Thread: Lathe alignment bar

Hi Mike, thanks for your input.

In my case i would disagree on the S/H worn out lathe being a better option to a new Chinese. My Chinese lathe is far more capable of accurate work to a worn out well built lathe of yesteryear. Sure, I can attempt to buy an old one and re-furbish it. I may even succeed but at the time I aquired my Chinese lathe, I had no lathe work experience. How would someone who is just starting out and wanting to learn turning be in any position to (1) know what they are looking at and (2) identify and know how to fix the issues?

Simon

Common saying by Jim Patterson, the owner of J.P. Engineering when I was working there over 30 years ago. "Anyone can do accurate work with a new machine, but it takes a lot of skill to do accurate work with an old worn machine."

Dean

Originally Posted by cba_melbourne

Pete, I did it without a test bar. First I tried the RDM method but gave up on it. Then the good old spool method. It took me several days, the best part of a week. That is why I can say with confidence that if the lathe requires torque shimming, the only practical way to go about is to buy a real test bar.

And no, I have no intention to discredit you or anyone else for that matter. You do that to yourself. I shall put my money to my mouth. Here is the deal: I buy a stock AL51G from machineryhouse. You use your test bar and spools and whatever to confirm its specs. If its less than the Schlesinger specs, we use those the machine shows. Then we take the headsock off, and you have 8 hours time at my place (I am just out of Melbourne, and you get coffee, the shop is heated) to realign it, using no test bar but just a level and your spools with an indicator. No scraping. It needs to meet Schlesinger specs or whatever specs we found before taking the headstock off. We check headstock parallel to the bed in both planes, and at tailstock height and the TS side offset marks roughly lining up. If you succeed, the lathe is yours to take. If you fail, the lathe is still yours but you owe me $1k (or whatever the lathe did cost me). Are you in?

Nice attempt at a dodge, but it has absolutely nothing to do with what you said. In fact this is what you said, and what I, and others have disputed.

That is about the one and only lathe adjustment that really, absolutely, requires the use of a test bar as in practice it cannot be done without.

That is crap. As was stated at the time, a test bar will make it convenient (ie faster) but is NOT required, as you claimed. So yes Chris I will take you up on your offer as a matter of fact. However there is NO time limit, because one was never specified in the first place. You have stated, categorically, that a test bar is "really, absolutely, required" to align a lathe's headstock. That's rubbish, so you can try and twist this around all you like and dig a deeper hole for yourself if you like. So here's the gig big fella, I will check with Phil to make sure it's ok by him, but yes if you insist on continuing to dig a hole for yourself, you will buy whatever lathe Simon has and have it freighted to Phil's place. There we'll check it as you suggest and take the headstock off. I will then reassemble the lathe and confirm it is within original spec without using a test bar. Once I've done that you will pay for the freight for that lathe to be sent to a charity in the area. If I cannot align that lathe, without using a test bar, as you claim is required, I will pay all the above and send the lathe to a charity. Nothing in it for me and I'm out of pocket significantly in getting down there etc, but one way or another a charity down there gets your lathe.

Edit: Oh, and the loser buys Phil a bottle of Scotch for his trouble, though I'm sure he would enjoy the entertainment regardless.

OK I have a question for the machine tool inspection gurus;

WRT lathe alignment bars, they all quote a runout figure. A typical figure is 0.01mm, I assume this would be at the extreme end of the bar. With a 3mt bar, that would be 0.01mm per roughly 200mm. Now if you have a really nice lathe and want it to turn nicely parallel at say 0.001mm per 100mm then this test bar has runout that is already beyond what you want to achieve.

Is there a limit to what level of parallelism you can achieve on your lathe alignment with such a bar? Do I assume you can only achieve 0.01mm per 200mm using that bar.

Also, I understand that the figure quoted is the worse case senario. Lets assume we got unlucky and got a bar made on friday arvo (POETS day) and it does indeed have the quoted runout.

Cheers,

Simon

Any further posts that contain non-family friendly language will be deleted, not edited.

When that happens this thread will be closed (as it probably should be anyway).

Originally Posted by simonl

OK I have a question for the machine tool inspection gurus;

WRT lathe alignment bars, they all quote a runout figure. A typical figure is 0.01mm, I assume this would be at the extreme end of the bar. With a 3mt bar, that would be 0.01mm per roughly 200mm. Now if you have a really nice lathe and want it to turn nicely parallel at say 0.001mm per 100mm then this test bar has runout that is already beyond what you want to achieve.

Is there a limit to what level of parallelism you can achieve on your lathe alignment with such a bar? Do I assume you can only achieve 0.01mm per 200mm using that bar.

Also, I understand that the figure quoted is the worse case senario. Lets assume we got unlucky and got a bar made on friday arvo (POETS day) and it does indeed have the quoted runout.

