If your faceplate work was right on the money with a tailstock centre and the only change was to place the work between centres then it must be your positioning between centres and/or the cutting forces used

For really accurate hobby stuff I always use a well lubricated dead centre in the tailstock

The faceplate job only needed pushed and relatively little support but the between centres job will need a decent support hole at each end

Edited By Ady1 on 13/09/2020 00:34:31

You need to set your tailstock offset adjustment so that it is turning dead parallel along the outside diameter of the job or a test peice before facing the ends. Otherwise, the ends will be conical to some extent. Trouble with old/cheap lathes is the tailstock can go out of line every time you move it or even tighten the barrel clamp.

The other factor with old/cheap lathes is the cross slide may well be cutting out of square to the main lathe axis, also giving a conical face to some extent.

Definitely +1 on use a dead centre in the tailstock, not revolving, for high precision work. Hobby grade revolving centres can be all over the place.

If you have already got with in half a thou over six inches, that is about as accurate as you can expect on a typical lathe. In industry, tolerances of less than a thou are usually done on grinding machines which are of a next-grade accuracy.

Edited By Hopper on 13/09/2020 00:50:06

Robin

Did you turn and face both ends between centres, or just face one end ?

You did well to achieve such accuracy with the 6" diameter piece, did you check to see if the end faces were square to the length, as a cylindrical square ?

Emgee

Edited By Emgee on 13/09/2020 01:01:43

.

Your visualisation of this process seems fine, Robin … but I fear that what you are physically doing may not meet its rules.

It is theoretically possible to achieve perfect parallelism of the two end faces by swapping the workpiece between dead centres [i.e. like the bodger’s pole-lathe, or the watchmaker’s turns] … it is ‘simply’ a matter of taking a ‘perpendicular to the axis’ cut twice.

In practise, however, there may be variation from that idealised situation:

• Your cross-slide action may not be precisely repeatable
• Your centres may be constraining the work to not rotate on two hypothetically pure points !

On that second bullet … I have previously been ridiculed for suggesting that someone should consider the action of ball-ended centres, in comparison to cones, but it is a useful ‘thought-experiment’

MichaelG.

Edited By Michael Gilligan on 13/09/2020 06:50:45

A plane is defined by three points in space. Two points make a line…..

Getting the centers sooooo close that one could 'flip' the part between the centers and have exactly the same plane????

That's what face plates are for……

Pete

But the ends will still be parallel within the vagaries of machine 'slop' and any possible errrors in flipping the piece?

Since this part is a bearer that only has to be parallel on it's circumferences then why not just stick a hefty stud into the center of each end and set it up between a chuck and tailstock such that both ends can be faced inwards as much as needed for the circumferential bearing surfaces without flipping the part and then remove studs, bore outwards on each face just below that bearing depth. The only variation you have is slop between left and right facing actions.

pgk

No. The ends will  be conical if you cut straight in square to the lathe axis while the job axis is at an angle to the lathe axis. The ends in the illustration are parallel, but they were turned BEFORE the tailstock was offset in this taper turning example. If you now face the tailstock end of the job, it will be a convex cone. If you then flip the job end for end and face the other end, it too will be convex. Two convex surfaces are never going to be parallel.

Edited By Hopper on 13/09/2020 08:32:48

The problem the OP had with the face plate thread was that he did not give the full details for the end product.

Drilling holes in the ends of the part and bolting the part to the faceplate would have been the easiest and most obvious way to do it (as opposed to one thin bolt through the centre). Same as facing the ends – there is only about 3cm around the circumference that actually needs to be parallel – and (clearly now) one of those was not needed to be done until the spigot was completed.

The boring, for the spigot, certainly could not be carried out between centres, for sure.🙂

Order of operations and any later covered-over ‘setting up for machining’ holes were simply not considered before starting the machining – as far as I can tell. Like there are three holes, for bolts on a PCD, required at some point? I would likely have made those after facing the first end!

Tailstock does not need to be moved between turning the part between centres – just wound back a cm or so to clear the centre at one end. We don’t even know how all these measurements were taken, and the precision is likely as good, or better, than the original machine anyway.🙂

That leaves three possible positions for the spacer to be fitted – meaning that should there be the tinyist allignment out-of-true, there would first be a choice of turning the spacer by 120 degrees or Possibly skimming one face by a hundredth of a mm or so. No real sweat on the whole job ,whose length is not actually a critical dimension.

This lot, above, is the result of hindsight. But unnecessary if some fore-thought had been given to the job before starting, I feel.

.

Robin’s previous thread on this matter went quiet after I commented [and blowlamp endorsed]:

…

For best results when turning between centres, you will [obviously ?] need to switch the workpiece end-for-end, rather than changing tools and saddle position. … You want both faces to end-up either flat or very slightly concave.

…

So we can hope that Robin accepts this … and his opening post here suggests that he does

MichaelG.

.

Ref.https://www.model-engineer.co.uk/forums/postings.asp?th=167097&p=2

Edited By Michael Gilligan on 13/09/2020 09:54:13

We have no way of knowing what Robin has accepted or done. But he does say in the first post of this thread that he can't visualize how an offset tailstock could cause the two end faces to be other than flat and parallel. My example illustration above is simply to help him visualise what is going on. Seems like quite a few others are having the same difficulty envisioning it. Anyhows, conical end faces may or may not be responsible for the variation in mike readings across the two faces. More testing and reporting back is required.

BTW Robin, how are you measuring the distance across the two faces? A 4" micrometer or with a dial indicator and stand on a surface table?