All squared away.

[DavidA]

The headstock on my red lathe can be removed. and it is located by a pivot bolt at the gear end and a clamp bolt at the chuck end. As it can swivel on the gear end bolt there is a pair of coinciding parks at eh front to assist in re-aligning it.

But let's assume there is some doubt about the marks.

How would the members ensure that the head is truly concentric with the lathe centre line.  If it isn't the chucks and face plate will not be square to the axis.

To add to the interest let us assume that you don't already have a face plate to assist in this operation.

I know how I would do it,  but am curious as to how others see the problem.

Dave.

[chipenter]

Test bar http://www.arceurotrade.co.uk/Catalogue/Measurement/Parallels-Angle-Blocks-Test-Bars and a dial guage in the toolpost , or iff the tailstock is spot on a bar between centres .

[RussellT]

If you put a round bar in the chuck and then take average readings with a dial gauge it should be possible to align it without having to buy an expensive test bar.

Russell

[Arbalist]

Does this help?

http://igor.chudov.com/manuals/Rollies-Dads-Method-of-Lathe-Alignment.pdf

[steampunkpete]

If I had them, I'd use a round bar in a collet rather than a chuck.

I'd use the longest bar that I had, as the longer the bar, the more accurate the achievable alignment. If I had some straight tube I'd use that for preference, but it would have to be closely round tube, not oval.

I'd then mount a dial gauge on the saddle. I'd then check that the bar was straight by setting the dial gauge at the end of the bar furthest from the headstock and rotating the chuck by hand. I'd also check at a couple of intermediate points in case in case the first reading was at a null point  on a banana shaped bar.

If then the bar proved to be straight, I'd take dial gauge readings along the bar by traversing the saddle.

Adjustment to the headstock by splitting the difference in traditional form. The headstock to be bolted down to a known torque between adjustments.

This all assumes that the mis-alignment is only in the horizontal plane. Given the risk of dirt trapped beneath the headstock, distortion etc. I'd also check alignment in the vertical plane by setting the dial gauge on top of the bar and repeating the alignment check along its length. If there was a problem there, I'd weep and then bang my head on the headstock until one of them was properly aligned.

By this time I'd be in need of a cuppa or a beer.

What have I missed?

[DavidA]

Everyone seems to have a version of the same idea.

Pete,

..I'd then check that the bar was straight by setting the dial gauge at the end of the bar furthest from the headstock and rotating the chuck by hand. ...

This would not necessarily show the bar was true. You can have a situation where the headstock spindle was absolutely spot on (I should be so lucky ) but any error in the chuck or the chuck back plate would give you the end of the bar rotating around a (small) circle.
You would still need to do the 'rotate and half the error' operation at both ends to find out if the bar or the chuck was causing the difference.

Thanks all for the advice.

Dave

Still pursuing the fabled micron run- out.

[Fergus OMore]

Not me Three or 4 jaw chuck with the longest piece that will swing in the gap. Same as a faceplate.
Face the extreme outer edges and clock . If the up and down worries, put a block the saddle and fit the top slide at 90 degrees. Do the same with the clock or even a bit of fag paper!
It's about the 6th Proposition of Euclid- if I recall.

It's amazing what can be done with a bit of metal 'twer that size and a bit of spit and fag paper.

Norman

[edward]

I have the same problem with my ancient drummond. My chuck has a bit of a wobble so plan to turn a bit of something down and mic it. We shall see.....

[RussellT]

I think one of the advantages of these methods is that the bar doesn't need to be straight or parallel.  It does need to be round though.

The averaging method takes away all the errors and looks at the axis of rotation of the spindle.

If the bar is a bit wonky then a big foot on the dial gauge might help to eliminate any error due to movement in the wrong plane.

Russell

[Fergus OMore]

Sorry, but it is wrong. If you clock a test bar ( ignoring a wiggly chuck or whatever)  your record a combination of out of truth for the spindle- and the wear of the bed.  This cutting of a test piece is all well and good for a new lathe. It is in Myford's setting up instructions for their setting up 'thing'. I've got a test bar that has a No2 MT shank ground from a worn out reamer- great on my lathe which was slideways ground fairly recently - but bloody awful for the same use before that.

The first thing is to assume that moving the saddle is the first step to a quick way to a Moonbeam from a Lesser Lunacy.

Probably the best way is to get the two classic books out. One is Connolly and the other is Schlesinger.  They are the 'modern' editions of Euclid.

And if you have done all this, you still have to do something else- and that is compensate for the deflection of a lathe tool.

Regards

Norman

[vtsteam]

But If the spindle bearings and bed are worn enough, then making the spindle parallel with the bed is easy:
 Don't do anything! 

(Joking, but if wear is signifigant than there is no meaning to the words "making them parallel". )

David is presenting a hypothetical case of an assumed straight bed and unworn spindle, for fun consderation here, I think.

[Fergus OMore]

David is presenting a hypothetical case of an assumed straight bed and unworn spindle, for fun consderation here, I think.

