Rotary table as a dividing head without the dividing plates

[Andrew SchofieldParticipant @andrewschofield]

Can I use my rotary table as a dividing head without the expense of buying dividing plates and use the built in 360 degree scale to set the spacing?

[Clive FosterParticipant @clivefoster55965]

You can but it’s hard to be consistently accurate. Especially with numbers that don’t divide easily into 360°.

Generally reckoned that if you don’t wish to purchase dividing plates or make one of the digital, stepper motor driven, drive devices your efforts are best spent on making plates one row at time as needed. To a first approximation the residual errors in making the plate can be considered to be reduced by a factor equal to the drive ratio. So anything produced via a plate made on the rotary table ought to be that much more accurate than an equivalent made directly on the dividing head.

If you have a suitable DRO the built in PCD functions are pretty effective at producing dividing plate holes up to  moderate number counts.

Clive

[JasonBModerator @jasonb]

Yes you can just divide 360 by the number of divisions you want and then wind it around to the angular measurements. better to work out the angle for each rather than just the first and keep adding them together as that can give cumulative errors. Depending on the rotary table you may have to best guess the seconds as some only go down to minutes on the handweel, others to 20sec spacing. Always try to avoid backlash.

[BazyleParticipant @bazyle]

You can get a long way with a set of ‘holes’ printed on a piece of paper (there are website that do that for whatever hole count you want) if you know how to mount it on your rotary table’s handwheel.
If you just want to use the 360 marks you can probably set to ‘half a mark’ quite easily ie 2×360 so it that accurate enough?. To go a bit further with just the markings you can make a vernier scale for it.

[Nigel Graham 2Participant @nigelgraham2]

If you need hole counts not integer divisions of 360, and assuming your table has a Minutes dial on the handle, I  would approach those by first creating a table of Hole Numbers’ angles in Degrees and Minutes.

An ‘Excel’ (or equivalent) spreadsheet is ideal for this, but remember that computers use Radians so treat the values purely as numbers not angles. So convert each division-sum’s remainder by multiplying it (if a decimal fraction) by 60 to gain the degrees then minutes then seconds of arc.

I would calculate each increment individually to keep the rounding errors individual and restricted to each step.

All that repetition is why I suggest a spreadsheet!

[SillyOldDufferModerator @sillyoldduffer]

Yes, but it gets tedious and error prone as the number of divisions rises and especially if the required turn angle isn’t a factor of 360°.

No problem doing a hex bolt head because 360/6 = 60, which the dial does easily and without brain strain.   But cutting prime numbered toothed gears is a nightmare.   For a 31 toothed gear, 31 repeat turns of 11.612903226° are needed and  11.612903226° can only be approximated on the dial.   Misreading the dial or losing count ruins the gear, and I wouldn’t attempt it without Nigel’s spreadsheet.  Dividing wheels are a handy memory aid, but, blush, the operator can mess them up too.  For that reason microcontroller driven stepper motors are popular.  The computer does all the sums has a perfect memory and isn’t distracted!

Dave

 

[Nigel Graham 2Participant @nigelgraham2]

How approximate? Worth investigating as a handle on the accuracies you can expect.

I read Andrew’s question as using a rotary-table directly as a rotary-table, with its own divisions; not as a dividing-head in which the internal gear ratio is part of the calculation.

If I’m right we don’t need consider that. So taking your example, the 31T count, and please review my arithmetic!

360 / 31 = 11.61290 Degrees.

Now, if we assume the rotary table has a Minutes dial on its handle:

60 X 0.61290 = 36.774 Minutes  (NB multiply as the remainder is a fraction.)

Can we go down to Seconds of arc? Probably not, on a conventional rotary-table but it might have a vernier-scale giving perhaps 0.1 min (or 6s).

60 X 0.774 = 46.44s.

So 11º 36′ 46″ per division.

So if the table lets you read to Minutes you can estimate that about two-thirds minute and that may well be within tolerance for most work.

.

What is that 31T for? A prime-number to give a “hunting tooth” in a power transmission-gear? To give a close approximation for a change-wheel set for cutting metric threads where a 127T wheel won’t physically fit (a 32 or 64 T wheel is closer)? Or a random number?

.

There is an alternative and a well-established one at that, though usually associated with using the lathe as a dividing-head.

Determine what dividing-plate hole-counts you actually need, then make a simple dividing-plate divider by using suitable gears, such as lathe-change wheels or any of sensible size really, mounted on an appropriate mandrel and fitted with a detent. The gear’s own counts need not necessarily match the counts you want as long as they are integer multiples of it.

Since this would appear to be all milling-machine work, build the divider as a separate stub-axle carrier and detent carrier, to fit the table T-slots.

