Experiments with Chuck Centering

The recent discussion about the size of the register and the centering of the chuck on a threaded spindle led me down quite a rabbit hole.

I am a believer in the "threads center the chuck" theory, so I went to check the data I had on clearances on my chucks. I'd done this a few years ago in the context of calculating the thread engagement on various chucks and never tripped to the difference between my measured spindle diameter and the South Bend specification. At that time, I had measured my spindle at 1.498 and went on my way without realizing that the South Bend specification was 1.509.



I went back to check this and did a number of measurements using a B&S tenths reading 1-2" micrometer that had been zero set on a 1.5000" stack of gauge blocks. The overall average of the readings was 1.4969, substantially under the specified diameter of 1.509. I then took a close look at the area near the chuck seat, and it looks like the spindle may have been ground. I'm not sure what a South Bend spindle is supposed to look like in this area.



It's a very clean 1978 10K 4 1/2' underdrive with hardened ways that seems to be in excellent condition other than this spindle diameter surprise.

I realized this situation could be a good way to check chuck centering without a close fit to the register. I turned a slug of mystery metal in the 3 jaw chuck that has a register bore diameter .034 larger than my spindle. I then measured the turned area using a +/-.002" range, .00005" resolution Hamilton Watch dial test indicator set in the tool post. Four readings were taken at 90 degree locations by rotating the chuck by hand with the drive belt disengaged. Four repeats of the cycle were made.



The surface finish was not very good, but I didn't want to do anything that might alter the concentricty of the turned area. This led to some variability in the readings. The averaged readings on the indicator were converted to runout by subtracting the average of the measurements from each measurements. The runout measured .00016, .00000", -.00007", and -.00009".

This chart shows the variability in the measurements expressed as 3 sigma limits.



I then removed the chuck without touching the slug in the chuck. I screwed the chuck back onto the spindle and seated it firmly without using any momentum to seat the chuck. The measurement sequence was repeated, and the runout was calculated at 00011, .00001, .00001, and -.00014.

The chuck was removed a second time, and was seated by snapping the chuck approximately 20 degrees against the seat. The runout was calculated at -.00001, -.00008, .00012, and -.00002.

Here's the data presented in a graph.



I was pleasantly surprised by the low runout measured on the turned slug before the chuck was removed and reinstalled. The runout did vary with the two chuck changes, but even so, the overall runout remained below a total runout of .00025.

This chuck seems to center with a runout of .00025", with a register clearance of .034", over 100 times larger. It is clearly not impacting the centering of the chuck under these conditions.

I need to get a better combination of tooling and materials to get a higher quality surface finish and repeat the test, as I believe a large portion of the variability seen in the runout is due to surface finish, not chuck positioning.

Looking at the spindle picture, there is a lot of swarf on it, with some burrs on it too. These can be significant contributors to the runout you are seeing.

What do you mean by "snapping the chuck"?

And don't call that area behind the threads "a register", because it is not.

Wheels17: Thanks for posting that data, it looks like a LOT of time spent to provide this useful information. This clearly shows that the spindle/chuck "register" (no disrespect meant to SKL) actually has very little to do with runout. I know just enough about lathe work to get me into trouble, but from looking at your data, it is pretty clear to me that the "register" (sorry SKL, don't know what else to call it) diameter has FAR less to do with concentricity than the quality of the spindle/chuck threads, and how square the mating surfaces of the chuck and spindle ARE.

As an aside, that is an amazing dial indicator. I have never seen one that can accurately measure runout that small (my "best" one is a .0005"). I doubt if I will ever make anything on my SB9 that requires that degree (.00005")of accuracy. I'm not sure if my lathe is even capable of machining to that tolerance

Your data was a very interesting read, and I thank you for posting.

SLK001, thank you for commenting. We agree on the register idea! I'm answering your questions in reverse order.

There is a lot of controversy about the need to closely fit the chuck and spindle in the un-threaded portions of the spindle and chuck commonly referred to as the "register". Try a search! Many people on this board who are far more competent than I, and whose opinions I value very highly express the need for this close fit. I don't believe it. So I decided to get some actual data.

By "snapping", I mean the highly incorrect procedure of spinning the chuck into a sudden and hard stop against the shoulder.

There is a gap of .017 between the "register" of the spindle and the "register" of the chuck. There's no possibility that any centering is taking place based on these portions of the spindle and chuck. Please note the runout numbers. The maximum runout after the chuck was removed and replaced two times was .00025". 2 1/2 tenths on a lathe that is not perceived to be a high precision lathe. If only the rest of the lathe and my skills were that capable.

