Runout turning between centres

[HevansccParticipant @hevanscc]

I know that this problem comes up frequently, and I have scoured the forum to arrive at an explanation, but not now feel the need to ask for direction.

I needed to make a shaft with a close running fit in two oilite bearings separated by about 80 mm, so in anticipation I set out to align my lathe (an ex-school Viceroy of 1970s vintage) using the Harold Hall method of altering the bed twist using the adjustable feet, and then aligning the tailstock. All seemed to go well and I produced a test bar which varied in diameter from 26.785 to 26.795 mm along a 180 mm length, a difference which is at the limit of my measuring capability with a micrometer.

This was done with a live centre in the headstock and a running centre in the tailstock, and before I removed it I clocked it with a DTI and got about 0.01 mm deflection over the length. Fast forward several weeks and I turned a shaft of 14 mm nominal diameter using the same between centres setup to produce the following:

Runout over 12 cm: Tailstock 13.98 mm, Headstock 14.10 = 0.12 mm

Clearly not good enough, so I removed the shaft and put it in the three jaw chuck and turned it to get:

Runout over 12 cm: Tailstock 13.92 mm Headstock 13.94 mm = 0.02 mm

Which is much better and probably as good as can be expected on my lathe with my chuck.

So then I put the test bar back in with the original set up of live centre and running centre and clocked it with the DTI to get 0.075 mm defelction along its 180 mm length – which would explain the 0.12 mm taper on the shaft. I removed the running centre and replaced it with a dead centre, clocked it again and got 0.030 mm runout, which is better but not perfect.

So, my questions are:

1. Is it likely that using the running centre accounts for the runout, or tailstock alignment, or both?

2. Am I expecting too much from what is an old school lathe?

3. Should I have just turned the bearing surfaces of the shaft separately over short distances rather than the whole shaft?

Any advice or observations gratefully received.

Hywel

[HevansccParticipant @hevanscc]

[MWParticipant @mw27036]

Hi

I do remember people saying a while back that when they needed to get a perfectly running centre, it's done using a soft centre and the point of the centre turned in situ, to make sure there's no chance of any runout?

I don't think you mentioned it but I trust you made sure you weren't getting any play in the gibs as well? 

Michael W

Edited By Michael-w on 23/11/2017 12:00:00

[Bob RodgersonParticipant @bobrodgerson97362]

I have found that it tends to be the tailstock alignment that causes most grief. If I am turning between centres and the job requires accuracy I set up and run a test on a bar of similar length, adjust the tailstock as required and then without unclamping the tailstock place the workpiece between centres and finish it in one setting.

[HevansccParticipant @hevanscc]

Posted by Michael-w on 23/11/2017 11:58:22:

Hi

I do remember people saying a while back that when they needed to get a perfectly running centre, it's done using a soft centre and the point of the centre turned in situ, to make sure there's no chance of any runout?

I don't think you mentioned it but I trust you made sure you weren't getting any play in the gibs as well?

Michael W

Edited By Michael-w on 23/11/2017 12:00:00

Good point, I must confess to not having adjusted the gibs since I got the lathe, but I managed to produce the test bar – maybe by luck rather than judgement?

[HevansccParticipant @hevanscc]

Posted by Bob Rodgerson on 23/11/2017 12:04:14:

I have found that it tends to be the tailstock alignment that causes most grief. If I am turning between centres and the job requires accuracy I set up and run a test on a bar of similar length, adjust the tailstock as required and then without unclamping the tailstock place the workpiece between centres and finish it in one setting.

Yes, I probably should have checked alignment with my test bar beforehand and then done as you suggest. I naively expected things to remain the same after aligning and producing the test bar

[OldironParticipant @oldiron]

I agree with Bob. Unclamping the tailstock has probably given you the run out between jobs.

As Michael says, use a soft centre and turn a new taper then leave it in place until the job is done.

Its so easy to think that a setup will stay constant after spending so much time doing it but only one small thing has to change and its all gone to pot.

regards

[Martin KyteParticipant @martinkyte99762]

Turn the soft centre in the headstock on situe. Don't finish turn with a live centre in the tailstock. Use the best hard centre you have. Any bearing is going to have some run out but a fixed point will not have any by definition.

regards Martin

[not done it yetParticipant @notdoneityet]

You do not indicate how the lathe is situated. Is it on its original stand or now bench mounted? What is the floor? Concrete or other?

