Rotary Table Questions

[RobinHood]

@Susquatch, I do agree with your statements about accuracy.

Lots of gear trains have enough backlash that the gears can have some tooth form error (spacing differences, profile tolerances, and run out allowance). Plus I think most systems also have a run-in period where the elements in a gear train wear in as they mesh with each other for the first time.

Here is the test report of my Japanese made Yamatokoki Co. ‘NEWS‘ Universal dividing head from 09/ Sep/1968:


As you can see, the dividing accuracy of the worm drive is +/- 46”.

This is a very well made precision DH and probably would cost you in excess of $5000 today. The “News” brand of tooling was a division of YUASA.

If you have to force the crank handle so that it would bend in order for the spindle to turn, it is time for a serious inspection of the tool. There should be very little to no resistance whith the lock released.

The plain bearings are in very good condition and ride on a very thin film of oil. If the handle had a bit more mass, it would keep turning if you gave it a good spin - like it does on my BP 12” RT. The handle runs on for about one turn after being released. (Unfortunately I don’t have a report sheet for the BP RT).

You mention plunger play - there is no discernible radial play on the DH. The parts seem lapped together. There was just some dried out lubricant preventing smooth operation when I got it. Careful disassembly, cleaning, lube and reassembly took care of that.

The plates are a very close sliding fit onto the crank spindle. There are countersunk fasteners that guarantee zero play attaching.

The crank itself is keyed. There are two lash eliminating screws to reduce play to zero after installation.

I already mentioned the adjustable worm/worm wheel arrangement.

They went to great lengths to make this as precise as possible.

I fully trust this tool. As stated above, if I did not have one, I would certainly consider electronics together with a RT to address my dividing needs.

 

[Susquatch]

@RobinHood - As always, I write things with what I mean in my head and then others read something else in them.

I didn't mean that the crank handle bends. I meant the handle for the plate pin is thinner than the regular round crank. All metals no matter how big or small bend. But the smaller they are the more they bend. Stress is metal bending in response to force. Steel is strong, but it bends too. No matter how smooth the table is or how little friction the bearings have, the handle has to be bent to retract the plunger against its spring force. I'd bet there is a few thou of bending associated with that. A few thou is quite a bit more than a minute of arc at that radius.

I like your calibration print out. I'm certainly impressed to see a precision under a minute of arc. Mine is certainly not that good. At best, +/- one minute ASSUMING no other errors. But that is for the main worm drive in both our cases. I believe that's as good as it can get and any indexing or secondary driving system will probably be worse.

I don't understand how anything can slide without some clearance to allow it. The total length on my plunger assembly is about 2 inches with about half an inch in the spring sliding mechanism. It's loose enough to slide freely. I doubt it's lapped. But even if it was, there is going to be a little movement. Any movement (no matter how small) is a deterioration from the direct indexing of the direct drive system.

I'm not saying it's terrible. I'm just saying I believe the dividing plate system is not as good as the direct drive system. But both are perfectly usable for any gear making activity.

It's important to understand why I was writing about this in the first place. It wasn't meant to knock dividing plates - although I can easily see why others might think so. It was meant to say there is no reason to be concerned about the accuracy of the regular crank system. In my mind, plates are not a requirement, they are a convenience. The standard drive angular measurement system is not worse. In fact, as I described, it is actually a bit better. But not meaningfully so.

I too would totally trust either system to make a gear.

 

[Susquatch]

807 USD ouch
https://www.sherline.com/product/8700-cnc-4-rotary-table-indexer/

Where can I buy just the electronic stuff.... for say... $100 lol

I think the thread got a bit mixed up on your question. But in case you missed it.....

Replace that calculator looking thingy with an Arduino Nano the size of a thumb drive and an input reostat to generate two or three digits for the number of divisions, some clever programming, and voila - your desired system for well under a hundred bucks.

I'm gunna do it too.

@Johnwa - I can't seem to find the latest version of that software that you mentioned. Could you please link to it or help me find it?

 

[PeterT]

Does anyone here really believe that any of those holes on that 4" plate is within 0.000004 of an inch from the correct location? Then there are all the other errors too. Just lining up that hole pin and stressing the mechanism being a huge one.......IMHO, rounded minutes of angle using a plain old calculator, a programmable calculator, or a spreadsheet are no doubt more accurate than the plate despite its magical prime number membership.

