Lathe levelling

By accident I came across a video about lathe levelling by means of an engineer's sensitive level. The person demonstrating it took great pains to clean the ways and then place packing on them to create a shelf with which to support the level. It occurred to me that I'd seen similar videos where the level was supported on packing blocks on the lathe ways. But why?

Why not use the saddle to support the level? It's the saddle that guides the cutting tool, so why not use the level to see how the saddle itself leans because of twisted ways? Using the saddle as the support also allows observation at any point along the bed. Is my reasoning faulty?

Because you want to "level" the bed itself and not the saddle. Saddle top surface may not necessarily be parallel to the parts that run on the ways. Also, bits of swarf or even oil between bed and saddle ways can give a false reading.

It's all a bit academic anyway. The only way to be certain that your lathe is set up to turn parallel is to turn a test piece held in the chuck and with no tailstock centre in place and measure it. A level bed is not always a guarantee of parallel turning under load, particularly on a used/worn machine.

Why not use the saddle to support the level?

It probably depends upon the bed formation of the lathe in question, but you would also see bed wear from readings taken off the saddle & possibly get a false reading ? Could you link to the video(s) concerned ?

Nigel B.

The Smart & Brown model A at the museum has at least 0.025" wear in the ways at the headstock end, probably more as the saddle was running on the top of the rack at the left end. Any attempts at using a spirit level would be futile. Only actual measurement of cut diameters would have any meaning. As it is, a length of brass bar about 1" diameter and projecting 4" from the chuck was 0.0001" larger at the free end which is good enough for me. The rack has been tilted down by 0.025" at the left end to help with the gear engagement and so it doesn't interfere with the saddle. The instructions for the Atlas 12 X 24 which has a plain bed like a Myford are to use a sensitive level to ensure both ends of the bed are parallel front to back. We have built adjustability into the bed mounts to achieve this. It would still need to be proved by an actual test of diameters machined.

Conventional levelling is not practicable with my Willson slant bed. The manual advises comparing readings from the saddle at the extremes of travel.

I totally agree that the final arbiter on lathe levelling is measuring taper, but that is not what my posting was about. I want to know why measuring bed twist with a level supported on packing from the bed ways is considered to be more accurate than using the same level actually supported by the saddle. Surely its the saddle that ultimately guides the cutting tool, so its use must be more accurate. Am I wrong?

One of the videos can be seen **HERE**

Yes you are wrong, the lathe bed guides whether the tool cuts parallel or taper.

Tony

I think you have a good point here. Normally I would suggest the blocks were to get clear of the V bed ways. Now if the Vs are worn at one end and asymmetrically then levelling on the unworn flats would not actually get you very far. The saddle would be telling you what it was seeing and in the end that is what is really important.

Also if your bed is bent like a banana on its side levelling again is a waste of time and the turning test of ends of a bar only doesn't tell the whole story.

Edited By Bazyle on 21/03/2020 15:47:29

If the bed is twisted, you cannot be certain that the saddle will not remain fully in contact. There has to be clearance between the Saddle and the bed ways, or it would not move along!

What if the Saddle were distorted?

Working off the Saddle introduces another possible source of error, needlessly.

The simpler that you keep things, the fewer possibilities of error.

If the bed is twisted, you are most unlikely ever to cut parallell.

My lathe has a prism / flat surface for the Saddle to run on, and similar for the Tailstock

I pack a parallel, from the flat ways, and adjust until the level reads the same at Headstock and Tailstock ends.

Howard

Gary, you put a sensitive level at each end of the bed to make sure not that they are 'level' but that they are level with each other. Any pedestal lathe sitting on the floor but not levelled will almost certainly have a twist in the bed. If it's on castors even more likely.

Hopefully this sketch will show why you would level a lathe.

I've knocked up the above image and laid one over the other but rotated 1 degree to simulate an exaggerated twist in the bed. The images depict the saddle at one end of the ways and then at the other. If you look at the tip of the tool you'll see that it's closer to the 'work' at one end than at the other because the twist has rotated the whole assembly clockwise. This would cause you to cut a taper in your work.

Edited By Pete Rimmer on 21/03/2020 16:41:45

When you are dealing with a lathe that has a lot of wear at one end of the bed and none at the other, there is no way that the saddle will make full contact at either end, or at any point for that matter. The S & B saddle sits on an inverted vee at the front and a plain surface at the back. most of the wear is in the prism, so a bubble would be at the rear of the level at the left end of the bed even if it was central at the right end. A similar situation would arise if the level was on the saddle. I have the saddle set so that it is very tight at the last 3" at the right hand end. It is the best I can achieve.

The errors in turning vary with the diameter of the work as a different tool height at either end would be a greater percentage of small diameters.

Pete, your diagram illustrates the advantage of putting the level on the saddle assuming there was no wear in the lathe. It also highlights the importance of levelling the typical Chinese lathe which has excessive centre height for the basic dimensions of the lathe to make the swing larger. 

