3jaw chuck alignment problem

[Anatol]

The below link shows runout of a shaft in a lathe. Is this what was observed?

yup, thats almost EXACTLY what I observed.
thanks.

 

[petertha]

No harm at all verifying lathe twist first. Get that established beforehand. But don't proceed to jacking a foot & introducing compensation twist until you validate headstock/spindle alignment. If its out even a smidge, it can have a significantly larger impact & you will be chasing your tail twisting the bed or worse. To put numbers to it, if you measure 5 thou of deviation on a 12" test bar, that translates into a headstock angle rotation of only 0.024 degrees relative to the bed axis. Tiny! Those head stock jacking set screws are very sensitive. When I adjusted mine I figured out which way the head would unload & just by cracking one set screw & not the other it aligned itself perfectly. Yours may have been incorrectly set at the factory, drifted on its own through use or I have heard of a it developing after a machine move with a sling around the head.

Well meaning people are often eager to recommend shimming a foot (twist compensation) to reduce 'head stock taper cutting'. But I suspect this advice dates back to machines where the head stock is integrally cast to the bed & spindle bored 'very close' to lathe bed to begin with. If you have a lathe like that, great. But that is actually quite uncommon on typical current Asian hobby lathe offerings where the spindle is contained within a free floating gearbox block until its bolted down into position. So jacking the bed is introducing a second wrong to try & make a right. Now you have a twisted bed and a cocked head stock. So that's why I asked that question up front. Notice that none of this involves rotating the spindle. It is just measuring angular deviation along the center plane. The spindle could be running with zero TIR but that's not addressing a potential head stock alignment problem which is causing taper cutting which is relative to the bed ways.

A cylindrical test bar will work fine enough for what you are trying to solve. Just realize that you are introducing the chuck & jaw grip into the error circle. So re-clamp a few times & look for consistent measurements. I bought a very good value test bar off ebay like this. Not only is it good for headstock work, it may also fit your tailstock hopefully & be beneficial there too which is the next thing to dial in. But first things first

 

[petertha]

The below link shows runout of a shaft in a lathe. Is this what was observed? This is not the same as a twisted bed or misaligned headstock. I'm sure more suggestions will help, but let's make sure we're solving the right problem.

Hmm.. now that is interesting. But I wonder what we are looking at? It doesn't look like a test bar, it looks like stock that was chucked, heavily turned & now measuring deviation. Did it just stress relieve and go 'bwang'? That would account for DTI accurately mapping a banana. You can see the remnant of a nub on the end of of the bar. Maybe the parting operation applied bending force & it distorted? We need to see Season-1. Where was tail stock in all this? How hot were the chips? Who is the mysterious alloy? LOL

 

[petertha]

My lathe if its helpful. The grainy pic with the 2 arrows are the jacking screws.

 

[goldstar31]

If I recall, Georg Schlesinger advocated a hollow test bar to minimise the droop from the weight of the test bar.

Again, he produced a set of tests for machine tools including lathes but and this is a BUT, never included lathe chucks and especially second hand chucks of unknown provenance!

So, as I said before, let us follow an internationally recognised test procedure rather than the seemingly 'hit and miss' procedure which most of us are trying to examine and hopefully , correct.

Norm

 

[mu38&Bg#]

The video was simply to make sure the correct terminology is being used to describe the problem (runout increasing away from the headstock per post #1). The problem in that video is severe, but it was short and to the point. I don't care how it got that bad.

Why muddle the real problem which is spindle/backplate/chuck runout, with aligning the bed/headstock? As I said in my first post, runout and alignment are UNRELATED. I'd like to see someone use an indicator on at test bar that runs out .030"at 4" from the chuck (per post #1) to align a bed/headstock, what fun! Remove the headstock, ship it to me and I'll measure it to find the problem, no bed required.

I assumed( in error) that the intent of this thread was solve the actual problem. I understand the lathe setup process, though I can't throw any of these old names out there (except Moore).

 

[petertha]

I think we are lacking a clear picture of what deviation is actually being measured, which leads to guessing or interpretation. Throw in some non-standard terminology, or possibly standard terminology applied incorrectly and things can get messy.