I don't think anyone has suggested hole plates are more accurate. I suspect the intent is more about reducing operator error. And maybe that's debatable if you personally favor aligning a Vernier scale 127 times vs rotating a sector arm & pinning a hole 127 times. Both are mind numbing if you ask me. There are entire textbooks written about jig design principles pertaining to the human element (typically assembly line work). So I'd guess someone has studied this long before we came along & concluded plates are a good thing, otherwise why bother. And if you want to talk about accuracy then its probably prudent to consider all other factors that contribute to the final part production - alignment & concentricity of the blank, cutter tool geometry... Its an interesting topic. When you see videos of how industrial precision gears are made, those machines look nothing like what is in our shops. For one thing they typically are hardened & ground. Cutting or hobbing are usually roughing operations before they go in the oven.

Another thing I was just thinking about (always dangerous). How accurate is the worm & gear inside the RT? I'm not talking backlash or diminished effect of gear ratio, I mean the gear train accuracy itself. Yes, you have visually aligned the vernier scale all right, but does that position really equate to an angle accurate to arc seconds? By analogy, I can turn the knob on my 20$ Chinese micrometer so the vernier 10ths mark lines up & choose to believe its measuring 1.0000". But I'm pretty sure if I compared it to gage block it could read 0.999 to 1.001 on a good day. So much for tenths. What kind of barrel screw thread was I expecting for that price LOL. Maybe for fun & excitement you should rig up a repeatable stop on the RT table referenced to a zeroed DTI. Crank rotate the table 10 times & terminate at the same DTI reading. If its exactly 360 degrees c/w zero arc seconds, we have a winner. If Its 357 degrees or 362 degrees, well now we know more about our gears & true angular resolution on the scale.

 

[Susquatch]

I don't think anyone has suggested hole plates are more accurate. I suspect the intent is more about reducing operator error. And maybe that's debatable if you personally favor aligning a Vernier scale 127 times vs rotating a sector arm & pinning a hole 127 times. Both are mind numbing if you ask me. There are entire textbooks written about jig design principles pertaining to the human element (typically assembly line work). So I'd guess someone has studied this long before we came along & concluded plates are a good thing, otherwise why bother. And if you want to talk about accuracy then its probably prudent to consider all other factors that contribute to the final part production - alignment & concentricity of the blank, cutter tool geometry... Its an interesting topic. When you see videos of how industrial precision gears are made, those machines look nothing like what is in our shops. For one thing they typically are hardened & ground. Cutting or hobbing are usually roughing operations before they go in the oven.

Another thing I was just thinking about (always dangerous). How accurate is the worm & gear inside the RT? I'm not talking backlash or diminished effect of gear ratio, I mean the gear train accuracy itself. Yes, you have visually aligned the vernier scale all right, but does that position really equate to an angular position accurate to arc seconds? By analogy, I can turn the knob on my 20$ Chinese micrometer so the vernier 10ths mark lines up & choose to believe its measuring 1.0000". But I'm pretty sure if I compared it to gage block it could read 0.999 to 1.001 on a good day. So much for tenths. What kind of barrel screw thread was I expecting for that price LOL. Maybe for fun & excitement you should rig up a repeatable stop on the RT table referenced to a zeroed DTI. Crank rotate the table 10 times & terminate at the same DTI reading. If its exactly 360 degrees c/w zero arc seconds, we have a winner. If Its 357 degrees or 362 degrees, well now we know more about our gears & true angular resolution on the scale.

Bang on.

You are singing my song now!

And yes, I did feel like some were saying (or implying) that a dividing plate was a requirement to achieve certain angles. Which I did not agree with.

But I do agree with every point you have made.

Now, let's get on with that stepper motor design. Which will be even worse but still will make great gears!

And for what it's worth, I think making a gear is mind numbing regardless of what method is used to set the angles! However, the resulting gear is almost always a thing of pride!

 

[Susquatch]

@PeterT - I forgot to mention the 127 hole plate you drew up. I was glad to see that. In my minds eye it was even worse.

But it did make me wonder. Why couldnt one make a plate with say three separate pins that are offset radially such that they would align with three circles of holes instead of just one. "Divide" the beast up so to speak. I'm having a bit of trouble visualizing that too since you cannot divide 127 by 3. But I assume it could still be done by skipping a few holes on one or more of the circles and a bit of dancing on the sequence.

 

[Johnwa]

Kquiggle‘s documentation and code is avaiable at
https://sites.google.com/site/lagadoacademy/miscellaneous-projects/stepper-motor---accelstepper-h

I’m using the code I posted in message 149 of the HMEM thread.
I intend to migrate to Kquiggle‘s version but I want to add a rotary encoder that will at least give an error or possibly even correct for missed steps.
I have an $5 encoder that can be easily mounted on a stepper. The specs suggest it should be accurate to 0.1 degree but my test show it’s more like 0.3. With a 90:1 rotary table it would be fine but with my 4:1 indexer its a bit large.