You could of course carry out this project on the lathe, if you can mount suitable gears on the back of the spindle, and detent to match, using a milling/drilling attachment on the cross-slide. The method is well described in books on turning. I have not seen described, let alone have I tried, extending that by fitting a change-wheel train to act as a compound divider for awkward hole-counts.

 

[Clive FosterParticipant @clivefoster55965]

Even with something like the OMT 16″ rotary table with optical scales shown about 3/4 of the way down this page :-

https://www.lathes.co.uk/omt/page2.html

and a spreadsheet table lots of numbers are tricky, some impossible.

I know I’ve done such precise things in the optics lab with one just like that (majorly heavy brute) for specialist test purposes. You really, really don’t want to voluntarily do it to the “tell me 3 times technique” to be reasonably confident in the results let alone enough repeats for statistical analysis. (But 6 months pay is good motivator!)

Expecting to accurately do anything more than fairly friendly numbers on the typical Model Engineers rotary table such as a Vertex is unreasonably optimistic. I have an 8″ Vertex and it’s half way decent by any standards, wonderful for the price, but I’d not dream of trying anything beyond the simplest divisions on it. My 10″ and 12″ tables are unusual in that they have degree markings on the outside of the table with a linear, adjustable, vernier scale on the body. So splitting divisions is much easier and far more reliable than with the usual type having the fine scale on the handle. No backlash worry for starters. If I had to I could arrange to split a vernier interval into 5 without totally unreasonable effort. But life is way too short.

The printed hole suggestion from Bazle is excellent so long as you have a properly accurate printer, preferably laser as some inkjets and some paper stretch oddly. But why stop at a print. Stick it on a plate and drill the darn holes. Then use that plate to make a second generation one. Takes a bit of time but it’s all straightforward. Not the “one slip and the job is toast” situation all to easily hit when trying to push a rotary table too far out of the comfort zone.

Before considering “unfriendly” numbers for direct division on a rotary table using the scales it’s important to consider what the tolerances you are working to are. If its bolt holes in specified standard metric clearance sizes the tolerances are big enough that it’s hard to get it far enough wrong for things not to fit. If it’s a gear then it might seem that Daves 9 decimal places are needful. But even my top quality Cincinnati clone dividing head with its wide range plates isn’t that good. Working off the scales you have to be sensible and generally need to verify appropriate tolerances in terms of standard cartesian X-Y co-ordinate error distances and round off appropriately.

This is the sort of topic where amateurs tend to get hung up on unreasonable mathematical precision whilst professionals concentrate on getting stuff within the tolerance range needed to make things work. Even if that means slightly uneven spacing. Theoretically Geometrical Design and Tolerancing (GD&T) is the way to handle such things. It’s also a monumental headache for normal folk and best avoided. Simple X-Y estimation is good enough for anything we do.

Jason and I have friendly (I hope) differences on several aspects of what is, for the home shop person, most appropriate technique but this is one where I really have to firmly disagree. Dividing circles directly off rotary table scales is something that should be done only when there is no other alternative and the numbers are friendly. It’s not to be considered a general technique. Even if it did work fine the first time. Been there, dunnit, got the Tee shirt and thrown the failures away hard enough to nearly make orbit. That was on a BCA jig borer which is designed to make such things reasonably possible. For the highly skilled and well practiced worker. A far better man than I.

On typical Model Engineer / Home Workshop kit its that much more difficult.

The very experienced among us tend to forget that it’s all much harder for the inexperienced, especially those whose only mentor is the internet. It’s the inexperienced who generally won’t have the preferable kit and hope to push the performance of what they have beyond reasonable aspirations of success given their current skills.

I’ve long felt that there is a need for an accessible treatment of tolerances and real world variation in terms of what can reasonably be done with typical home shop gear and what is needful for typical examples of what is done in the home shop. Condensed re-gurgitation of Machineries Handbook et al staples won’t do.

Too much assumed background.

I might have the references and knowledge but if I tried to write it the results would make that “Eating of Elephants” thing look a mere precis.

Clive.

PS Nigel posted whilst I was writing. Excellent example Nigel.

[peak4Participant @peak4]

On the other hand, as discussed above, you could knock up your own dividing plates.
Dr Al’s site’s pretty good for printable division circles
https://www.cgtk.co.uk/metalwork/calculators/numberedwheel

Also has a dividing head calculator
https://www.cgtk.co.uk/metalwork/calculators/dividing

Bill

[JasonBModerator @jasonb]

I’m with Nigel on this how accurate do you really need. Even with plates there is backlash in hobby rotary tables and a lot more than a second or two that you would be “guessing” between handwheel scales. You can get backlash when using plates particularly if you go a bit past and then move the arm back to the hole or as you clamp the table to hold it firm for the cut

If it’s a timing gear for an engine then a minute or two difference between gear spacing is not going to affect the timing as the teeth will still mesh. Thinkness of a line or dots on paper would be similar. So long as you avoid cumulative error you won’t end up with 31.5 teeth

Better to get out in the workshop and get it done with what you have to hand than sit in the warm thinking of reasons why it won’t work. As funds  or time allow buy the plates (and arms etc) or make them. My first rotary table did not even take plates or have very small handwheel divisions but I managed.

