[How do I?] "map" The Errors On A Older Lathe?

It is easy to become obsessed with accuracy on older machines, but the truth is, unless you know what the dimensional accuracy of the machine was when it left the factory, you cannot know how far it has deviated from that point. In order to make any useful measurements on bed wear it is absolutely useless to start from the 3 jaw chuck which is very likely the most worn and inaccurate part of the machine. Even a brand new 3 jaw chuck of good quality will run out .003" to .004". If you really want to do an objective test, you need to start from the machine spindle, not a chuck. Remove the chuck and carefully clean both the headstock and tailstock tapers Get a test bar, and fit this between centres of known good accuracy (or new, with a certificate of tolerance), and fix a good DTI in the tool post. Move the compound slide until the V ways are fully hidden, then tighten the gyb screws to lock the top slide. Zero the DTI on the test bar and lock the cross slide or tighten the gybs. Run the DTI along the side axis of the test bar, then adjust the tailstock set over till you get the least deviation possible on the side axis. If you now run along the top axis you will get a reading of the fall between the headstock centre and the tailstock centre. Make a note of this. Move the DTI to the headstock end of the test bar, lock the carriage, and rotate the lathe slowly by hand, and note down the deviation Mark the high and low points with sharpie on the bar. Move the DTI to the tailstock end of the test bar, and repeat, turning by hand and note the reading. Now set the test bar so that the marked high and low points are on the top and bottom faces of the bar, so that the finger of the DTI runs along the null point of the front axis (face) of the bar. Wind slowly and gently, if you wind too hard the pressure on the gears will move the carriage slightly, remember that cutting force pushes the saddle onto the bed Vees. Note down the deviation. Move the carriage to the tailstock end of the bed, where the wear will be least and try to rock the carriage across the lathe using A LIGHT push pull motion. Remember there are no cutting forces present to push the carriage hard down on to the bed, and you are trying to detect front to rear movement on the carriage without forcing the carriage up the vee ways of the bed, note the deviation, and repeat at the headstock end, and in the centre. You now have some idea of the wear in the bed, but remember, the harder you push, the more deviation you will see, but what you are seeing is not necessarily wear, it is just the carriage moving because there is no cutting force present to hold it down onto the vee ways.

Now you have a simple choice, either spend time and money correcting it, or live with it. Personally, I live with it unless it begins to affect what I make, then I find the problem, and find a way to work around it. The only thing a super accurate lathe does is grant you bragging rights in the bar, I know many turners who turn out accurate professional work as employees using machinery that is far from new. Most accuracy comes from the operator knowing his machine, and making allowances for any wear present. I used to use a 3 jaw chuck which ran out about .015" if you put a bar in it and tightened it up. With some care and tapping, I could get the error down to about .006", but if the component I was making was being turned on all surfaces, finished and then parted off the bar, the fact that there was .006 difference in concentricity with the bar left in the chuck is completely irrelevant. This is why you should always start with a bar slightly larger than your finished dimensions and turn all to size, rather than start with a bar with the finished outside diameter. Even a brand new quality lathe is built to tolerances, and I know that my worn machines are still more accurate than I am! Remember if everything was turned to 0.0000 (were it possible to do this) as soon as the temperature changed, it would seize up!
 
That was very well written Hermetic. I agree. Al
 
If you really want to do an objective test, you need to start from the machine spindle, not a chuck.
Hermetic, First and foremost, thank you for this very thorough and well written information.

As a supplemental to my understanding, if I chuck something (2 1/2" round stock)and turn it with a light cut, upon measurement, I will see the error in that portion of the ways that the saddle/carriage traverse. True? Understanding that there are some influences such as gib adjustment and so forth.

My main goal at this point is to fully understand all the dynamics of the current "state of affairs", thus I can be a better operator of the machine and consequently know how best to proceed. Proceed with
some corrective action or work with the machine as is.
 
Hermetic,

I would like to hear your ideas on the diameter of the test cut being smaller at the chuck end of the test cut bar.

Would the automatic feeding of the carriage being pulled tend to wear the ways ahead if the saddle? The leading wipers were not replaced and had imbedded cutting chips, and so forth. Cutting chips can be like small hardened material that scrapeand wear the cooresponding ways.
 