It's all to do with hides- flat or otherwise. The squaw on the hippotamus hide is equal the son of the squaws of the opposite two hides. Easy as Pi. 
Norman

[vtsteam]

We had pizza and cherry pie to celebrate, Norman, you?

ps. don't forget deflection of the test bar due to clock spring if you're into micron lathes

pps. do you use an air bearing traveling steady on a micron lathe? questions, questions......

[Fergus OMore]

Gave my wife the Book of Pi which was about a guy with a  tiger on a lifeboat. Very confusing!

But I got utterly confused with Holzapffel- the lathe guy who was actually using hippo hide to flatten metal. Seriously, I'm not joking. Then he sold an ornamental turning lathe to the Tsar of Russia and you know the problems Putiin it into today's topics. A sort of Crime here or was there?

I had a really fun. It was how the Planet Venus was used to do the Megalithic Yard and it's almost the Vernal Equinox on which to do the calculations. Look it up- Prof Thom and  Knight and Lomas- University of Bradford

I'm not the only one

Norman

Sorry I wrote Hall and Knight- wrong =Knight wrote my School Algebra book

N

                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                   

[steampunkpete]

(Joking, but if wear is signifigant than there is no meaning to the words "making them parallel". )

David is presenting a hypothetical case of an assumed straight bed and unworn spindle, for fun consderation here, I think

I think that Mr. VTSteam is quite right again, and if you think about it, we aren't trying to align the spindle axis with the bed of the lathe, we are trying to align it parallel to the locus of some point on the saddle, so there is no real reason why we shouldn't move the saddle.

The first thing is to assume that moving the saddle is the first step to a quick way to a Moonbeam from a Lesser Lunacy.

I'm not certain what is intended by this, and I would welcome an explanation as it might help me to understand.

I'm not sure why there is a consensus on using a chuck rather than a collet. My experience is that collets are more accurate (although they can't be adjusted in the same way as an independent jaw chuck) so what is the advantage of starting the job with a chuck?

[philf]

I know how I would do it.

Dave,

Are you brave enough to tell us yet?

Cheers.

Phil.

[Fergus OMore]

Actually, you can do it with a child's lazer pen- hence my Moonbeams

Norman

[AdeV]

Actually, you can do it with a child's lazer pen- hence my Moonbeams

First, catch your child....

[DavidA]

What I did is fit my faceplate then take a very light skip across to the centre.

(That's why I excluded the use of a faceplate in the hypothetical case above)

I then placed a straight edge right cross the plate.
Only if the headstock was square  to the bed would there be no gaps at any point.

If the chuck end of the headstock was toward me then I would get a concave cone,  if it was away from me I'd get convex cone.

Of course, this won't do much good it your chuck back plate isn't square. So I suppose if you treated that the same way then everything should fall into place.

Then you can put a bar in the chuck and do the tests mentioned above.

The truth of the back plate and how well it fits the register and the thread seems to be critical to all this.

Dave.

[vtsteam]

Assuming the saddle slide is square to the ways.

All these problems can come up in building a new lathe.

[Will_D]

Surely there must be a logical approach to all this?

Like:

1. Check the headstock play.
     Fit big chuck, Long length of bar sticking out (unsupported)
     Clock it in the two axes and see how much play there is Y axis (cross slide feed) and Z axis ( Up and down)

2. Check any end float in headstock (X axis): clock on end of bar, use a lever on the tool post and back of chuck

When these have been measured and corrected then move onto the saddle and lathe bed alignment checks:

This is as detailed in the Myford set up instructions and uses a machinable test bar (unsupported so use alloy) and take some very fine cuts about 10 " apart. Adust sadle and bed supports until both diameters are the same.

3. Next is the tailstock!

[DavidA]

VT,

...Assuming the saddle slide is square to the ways...

That's the easiest bit.

Stand your clock on the bed and traverse the cross slide from side to side.

Maybe clamp a good straight edge to the cross slide table first and clock against that.

My own cross slide shows less than a tenth of a thou' over it's five inch of travel. That is my reference for everything else.

Dave.

[vtsteam]

Sorry David, I don't get it....from your word description maybe.


Clock on slide maybe, against square on bed.

[Will_D]

The saddle has to be square to the bed! It just slides along the reference bedway (either the "v" on the posh lathes or the bed edge (like on ML7s).

What may not be square is the cross slide ways in respect to the bed. This at the moment is not important as we are concentrating on the headstock alignment and making sire there is no twist in the bed.

I have heard that the cross slide is deliberately not set at exactly 90 degrees to the bed! That way when you face a large piece the centre of the work is concave by a tiny amount. Concave is better than convex when facing!

[vtsteam]

will, I did say the slide. And it definitely is a concern in new lathe construction, also as I said. And wear can alter the relationship.

[DavidA]

Will,

I didn't mention the saddle.  Clearly that follows the bed ways and it is an imaginary line down the middle of the bedways that we are trying to get everything lied up with.

If the cross ways are not dead square to the bedway then you are screwed right from the start. You may be able to line the headstock (and tailstock) with the bed, but will never get a square facing cut across any job as the tool will be traveling at an angle to the idealised centre line of the bed.