 

[Dabbler]

@Johnwa now that hybrid stepper motors and controllers are getting so cheap, perhaps someone starting out new might use one? Just a thought...

 

[Susquatch]

@Johnwa now that hybrid stepper motors and controllers are getting so cheap, perhaps someone starting out new might use one? Just a thought...

I was thinking exactly that!

Based on a quick review of stepper technology late last night (a welcome break from the stress of moving my mother), I found myself disappointed with what the regular and even half step Motors could do. (1.8 degrees and 0.9 degrees per step.) a far cry from the one minute my RT can do mechanically. But then again, maybe enough...... LOL!

The hybrid system were also interesting, but I couldn't find anything that said they worked with static systems - only low speed applications. More research needed.

That got me wondering about alternatives and of course also thinking about the many issues that any stepper system would have to address.

I wondered about a gear drive to improve stepper resolution, and stumbled onto a 5:1 sun gear stepper on Amazon that struck me as an obvious improvement and only $60. That's at least in the 10 minute neighbourhood. Maybe there are 10:1 or even finer Motors out there too. I don't know - I didn't look.

https://www.amazon.ca/dp/B075KCC1QS/ref=cm_sw_r_apan_glt_i_Q5PJNESBRNRH2J74QEA9?_encoding=UTF8&psc=1

The encoder problem is a big one too. Again, I wondered why a stepper motor doesn't always include that built in right off the starter blocks? Maybe some do.....

I think I already mentioned this, but it seems to me that a simple reostat could be used to dial in a number of steps displayed the same way as an rpm readout - or maybe a bump up/down button.

To your point, when I was still working, we would routinely throw a newbie at a project just to see what whacko ideas they might come up with that might be better than what people who "always did it that way" might do.

But sometimes it's also good to be lazy and just do it the old way cuz it works. Take the good where you find it and weed out the bad.

 

[Dabbler]

Almost all of the stepper drivers will microstep to at least 16 microsteps, and the better ones lock correctly up to 64 microsteps. Most 3D printers use the TM220x series stepper motor drivers that do all of that, and silently. Doing the correct torque/current calcs woth appropriate acceleration/deceleration factors it would be very rare to ever lose a step...

 

[Johnwa]

Hybrid Stepper motors would certainly be an improvement. Loose set screws and slipping couplers can still get you though.@#$&!!!!

If my calculations are correct a stepper(at 200 steps) directly driving a 90:1 table will give positional error of less than 0.001” on a 1 inch diameter gear. That’s more than adequate for me.

 

[Susquatch]

Hybrid Stepper motors would certainly be an improvement. Loose set screws and slipping couplers can still get you though.@#$&!!!!

If my calculations are correct a stepper(at 200 steps) directly driving a 90:1 table will give positional error of less than 0.001” on a 1 inch diameter gear. That’s more than adequate for me.

My calculations support yours. At a half inch radius, one full cog error is only +/- a half a thou. I'd sure be happy with that too.

 

[Susquatch]

As a result of dreaming about stepper motors, I did a little more research. Here is what I thought was a really good evaluation.

https://hackaday.com/2016/08/29/how-accurate-is-microstepping-really/

In particular, I loved the idea of mounting a laser on the motor spindle to evaluate arc accuracy at a distance to magnify any error. This could be applied equally well to the Rotary Table itself to calibrate it. I would use green instead of red (for visibility in daylight conditions) and could probably get a 20ft radius inside my shop or 100 yards out the barn door to the forest. Basically lots of space to do a downtown evaluation. I'll have to do some math to see what is really required to do a proper evaluation.

Anyway, I thought some of you might find the article as interesting as I did.

 

[Susquatch]

Or how about a rifle scope? Even better than a laser. They are typically setup to provide 1/4 moa adjustments (1/4 inch at 100 yards). It would be easy to mount a scope on the RT and sight on a piece of calibrated paper outside the shop door. At 100 yards, 1 MOA is approximately 1 inch. At 50 yards it is 1/2 inch, etc. If I put a scale on the target paper, I don't even need to go for a walk. The scope will provide the required magnification to read it right from the Rotary Table.

 

[RobinHood]

Or how about a rifle scope? Even better than a laser. They are typically setup to provide 1/4 moa adjustments (1/4 inch at 100 yards). It would be easy to mount a scope on the RT and sight on a piece of calibrated paper outside the shop door. At 100 yards, 1 MOA is approximately 1 inch. At 50 yards it is 1/2 inch, etc. If I put a scale on the target paper, I don't even need to go for a walk. The scope will provide the required magnification to read it right from the Rotary Table.

