Lathe Accuracy Problems

[ChrisHParticipant @chrish]

 

Am I well wee'd off with this? Logged in, wrote a long thread, went to post it and guess what, a clever note saying I had to be logged in to use this facility. I WAS logged in or else I couldn't have written the thread. Why hasn't this been sorted by now? Moderators/web chiefs, please note. It is seriously annoying as well as a complete waste of the threadwriters time.

Rant over, now:

Anyway, what I was asking was HELP! Could someone please point me in the right direction and advise where I am going wrong.

I am trying to set up my lathe to turn parallel as well as then making one of Harold Hall's cylinder squares, but am getting nowhere fast.

My lathe is a German made Weiler lathe, about 35-40 years old. OK, it is a bit worn but quite OK within the area in which I work, very manageable. The lathe sits on a very robust welded mild steel bench made by Weiler, and is held down by four holding down bolts, one at each corner, of the headstock, and by another four at each corner of the tailstock. When I first set the lathe on the bench I set out to minimise the twist as far as possible with 6thou shims under the holding down bolts, as that was all I had at the time, being guided by a reasonably accurate level.

Today I set out to make it turn parallel and make a cylinder square in the process. I set a length of 30mm diameter bar with about 30mm in the chuck and a tad over 90mm as turned length. As originally set up, I got an out of parallel of 0.05mm, being larger at the outer end. In this case the manual says to harden down the headstock front right bolt, followed by the tailstock rear left bolt if necessary. As all bolts were already hardened down I thought again. Using another 6mm shim which was already cut to fit, I inserted it under the rear right headstock bolts and tried again with another cut. This time I got out of parallel of 0.04mm, but this time the other way round, the chuck end was bigger.

Simples I thought, substiture the 6thou shim for a 3thou shim and she'd be right. Wrong! Now it was out 0.02mm.

The lathe speed is about 300rpm, a very slow feed is being used, the tool is HSS ground on an off-hand grinder and honed with a diamond hone, it's 1/4in square and protudes from the tool post about 15mm, the tool is like a knife tool with a bit of approach rake (forget the correct term!).

I played with bit of different shim all afternoon getting nowhere, then I noticed an oddity which baffled me. My practice was to take a 2thou cut down to the chuck, and without touching anything bar reverse the feed, cut back out again. I noticed that on the cut towards the tailstock the tool was taking a progressively heavier cut, heavier being a relative term given the original depth of cut.

So I took another cut towards the chuck and stopped the lathe when the tool reached the chuck, and micrometered the bar. It was 0.09mm larger at the tailstock end. So I just reversed the feed and cut back out again, and remeasured after that cut. Still out, but now down to 0.035mm.

All this has, as I say, me baffled, the little grey cells are getting ever older! My thought process have seem to have been driven down a dead-end and have just given up tonight.

So can anyone, please please, tell me what I am doing wrong, why/how is the lathe cutting like that, and what I need to do to get back on track – I need pointing in the right direction!

Chris

PS, this damned website only wanted me logged in AGAIN, even though I was. Had copied what I had wrote this time but….

Edited By ChrisH on 09/04/2013 21:50:07

[ChrisHParticipant @chrish]

[Sub MandrelParticipant @submandrel]

Sounds like the saddle is twisting slightly, direction depending on which way the feed is driving it. Sort that before trying to eliminate any twist.

Neil

[ChrisHParticipant @chrish]

Neil, any ideas on how to do that?

Chris

[Ian PParticipant @ianp]

I sympathise with your editing problems. Usual advice I have been given iis to create text in another programme (Word or similar) then copy it to this site. Its 100% going to work but its just a pain and take any spontaneity out of the equation.

My advice regarding the lathe alignment might raise a howl of protest from people that own and use precision levels but hey ho this is my 2 pen'orth.

Undo the two bolts at the tailstock end and slightly loosen the two at the headstock. This will remove any existing tension being fed into the bed via the stand.

Then, repeat your turning test and start over again.

