What's new
What's new

Accuracy of Dividing Head Rotary Table?

The fact that you can see movement with the locks off is why they need to be locked for the most precise work. In fact, it was by putting an indicator on a mill's table and observing that movement that brought me to this conclusion.

You do not start using the locks in the middle of making a part. If you do, then the features before that point are questionable because the table could be anywhere that the tightness of the gibs would allow.

The idea is to use the locks to the greatest degree possible from the very start of a part until it's last feature is cut. All operations are performed with the table and therefore the part, locked down to prevent as much movement as possible. Yes locking does shift the part a bit. That is the whole idea: you shift it to a known and REPEATABLE position. Loose gibs can leave the part anywhere in their range of sideways movement.

I see discussions about milling vises and the problems associated with clamping parts in them. Movable jaw lift, fixed jaw bending/lift, even the base flexing (I have a small vise that shows this). And many of these problems are measured in tenths. Well, I have seen table movements in the range of 2, 3, and even more thousandths. They can be at least an order of magnitude larger than typical milling vise problems.

So, do I make parts with the locks off? Of course I do. Unlocking before and locking after every table movement is a PITA. No one likes doing that and if you are in a commercial shop making parts with broad tolerances; it will cost time, which is money.

But I say if your part needs the highest accuracy, then you need to keep the gibs tight and use the locks every step of the way.

In fact I wonder if some of the problems discussed here are not caused by unknown table movements due to it being unlocked instead of the vise or other factors that we are so fast to blame.

By the way, it is easiest to see table movement at the ends where it is amplified by the lever arm principle. But it will most likely cause problems at the center point where those two sets of dovetails cross. That central area is where 99% of the cutting takes place on a mill. A cutter with a single cutting edge (boring bar or fly cutter) could move the table first to one side and then to the other, thus creating an oval hole or other problems.

Gentlemen, LOCK your slides!

If you want the best, most accurate work, that is.

.....

I am going out to the shop now to complete a LOOSE tolerance part (+/-0.005" to +/- 1/16") and I will not be locking anything except the quill feed. This part just does not require the extra time.



I don't agree with this - some of it anyway. I do agree that actuating a table lock on most machines is likely to shift the work a bit, especially if the machine is worn. I do not agree that the locks should always be left off though. With clamps off I've had bored holes go out of round, unintentional movement causing dips in milled surfaces, etc.

What I do is compromise. When I edge find, indicate, or what have you, I bounce the lock on and off as I bring the table into location, finishing with the lock on. Same when approaching hole locations. For milling, I will generally leave the clamps snug on the axis that isn't moving. A quick mic to check center sometimes shows a shift in centerline when milling with the clamps on but the holes will be right on the money, that's very machine dependent. One of those things you've just got to be aware of and compensate for if necessary.
 
Last edited:
Dividing heads and rotary tables rely on worm gears and divisions on their hand wheels. At least, those are the two primary factors. And, like everything else that we use, they can differ in accuracy. You ask about those angle setting devices vs. a DRO. Here are the considerations I see for both ways.

DRO:
With a DRO you are working in the X-Y plane with movements in those two directions. The accuracy depends on several factors. First, the accuracy of the DRO scales. That should be published by their OEM. But remember the least division they can show is NOT the accuracy. A DRO that reads down to 0.0005" may only be accurate +/-0.002" in a foot.

Another consideration is how parallel the scales are to the X and Y axis and how perpendicular they are to each other.

Finally, errors can occur if the gibs are not tight. I have seen mill tables wiggle 0.003" or more when the table locks were not clamped down. This was due to the gibs being adjusted for easy and free movement. For maximum accuracy you should lock the table down in both X and Y axis before drilling a hole.

Rotary table or indexing head:
These devices are primarily for setting angles. And, instead of working in the X-Y plane as you are accustomed on a mill, you will be working in a R (radius) and a (angle) coordinate system.

In practice, this means that the radius of a hole circle will be set using the table feeds and the DRO if you have one. First you locate the center and then you move the table for a distance equal to the radius of the hole circle. The same factors as discussed above for a DRO apply to this dimension.