If you are going to go to that length of evaluating your RT, you could also use a surveyor’s theodolite - kind of like Alex does here…

 

[Susquatch]

If you are going to go to that length of evaluating your RT, you could also use a surveyor’s theodolite - kind of like Alex does here…

I happen to have a good digital theodolite so I did consider that approach just a few cerebral farts after the rifle scope idea. But I quickly concluded that a good scope has much better optical quality and the cross hairs are finer. That said, I also thought that the RT and scope might be a good way to calibrate the theodolite!

After discussing the pros and cons of dividing plates with you, I found myself insanely curious about all this stuff. I want to know more than I do if for no other reason than to confirm for myself that it either does or doesn't have any practical impact on my own applications. As it is, I suspect it's all pretty academic.

 

[RobinHood]

Here is my take on dividing plates: they have been around a very long time. Dividing heads (and RTs) were the only means of making accurate divisions of a circle for many decades. Remember, even the DROs were not available - so no bolt circle function. Parts still needed to be produced in large quantities to certain accuracy levels (depending on what the end use of the part was). So workers in factories were tasked with the monotonous work of cutting gears, for example. Probably to reduce the number of spoiled parts, dividing plates were used to keep the parts within tolerances and reducing errors vs setting a bunch of angles (numbers) read out of a table.

I think either method can be as accurate as the other given the right size of dial scale and fine enough graduations or accurately spaced holes on a hole circle. The rest is down to the operator.

Electronics changed everything. I bet you if stepper motors were available before the first RT or DH with vernier scales / hole plates were invented, we’d only see steppers and the associated electronics.

I think the same can be said about theodolites vs electronic (gps) surveying equipment. Why bother with super precise and hugely expensive Wild Heerbrugg or Leica manual instruments if the same can be done with electronics but only better? Because the electronic ones came later.

So for no other reason than that, I had my manual BP RT (with a dial vernier to arc seconds) and my News Universal DH with plates before I got any other circle dividing device.

Very good discussion indeed. As stated previously, I can foresee an electronic 4th axis for my mill - especially after seeing Tom Lipton cut his spiral fluted plastic drill with such ease. Very cool.

 

[Susquatch]

Yup. I think you are absolutely right about all of that.

Maybe because I spent the early part of my career researching and designing electronic sensors and systems to control or replace mechanical systems in cars and trucks, I might have a slightly different perspective of what is more reliable though. I like electronics as much or more than most people. One look at my electronics lab and all my electronic gadgets would quickly tell you how much I love electronics. But with knowledge also comes contempt. Ya, it's possible to do things electronically that you cannot do mechanically. But I just don't trust electronics nearly as much as I trust good old fashioned mechanical systems. Most of my distrust results from my experiences with the skill set of the programmers and circuit designers. The vast majority have zero or very low experience or knowledge of the mechanical systems they seek to emulate or control electronically - alternatively, some are experts mechanically but know little about the electronics side. With few exceptions, skilled mechanical people don't do electronics or programming. The result is often amazing but garbage in is still garbage out.

And perhaps equally troublesome is the gullability of us users. If it's digital and has a number we can read, too many of us trust it as if our life depended on it. Me, not so much.

I have a saying that I love to use for electronic devices. "Trust is earned, not given". Of course that saying applies equally well to many other things including a lot of people!

Btw, does your RT really read arc seconds or was that a typo and you meant minutes? If it really is seconds, I'd love to know more about how it does that.

Anyway, I don't mean to come across as down on electronics. I'm really not. Good designs can do absolutely amazing things that would be impossible otherwise. I guess it would be better to say that I don't have blind faith.

Yes, spiral milling would be one of those absolutely amazing things wouldn't it!

 

[RobinHood]

Btw, does your RT really read arc seconds or was that a typo and you meant minutes? If it really is seconds, I'd love to know more about how it does that.

Amazingly, yes it does.

To 5 arc seconds on the vernier divisions (with some interpolation accurately to at least 2.5 arc seconds).

One turn on the dial is 3*. Hard to take a picture because the dial is so large, but I think you can just see the 0* mark on the right and the 1* mark on the left with the 60 minutes division in between.

 

[Susquatch]

Amazingly, yes it does.

To 5 arc seconds on the vernier divisions (with some interpolation accurately to at least 2.5 arc seconds).

One turn on the dial is 3*. Hard to take a picture because the dial is so large, but I think you can just see the 0* mark on the right and the 1* mark on the left with the 60 minutes division in between.

View attachment 17773

Wow! Yes, I see the gradations you mention and I can also see the minutes, the 5 second vernier, and even the 2.5 second interpolation. That's one very impressive RT!

Amazing what a big diameter dial can get you!

More importantly, properly used with backlash properly handled, I would absolutely trust that!

Bet you love it!

And ya, I'm jealous!