Cheers,

Simon

Yes the Chinese test bars on eBay are all specified 0.01mm. That is sufficient to achieve Schlesinger standards which I think (I did not lkook it up) is something around 0.02mm/200mm. Mind you, 0.01mm over 200mm would already be a very very well aligned machine. If your garage slab moves like mine does with the seasons, alignment won't stay that accurate for more than some months unless its a heavy industrial lathe with a very stiff bed.

With a light bench lathe, when you need the best possible parallelity, you measure your workpiece at both ends as you approach the final dimension. Then correct by slightly changing the front right levelling adjuster between the lathe bed and the bench or cabinet. If your lathe has no levelling adjusters, you can slightly loosen or tighten the front right mounting bolt. It takes very little to twist/untwist the lathe bed such as to cause a change in parallelity of your workpiece by .01mm.

If you wanted to turn somethig that is as you say parallel to 0.001mm/100mm, I believe that is not something you can achieve with any repeatability on a lathe. Just to measure 0.001mm with a micrometer requires a constant temperature room, wearing gloves to keep the heat from your fingers away from the micrometer, not breathing onto the micrometer, let the workpiece temperature settle after a cut...... The spindle bearings in your lathe have already more runout if they are only automotive grade, and the surface imperfections of your bedways grinding will already exceed 0.001mm too (some Chinese lathe beds are ground with rather coarse grain wheels).

Thanks Chris.

The reason I ask is because I remember reading one of Harold Halls books relating to lathe work and one of the projects is a cylindrical parallel. He describes using only a 3 jaw and no support, using light cuts with a finishing tool to produce a parallel with a taper of less than 0.1 thou per 6" (from memory) and suggests that this should be achievable with a standard lathe. That is indeed a very well aligned lathe he has. I would have to re-read the article, he may have suggested tweaking the bed level to achieve that but I can't remember. But you suggest that even if you did get it that good, it wouldn't stay like that for long?

Sometimes I think I chase rainbows!

Simon

To avoid Schlesinger's name being taken in vain (and in this case being mis-quoted), the limits are -

Finish turning lathes up to 400mm centre height-
Next to the spindle, permissible run out is 0.01mm
300mm from spindle, permissible run out is 0.03mm

Toolroom lathes up to 200mm centre height-
Next to spindle, permissible run out is 0.008mm
300mm from spindle, permissible run out is 0.015mm

Hobby lathes and assorted soggy noodles-
Not mentioned

Taken from Testing Machine Tools (7E), G Schlesinger, 1970

I agree with Fred that this thread is getting out of hand; some of the statements being made are ridiculous and contradict years of good practice. Invoking Schlesinger to 'prove' that a poorly bedded in machine is top notch is spurious and a distortion of the intent of the standards, which was to introduce common acceptable standards for machine tools.
Connelly (Machine tool reconditioning, 1955) states (when speaking of headstock alignment to the bed) specifically that 'slides to be fitted to the bed ways with a surface quality of 10 - 15 bearing spots per square inch'. I interpret that as basically meaning that even attempting to align a headstock without decent contact would be a bodge job. Neither of them mention the concept of 'torque shimming' and looking briefly on the internet I could not find a mention of it. From that I conclude that although some manufacturers may use it as a cheap and nasty way of aligning headstocks it is not a generally accepted method and because of that should not be spoken of as a general panacea for lathe construction defects.
As I and others have stated before, this forum is not just visited by members who know what they are looking at but by others new to machine tools as well as people (students?) researching various topics. To persist with propagating poor practice does everyone a disservice, including damage to the reputation of the person promoting the technique.

Michael

Originally Posted by bob ward

Sadly yes. Although it turns out the headstock is not solid shimmed but “torque shimmed”.


I’ve not given this much thought before, but I guess from the lathe maker’s point of view there are 3 methods of mounting and aligning the HS.


One is machining the underside of the headstock and the corresponding portion of the lathe bed so the mounting points are coplanar. Probably the most expensive of the 3 options as it requires a degree of quality control, stress relieved castings etc.


Two is to machine everything close enough and then solid shim. It would be fairly time consuming to mount and demount the HS to get the shims right. Solid shimming to me would be acceptable if not ideal.


Three is to machine everything close enough then spend a minute or two (its a production line, you would soon become very quick at it) tweaking the hold down bolts (AKA torque shimming) sufficiently to keep Mr Schlesinger happy. Cheap and very nasty.


Method 3 is the one that wins of course. Why that should be is the subject of another discussion completely, but suffice to say the importers bear most of the responsibility.

Thanks for that Bob. That torque shimming terminology is a bit wide of the mark in my very humble opinion - had visions of thin shims being inserted at various points in the interface to tweak alignment and then some sort of torque spec applied, which had me thinking, what the, kill me now!

Even if Chinese QA could be trusted, I'd wonder how good it would be once it arrived.

Cheers,
Bill.