Electronic is not fool proof, easy enough to enter a wrong number or not properly change a previous one. I’ve done it on the DRO when doing PCD more than once.

Add in that the R/T may have few microns run out and any chuck fitted to it the same. Far eastern gear cutter may be a bit off and any arbour and mill spindle may also be a bit off. Depth of cut set with the handwheel from some kind of touching off, etc and your gear is going to show some variation in depth of cut so you are not going to get the perfect gear. Gapping the gears to run rather than drilling the ctrs at book distance will give an engine that runs rather than one that never gets built.

[JasonBModerator @jasonb]

Idon’t know what others are expecting in the way of accuracy in the home shop but here is an example.

 

Daves 31T gear

lets say 1MOD which is a bit big for some models but might be OK for a larger hit and miss.

Say our handwheels and vernier scale are 20second divisions (typical vertex and Soba 6″ R/T) so could be 10seconds either way assuming no backlash or other losses. Less if you visually split the divisions on the handwheel or have smaller divisions such as my ARC one that has 10second divisions.

At the 31mm PCD of the gear a 10 second error equates to 0.000751mm or 3/4 of a micron or 0.00002956693″ or  3/100 of a thou. Or half that on a hobby ARC rotary table.

What tolerence do people actually need and what are you making that needs it?

[Dave HalfordParticipant @davehalford22513]

All the above – plus trial it first using a fine sharpie to mark the work, if that’s good then when cutting you have a double check.

[Howard LewisParticipant @howardlewis46836]

If you have a good supply of redundant CDs or DVDs, you could make your own Dividing Plates, and the Handle and “fingers”.

Some time ago, (A long time ago) there was an article in MEW on doing just this.

The first discs are drilled with the hole circles that can easily be produced, so probably up to 90 holes, (Many odd rather than even numbers) with a few exceptions..

Then they are used to make another set of duplicate discs, This time the hole positions will be more accurate by the ratio of the Rotary Table (The Vertex HV6 and a Soba have a ratio of 90:1, so a big improvement in accuracy)

You can reiterate that process as many times as you like, until you have achieved your required level of accuracy.  The second disc should be more accurate than the first by a factor 8100, for a 90:1 Table

Then,using the last disc, you drill the metal Dividing Plates that is going to be used for real.

How many holes? The same as those on the plates that you do not want to buy; a lot of hole drilling, but you will get there after probably a couple of discs for each plate.

The chances are that 3 plates will cover many of the divisions that you want, a fourth will cover some of those missing so far.

The hole counts for the three plates for my HV6, are:

A) 15,16,17,18,19,20

B) 21, 23, 27, 29, 31, 33

C) 37, 39, 41, 43, 49

These will not cover 44, 52, 53, 56, 59, 61, 64, 67, 68, 71, 73, 76, 77, 79, 83, 84, 88, 89, 91, 92, or 97 divisions.

An EXCEL spreadsheet could tell you what hole counts would be needed to cover some of the omissions.

One day, I may get round to making a fourth or fifth plate for my HV6!

Howard

[Pete RimmerParticipant @peterimmer30576]

Any error in hole diameter or drill wander will be more than the errors described above, and that’s before you consider deflection and machining errors in the part you’re indexing.

[Howard LewisParticipant @howardlewis46836]

Hopefully, drilling thin plastic discs will not produce much drill wander.

The important thing is the hole position, so the handle and its spring loaded peg need to be made to be close fits, so that the spring loaded conical ended peg in the handle can cope with any difference on hole diameter.

As Jason said, we need to bear in mind the degree of accuracy that we are trying to obtain.

A slight inaccuracy in gear tooth spacing might be exceeded by the lack of precision in blank diameter, depth of cut, centre distance, and bearing clearances. For most of what we doing, we can probably live with those.

And that disregards any inaccuracies of components within the Rotary Table. Hobby tables are not going to be of the same precision (or cost!) as the OMT optical ones in industrial Standards Rooms.

And unless the calibrated measuring instruments, and material have soaked in an accurately controlled 20’C for 24 hours, our measurements will not be absolutely precise

Let us live in the real world, rather than getting hung up on delusions of accuracy!

Howard

[JasonBModerator @jasonb]

It’s not so much the drill wander but the marking out of the holes and how accurate your punch marks are unless you have a DRO with PCD function.

A few other fun theoretical facts.

