but first you have to become a first class Metrologist. You need to quantify your errors and identify their source, before you can approach perfection.

[Andre' B]

Not sure I fully understand what you want but I think you need to look at somthing like this.
http://www.subtool.com/st/MG5CVS1.shtml
Here is anouther good one.
http://www.harigmfg.com/ultragrind5cv.html
They are a bit over $2000 new.

Here is a cheaper version, not sure if it does the sine bar thing for inbetween angles. I think I would look for a used Harig or Suburban.
http://www.tccatalog.com/review/product/list/id/62118/

[CalG]

These references are the names of precision optical and mechanical metrology instrument manufacturers,(some gone now, but the equipment and knowledge lives on!

"If it can be measured, It can be made"

CalG

[Odin]

Not sure I fully understand what you want but I think you need to look at somthing like this.
http://www.subtool.com/st/MG5CVS1.shtml
Here is anouther good one.
http://www.harigmfg.com/ultragrind5cv.html
They are a bit over $2000 new.

Here is a cheaper version, not sure if it does the sine bar thing for inbetween angles. I think I would look for a used Harig or Suburban.
http://www.tccatalog.com/review/product/list/id/62118/

Wow. 2000$ for a division plate. I'm sure it's great quality, but....WTF? Only 72 lockable positions for the first one. After that, you have to use a sine bar to divide further. Now forgive me, I have no idea what a sine bar is, or how it works. But I bet that means you have to adjust manually for each new sub angle.

I need to make gears with tooth counts in the HUNDREDS. 300 and below or so.

Here's what I'm thinking now- have a large (say, a couple feet in diameter) division plate made for me, with someone else's much more accurate machinery, and have index holes for as many teeth maximum as I need. Mount this with a custom fixture I make myself to sit off the end of the Taig workslide.

At one point, I thought to myself, a well made dividing plate, deep enough so the index holds with no movement at all (Say, an inch deep), that was maybe 5 feet in diameter, mounted to the wall & center trued to the machine, could have accurate holes indexed enough for even tenths of a degree. Perhaps that's beyond overkill. I like ideas scaling large accuracy to small levels, my mind is always applying a pantograph mentality to things with large stuff like that plate.

Let's just say now, I feel I'm going to get that Sherline anyway, and use it to make whatever more accurate rotary table concept I decide on.

I really liked the metal tape drive idea, though- that, rotating through only a bit over 360 degrees would have been PERFECT for what I want to do. Thing is, I have no idea how much one costs. Do you buy the whole thing, or just the metal tape?

For that matter, has anyone in all of CNC Zone built a metal tape indexing table, and thus knows costs? Despite the high likelyhood of me getting that Sherline rotary table now, I'm still very, very interested in the metal tape drive setup.

And besides- not everything I plan on making needs to be as precise as those gears- A standard, good rotary table would let me make all sorts of useful things.

[NEATman]

Odin-
You may be able to take a rotary encoder with a thousand counts or more, attach a friction wheel to it that rides on the OD of the index wheel, which would get you a high line count rotary encoder with no backlash. Simmilar to the encoders used on a Prototrak.
Another option for a highly accurate rotation is a capstan drive table. Think of a flexible steel disk (like a curcular saw blade) that has a smaller set of pinch wheels on opposite sides of the disk. The axes of these two wheels is perpindicular to the main axis of rotation. The flex in the disk allows slight misalignment, and the friction drive gives you zero backlash. You could then add an encoder to this arrangement, and get high resolution output. If there is any slippage however, it will not repeat.

Ultimately, it looks like you would like something just like this:
http://www.abtechmfg.com/pdfs/abtech...ndex_table.pdf
Manual worm wheel drive, high accuracy encoder, etc.

You may also want to check ebay for this type of encoder:
http://www.renishaw.com/en/6434.aspx
I may have one or two 100mm rings in my pile of stuff...

Also, take a close look at the "air-ball" tables that are offered by ABTech as well. The key point is that when you turn off the air, the gap closes and it locks in place. The air bearing allows near frictionless motion, while the removal of air pressure causes the two to lock together. Perhaps you could even spring load this for a horizontal application.

Another cool drive mechanism is this one:
http://www.sagebrushtech.com/downloa...ts/rotolok.pdf

Perhaps combine one of these drives with a high accuracy encoder and you'll have what you are looking for.

Keith