Precision division plates

I just used a change gear wheel off a Myford lathe and turned up a worm with thinned thread to fit it. All done in the shade.

I think John Stevenson used a CNC mill to drill the division plates he sold but making the worm for a dividing device is another game. Making one for the home workshop is probably going to be rather different than making one for a high end rotary table. Making gears that work effectively doesn’t mean they would pass the most stringent tests of a metrology lab but they can work effectively. I suppose you have to define the spec of your finished product before you select the method of making it. Many of the things we Model were crudely made by the highest standards of today but they were mighty effective. Cutting a worm on a screw cutting lathe with a hand made tool is not going to match a CNC ground worm but for most people most of the time it will work fine.

Mike

What was the book Mike?

Thanks

Bill

Hi Mike Poole

When I think of high accuracy in the 20th century I think of Moore Tool
**LINK**

This company designed one of the highest accuracy mechanical rotary tables ever made. It uses a hirth coupling to form the 1440 divisions it divided a circle into than a further vernier for the spaces between the divisions.

I had not seen the ball method of dividing that looks rather interesting! If the balls were set in two precision grooved plates and epoxied in place, they would have to be absolutely level in one plane, you would have the two sides of a hirth type coupling.

A coupling of this type when used for indexing will tend to average out any errors. Moore recognised this and used it for their highest accuracy tables. It is described in depth in the Book Foundations of mechanical accuracy. One of the great mechanical engineering reads.
**LINK**

Hirth Coupling.
**LINK**

Gee I love first principles in engineering.

Regards
John

Edited By John McNamara on 07/03/2019 07:20:15

Bill, The book is Laboratory and Workshop Notes compiled and edited by Ruth Lang My copy is published in 1949 but other editions seem to have been published in the 50’s which I have not seen so cannot comment on the contents.

John, those are some interesting links, I will look out for a copy of the book for a no doubt fascinating read.

Mike

Edited By Mike Poole on 07/03/2019 09:47:14

Regular dividing plates are suitable for all but the very finest of work but one often overlooked aspect of them is the concentricity of the hole circles to the centre hole. If it's not the indexing will be all over the place.

I bought a set of three used plates for my IXL head and found that although all the hole circles were concentric to each other the centre hole was a good 5 thou out. Thankfully I had to bore it bigger to fit and had enough meat on it to get it central.

Tenths of an Arc minute?

My mount is giving me grief because I'm struggling to keep it guiding to better than a arc-second, I should be able to get it down to 0.6 - 0.7"

Then I can attach a gear blank to my telescope...

Neil

To be fair, the uncorrected periodic error of the worm and wheel is apparently between 4 and 60 arc-seconds, which gives pretty fair idea of the periodic error of a rotary table. Software can either map the PEC or 'model it' (the latter takes more processing but works very well if you have a portable setup where mapping PEC every time you set up is a bit OTT).

Also there will be an error use to eccentricity of the worm wheel, not sure of the magnitude as it's largely irrelevant for astro where you typically only move over 15-30 degrees during tracking and is very easily guided out as it is so slow.

Edited By Neil Wyatt on 07/03/2019 21:06:12

It would be nice to see inside one of the Moore indexing tables. Obviously it is not a rotary table that you could (or would!) use for machining the radius on a connecting rod, but the intriguing bit is how they achieve the repeatability of the axial movement, without any backlash presumably.

Ian P

Neil,

If you can find a way to use 'telescope mount style' Periodic Error Correction to generate/calibrate our indexing devices ... The likes of Taylor Hobson will beat a path to your door:

**LINK**

https://www.taylor-hobson.com/pressreleases/news/2015/september/polygon-indexing-head-checking

MichaelG.

Some long time ago, there was an article in MEW about making Division Plates. IIRC it involved making a plate from a surplus CD, and then using that to make another one, which would be more accurate, by a factor of the ratio of the Dividing head / Rotary table used. After several iterations, a plate of acceptable accuracy would result.