That 10seconds error from just using 20second divisions will equate to an angular error of 0.0008%. Plenty here seem to be happy to cut approx metric/imperial threads with a higher pitch error than that and the thread error is cumulative unlike the gear cutting.

The beginner reading some of the replies here may be put off having a go at cutting their own gear and decide to just buy them in from the likes of HPC. Now a quick look on HPC’s website tells us that their gears are made to give a backlash of between 0.07 and 0.3mm depending on size ( 0.25Mod to 6MOD). So for our 31T MOD1 example that would be about 0.1mm backlash. Half that is the reduction on each of the two mating tooth surfaces so that is 0.05mm of 50 microns Theerfore our “error” of 0.75microns is not going to make a monkeys difference in practice.

Mention was made of the gear possibly being a lathe change gear. General advice when setting up the gear train is to put a strip of 80gsm copy paper between the gears to set the backlash. Well what do you know 80gsm paper is 0.1mm thick, same as the backlash HPC machine to so again that tiny theoretical error is not worth loosing sleep over

I’ll leave it to someone else to work out what the consequences of using an involute cutter with its profile being used to cover a range of tooth counts is over the correct involute profile but probably more than our angular one.

[Peter Cook 6Participant @petercook6]

If you only need it for a specific job, or as the basis to make a metal set, it is easily possible to 3D print an index plate, and the modelling programmes are fairly good at PCD’s. Or there is what seems to be a customisable version  here on Thingverse to create the relevant files. But of course you need access to a 3D printer.

[old martParticipant @oldmart]

We needed to produce 118 divisions at the museum when I used a brand new 3mm pitch leadscrew to replace the worn out 8tpi cross slide one on the Smart & Brown model A lathe. The leadscrew came with two bronze nuts and machining the end to make it fit the lathe was easy, but changing the scale was not. I made a new scaleblank and Dick printed a list of 118 divisions per turn using a spreadsheet from Microsoft Excel, in degrees and the nearest minute. This was because the 3 plates that I had with the 6″ RT could not manage 118. The lathe is imperial, so the 118 per turn is accurate to about 0.0001″ per turn which is good enough for government work.

It is more time consuming than plates but perfectly possible. If I was doing 2,3,4,5, 6 or 8 divisions, I wouldn’t ever bother with plates

 

 

 

 

 

 

 

 

 

[JasonB]

On JasonB Said:

Idon’t know what others are expecting in the way of accuracy in the home shop but here is an example.

 

Daves 31T gear

..
What tolerence do people actually need and what are you making that needs it?

Wasn’t me who introduced accuracy! Presumably cropped up because my example 360/31 is 11.612903225806451 recurring?  I assumed, silly me, it was understood that the number of digits taken from a decimal number are adjusted to meet the needs of the job. Accuracy sufficient to meet the needs of the day, and no more.  In a 31 toothed gear rounding to:

• 11 causes a 19° error – unacceptable
• 11.6 causes a 0.4° error – might be OK on a traction engine, not on a DTI gear
• 11.61 reduces the error to 0.09° – still high for a precision gear
• 11.612 reduces the error to 0.028° – much more like it, at least on a small diameter
• 11.6129 rounded up to 11.613 gets the error down to minus 0.003°.  Good enough for most mechanical applications.  Hopeless for Satnav though: a 0.003° error is over 330metres!

Another problem to watch for is accumulating errors.

Going back to the “do I need dividing wheels” question, that’s a different case.   Index wheels on a rotary don’t improve accuracy; they reduce human error.  Without them cutting a 31 toothed gear requires the operator to calculate and meticulously follow a table like this one:

Gets worse as the tooth count rises, here’s 127 teeth:

Assuming the manual is correct, dividing wheels get rid of the possibility the operator will get the sums wrong, and the clock/plunger/indexer reduces the risk he’ll miscount or loose track.  This isn’t accuracy as addressed by Jason, this is accuracy improved by avoiding mistakes!

Having dividing wheels for a rotary table would be a no-brainer except better is available.  Turning the rotary table with a microcontroller further reduces the chance of human error, and  they’re much less tedious than manually clocking round an index.  Doing without either is possible, but bothersome unless the divisions are all simple or the table is rarely used for anything complicated.

Dave

[JasonBModerator @jasonb]

Good job you don’t cut any cam’s Dave, two axis to alter per cut and possibly 150 sets per cam lobe😁

11 causes a 19° error – unacceptable………………

We have all said avoid cumalative error so that 19deg total over 31 cuts come down considerably, using whole degrees you would only be a max of 0.5 degrees off.

Luckily I don’t nee dto use satnave to find my way to the workshop so will ignor the error as not related to the OP’s question.

Electronic division is nice but if the OP can’t justify the cost of a set of plates and quadrant going the division master route won’t help him either as that is more expensive than the zerp additional cost to use the handwheel he has got.

[Author]

Posts