VT,

What I mean place your dial indicator on one of the bedways (or maybe on a bar clamped across the bed) then clock against a straight edge that is clamped to the cross slide.

If the cross slide is square to the bed than you will be able to find a position of zero dial movement when you move the cross slide in and out.

If you can't fins this point then it is either wear or you are measuring a slight angle in the cross slide ways.

I also have heard this about the cross slide deliberately being left out of true. Sounds rather odd to me. According to that theory you an never cut a truly flat surface.

Dave

[philf]

VT,

...Assuming the saddle slide is square to the ways...

That's the easiest bit.

Stand your clock on the bed and traverse the cross slide from side to side.

Maybe clamp a good straight edge to the cross slide table first and clock against that.

My own cross slide shows less than a tenth of a thou' over it's five inch of travel. That is my reference for everything else.

Dave.

Dave,

I have to disagree.

The cross slide could be 15 degrees out of square to the bed and a clock fixed to the bed would only show the out of parallelism between the cross slide dovetail and the side of the cross slide (which may be zero).

Taking a cut on the faceplate isn't the 'approved' way of doing things. According to Connolly or Schlesinger (which Norman quoted as the bibles) the method is to clamp a substantial parallel lightly in a 4 jaw chuck. With a dial mounted from the cross slide take a reading with the gauge at one end of the horizontal parallel. Without moving the cross slide rotate the chuck through 180 degrees and take another dial reading. Keep repeating this, lightly tapping the parallel, until you get identical readings at both ends. Then you can move the cross slide scanning the horizontal parallel to get an out of square reading.

You still don't know however if it's the cross slide or headstock which is out of square.

There are many checks and adjustments to make before the above. The first being to level the bed to ensure there is no twist. After all (as Will pointed out) a Myford is adjusted to turn parallel by twisting the bed (or perhaps more accurately untwisting it).

Cheers.

Phil.

[philf]

I have heard that the cross slide is deliberately not set at exactly 90 degrees to the bed! That way when you face a large piece the centre of the work is concave by a tiny amount. Concave is better than convex when facing!

Will,

There was a detailed discussion on this subject some time ago:

http://madmodder.net/index.php/topic,7066.msg75763.html#msg75763

Phil.

[DavidA]

Phil,

You're correct, I described it the wrong way around.

You use a square to position a straight (preferably a parallel) edge across the bed. Then fix the clock to the cross slide and traverse it across.
If there is an error then the clock will move either towards or away from the fixed bar.

Sorry about that. Old age setting in.

Dave.

[vtsteam]

Like I said.....
 

[Fergus OMore]

Centrecam anyone? It had only one fault and that was computer programme.

The rest was a poor imitation of what professional car body repairers were doing with laser alignment.  Car chassis were being made with Off Centre alignment- not straight lines as of yore. Anyway, I did a bit of this- hence my interest in Mike Trethewey's Centrecam and also Peter Rawlinson's excellent articles in ME.

It was- and is a lot, lot simpler than first believed.

Norman

[philf]

A few days ago there was a discussion about testing the parallelism of a lathe spindle to the bed.

Fergus mentioned two books by Schlesinger & Connelly (which I'm very fortunate to have hard copies of). I've just found pdf copies of the two books as well as two more very interesting looking machine tool testing books on: http://totallyscrewedmachineshop.com/documents/documents.html.

The Connelly book gives a huge amount of info on scraping and correcting all sorts of errors on many different types of machine tools. The Schlesinger book is more about testing than remedy.

I thought Vtsteam might find them of help with his 'New Lathe' project.

Phil.

[vtsteam]

Thanks Phil! Yes I will. 

[Fergus OMore]

Thanks Philf!

Two( I think) points arise. When these worthies did their thing, there was no such thing as easily available and cheap laser stuff.
Somewhere in my laughingly called workshop are two or is it three lasers. One is a a plastic affair which is almost a baby rotary table with the ability to send a beam at an angle.  I bought it to supplement my machinist's spirit level whilst aligning my lathe. OK, it somewhare with cheap surveyor's laser level etc.

So going back to CentreCam, it was in its infancy. Trethewey WAS going to bring out a new program or update it- but hasn't.  Peter Rawlinson wrote a lot about lasers in ME or MEW but I forget which. However, he sort of ran into problems of getting the laser pinpoint- to be a pin point.
OK, I'm rambling a bit but another issue is that cars 30 years ago were being re-aligned after collisions with lasers going through clear plastic graticules and being turned- with prisms.
It is 30 or so years since I was being taught as a manure student by Alan Robinson who wrote the Repair of Vehicle Bodies- I had the first or second edition at Gateshead Tech. Of course, this was all part of City and Guilds qualification.

OK- can all this be modified to a cheap alignment system for --us?


Phew!

Norman

[Fergus OMore]

Just unearthed these laser things.

One is Strait-Line laser level 120 and the other is a Laser Straight. Both switched on after yonks( Geordie- meaning for ever) but I am getting straight lines. Any bright( oops) ideas to tuurn them into a round pinhole, please?

Norman