Facing parallel between centres.

[HopperParticipant @hopper]

You win. You get the last word in. No. Wait. I just did that.

[Pete RimmerParticipant @peterimmer30576]

Posted by Hopper on 13/09/2020 09:51:00:

Posted by Pete Rimmer on 13/09/2020 08:44:36:

The faces will not be parallel but the ends of the part will be co-planar which is the desired result.

Both faces will be convex. Call it what you like. Try perching your mill column on top of a disc with convex faces top and bottom and see if it acts like a ball joint or not.

I missed this reply previously.

Firstly, the faces might be square, or convex, or concave but convex is not guaranteed. That depends on which was your setup is biased.

For a riser you would not machine the whole face flat but a bearing face on the periphery with the centre relieved – just like any quality cylindrical square. To do otherwise would be inviting trouble from dings, burrs or trapped debris on the face – including any 'convex' issues – during assembly or at any time in the future. These are all negated with a relieved centre section and if you do it, any tiny amount of cone on the end faces (and it will be tiny) is irrelevant for the application.

What is most difficult is figuring out where Robin is measuring his discrepancy. He states "about a thou and a half on 4" diameter". I suspect he means 'over 4" length' but if it's diameter he's measuring it's not really relevant except to identify that his centres need adjusting by 7-8 tenths.

[BazParticipant @baz89810]

In my humble opinion the only way to get both faces parallel is to machine both ends at one setting. Hold the workpiece on a mandrel between centres, face the tailstock end and then without moving the work face the headstock end, failing that find a friendly engineering company who will kiss the ends over on their surface grinder, a few beer tokens should seal the deal.

[OldironParticipant @oldiron]

Maybe I missed something here. Why not get the faces parallel on the mill after turning the diameter and facing one end on the lathe ? All it needs is a through hole and counterbore for a centre hold down bolt on the rough end.

regards

[HopperParticipant @hopper]

Posted by Pete Rimmer on 13/09/2020 11:14:18:

Posted by Hopper on 13/09/2020 09:51:00:

Posted by Pete Rimmer on 13/09/2020 08:44:36:

The faces will not be parallel but the ends of the part will be co-planar which is the desired result.

Both faces will be convex. Call it what you like. Try perching your mill column on top of a disc with convex faces top and bottom and see if it acts like a ball joint or not.

I missed this reply previously.

Firstly, the faces might be square, or convex, or concave but convex is not guaranteed. That depends on which was your setup is biased.

I was referring specifically to the diagram I had posted above. Which was in response to Robin's original post saying he could not visualise how offset tailstock could make the end faces be out of parallel. It was purely an exaggerated example to show visually how it all works.Sorry if we seem to be talking at cross purposes here.

 

Edited By Hopper on 13/09/2020 11:44:16

[HopperParticipant @hopper]

Posted by Oldiron on 13/09/2020 11:26:38:

Maybe I missed something here. Why not get the faces parallel on the mill after turning the diameter and facing one end on the lathe ? All it needs is a through hole and counterbore for a centre hold down bolt on the rough end.

regards

LOL. Yes you missed the previous two or three threads on this job. He got it within half a thou across 4 inches but now wants to get to better than that if possible, You'd be lucky to get better on most mills or hobby lathes IMHO. Baz is right where he says above that grinding is the solution for precision of sub-thou increments.

[Pete RimmerParticipant @peterimmer30576]

Ok a quick calculation. Let's suppose that Robin's part WAS turned between centres and has 1.5 thou taper over 4" – that's 8 tenths off-set for the centres. That would put the end faces off-square by 0.011 degrees, or, nothing in the real world.

If you faced the 6" diameter part and didn't relieve the centre you'll have just over half a thou convex or concavity on the end (actually 0.00052&quot over the full face but if you relieved the middle 4" of that 6" face to leave a 1" annular land that 'concavity' is reduced to 0.00013" over the width of the land. In other words it will blue up almost perfectly.

Who would argue that a face isn't flat when it blues up on a surface plate?

[Pete RimmerParticipant @peterimmer30576]

Posted by Hopper on 13/09/2020 11:47:59:

LOL. Yes you missed the previous two or three threads on this job. He got it within half a thou across 4 inches but now wants to get to better than that if possible, You'd be lucky to get better on most mills or hobby lathes IMHO. Baz is right where he says above that grinding is the solution for precision of sub-thou increments.

I'd scrape that part parallel in no time.

[blowlampParticipant @blowlamp]

Assuming the part has been fully roughed out and now only needs the finishing cuts, this would be my checklist:

 

1/ Ensure the lathe is facing square & flat:

Do a light facing cut on some (scrap maybe?) material mounted in a chuck and place a straight edge over the face to ensure it's been machined flat or slightly hollow. This is desirable to prevent rocking when assembled to the milling machine.