One day, I will get round to making one or more plates to extend the range of my Vertex HV6, which has a a 90:1 ratio..

Also, it would be nice to have one which would allow me to make 127 divisions, not that i am likely to cut a 127T gears for the mini lathe, since I already have a 63T; SHOULD I ever feel a need to screwcut an Imperial thread.

Just a "wish list" nice to have. By the time that I've made the plate, will probably forgotten what it's intended use!

Howard

.

It would indeed be nice, Ian ... but I suspect the 'ticket price' is way out of my league.

However: The principle, as John has already described, is simple enough:

[quote] It uses a hirth coupling to form the 1440 divisions ... [/quote]

Here, courtesy of Wikipedia, is that magical device: **LINK**

https://en.m.wikipedia.org/wiki/Hirth_joint#/media/File%3AHirth_Verzahnungsringe.JPG

Moore's genius was to recognise that if his 1440 teeth all meshed correctly, at every possible pairing, then the indexing must be accurate to a level which most of us can barely comprehend.

MichaelG.

Howard, one problem with a 127 dividing plate is you often need a larger than standard plate to drill all the holes on one diameter. 3mm holes with 4mm spacing requires a diameter of about 160mm. Not a problem if you can fit one that size to your rotary table but it may also require a longer crank and sector arms. It may be possible to do two rows of holes but then you may spend a lot of time reseting the setting pin to a different radius.

Martin C

.

... and another [perhaps more fundamental] is that you cannot [*] take advantage of that magical 'improvement in accuracy' that multiple generations made via the worm & wheel might give for some other numbers.

MichaelG.

.

[*] or at least ... I have never thought of a way !

This may be of interest:

**LINK**

https://www.researchgate.net/publication/328921911_High_precision_calibration_of_polygons_for_emerging_demands/fulltext/5beb74d5299bf1124fd0e855/328921911_High_precision_calibration_of_polygons_for_emerging_demands.pdf?origin=publication_detail

MichaelG.

.

Edit: But of course ... 

if Neil can find a way to use his telescope mount, then all of this will become obsolete.

Edited By Michael Gilligan on 09/03/2019 09:01:52

I understand the how the 1440 divisions are accurately defined but since the two parts of the coupling move apart axially to disengage, I wondered how they guided the moving half without introducing any angular uncertainty.

Ian P

Used many Hirth couplings over the years on rotary indexing applications, most were in the region of 600mm diameter. The parts carrying the two rings were guided in fairly loose plain bearings allowing the coupling to make position. The concept of the Hitrth coupling is that the accuracy increases slightly as the halves bed into each other. I mostly used the lower tooth counts too suit the index required which was usually multiples of two. Expensive pieces but simple. As well as two piece there are three piece which have an inner and an outer ring joined by a single ring on the other side, this allows one ring to remain in contact with the fixed ring at all times during indexing.

.

Surely the coupling self-aligns ... or am I missing the point again ?

MichaelG.

.

Please excuse the nonsense that I posted this morning at 08:45:23

I can only blame 'caffeine deficiency'

repeatedly indexing 90 steps on a 127 circle will do the magic

.

MichaelG.

The 1440 precise divisions are between the two halves of the coupling, one half rotates with the table so is rigidly coupled to the job, the lower half (presumably) has to be rigidly coupled to the main body but with the axial movement to engage and disengage the coupling teeth. Any angular play would destroy the accuracy so wonder what sort of zero backlash splines or guide was used. If as mentioned earlier, the coupling teeth 'bed in' over time then the axial, engaged/locked position will move slightly but incorporating any sort of spring loading sort of 'feels' wrong.

Ian P

.

I think we are still at cross-purposes, Ian

In the book "Foundations of Mechanical Accuracy"

Wayne R. Moore credits the A. A. Gage Co. with the concept

... so let's go to a page which briefly describes their Ultradex and includes a couple of useful images: **LINK**

http://www.aki-alltech.co.jp/english/right9.files/news9_13.htm

MichaelG.