Here is where I'm coming from. Going back to post#1 he says: A rod set in the jaws describes a cone. Close to the jaws, the rod is 10-15 thou out, but 4” down the rod the error increases to 30 thou or so. Tracing back along the rod into the chuck, this would suggest the minimum error on the rod - the point of the cone - would be back inside the chuck - if you could measure it.


describes a cone. A cone is a 3D object so kind of infers the test bar is rotating. But he never actually says rotating. He took a measurement close to the jaws and 4" down the rod. I can visualize a cone if
1. the test bar was straight but gripped in the chuck at an angle offset to spindle axis. This could be the jaws, the chuck, the mounting back, the spindle nose.... He said he replaced the chuck with another & got similar results. So maybe the mounting back or spindle nose is suspect if they are common to both chucks. I suggested removing all the rotating hardware to eliminate them & insert an MT test bar into the spindle socket. He doesn't have one, so the jury is out
2. the test bar is bent outboard of the chuck. I assume this is not the case because even rolling on a plate would show 30 thou big time


tracing back along the rod into the chuck, this would suggest the minimum error on the rod - the point of the cone - would be back inside the chuck - if you could measure it.
- this is just a guess, maybe I'm 100% wrong. It suggests he is just tracking a DTI down the bar, no spindle rotation. So I will volunteer this sketch as one possible interpretation. This is why I asked if he has a detachable head stock, which he has.
- this form of misalignment under spindle rotation does not describe a cone, but it would certainly cut a CONical taper. Is that what he meant?
- point 1 on sketch, is that what he means by point of cone would be back inside...?
- now just considering this mode of yawed head stock, how do we interpret when someone says 'runout'? If the test bar is 100% concentric to spindle axis, there is no runout between those components on that basis. But there is runout of the test bar relative to lathe bed axis if we introduce that aspect in real life machining. So which are we talking about? There probably is a precise definition or we should use extra words to be clarify. Google explanations are somewhat ambiguous.

Hopefully the OP will delve into issue & offer some more insight.

 

[Cogsy]

Time to throw my 0.02 in. The way I read the first post makes it clear that these measurements are being performed on a rotating bar. Firstly, Anatol says "A rod set in the jaws describes a cone" which in normal speech would indicate revolution, however the real proof is his two statements "Close to the jaws, the rod is 10-15 thou out, but 4” down the rod the error increases to 30 thou or so" and " the minimum error on the rod - the point of the cone - would be back inside the chuck - if you could measure it". From this, we can see Anatol has not set his zero on his indicator at any point along the bar and simply measured misalignment, but is checking actual runout of the revolving bar at different points. If he was measuring misalignment of the bar over 4" of travel then the first point he measures would be 'zero' and the 'point' of the cone would not be within the chuck. He could not possibly have a figure of "10-15 thou" at the jaws, he could only give a difference figure like '20 thou over 4 inches'. Even if he was measuring a taper on a piece he had turned, the point closest to the jaws would still be his 'zero' point.

As Dieselpilot has been saying all along, no amount of headstock misalignment or bed twist can induce runout as Anatol is describing. It seems evident that the problem lies in the spindle nose, chuck backing plate or chuck itself. Except that he's likely checked for it already, I'd suspect a burr or maybe a small chip in the backing plate/spindle nose register could be a likely cause.

 

[Anatol]

I assumed( in error) that the intent of this thread was solve the actual problem. I understand the lathe setup process, though I can't throw any of these old names out there (except Moore).

Tanks Dieselpilot - No error. I am very keen to get my lathe straight again (otherwise its useless). I'd like to get it very straight - straighter than in was. (I am also interested to learn the history of this art.)

To recap - I am afraid I induced this error by jacking up the headstock with a prybar, as noted in a previous post.
To clarify - the measurements of 'runout' and the description of the cone described by the bar were with the bar in rotation.

Cogsy said "As Dieselpilot has been saying all along, no amount of headstock misalignment or bed twist can induce runout as Anatol is describing. It seems evident that the problem lies in the spindle nose, chuck backing plate or chuck itself. Except that he's likely checked for it already, I'd suspect a burr or maybe a small chip in the backing plate/spindle nose register could be a likely cause."

The backing plate (camlock D1-4), as noted, runs quite true, within 2thou radically and axially. But two different 3 jaw chucks present very similar measurements, so a bit of grit in the chuck jaws or backing plate seems unlikely, especially since I had the jaws off one chuck and cleaned and reinstalled them. But I'll check again.

I'd like to pause this discussion until I have time to level and measure and calibrate the whole lathe.
Thanks to everyone who's lent their experience and intelligence to this thread.

 

[goldstar31]

I strongly suspect that your mention of a whole 2 thous on the backplate is way way--excessive.
Again, I seriously wonder about whether the spindle is bent, slack or worn or perhaps the bearings are 'shot'

Returning to your desire to 'learn more', I would print out the recommended tests for lathes in Georg Schlesinger's book which is available as a pdf on the 'net.

Looking at my library of measuring 'tackle', I have a properly ground test bar made from a worn out reamer with a good No2 Morse taper shank but I have also one of these bars of any old piece of ms rod which has two 'rings' machined in it- one close to the chuck end and the other as far out as bar itself which is cut without tailstock support and then both rings 'miked' to determine whether the they of the same size .

If they mike OK, all is well but if they don't you have a long series of tests to determine what is causing the dissimilarity in results.