Hi jjtgrinder,
What you need to do first is establish the runout on the chuck, by putting a dti onto a piece of ground round stock held in the chuck. The drawback with doing this is that the three jaw chucks are not repeatable, that is, if you take the bar out of the chuck and then replace it and test again, you will probably get a different set of readings, and the more worn and bell mouthed the chuck is, the more variation you will get. That is why I suggested to take the chuck off and work from the spindle. Really (in my above post) you should put a centre in the spindle, then check that with a tenths DTI mounted in a locked carriage , when you have confirmed that the centre is concentric with the spindle, fit the test bar, centre the tailstock, and then with a DTI in the SPINDLE run round the tailstock centre to see if that is concentric with the spindle! In other words, how far do you want to go? there will be minor inaccuracies in ALL of these readings, but you can learn to live with them. If of course you find that the centre is NOT concentric with the spindle you must stop the process, remove the centre and look for the reason it is not concentric, such as damage to the spindle bore taper, dirt or swarf etc. When this is removed/repaired and the centre is concentrically seated in the spindle, (remember there WILL be runout on the spindle, it will be small, but measurable, and can only really be measured on the internal taper of the spindle, or the outside of the chuck register if there is one, not easy!) You continue to the next stage, removing as much error as you can at each stage. If you are going to use a chuck, first strip it and clean it, and check for wear on the scroll (the spiral which operates the jaws) and check the fit of the scroll plate in the chuck body, because if there is more play (wear) than there should be, when you tighten the chuck, the scroll plate moves slightly sideways and the jaws tighten unevenly. You must also check the fit of the chuck on the spindle and that the chuck backplate is concentric with the spindle when fitted. Any damage to the register behind the thread is critical to the chuck being concentric with the machine, it is the register which centres the chuck, not the thread. Likewise if you have a chuck that fits to a taper fitting like the L0 type, that taper must be undamaged for the chuck BODY to be concentric with the spindle, this does not mean that the jaws will be concentric with the spindle for the reasons given above, this is why it is best to work from the spindle, you are removing several areas of possible inaccuracy that you could eliminate instantly by buying a new or better chuck. Three jaw chucks are workhorses, they are not ever accurate to close tolerances, and they are NOT repeatable. That's why the most accurate chuck you will ever own is a 4 jaw! (Or a collet!)
Phil
 
Hi Again, you posted while I was typing.............. we need to get the terminology clear here, most important. If you mean by "the diameter of the test cut being smaller at the chuck end of the test cut bar." You mean that the lathe removed more metal nearer the chuck, or to put it another way the diameter of the test piece is smaller nearer the chuck, and gets larger as you move away from the chuck. If this is the case this would more likely be a slight twist in the bed, which can be corrected, and would mean that if you stand at the headstock end of the lathe, looking down the length of the bed, the twist is clockwise, IE the vees get slightly further away from the centre as you move away from the headstock.. Embedded chips I don't like the sound of! Is there visible scoring on the Vees, if so as long as it doesn't cover the whole of the vee, the bed should not have dropped, and you can correct wear (to a certain small amount, by slightly twisting the bed in the opposite direction, but we are getting way ahead of ourselves here) can you post up some pics of the lathe, and some closeups of the bed, especially the areas you think are the most worn.
Phil
 
if I chuck something (2 1/2" round stock)and turn it with a light cut, upon measurement, I will see the error in that portion of the ways that the saddle/carriage traverse. True? Understanding that there are some influences such as gib adjustment and so forth.

Yes, this is true...................to an extent, what you need to know is where the inaccuracy is, and the problem is that what you are seeing is cumulative errors, a bit here and a bit there, it depends on the type of lathe. Once you have taken a full cut from the outside of the bar, the bar will be concentric with the spindle (subject to any spindle error) but may not be (probably won't be) parallel. If the lathe is bolted to its original cabinet stand and on a fairly level floor there should be no bed twist, but if there is, you would correct this by shimming where the bed bolts to the cabinet stand. If the lathe is the type with separate legs at either end, it must be shimmed to the floor , which should be as level as possible, and then bolted down with the bolts just snugged against the shims, then you do a turning check and add or remove shims till you get as near perfect as possible, or you run out of patience and say "near enough"! What I am trying to say is that you cannot make sense of wear in the bed until you know that what you are seeing is wear, and not twist, or a combination of the two, and if you have the separate leg type of lathe, or a bench lathe bolted to a wooden bench, you MUST untwist the bed before you can make any other measurements, Thats why I have asked for some pics.
Phil
 
When I did a test cut on my 20" L & S lathe a while back, it did the same thing, taper smaller toward the chuck. Measured .0015" in 12". Bar diameter was at 3" OD. Material was ductile iron, something that was handy at the time. I made sure the bed was level with no twist at the time I made the test cuts. Tool push off, I doubt it! Using positive rake tooling, very stout setup, freshly rebuilt lathe, including a re-ground bed. I'm not a bit disappointed! In fact tickled to death that it cuts that close and good for a 60 year old lathe! If you follow Dr. Schlesinger's guide to "Testing Machine Tools" accuracy, he says .0008" in 12" allowable in either direction! Ouch! Do we really need that kind of accuracy in the lathes we play with?

Ken
 
No Ken, we don't, that is good work, and a machine that is as accurate as it needs to be, in fact probably more so! Of course you could try a .0015" shim under the front tailstock end foot....................................here we go again!!
 
I have two lathes, an early sixties Colchester Mk1 Student, and a Covmac. I have just had the workshop floor in what is going to be the new "machine shop" re concreted, and it is very flat and level, thanks mainly to the fact that I didn't do it! I got a pro in with his lads and a powerfloat. My Student is a straight bed model on an original steel cabinet, but the Covmac is on separate legs and so it will definitely need shimming and bolting down in order to get it to turn parallel.
 
Back
Top