 

2/ Ensure the lathe cuts parallel between centres:

Machine the length of some bar (between centres) to see if both ends come out at the same diameter. If not, then adjust the tailstock until they do.

Note. If your facing cut gave a slightly humped result, you could compensate this out by setting the tailstock over towards the operator to make the lathe cut a slightly smaller diameter at the tailstock end. Doing this will tend to make facing cuts hollow too.

 

Now the lathe is set up properly, mount the job between centres. Face & turn to diameter (at the tailstock end only), then swap the job end for end and repeat.

 

 

Martin.

Edited By blowlamp on 13/09/2020 12:18:57

Edited By blowlamp on 13/09/2020 12:19:44

[HopperParticipant @hopper]

Posted by Pete Rimmer on 13/09/2020 11:50:00:

Posted by Hopper on 13/09/2020 11:47:59:

LOL. Yes you missed the previous two or three threads on this job. He got it within half a thou across 4 inches but now wants to get to better than that if possible, You'd be lucky to get better on most mills or hobby lathes IMHO. Baz is right where he says above that grinding is the solution for precision of sub-thou increments.

I'd scrape that part parallel in no time.

I dont think Robin has your years of experience at scraping though.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Baz on 13/09/2020 11:23:36:

In my humble opinion the only way to get both faces parallel is to machine both ends at one setting. Hold the workpiece on a mandrel between centres, face the tailstock end and then without moving the work face the headstock end […]]

.

Unless the lathe produces faces that are perfectly square to the axis … that would result in one convex and one concave face.

Manufacturers generally accept that perfection is not achievable; so they set the facing to be very slightly concave

MichaelG.

[Graham MeekParticipant @grahammeek88282]

If you look back to the post of the exaggerated tailstock set-over drawing previously. It will be noticed that the previously square base of the cone nearest to the tailstock is now at an angle to the plane of the cross-slide movement.

A cut taken across this face will as one poster pointed out produce a conical face. As the tailstock is brought closer to the true lathe centre line this conical face becomes less conical.

There comes a position when a point is reached that this face becomes truly flat. Beyond this point and moving the tailstock towards the operator the face will again go conical but in the opposite sense, Concavity.

Any one who has had to produce a flat face on a component between centres on a Cylindrical Grinder will be familiar with this technique. I hasten to add that the process is reversed on a Cylindrical Grinder, as the cutting edge in this instance is the edge,or corner of a disc, the grinding wheel.

This was the reason for my original advise for doing this job between centres. As the lathes in-ability to face flat can be worked around by altering the tailstock setting. I wrongly made the assumption the Originator knew this. Plus when setting the tailstock up to do such a job, eye balling the centres is not good enough. Time to get out the Verdict clock and make sure, by clocking the taper socket in the tailstock.

A Tip, Jubilee Clips make good driving devices for large components where light cuts are being taken, especially when cylindrical grinding. The out of balance can be minimal, and by using two Jubilee clips opposite one another the part can be brought more into balance.

Regards

Gray,

[Howard LewisParticipant @howardlewis46836]

Unless I misunderstand the situation, the objective is to produce a riser for a turret mill.

In my mind the parallelism of the OD is far less important than having the end faces parallel.

If they are not parallel, the now raised column will be at a slight angle, rather than perpendicular, so that any cut made with an End or Face Mill, wlil be slightly concave..

Possibly over simplifying things, I would take a very light skim across the Faceplate to ensure that there is a datum surface hopefully square to the lathe axis.

Then clamp the work to the face plate, whilst pressed against it by the Tailstock Centre, before taking the lightest possible cut to clean up the one end.

Then change the workpiece, end for end and clamp up again, in the same way.

Take the lightest possible cut to clean up the second end.

This should, in theory, result in both ends being parallel, .

You can't keep taking cuts off the ends, or you will finish with a large diameter washer!

Howard

[Martin KyteParticipant @martinkyte99762]

With the workpiece rotating between centres the lathe tool tip will trace out a surface which can be flat, convex or concave. Swapping the work end for end will produce an identical surface as neither the centres or the lathe settings have been touched.So if you end up with a perfect flat on one end (which is unlikely) you will get flat surface in the other end and in this case and only this case will the two be parallel on the basis that only flat surfaces can be parallel.

Moving the saddle down the bed and changing the tool to cut the underside of the first surface will produce the inverse of the first surface. If the first surface was flat the second will be flat (Hoorah, but unlikely). If the first surface was concave the second will be convex.

For risers or cylindrical squares we don't need parallel surfaces but two extremely flat coaxial cones tip to tip such that the tip centres are fractionally closer together than the peripheries. As has been mentioned its the peripheries that are parallel not the faces which need to be slightly hollow.