That leaves the angular accuracy of these devises and how that translates to the accuracy of the hole locations in the tangential direction (the direction tangent to the circle of the hole circle). To translate the angular accuracy to the linear, tangential accuracy a bit of math is needed. This calculation requires that you know the radius of the hole circle and the angular accuracy of the rotary table or indexing head.
So if:
R = radius of bolt circle
da = angular accuracy of the device
Then:
Tangential error = R * tan(da)

On a scientific style calculator enter the angular accuracy (da) and press the Tan button. Then press X for multipication and enter R and then =. The calculator should then display the linear accuracy in the tangential direction.

Caution: most calculators only accept angle values in degrees or some other primary angular unit. So entries in degrees, radians, or other primary unit will be accepted if you can select that unit for your calculator. But arc minutes and arc seconds are not acceptable and must be converted to a primary unit, like degrees, before taking the tangent of the angle. I show this in the example below.

Example:
R = 6" (hole circle with 12" diameter)
da = +/- 10 arc seconds (from OEM specifications)

Tangental Error = 6 * tan(10")

On most scientific calculators you will enter 10 but that is arc seconds, not degrees so you must convert that to degrees first. Arc seconds are 1/60 of an arc minute so you divide that +/-10 by 60 to get +/-0.1666... arc minutes. But that's still not degrees.

Likewise arc minutes are 1/60 of a degree so you divide by 60 a second time to get +/-0.002777... degrees. Now you can press Tan to get the tangent of that angle in degrees: +/-0.00004848... And multiplying by 6" you get +/-0.00029" or just about +/-3 ten thousandths.

Notice that I put a +/- in front of the intermediate figures and the final error figure. That is because the OEM's angular error is specified as a +/- number and that condition carries through the entire calculation.

Also notice that the radius of the hole circle is a factor in the final multiplication. This means that a 2" radius will have twice as much error as a 1" radius. And a 6" radius will have three times as much as a 2" one. The error increases as you move outwards from the center.

Since I used reasonable numbers in that example, chances are that a decent rotary table or indexing head will provide less error due to the angular spacing than a DRO will provide in the radius of a bolt circle. But your device may vary (YDMV).

Also, if you want the best accuracy when using a mill, then LOCK as many movement axis as you can for the current operation.
What he said.

Lets put down the beer and use both hands on the calculator for a second. It's EIGHT holes on 12" bolt circle. If they're evenly spaced that means FOUR of the holes are in your primary X-Y directions: 0,6 6,0 0,-6 and -6,0. Those are whole numbers, no decimals and no error possible in calculation. They is what they is.

That leaves the four remaining holes at 45 degree spacing. If there were more holes and they were at shallow angles with respect to either axis, you could maybe concern yourself with rounding errors of the DRO but, they're not. They're at 45 degrees. Any inaccuracy is shared between X and Y. It's the best location possible.

Calculate the theoretical position to 5 decimal places:
Sin 45 = 0.70710678
0.70710678 X 6"= 4.24264" (for both x and y positions)

That's the perfect location. If your DRO reads to half a thousandth you'd dial it to 4.2425" and call it a day. Your theoretcial positional error would be 0.00014".

It now comes down to: how square is your machine and how accurate is your DRO? Academically, a rotary table might be more capable of making roundy parts round. Without a DRO on it, you're not going to get the table to stop within a ten-thousandth of an inch of true position at 6". Even if you had a CNC rotary table, I doubt any process you'll apply at that location will be that close when it's done either.
 
The OP is teetering right on the edge of a very deep rabbit hole. It's a whole lot of reading, but I don't know of a better source to explain accuracy and coordinate location than the Moore Tools book with a PDF of it found here. https://ia800104.us.archive.org/20/...curacy/Foundations_of_Mechanical_Accuracy.pdf

It's pretty much how they got famous for there Jig borers, grinders and coordinate measuring machines. And it isn't quite as simple as one would think.
 
A rotary table will do good work if you pay attention to the handwheel micrometer/angle indicator, and set it as close as possible each time you make a turn to a new position.

I use the locks on anything I do. I have a small, less than rigid, machine. I need all the help I can get.

pivot roller210.jpg

wheel19.jpg
The rotary table, although not the best, has been a real boon to me. I have a set of dividing plates that fit it, but never have used them. Just do the math, and watch the marks on the handwheel.