Ian P

[Ian PParticipant @ianp]

After reading Neils post I think he may have a point about the saddle moving, BUT, there are other factor that might give the same effect and confuse the issue.

The shape and sharpness of the tip are not likely to be exactly symmetrical 'in both directions' so the depth of cut might be different depending on which way its travelling. You need to cut in one direction only for the purpose of these tests, That will at least eliminate one factor.

Ian P

[Anonymous]

A few random thoughts:

1. 15mm seems like quite an overhang for a 1/4" square toolbit

2. The cut will act differently in each direction, as the cutting face will have different angles and length presented to the work in each direction

3. Unless your chuck is perfect, with jaws that are all exactly aligned with the spindle axis, it is possible that the work is pivoting in the chuck very slightly under the cutting forces

4. Try eliminating some variables by locking the topslide and cross slide during a cut

5. Try a bigger depth of cut; 2 thou is fairly small and depending on the rigidity of the set up and the bar material the tool might not always actually be cutting

Approach the problem logically. Make a cut, take measurements, think about it.Then make a change and see what happens. If it doesn't work out go back to the original setup and recheck. Write it down. Don't get into the vicious circle of cut 'n' change at random. If a change doesn't have the expected result, try and think why.

Regards,

Andrew

[NobbyParticipant @nobby]

Hi Chriss
It may be posible to turn you "bottle" square between centres as you may have more control, what steel is it can you get it hardend & ground ?

Nobby

[Ady1Participant @ady1]

Are you using a centre to support the tailstock end?

[HopperParticipant @hopper]

+1 on 15mm being way too much overhang for a 1/4" tool bit. You should aim at more like 6mm or a shade more.

Once you have got the bed twist sorted out, or eliminated, by Ian's suggested method and careful shimming etc, if there is any error, you can correct it by offsetting the tail stock a thou or two.

Carriage twist can be minimised by nipping the gib adjusting screws a bit.

It is best to do parallel turning tests between centres, after checking with a dial indiactor that the centres run true. Holding one end in the chuck introduces the possibility of error from run-out and poor alignment in the chuck itself.

Harrold Hall's website and lathe book have a good run through on the procedure, without requiring an expensive level.

[David JuppParticipant @davidjupp51506]

Have you checked for play in the spindle bearings (when at normal operating temperature)?

[ChrisHParticipant @chrish]

Wow, what an amazing response, thanks a lot! I will try and answer the points raised above:

I will check/nip up all the gibstrips to ensure minimum play can exist on the carriage etc. as a first step. Will also lock the topslide and cross slide when making a cut.

I agree that one cut one way should suffice, was concerned when I got the results indicated in the original post. Hopefully when it is set up right the lathe will not do that.

I was thinking that the tool overhang was too much, now confirmed. I intend to grind a new tool from a new piece of 1/2in HSS tool steel which should deal with any tool fexibility problems.

I did write down what I was doing and the results obtained, I have a note pad in the workshop which I use to write down stuff as I go along as I find it all too easy to forget what I have done – old age races on!

The spindle bearings are two back to back taper roller bearings and all is solid there, play wise.

I am not using a tailstock centre to support the work, just following Harold Hall's instruction in his book on the lathe. That is why I was also taking fine cuts, but I appreciate the point of perhaps the tool not cutting at times, will check on that. The steel is EN1A (or 230M07, whichever you prefer)

I think I will take out the shims that I inserted yesterday, slacken all holding down bolts off then just tighten them hand tight and start again, and report back.

Sorry about the mix of units by the way, I am of an age when we used feet, inches, thous, pounds, ounces, lb/sq.in, deg.F. I am now also used to working in meters, mm, kilogrammes, grams, and deg.C, (but cannot get used to metric pressure units, I use them if I have to but they just don't mean the same to me as psi!), my lathe although German has dials graduated in thous, tool steel is imperial but my mic's are metric as is my rule, such is modern life!