Any timber involved can cause changes with the seasons.

All good points above. I would turn the bearing surfaces as two separate cuts. The only time I would make a long shaft like that is if any gears needed to run on the same diameter, which is unlikely. A little polishing with abrasive strip would provide the same dimensions for each bearing. Even the finish provided by the cutter might give some variance at each end of the cut (measuring to 0.005mm).

[Clive FosterParticipant @clivefoster55965]

Did you actually twist the bed when first setting up or merely adjust things so the bed was unstressed with the right load on each foot for it to take up its natural position?

If you have actually managed to twist it a bit all bets are off. The detail alignment behaviour of a stressed, especially torsionally, complex beam structure (which is what a lathe bed is) not simple. In general small variations due to temperature, load history, etc. etc are inevitable as the bed shuffles around so its unlikely that the tailstock alignment will repeat to the sort of accuracy you are looking for.

If it is in its natural position then things will behave themselves but wear and inevitable production tolerances mean you are unlikely to be dead on at any random position. A Viceroy is a decently made machine but variations of a thou or two with tailstock position should be considered normal. Even on machines of the very highest class its unwise to assume that perfect alignment can be held all along the bed. Obviously use minimum tailstock poppet extension.

If you plan to do much between centres work its well worth making a set of centred test bars with replaceable collars at each end to do a simple turn to size and adjust check before starting work. Best practice is to swop end for end and measure run out before deciding the alignment is good. Probably the least painful way of producing a set that suits what you do is to make a list of suitable lengths and, if you don't already have it, make up the appropriate one from the list each time you do between centres job. I'm up to three by now. Need to keep the makings on hand of course. I use 1" / 25 mm round BMS steel bar, checked for straightness, with 2" / 50 mm delrin collars "inch ish" wide stiff push fitted A grub screw holds them in place against turning forces. Material isn't critical but it does need to be easy to get a really good finish on the collars. I feel that the lower forces involved when working delrin help accuracy.

Don't forget to recess tech centres to protect them from damage. Arrange storage so the centres stay clean. A bit of unnoticed muck or tiddly piece of swarf inside a centre can ruin your whole day. Or at least involve workshop esperanto. How do I know?

Clive.

[HevansccParticipant @hevanscc]

Thanks, lots of excellent info for someone who has no training in this field.

The lathe is on its original stand on a concrete floor (not bolted down). I have no idea if it is twisted because did not set it up with a level originally – basically just plonked on the stand and semi-aligned using the method in the Sparey book. The Viceroy has nuts below the cabinet which lower or raise the bed so I have been adjusting these, but no idea if correctly because the 'manual' is silent on this.

It is good to know that these variations are unexpected given the circumstances I ahve been operating because I don't want to chase decimal places for no reason.

Checking alignment beforehand, turning centres in situ and then keeping the setup constant for the job at hand is certainly something I will be doing in future; and investing in a good dead centre for accurate work.

Regards

Hywel

[Russell EberhardtParticipant @russelleberhardt48058]

How far was the tailstock barrel out of the body? Did you use the barrel lock?

Older lathes can have wear in the that makes the barrel a bit loose when extended fully. On my 1950s lathe i get no measurable movement for the first 20 mm or so but with the barrel fully extended I can move the tailstock centre by about 0.05 mm from back to front .

Just a thought.

Russell

[Martin ConnellyParticipant @martinconnelly55370]

Hywel, as an exercise try the following.

With the tailstock barrel just short of fully retracted set up an indicator on the side of the barrel and zero it. Then clamp and unclamp the barrel and see what happens to the reading at different degrees of clamping.

Then extend the barrel to about midway and move the tailstock along the bed away from the indicator by the same amount compare the readings and repeat the clamping unclamping again.

Finally repeat with the barrel almost fully extended and get a third set of readings.

In an ideal world the readings would be zero at all positions but there is no such thing in reality. It would be interesting to know how much variation there is between the lowest and highest readings and to see if there is enough to account for your taper.

Martin C

[Neil WyattModerator @neilwyatt]

Even if your lathe is a near basket-case (which it obviously isn't) a way around this is to leave both journals just oversize, then skim the one at the tailstock end, then without moving the tool turn the bar end for end and skim the other journal at the same setting.