Accuracy is a function of money I guess. Depends on how deep your pockets are.
 
Thanks for all the feedback! I am super happy that I asked this question because it sounds like I can skin this cat a few different ways. I was leaning more towards the indexable jig but I think I am going to have a crack at it with my DRO. If that fails to meet the tolerances this part requires, then I am going to buy a nice rotary table with a dividing head and see how that compares. Thanks again!
 
I was pretty impressed by .0017" at 12" of diameter with a rotary table. I am not sure if my DRO and Mill combo has that accuracy.. Maybe it does.? This is a new part for me so we will see.
 
Last edited:
Wasn't it Moore who had the ad where they had several plates with hole patterns made on their machinery in several different shops across the country and you could drop all the pins in the holes with some few tenths clearance and they would drop through all of them no matter which way they were rotated? That's precision. Way overkill for most general stuff though, and you sure aren't going to get there with drilled holes... I'd guess that just about any rotary table or especially dividing head from an actual name brand would be plenty accurate enough for a drilled hole pattern.
 
WOW that is impressive.. I'd like that kind of accuracy for the part in question, but I do not "need" it..

I would say point to point is my highest priority. Hole diameter/roundness/etc matters, but I want to start off with the centers in the right spot... I wasn't sure my DRO/Mill combo would do it. I usually dive in head first, but I figured this time I would ask those who are more knowledgeable than me.

Thanks again!
 
Wasn't it Moore who had the ad where they had several plates with hole patterns made on their machinery in several different shops across the country and you could drop all the pins in the holes with some few tenths clearance and they would drop through all of them no matter which way they were rotated? That's precision. Way overkill for most general stuff though, and you sure aren't going to get there with drilled holes... I'd guess that just about any rotary table or especially dividing head from an actual name brand would be plenty accurate enough for a drilled hole pattern.
I seem to recall that was a full page ad for either P & W or maybe B & S jig borers that was done with somewhere between 7 or maybe it was 9 companies using there model of jig borer. And given the time period involved, around the 1930's I think, it was all done using manual machining coordinates. There were a very few holes in one or two of the plates that wouldn't pass. But I'm sure someone here remembers the more exact details better than I am.
 
One thing to think about is that a lot of RT's are 90-1 where as most common on the smaller DH's is 40-1. So if any thing because of the lower helix angle of the gears the 90-1 RT could be more accurate with a set of hole plates.
 
One thing to think about is that a lot of RT's are 90-1 where as most common on the smaller DH's is 40-1. So if any thing because of the lower helix angle of the gears the 90-1 RT could be more accurate with a set of hole plates.
I worked at L&W chuck from 1984 ot 1986, we made dividing heads among other things. We made two sizes, the small had a wormwheel about 3: dia and the larger one the wormwheel was about 6". The older man who assembled the dividing heads and made them ready to ship had a master that he used to inspect them once assembled. As far as I know the standards were the same for both sizes of dividing heads. From memory the master had about 10 notches on it's OD and he would index to each spot and then check that notch with a dial indicator. No idea what his exact pass fail numbers were.

We made our own worms and wormwheels, the wormwheels were cut on a gear hob, they were made out of cast iron. We would hand feed to depth then blow off any chips and then let the machine run with a piece of clean typing paper under the cutter and use a stopwatch to determine when absolutely no cutting was still going on, "sparking out" kinda LOL. Then if that time period was 10 minutes afrer we reached full depth we would just "spark out" the whole batch of them that long on each part.

Some wormwheels did not pass final inspection after assembly, we would carefully realign those on the arbor and recut them.

The ID hole in the wormwheels was carefully made as well, we had a broach we used that had a series of burnishers after the cutting teeth that was the 1960's I suppose way to get the same exact hole size every time, they were a snug fit on the spindle of the dividing head.

I have only had a few rotary tables apart but the wormwheel on them was typically machined into the table casting ?

Point is that IMHO there are a LOT of factors involved in accuracy that may not be as clearly defined as just the gear ratio of the device.
 








 
Back
Top