Graham has explained the adjustment of the lathe centres to approch flatness from the concave condition. Aim for nearly but not quite flat tending to concave. Swap the work end to end and do the other face.

regards Martin

Edited By Martin Kyte on 13/09/2020 15:30:07

[BazParticipant @baz89810]

I think you all need reminding that this is a riser for a cheap as chips Chinese / Taiwan produced mill, we are getting close to getting the optical flats out here and discussing how flat it should be to how many light bands, remember where this bit goes and the accuracy of manufacturing, or lack of accuracy in the case of these machines, and yes I talk from experience, I own a Myford VMC, and I can assure you that it wins no prizes for build accuracy, it is good enough to be called a milling machine and it mills to milling machine accuracy, it seems to me that model engineers buy a lathe or mill and think they can split atoms on it, you can’t, in industry if you want better than a couple of thou you grind, Thank you to Graham who has explained how to adjust a tailstock.

[egaParticipant @ega]

Could a riser with a small out of parallel actually be used to correct an inaccurate machine?

[JasonBModerator @jasonb]

Indeed it could, with the high point front or back it could correct nod issues r placed at the side tram, though with a swivel head I'm not sure if it's worth worrying about as you would tram head to table anyway.

[MadMikeParticipant @madmike]

Well I have read and re-read this entire thread. What machine are you turning the riser on? Is it a new and calibrated tool room lathe? Or a Chinese hobby lathe or maybe a Myford?

Before all the discussion about concave faces, turning between centres etc etc it would be reasonable to ask if the face plate to which the riser was bolted has been clocked to determine if it has no run out. Then of course the inevitable question has been missed………just how accurate do you expect it to be and what tolerences are you working to?

I am amazed that as "engineers" nobody has even attempted to get to the bottom of the component tolerancing and the machine and even the machinists capability.

[Robin GrahamParticipant @robingraham42208]

Many thanks for replies.

Looking at Hopper's diagram I see where I went wrong in my thinking – my mental model was that in an ideal world where the cross slide is perfectly perpendicular to the lathe axis it would have to cut a plane perpendicular to the lathe axis. Obviously (now) not true across the diameter – I did say I have problems with 3D visualisation! It's embarrassing as I know full well that the lathe faces slightly concave under the best conditions and understand the how and why of that. I knew there was something suspicious about my reasoning, just couldn't decide what and missed the wood for the trees. I blame the guy who taught me about conic sections!

I finished the riser by bolting to the face plate with M8 studding though the spindle and tailstock support:

I'm happy, in the light of comments, that I got as good as half a thou. On a Far Eastern  lathe!

Robin

 

 

 

Edited By Robin Graham on 14/09/2020 00:31:51

Edited By Robin Graham on 14/09/2020 00:34:42

[JasonBModerator @jasonb]

Posted by MadMike on 13/09/2020 22:58:28:

Well I have read and re-read this entire thread. What machine are you turning the riser on? Is it a new and calibrated tool room lathe? Or a Chinese hobby lathe or maybe a Myford?

………………………………………………….

e component tolerencing and the machine and even the machinists capability.

A clue was the opening line "I've posted recently" a quick look at Robin's posts will see it started with this then this to this where lathe and accuracy of faceplate was established early on and the suggestion of doing it between ctrs was first raised, now it seems most are suggesting this way won't be accurate

 

 

Edited By JasonB on 14/09/2020 08:21:08

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Baz on 13/09/2020 16:10:46:

[…]

… in industry if you want better than a couple of thou you grind, …

.

But, in his home workshop, Robin has already done better than that … and is now exploring the limits.

Bravo !!

MichaelG.

[BazParticipant @baz89810]

Yes exactly, he got to about half a thou over six inch diameter four inches from the chuck, plenty good enough in my opinion.

[Nigel McBurney 1Participant @nigelmcburney1]

Best thing to come out of this discussion was the suggestion to use a jubilee clip as a driver where a conventional driver cannot be used. To turn a part to obtain good parallism use a substantial cast iron sub plate bolted to the faceplate and then lightly skimmed flat, I use a scrap rotary table casting then bolt the work to this subplate , for brake discs they are bolted to the subplate and then skimmed on both sides at one setting, face plates are a useful tool though they will distort if clamps are over tightened, or the work mounted to the plate is not flat, Face plates in practice cannot be made too thick or rigid as they would reduce the lathe gap clearance when used on general work.

[blowlampParticipant @blowlamp]

So am I correct in thinking that you started this thread after you had actually completed the job

Martin.

[Mike PooleParticipant @mikepoole82104]

I have had my squared paper out and exaggerated the centres being offset in both directions, if the saddle is set to an exaggerated look in it faces concave with both offsets, turning end for end gives me parallel faces that are faced slightly concave, I don’t see how if the saddle is looking in how the centre position can effect it cutting anything but concave.

Mike

[Author]

Posts