Seemingly people have forgotten that a soft centre in the spindle is always recut before working between centres.

Ot. LOL people don't know that 'soft centres' are available- apart from in boxes of chocolates LOL.

Norm

 

[ShopShoe]

Anatol,

I've been following this discussion and I applaud your decision to stop and reconsider what all is going on and to establish some kind of a baseline for further changes.

It's too easy to "chase your tail" trying one thing, then another, until you get things so muddled that it seems nothing will work out.

One thing I don't see in your description is the history of your lathe and if is possible some of the problems may have come from a previous owner. Also, was there an "event" in the lathe's operation before you measured the originally-posted problem. I know you said there was not a crash, but maybe something else. (And I'm sorry if you did post those things and I missed them.)

Everyone else,

(A rant of my own, more like a gentle chiding, and meant to be taken lightly)

Everyone's input is valued, as that is what makes us all better at what we do. However (gulp), I find it tedius when the original poster in any thread gets relegated to third person while independent debates get started on the "proper" way to do someting that may or may not be an actual part of the issue under discussion. Thank you everyone for the reference works cited and for the experiences recounted, but let's watch Anatol work through this.

I had to say it, please don't kill me

--ShopShoe

 

[bazmak]

You don't need an accurate level to check the bed twist.Just do the 2 ring machining
and adjust the /remove the bed twist accordingly.You will need riser blocks.
When I did my 3 no ml7s and 1 no super 7 recently I only needed to adjust the risers twice
for each lathe.When the bed twist is removed the lathe will machine parallel so you know the twist is removed
why then check it with a level.

 

[Anatol]

Thanks all
Thanks Norm - Schlesinger downloaded
FYI - no nonsense link here-https://springfieldarsenal.files.wor...hine_tools.pdf
Thanks Shopshoe for sage words.
signing off on this thread till I have more tangible results.

 

[tjwal]

Your run out problem is related to the spindle nose and chuck and not spindle alignment. Others have posted what to look for in that area.

In any case you may want to check your alignment as well. I’d do that as a first step. I use “Rollie’s Dad’s Method” to check and adjust the spindle alignment. Simply put the method determines the axis of the cone that you described in your first posting.
You need a dial indicator, micrometer and a reasonably straight bar. Although a nice round straight bar helps it is not necessary. If it is ‘near’ precision ground then you don’t even need the micrometer.

 

[petertha]

I read the first post makes it clear that these measurements are being performed on a rotating bar. Firstly, Anatol says "A rod set in the jaws describes a cone" which in normal speech would indicate revolution, however the real proof is his two statements "Close to the jaws, the rod is 10-15 thou out, but 4” down the rod the error increases to 30 thou or so" and " the minimum error on the rod - the point of the cone - would be back inside the chuck - if you could measure it". From this, we can see Anatol has not set his zero on his indicator at any point along the bar and simply measured misalignment, but is checking actual runout of the revolving bar at different points. If he was measuring misalignment of the bar over 4" of travel then the first point he measures would be 'zero' and the 'point' of the cone would not be within the chuck. He could not possibly have a figure of "10-15 thou" at the jaws, he could only give a difference figure like '20 thou over 4 inches'. Even if he was measuring a taper on a piece he had turned, the point closest to the jaws would still be his 'zero' point.

I think the scenario you are describing is what I've suggested might be one possible explanation. Here is a corresponding sketch to help visualize that specific scenario. And even this is simplified. The test bar could be gripping out of alignment in a 2D plane, or just as easily in 3D space.

petertha> I can visualize a cone if 1. the test bar was straight but gripped in the chuck at an angle offset to spindle axis. This could be the jaws, the chuck, the mounting back, the spindle nose.... He said he replaced the chuck with another & got similar results. So maybe the mounting back or spindle nose is suspect if they are common to both chucks. I suggested removing all the rotating hardware to eliminate them & insert an MT test bar into the spindle socket. He doesn't have one, so the jury is out​

Now if the OP confirms this is indeed how he conducted the test & measured results, then we are getting somewhere. Maybe not a solution but going down the right path. If he says, No that's not what I meant, then he needs to elaborate or clarify the test conditions so that we get a better picture. Or through Q&A perform a different confirming test to further narrow down the problem. If this particular scenario turns out to be the sole culprit, then yes, it has bugger all to do with lathe twist or spindle axis misalignment. Those 2 conditions could be perfect & still exhibit the same measurement deviation right? But they are easy checks to perform & potentially eliminates other problems because of possible interactions.

 

[petertha]

The backing plate (camlock D1-4), as noted, runs quite true, within 2thou radically and axially. But two different 3 jaw chucks present very similar measurements, so a bit of grit in the chuck jaws or backing plate seems unlikely, especially since I had the jaws off one chuck and cleaned and reinstalled them.