Obviously not a perfect solution if you have to thread the whole bar through one of the bearings.

Neil

[Rick Kirkland 1Participant @rickkirkland1]

I like Hevanscc. . He knows what a live centre and a running centre are AND he knows where they go. That is so rare these days. What a heartwarming posting to read.

[Clive FosterParticipant @clivefoster55965]

Hywel

As its on the original stand there should be no great twist issues.

Best practice is to get the stand sitting evenly on its feet and bolt down so there is no strain on the stand. Then you sit the lathe on the adjusting nuts and move them up or down to get it evenly balanced. One of those "develop a feel for" things. One way of getting the pairs at each end pretty close is to use one lifting nut to just release the weight from its mate on the other side. When it takes the same amount of turn on each to just take the weight off the other things are very close indeed. If you have both ends like that and the bed is adequately level, builders level will do here, on the long axis it will be very close when nipped down. A proper precision level or, better, accurate clinometer, will show you the shifts as the top nuts are tightened down. The "equal turns to release weight on the other side" technique is quite capable of getting you into the couple of minutes of arc error band.

With these strong, welded box, style stands its arguable whether the lathe bed stiffens the stand or the stand stiffens the lathe when all is bolted up tight. Primarily due to that issue some folk advocate merely nipping up the top nuts enough to stop things self adjusting rather than bolting down really tight.

Martin C makes an excellent suggestion of how to check if, and how much of problem you have with tailstock wear. Given that we are dealing with small errors so correspondingly great care is needed when setting up to measure them directly it might well be quicker overall to just make some test bar and collar set-ups and use them between centres. Time saved doing direct measurements being put into making the test bars and collars. I'd probably make ones with between collar measurements of 3", 6", 12" and 18". 9" would be useful too. Test bar length of around 3" more than the between collar distance should be fine. Set up true for same diameter collars with the tailstock barrel fully retracted then loosen the tailstock so it slides back as you wind the barrel out further. Lock the tailstock and either put a dial gauge on the tailstock end collar to see what the total error is when you spin the headstock by hand. It was true when you made it. Or simply measure the collars again and take another skim. Remeasure the diameters and the difference tells you the error.

Write down your results so you can use a dial gauge to set things pretty close before messing around with collards to get them dead right. Generally if you set the tailstock right for about 2/3 rd of the way down the bed it will be good enough for most jobs.

Clive.

Edited By Clive Foster on 23/11/2017 18:11:57

[HevansccParticipant @hevanscc]

Posted by Martin Connelly on 23/11/2017 14:25:16:

Hywel, as an exercise try the following.

With the tailstock barrel just short of fully retracted set up an indicator on the side of the barrel and zero it. Then clamp and unclamp the barrel and see what happens to the reading at different degrees of clamping.

Then extend the barrel to about midway and move the tailstock along the bed away from the indicator by the same amount compare the readings and repeat the clamping unclamping again.

Finally repeat with the barrel almost fully extended and get a third set of readings.

In an ideal world the readings would be zero at all positions but there is no such thing in reality. It would be interesting to know how much variation there is between the lowest and highest readings and to see if there is enough to account for your taper.

Martin C

So, into the shed aft work to make the measurements:

Barrel retracted: clamped 0, unclamped 0 mm

Barrel half way: clamped 0.025, unclamped 0.025 mm

Barrel fully out: clamped 0.075, unclamped 0.075 mm

In use, I have only extended the barrel the shortest distance, 1/4 max.

Hywel

[HevansccParticipant @hevanscc]

Posted by Clive Foster on 23/11/2017 18:11:40:

Hywel

As its on the original stand there should be no great twist issues.

Best practice is to get the stand sitting evenly on its feet and bolt down so there is no strain on the stand. Then you sit the lathe on the adjusting nuts and move them up or down to get it evenly balanced. One of those "develop a feel for" things. One way of getting the pairs at each end pretty close is to use one lifting nut to just release the weight from its mate on the other side. When it takes the same amount of turn on each to just take the weight off the other things are very close indeed. If you have both ends like that and the bed is adequately level, builders level will do here, on the long axis it will be very close when nipped down. A proper precision level or, better, accurate clinometer, will show you the shifts as the top nuts are tightened down. The "equal turns to release weight on the other side" technique is quite capable of getting you into the couple of minutes of arc error band.