Edited By old mart on 21/03/2020 16:52:13

Yes, the tall stack but narrow bed lathes suffer the most from twist plus of course they tend to be less rigid in the first place.

I don't advocate putting a level on the top of the saddle simply because it will only give you the angle of the saddle, not the way. Anyone who has scraped in the underside ways of a saddle would know that the biggest wear on a saddle is on the front way nearest the chuck - the wear there is going to be more than the wear in the actual bed and it's normal for saddle ways to be worn both ends and 'high' in the middle causing them to rock.

For a badly worn lathe that is cutting a taper you might use inducing a bit of twist as a get-you-by mind - some manufacturers actually specify setting a twist to remove taper.

Where it REALLY gets complicated is for lathes like Colchesters which have adjustable headstocks via the rear-end jacking screws. You could get into a horrible situation where you adjusted the headstock to correct for cutting taper in a part when thee cause was actually the bed has a twist. Now you have adjusted the headstock out of axial alignment with the tailstock so you'll be drilling off-centre and causing reamers to cut on one side.

If you're going to work on your lathe alignment the first thing you do - the very first thing, is level the bed.

Old Mart, only the tail stock runs on the inverted V on a S&B model M. The saddle runs on the two flat topped ways. Model A is as you describe.

Martin C

As Howard, partly. The carriage will only give you an average over its length. Measuring across the ways is the truth at that, or those, points.

Far more important for lathes with multiple (more than two) supporting points - less so for most hobby lathes which may only be supported at three or four points, as the final test cutting will ignore any previous findings.

Mine is rigid at the headstock and ‘loose’ at the tailstock end, only supported on one spring loaded point with shims on one side or the other as appropriate.

I didn't know that the model M only ran the tailstock using the vee. As you know the A uses one vee for the TS and another for the saddle. Once I toyed with modifying the saddle to use the rear vee and flat front. A major undertaking, which I didn't quite have the nerve to do. Probably a good thing. For all its wear, the 71 year old lathe which was donated to the museum when Westlands left W-S-M rather than being scrapped cuts pretty well.

The Atlas 12 X 24 with the flat bed lends itself better to be aligned by matching the levels at the ends of the bed. It is much more flexible being lightweight, and there is no measurable wear at all in the bed. But then I would not see anything against the level on the saddle either which travels perfectly smoothly from end to end and has superb contact over its underside.

Edited By old mart on 21/03/2020 21:03:36

Whilst chasing out the last thou of taper on my Chester I discovered that even a very slight twist in the bed would result in waisting of the test piece. It was only a thou an a arf out but definitely the same dimension at both ends and slightly thinner in the middle of the cut. Thinking it through logically twisting will raise or lower the tool height at the stock end of the work, then as it travels along the cut it will dive or climb depending on the twist and be back on dead centre in the middle and thus remove the most material. It meant having to start all over again and shim the head to get rid of the very, very slight nod in the headstock.

I decided to make the levelling of the Atlas 12 X 24 user friendly at an early stage of its clean up and refurbishment. I had an unused block of aluminium 1 1/2" thick to fit under the headstock end which would give enough height to add adjustment to the tailstock end. This rough diagram shows what was done.

Gary, I think I would agree with you. The saddle slope should be a reflection of the bed slope.

I suppose in using the saddle one could get some small 2nd order errors creeping in, but these should be negligible in my mind, and certainly no worse that putting blocks (which have their own inherent geometric errors) in slightly different positions each time one makes a measurement along the bed.

Gerry

I did the same thing as Old Mart with my ML7 when I had it. Made removing twist SO much easier. Used exactly the same method when the bigger ( BL12-24 ) machine arrived, co incidentally, using 1/2 UNF setscrews and nuts, at both Headstock and Tailstock ends.. Unlike the ML7, iIt did not need to be raised.

Raglans made life so much easier, by only having one fixing at the Tailstock end

Howard

Well, after reading the comments I'm convinced that I'm right. My original post was only about levelling the bed - not about headstock and tailstock alignment. The objective of bed levelling is to ensure that the position of the cutting tool tip relative to the lathe axis remains constant as the saddle moves along the ways.

I really can't see why supporting a level on temporary blocks is more accurate than supporting it on a purpose-made block - ie. the saddle.

The diagram shows a level, 'A', supported on a raised 'V' way and a block, 'B', resting on a flat way. If the height of block 'B' exactly matches the difference in height of the two ways (as in '1' then fine, but rather unlikely. If it doesn't, then the horizontal position of block 'B' can have a noticeable effect on the orientation of 'A', as can be seen in the difference between diagrams '2' and '3'. Can you be certain of precisely positioning block 'B' for each measurement? I think not. But by using a purpose-made block (ie. the saddle), the uncertainty is completely eliminated.Once the best position of the level on the saddle has been found, then you just leave it there and wind the saddle along the ways to take the readings.

It's how the saddle rides the ways that's important.

Edited By Gary Wooding on 23/03/2020 10:31:56