Chris

Edited By ChrisH on 10/04/2013 11:18:24

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

Hi Chris H ,

Start in a different place .

Get a piece of true bar of reasonable diameter and mount in your four jaw chuck . Take as long as nescessary to set it to run dead true as tested with dial gauge in three different place along the length .

If you can't find a true bar set up next best bar as truly as possible and individually machine three bands – each end and halfway – to exact same diameter and turn away material in between by a small amount .

Now traverse your dial gauge using saddle along fullength of bar or over the three bands . Any anomalies will be obvious and won't depend on uncertain action of cutting tools .

Don't use tailstock support for this .

Don't follow the often given advice of correcting inaccurate turning by twisting the bed with shims unless you are absolutely certain that twisted bed is the problem . Even if it is the cause shimming should be minimal .

If you think about it any inaccuracies in diameter of piece you are turning over a short length near headstock would need an enormous twist at tailstock end to correct .

Regards ,

Michael Williams .

[John McNamaraParticipant @johnmcnamara74883]

Hi Chris H

Turning a perfect parallel over a longish piece, a cylindrical square, over 90mm or more. to a couple of tenths accuracy? In my experience not that easy.

On reversing the feed…. Assuming your lathe is fed from the front of the carriage via the apron, there is almost certainly a small amount of "crabbing" as the feed force is applied to the opposite ends of the apron.

The pushing force of the feed rack or screw reverses and the saddle will press against the side of the ways in the opposite direction. Remember the force is being applied to the front edge and is off centre.

The thin film of oil on the ways is enough to allow a little movement. and not necessarily even movement even if the lathe is in perfect condition. On older lathes like my 30 year old German VDF Ro80 I guess it is a given.

Cutting in one direction should help, and maybe you will get a fair result, possibly using the tail stock and turning between centres. I would use a fresh turned in place soft centre (For grip to cut and accuracy) in the 3 jaw and a live ball bearing centre in the tail stock. set over to turn the same diameter at each end of a 90mm piece. after testing If you have pre recessed and centre drilled the ends of the square you can finish turn the ends and the side in one setting That way the ends will be parallel to each other and very close to 90 degrees to the sides. Light cuts only and the soft driving centre should not slip,

However you may still get small concavity or convexity in the work if the bed or other conditions are not perfect.

In the end…. maybe marking an adjustable cast iron or other softer metal ring lap and lapping the piece carefully to make it parallel is the answer. If you do it in the lathe make sure you are not wearing clothing that may catch while the work is rotating and that the lap has smooth rounded corners. You do not want to be injured. plan to do it carefully and without distraction.

If you have access to hardening facilities you could harden the piece before lapping. Ideally having heated and allowing it to cool very slowly (To avoid hardening) before turning to normalise and remove stresses that may distort it on hardening later after turning.

Cheers

John

Edited By John McNamara on 10/04/2013 14:56:33

[ChrisHParticipant @chrish]

 

Well, having had a fair bit of time in the workshop today, on the lathe, I am not that far further forward.

 

I ground a new knife tool in 1/2in sq. HSS and diamond honed it. I tightened all the gib strips, for the carriage, the cross slide and for the top slide. I probably tightened them a tad too much, as the levers were hard to turn after that. I then put in a 4thou cut and got 0.065mm difference between outer and inner ends of the test bar.

 

So I slackened all 8 holding down bolts, (4 round the headstock, 4 round the tailstock), took out the shims I had put in yesterday, tightened up the bolts hard hand tight, and took another 4thou cut. Out of true down to 0.04mm, or about 1.6thou, over the 90 mm.

 

And that was basically as good as it got. Tightened down with a hammer as it says in the manual, first at the headstock, took a cut, then again at the tailstock, took a cut, unhammered the hammered bolts and added the a 3thou shim, tightened up hard hand tight again, took a cut, each time the result was the same, out of true 0.04mm.

 

Only bright point was at the end I put a dial gauge on that reads to 0.0001in, and ran it down the length of cut and it confirmed that my mic readings were good!