If your D1-4 back plate was out .002" radially and you happened to have a 100% perfectly radially concentric chuck + jaws + test bar, then you would see the same 0.002" radial runout on the test bar and it would be a constant 0.002" down the length. In reality all these mating surfaces have tolerances which could work collectively for you or against you depending on the arrangement orientation. If the back plate was out 0.002" and the chuck rear recess (or bolt up) was out 0.002", they could add up to max 0.004" runout on test bar if phased collectively, or min zero runout if phased in opposition, or anything in between depending on the orientation.

Now if you have angular discrepency between the back plate & chuck, this would cause more like what you are observing which is an increasing runout along the test bar. Similarly If the backplate + chuck body are perfect but the jaws are gripping the test bar at some angle, that would be a similar net effect.

What is confusing is you seem to be getting similar results with both chucks if I understand, so its pointing to problem with the backplate face. If you had (that same MT socket test bar we keep talking about! LOL) you could insert the test bar in the spindle, grip the test bar with the chuck jaws in reverse (the chuck is facing the spindle) and measure runout/deviation of the rear face of the chuck which at least confines the error to chuck body and/or jaws.

 

[Cogsy]

I think the scenario you are describing is what I've suggested might be one possible explanation. Here is a corresponding sketch to help visualize that specific scenario. And even this is simplified. The test bar could be gripping out of alignment in a 2D plane, or just as easily in 3D space.

​

​

This is exactly what I meant. Anatol has now confirmed his run out measurements are true measurements of run out (ie deviations in the shaft position through rotation). In your sketch the dotted red centre line is the axis of rotation and is (theoretically at least) the centre line of the spindle. This axis of rotation is not at all influenced by its alignment with the bed itself, nor bed twist, wear, etc., in terms of run out. The dial indicator making the measurements could be suspended from any structure rigid enough to support it and would yield the same information - namely the deviation of the shaft from the centre of rotation due to the shaft being misaligned with the axis of rotation. Misalignment with the bed/bed twist will result in deviances in turned shaft diameters but will not result in run out. We take this to the extreme when we use things like taper turning attachments to purposely turn non-parallel to the axis of rotation but the resulting shaft remains centered around the axis of rotation.

If this exact issue were being described in a mill rather than a lathe, with a cutter exhibiting more run out the further it was measured from the collet/holder, we would not expect tramming the mill to have any effect on the issue. Having a perfectly level and straight lathe bed is an admirable desire but it cannot help with this current problem.

 

[goldstar31]

Assuming everything eose on the lathe is correctly aligned and unworn, off setting the spindle is little difference to using a taper turning attachment.

Using a piece of bar 10" long instead of a sine bar and offsetting it by a quarter of an inch creates what Morse intended for his Morse tapers. Sadly, his operatives couldn't measure accurately but that is where it all started.

Another way of creating a truncated cone which Anatol is experiencing .
To think that years ago I was promoting how to cut a taper and now how to correct one.LOL

Norman

 

[DJP]

I'm still thinking that the problem is worn 3 jaw chucks. I have placed a shim under one jaw to correct runout and have bored the jaws to remove runout for that diameter of work piece. Multiple worn chucks doesn't help. My frustration was only solved with a new chuck or use of a 4 jaw. I would be interested if this runout problem being discussed still exists with a 4 jaw chuck. That might convince me that the problem is deeper in the machine bed or spindle.

 

[petertha]

Yes, worn jaws. Or jaws maybe inserted in the wrong numbered slots, although one would think that should be dramatically obvious. I heard a similar story where a guy bought a used chuck & having runout issues no matter which combo of jaw positions he tried, only to discover owner #1 replaced 'some' jaws with lookalike chuck. But the difference was he always had some runout, but it was constant down the test bar so the jaws were parallel but displaced from center differently between them. The OP's issue is increasing runout down the test bar suggesting the jaws are imparting an angle to the test bar when tightened. Some folks here have already touched on jaws but we really haven't heard how they were tested. The OP said same results on a second 3J chuck he tried, so what are the odds?

I was trying to think of a way to get a quick visual on jaw grip. Like maybe blue the test bar, clamp the jaws, remove & examine. If you don't see 3 consistent bands of grip mark impression or some kind of suspicious pattern like the toe of one jaw & heels of the other two, maybe that is telling you something? I've never done it so casting out as idea.

If the 4J were made to run true (ish) near the chuck jaws but had the same increasing runout on the outboard section test bar, then it would again point to something upstream like a back plate issue. Probably the simplest thing right now is just borrow a known good D1-4 chuck + backplate assembly & eliminate 2 of his middlemen altogether. If it has 0.002" constant runout on the borrow lathe, then it should probably be close to that on the OP's lathe if the spindle nose & D14 tightening mechanism is in good shape.