Originally Posted by Michael G

To avoid Schlesinger's name being taken in vain (and in this case being mis-quoted), the limits are -

Finish turning lathes up to 400mm centre height-
Next to the spindle, permissible run out is 0.01mm
300mm from spindle, permissible run out is 0.03mm

Toolroom lathes up to 200mm centre height-
Next to spindle, permissible run out is 0.008mm
300mm from spindle, permissible run out is 0.015mm

Hobby lathes and assorted soggy noodles-
Not mentioned

Taken from Testing Machine Tools (7E), G Schlesinger, 1970

Those specs are for the accuracy of the taper in the spindle nose itself..

The test you want is "Work spindle parallel with bed in horizontal plane (free end of arbor inclined towards direction of tool pressure)" The spec generally for the sized machines we deal with being 0 to 0.02mm per 300mm.

Originally Posted by bob ward

.........................

Three is to machine everything close enough then spend a minute or two (its a production line, you would soon become very quick at it) tweaking the hold down bolts (AKA torque shimming) sufficiently to keep Mr Schlesinger happy. Cheap and very nasty.

............

Thinking about this a little further, the class of lathe makers who *“torque shim” their headstocks into alignment wouldn't just be machining everything close enough as I say above, they would be deliberately machining the lathe bed and underside of headstock to ensure that the HS rocks.

It dawns that its not just my lathe or simonl’s lathe that have rocking headstocks, it must be ALL of that class of lathe, if you don’t have a rocking headstock how can you bodge it into alignment?

Maybe if they stuff up that process and end up with 2 flat mating surfaces they have to slip in a piece of shim.

It would be interesting to know if the high end import lathes, ie the toolroom and gun lathes also have torque shimmed headstocks. You would hope not, but who knows?

*”torque shimming” sounds quite high tech, its a cool sounding euphemism for something that’s not good at all. Its right up there with “friendly fire”, “non-positive growth”, “may be able to assist us with our enquiries”.

I successfully realigned my lathe using RDM. Just used a length of BMS 38mm or so. It had a ground finish with uniform diameter and was round enough for RDM purposes. It was mounted in the 3 jaw with about 800 mm stickout. RDM is a bit tedious because you need to find the high and low each time then average it out with a calculator, and write the results down so you can see how you're going, but it's just a matter of being a bit organised. RDM does not require the bar to be straight, parallel, round or anything like a precision piece. The averaging method takes care of all that. In my opinion it's more accurate than using a test bar, given that a test bar will have some runout and there will also be some inaccuracy in the spindle taper.

If I was using a test bar I'd use it to get "good enough", then use it with RDM to do the final adjustment. The better the bar, the easier it is to use RDM, you'll still need to do the averaging for runout, but you shouldn't need to compensate for diameter difference between your two chosen test points.

If I had to do it again, I'd have a laptop with a spreadsheet open, that'd speed things up enormously.

Where the test bar fails is with vertical alignment, over a couple of feet with a solid bar there's considerable sag, a surprising amount in fact. You can calculate that and allow for it fairly easily, but if you're doing that, then it's not much different to RDM time wise, except that RDM requires a high/low measurement each time. With a test bar you've still got the runout problem, which you'd use RDM to compensate for, so then the method becomes virtually the same (except that with a test bar you'd probably not have the diameter difference calculation, but then if it's in a spreadsheet that's automatic anyway)

Timewise I don't think RDM would be much different to using a test bar, and theoretically the result is perfect.

The big problem I had was doing the job singlehanded, making the adjustment without being able to watch the indicator is a real pain, I guess I could use my little webcam and laptop to watch the indicator remotely, that's something I'll do next time, along with a spreadsheet.

Having said all that, if anyone out there has a 7MT test bar surplus to requirements, for less than the Greek national debt I'd be very interested.

There is no difference whatsoever. The analysis method in part 1 of RDM is perfectly valid and I would suggest that this method is what was used for headstock alignment not long after the first metal lathe was produced 200 odd years ago. RDM, 2 collars, and test bar are all much of a muchness.

Unfortunately the mere mention of RDM (as typically published on HSM websites) causes some to foam at the mouth because they are unable to separate in their minds the perfectly valid alignment analysis part of RDM from the BS talk of twisting the ways.

We need to come up with a better name for part 1 of RDM to avoid its negative connotations. Perhaps we should call it Rotational Averages Alignment Theory or RAAT. If someone has old machining texts I’m sure that in them will be found the commonly used term that was once used to describe RAAT. It would be for the better if that term was reintroduced to the web and RDM was discarded.

Originally Posted by .RC.

Those specs are for the accuracy of the taper in the spindle nose itself..

The test you want is "Work spindle parallel with bed in horizontal plane (free end of arbor inclined towards direction of tool pressure)" The spec generally for the sized machines we deal with being 0 to 0.02mm per 300mm.

I chose that one because it used a test bar.

Michael