 

Now ‘wasted’ two days and still turning 1.6thou out over 90mm. I know it is not acceptable but am tempted to leave it at that!  My next move will be, I think, to really go back to square one, take all shims out from under any bolts that they may be lurking under and start the ‘levelling’ process from scratch again, and maybe in the process thus rectify the twist.  But that won’t happen now until next week, as because we have been married for a significant amount of time, Senior Management has decreed that we go and spend a significant amount of money in a hotel for the weekend to celebrate. I did venture the opinion that it was seriously curtailing vital workshop setting up work, but for some reason that opinion was not well received…………… At least the place does a decent pint of cask, I understand, so perhaps all will be well in the world after all, and I can have a think away from the job!

 

In the meantime, I am greatly indebted for all the comments added, it has and is giving me serious food for thought – many thanks.

 

Chris

Edited By ChrisH on 10/04/2013 18:39:45

[ChrisHParticipant @chrish]

A PS to my post just now, today have been only taking cuts one way, towards the chuck of course, and when I wind the carriage out, by hand with the lathe stopped, the tool is not leaving a mark on the turned bar, so that must be a plus point too!

Chris

Edited By ChrisH on 10/04/2013 18:50:37

[Sub MandrelParticipant @submandrel]

when I wind the carriage out, by hand with the lathe stopped, the tool is not leaving a mark on the turned bar, so that must be a plus point too!

That suggests to me the saddle IS twisting a little.

Others will probably disagree?

Neil

[ChrisHParticipant @chrish]

While having my evening soak in the bath, pondering the day’s play while supping a pint (how the retired live!) I got to pick up on a couple of things that John McNamara had said which made me wonder a few things.

Am I expecting to get too fine a degree of accuracy from a well used 35+ year old lathe? With the inevitable wear that such a machine has, am I actually going to achieve much better?

Is 1.6thou over 90 mm really unacceptable? That’s only ½thou over 1 1/4 inches, how often am I going to need greater accuracy in that in real life?

A lean over 1.6thou from, or to, the vertical over 90mm, what angle does that make, could you measure it (yes I know you could calculate it, but on the milling table)? Would you notice it?.

Am I wasting time and effort in trying to achieve a result which would be wonderful when what I have already got would be acceptable in 99% of the time, or more?

How accurate are other peoples lathes? (now I know, having asked that, someone will say they always get to turn ½ a thou over 10 inches or something!!). I would love to know what other people are working to.

What are the thoughts on the above?

Chris

[VersabossParticipant @versaboss]

Posted by ChrisH on 10/04/2013 18:36:39:

Only bright point was at the end I put a dial gauge on that reads to 0.0001in, and ran it down the length of cut and it confirmed that my mic readings were good!

Chris, that's something I don't really understand. When your turned bar is not parallel, and then you mount a dti (in the place of the tool I suppose), then the point of the dti should move along the same line as the tool's point before. So you cannot see a movement of the pointer, imho.

Now some questions. Which model of Weiler lathe is it? Weiler lathes are from the highest quality, and it is only 30 years old you say?

I own a Leinen of at least over 40 years old. Just a couple of days ago I turned a similar bar as you did, but the turned part (90 mm) was about 50-60 mm away from the chuck. As an experiment I used no tailstock support. The dia. difference was about 0.02 mm maximal (rather less iirc). This lathe cannot be 'corrected' with shims, as it is screwed down to the heavy stand since leaving the factory. Without having a look (rather difficult to do) I think there are only 2 screws, centrally arranged.

I know that this doesn't help you much, but I believe that you should get a better result from a Weiler.

Greetings, Hansrudolf

[Anonymous]

My lathe is over 30 years old, ex-industrial (Harrison M300). Turning a long casting (30 inches) with tailstock support I can't measure the difference over a turned length of about 3.5", may be a 10th of a thou? Near the headstock, with an unsupported bar over the same length I get about half a thou variation, give or take. I assume that's because the bed is rather more worn at the headstock than the tailstock end.