With these strong, welded box, style stands its arguable whether the lathe bed stiffens the stand or the stand stiffens the lathe when all is bolted up tight. Primarily due to that issue some folk advocate merely nipping up the top nuts enough to stop things self adjusting rather than bolting down really tight.

Martin C makes an excellent suggestion of how to check if, and how much of problem you have with tailstock wear. Given that we are dealing with small errors so correspondingly great care is needed when setting up to measure them directly it might well be quicker overall to just make some test bar and collar set-ups and use them between centres. Time saved doing direct measurements being put into making the test bars and collars. I'd probably make ones with between collar measurements of 3", 6", 12" and 18". 9" would be useful too. Test bar length of around 3" more than the between collar distance should be fine. Set up true for same diameter collars with the tailstock barrel fully retracted then loosen the tailstock so it slides back as you wind the barrel out further. Lock the tailstock and either put a dial gauge on the tailstock end collar to see what the total error is when you spin the headstock by hand. It was true when you made it. Or simply measure the collars again and take another skim. Remeasure the diameters and the difference tells you the error.

Write down your results so you can use a dial gauge to set things pretty close before messing around with collards to get them dead right. Generally if you set the tailstock right for about 2/3 rd of the way down the bed it will be good enough for most jobs.

Clive.

Edited By Clive Foster on 23/11/2017 18:11:57

You seem familiar with the Viceroy setup Clive. On mine there are threaded bolts on the bottom of the bed that extend down through the cabinet through the adjustment nuts. The adjustment nuts seem to screw into the underside of the cabinet and presumably extend through itmto lift the bed – I didn't look that closely when I assembled it and was guessing at the time. Then there are smaller nuts which go,on the ends of the threaded bolts and tighten down (or up,in this case) on the adjustment nuts – are these the top nuts you mean when you say to 'nip them up'?

Hywel

Edited By Hevanscc on 23/11/2017 19:40:51

[Clive FosterParticipant @clivefoster55965]

Hywel

Not directly familiar with the Viceroy I'm afraid.

But I have run into that sort of "screw up or down to bring level then do up the top nut or bolt so it stays put" adjuster several times on various pieces of equipment. Seem to be a number of variations but the rough setting principle of equalising the amount of movement needed to unload a mount is the same. What you actually have to shift and where the adjuster is varies. Special warm place in Hell reserved for the designer who puts the adjusters under the cabinet top plate on the other side of a re-enforcing angle with a fixed shelf not far below. So you have to do it all by feel whilst simultaneously wishing for longer arms.

Made my own home brew version for a SouthBend.

Clive.

[HevansccParticipant @hevanscc]

Posted by Clive Foster on 23/11/2017 20:04:48:

Hywel

Not directly familiar with the Viceroy I'm afraid.

But I have run into that sort of "screw up or down to bring level then do up the top nut or bolt so it stays put" adjuster several times on various pieces of equipment. Seem to be a number of variations but the rough setting principle of equalising the amount of movement needed to unload a mount is the same. What you actually have to shift and where the adjuster is varies. Special warm place in Hell reserved for the designer who puts the adjusters under the cabinet top plate on the other side of a re-enforcing angle with a fixed shelf not far below. So you have to do it all by feel whilst simultaneously wishing for longer arms.

Made my own home brew version for a SouthBend.

Clive.

Yes, you describe it correctly – Keyhole surgery where no spanner or socket seems to reach, while also trying to catch site of a DTI reading . . .

[Martin ConnellyParticipant @martinconnelly55370]

Hywel, the figures from the tailstock show that the tailstock position is important when setting the part between centres. They are only small values but if you are trying to achieve consistency it helps to know what errors the tailstock can introduce in the setup. Hopefully this will help chase down what is happening with your lathe.

Martin C

[Ady1Participant @ady1]

I've had a big improvement in centre work by stiffening it with a fixed steady

[Michael GilliganParticipant @michaelgilligan61133]

That shows proper ingenuity, Ady

It nicely 'beefs-up' the tailstock.

… I like it.

MichaelG.

[Author]

Posts