Regarding the swarf and burrs: The picture was taken before I started testing the runout. The spindle and chuck were carefully cleaned before starting the test. Due to the dimensions of the chuck and spindle, the damaged area of the thread is not used when threading the chucks onto the spindle. I suppose you could argue that a large gap between the spindle and the "register" is a good thing, since as long as there is no swarf on the threads in use or the shoulders, it has no impact.

QuickChange said:

...it is pretty clear to me that the "register" (sorry SKL, don't know what else to call it)

Click to expand...

Call it "the unthreaded area". More wordy, I know, but also more precise.

wheels17 said:

Regarding the swarf and burrs: The picture was taken before I started testing the runout. The spindle and chuck were carefully cleaned before starting the test. Due to the dimensions of the chuck and spindle, the damaged area of the thread is not used when threading the chucks onto the spindle. I suppose you could argue that a large gap between the spindle and the "register" is a good thing, since as long as there is no swarf on the threads in use or the shoulders, it has no impact.

Click to expand...

I assumed that you would do this. It's just that when I saw the burr on one of the threads, I thought that it would add a small amount of runout. You were measuring at so high of a resolution that you should be able to detect it.

wheels17 said:

There is a gap of .017 between the "register" of the spindle and the "register" of the chuck. There's no possibility that any centering is taking place based on these portions of the spindle and chuck. Please note the runout numbers. The maximum runout after the chuck was removed and replaced two times was .00025". 2 1/2 tenths on a lathe that is not perceived to be a high precision lathe. If only the rest of the lathe and my skills were that capable.

Click to expand...

I agree. My chuck has 0.020" between "the unthreaded area" and the chuck "clearance". This makes screwing my chuck on and off to be fairly easy all the way to the actual register, the shoulder. When it contacts the shoulder, the threads on the spindle will pull the threads on the chuck equally along the entire thread length. Any "slack" in a particular area (such as the lower part of the chuck due to gravity) will be dealt with by pulling it up the thread face, until all of the thread has been pulled to the max. At full engagement, the unthreaded area and the chuck clearance do NOT touch (or engage) so there is no possibility of the two registering.

Now if the unthreaded area had a taper and the chuck clearance also had a matching taper, then the threads would only push the chuck into the taper (the shoulder wouldn't be there) REGISTERING the two parts. But it doesn't, so we won't discuss that scenario.

It is a little tricky to wrap your head around what is really going on.
Turning the part and measuring it like shown, has nothing to do with how the chuck is threaded on the spindle or even if it is cocked at a slight angle.
The measuring/testing of the part in the spindle to tenths is not feasible because of the oil film and clearance on the plain bearings, and other.
The part needs to be inspected out of the machine on perhaps a surface plate with accessories, but what will that tell you? Interestingly, it will tell you nothing about your chuck or how its threaded on to the lathe. Testing the part properly will tell you how true the spindle bearings are.
The spindle shaft itself, can have the center hole running out .002", and the threads running out .002" the other way, but a part actually test cut can show near perfection in actual roundness...…
The remove and install with high accuracy, will usually work on a threaded spindle, but it can change more if the chuck is new to the machine or the threads freshly cut. If the chuck has more then .0015" run-out, vibration will cause the turned part to be larger on the end increasing with speed, out of balance chucks have the same effect.
The most important test for high accuracy using a chuck, is for the runout to be near the same 6" out from the jaws on a test bar.

Wheels17-

I missed the 'threads center the chuck' discussion, so apologies in advance if this retreads the same ground.

I believe SLK's comment about not calling that unthreaded area of the spindle a register may be part of the confusion. I agree, the unthreaded portion in your picture is not a register. I actually don't see a register on your spindle and wonder if it's been ground off for some reason.

I've got threaded-spindle Logans (so admittedly not a South Bend) but my understanding is that 'registers' that provide for repeatability when removing and reinstalling the chucks in this context are not the unthreaded major diameters of the spindle but rather a deliberately larger (significantly larger) diameter section of the spindle around which a relieved section of the back-plate mates. This means the threads are drawing the chuck onto the spindle but an actual register or journal is what provides for concentricity. And the face of the backplate pressing against the face of the register is what prevents the chuck from seating at an angle to the spindle.

Regardless, you're getting good repeatability and your indicator is pretty unusual.

-DD

The threads are very clean - try this with the typical chuck with chips embedded in the threads.
Also suggest you repeat the test, with an 8" long extension on the workpiece.

I once ordered a faceplate for a Logan 9B28 lathe. The machine had a 1.5”8 pitch screw spindle nose. When it arrived the neck measured 1.515”. I called Scott Logan and asked about it. He told me that he had to put some clearance in there because he sold a lot of parts that were being used on South Bends, and they are over sized. He also said making them that tight was not a good idea as they tend to get stuck on the spindle. The plate was perfect on my Logan and my SBL.
.