Regards,

Andrew

[ChrisHParticipant @chrish]

Neil, “ That suggests to me the saddle IS twisting a little” I would tend to agree with you, EXCEPT, that I have the gib strips so tight I can’t see how it could twist!

Hansrudolf, the lathe is Weiler LZ280. I think the age I gave must be wrong as I think it is a 1960’s lathe, certainly pre1970, which would make it 45 years old, or perhaps even more, my apologies.

The dial gauge I mounted on a magnetic base which was fixed to the carriage. I zero’d the gauge at the outer end of the cut length, then advanced the carriage towards the chuck. The reading went down, or up depending which way you look at it, by 1.6thou of an in.

The Weiler is a very good quality lathe, very well built, but it has also been well used in the past. I got it from a boatyard firm that went bust about 8-10 years ago. I make the “well used” observation from the state it was in when I got it!

I think I could possibly get a better accuracy, I would certainly like to think so, but the questions are could I, given its general wear? and how? and at what cost in time and effort? and at the end of the day is it worth it? What do others get? Yours is commendable certainly!

Chris

[jason udallParticipant @jasonudall57142]

Check me on this..

1 no tailstock

2 dti mounted on"saddle"

3 same motion used for dti travel as tool ( not compond for cut and not for dti )

4 dti shows same"change" as mike ( allowing for diameter vs radius)

5 dti probe same plane as tool tip motion

thus..4 rules out deflection from cutting forces either in saddle or tooltoolpost or chuck…

(very imressed here, to get same reading…assumption 5 above is hard to achive )

as to causes….well I think that spindle axis if off from saddle ..up/down or front/back.buy the measured amount…..so of the order off 1.6 thou in 90 mm or 1:2214 or 0.4 mRadians or 0.0259 degrees

Just goes to show what sort of accuracy we are asking of our machines..

[ChrisHParticipant @chrish]

Andrew – I am seriously impressed by the accuracy you are obtaining on you ‘old’ lathe, and green with envy!

Jason – forget what I wrote about the dti check, in the middle of the night it suddenly made no sense to me. What I had measured with the dti was the out of true along the top of the bar, or the radius, not the diameter as the mic showed. Sometimes your brain makes things fit what you want to believe as opposed to what is actually being shown, that obviously happened to me yesterday. Today I went and remeasured with vernier calipers and got a conformation of the readings the mic had shown; I can’t explain the dti reading at the moment but I am sure someone more clever than I will be able to, but perhaps it is as you suggest, the spindle axis is out of true of the saddle axis, either up/down or side-to-side. Perhaps I will check with the dti along the horizontal plane, see what that shows?

If the spindle axis is out of true with the saddle axis, what on earth do you do about that?

Chris

[brickyParticipant @bricky]

Hi Chris

I made a taper turning attachment for my lathe and instead of a dovetail follower I wanted an accurate bar of 16" in length.The bar stock that I had was not accurate enough for this use so I had to turn one form 1" bar between centres with a traveling steady.After much adjustment of the bed and the tailstock I arrived at .84700" to .84645" This took hours to achieve, so I can only suggest perseverance .All permutations have been suggested.Good luck.

Francis

[Tony Pratt 1Participant @tonypratt1]

Posted by ChrisH on 11/04/2013 10:49:31:

If the spindle axis is out of true with the saddle axis, what on earth do you do about that?

Chris

Hi Chris, don't despair; I would seriously doubt that the spindle axis is out to the bed. I would do the following:

Clean & lubricate all slides

Adjust saddle gibs to the best of your ability

Set up a newly sharpened knife tool

Take a small cut [.002”] towards headstock, stop lathe, wind tool back to end of bar, take a “spring cut” i.e. don’t put another cut on using the cross slide

Measure both ends of bar

Shim as manual instructs. try .001” at a time

Re cut and measure

Logically this should achieve the desired result [fingers crossed]

Tony

[Author]

Posts