PDA

View Full Version : Help getting parallel faces in a lathe



softtail
05-03-2010, 10:29 PM
I need to get a 3" dia, 4" long steel round to have parallel faces. I thought I nailed it pretty good, but on further inspection, somethings not right. I had it spinning via a nub in a 5c collet supported by a dead center on the other end, did both faces and the diameter (just for good measure) in one set up. The light finish cuts to the faces were done after taking the skin off the od. Both faces were finished with a very light power feed cut, carriage locked. The diameter doesn't matter, just that the faces are parallel to eachother.

Ideas what could cause this and how to get it right?

Thanks,


st

Dr Stan
05-03-2010, 10:43 PM
Try doing it between centers.

Tony Ennis
05-03-2010, 10:45 PM
What's not right? Are the faces not flat? Are they not parallel? How far off are they? What's your acceptable error?

J Tiers
05-03-2010, 11:05 PM
If you have not got the two centers accurately lined up, you may find that you have made two cones..... which may be parallel, but not at all what you had set off to do.

Check to see if the turned OD has given identical diameters at both ends.

If you don't need the entire face, maybe you can undercut the ends so that just a ring is faced, which will be easier to get "acceptably perfect".

And you must do something to deal with the portion next to the centers, which you can't reach with the tool without hitting the centers. So undercutting the end for that area will allow you a little leeway there.

tmc_31
05-03-2010, 11:32 PM
Softtail,

What kind of steel are you using? I know that cold rolled steel has internal stresses that are released when the "skin" or outer surface of the steel is removed causing the steel to squirm. Is it possible that this is causing your faces to be off? You might try some hot rolled.

Tim

Jim Shaper
05-03-2010, 11:55 PM
Edit: scratch what I originally wrote (if anyone saw it).

Put a 3 jaw in there and face one side, flip it, then put it against parallels against the face of the chuck, remove parallels after clamping and face it again.

Carld
05-04-2010, 12:38 AM
Use a four jaw chuck and since you have the OD turned you can chuck up an inch in the four jaw and then indicate both ends to "0" and face one end. Then flip it over and indicate both ends in again and do the other end.

You have to use a four jaw because you will never get it indicated to "0" in a 3 jaw chuck. It's important to get the end at the chuck and the free end running as close to "0" as you can possibly get it to have the faced ends parallel.

This is a time consuming job and nothing short of near perfection will do but remember, you will not get it perfect by machining it.

Paul Alciatore
05-04-2010, 01:21 AM
You didn't say what's wrong. Are the faces not parallel? Or are they conical? Or what?

From what you describe, I would expect two faces that are parallel, but each slightly conical. The face on the tail stock end should be slightly concave and the one at the headstock end should be slightly convex. This is due to the fact that no lathe can have the cross slide exactly perpendiular to the main ways and all manufacturers, well most anyway, will make sure that the error is towards a concave face where facing is normally performed, on the tail stock end. By "slightly" I mean perhaps 0.0005" or perhaps 0.001" for a part your size. If that is your problem, you will have to turn it around to face the second side.

I am doing a similar job at the present. My spacers are only one inch in diameter and vary from 1/4" to 1" long. They have a center hole which also must be concentric so a stud is out of the question. I am making about ten or so. Chasing tenths is a b***h. I'm using a shop made "collet" that grips the 1" OD. It was turned on the lathe's spindle with a flat face for the parts to seat against that is dead true (because it was turned in place). The gripping area is only 1/8" deep but it holds well enough for a light facing cut. I have to be sure that it is dead clean when mounting each part.

Probably the best way to do a piece like this would be to use a custom face plate. Give it a final truing cut when it is mounted. You can use a three or four jaw for the OD and the first face with no problems. You could rough out the second face the same way, leaving 0.010" or so for the final cut. The part is held with the first face against the face plate with a threaded rod through the spindle or with several bolts in a circle through the face plate. In either case, you will need one or more tapped holes in the part. Or a center bore and a nut in a recess. Use whatever is best for your part. Take light cuts.

This should insure that the two faces are dead parallel and also that BOTH are slightly concave.

If you need a better finish on the ends, you can use abrasives to lap them. But be careful to keep the faces parallel. If you do this under power on the lathe, it should be a very even process which would keep the faces parallel, but they will tend to become convex in the process. If you do it by hand it will be easier to keep them flat, but they may stray from parallel and frequent checks will be needed.

Of course, a surface grinder would make fast and accurate work of it. And no concave/convex. But I don't have one either.

Black_Moons
05-04-2010, 01:37 AM
You can allways face with your compound set to 90 degrees.
of course, Parallel implys 0.000" error.
Everything has error, so its more a matter of how long you wish to tweak and adjust and measure.

softtail
05-04-2010, 12:38 PM
A thousand ways to skin a cat it seems.

I may have gotten it as good as I can. Sounds like a surface grinder is the only way to get it dead nuts.

It's pretty damn parallel. I'm using it as a 'whipping post' mounted on a surface plate, welded parts are then mounted and measured off that to check alignment, some points are 3' away, so error is magnified.

As far as I can tell, the faces are flat (one face on the table, other has part mounted to it, bolts through each end), just not parallel. I can take a parallel and see a slight gap at the top of the post on the side, slight gap at the bottom of the side when swapped 180, so the sides arent' perp to the table, which in itself doesn't matter, but in this case tells me the faces aren't right either, and I 'swept' the table with a long tube mounted on the post, indexing the post 1/4 turns, taking readings at 3 areas of the table,then back to it's original position, it absolutely points to the two faces being out of parallel. The first and last position of the post were the same, and the difference of the combined variance was 0.00078" , which shows good repeatability.. but the three readings themselves, taken at 3 different areas of the table for a given post position varied as much as 0.179, so I was able to get the same crap readings for a given 'index' of the post. Better than getting totally different crap readings, but I need this thing to read the same across the table.

Can a surface grinder handle a part 4" high? May just take it down to the local shop and have the guy throw it on the mag chuck. I assumed I would be able to get it 100%. I have a digital height gauge, and can't pick up any variance with the post mounted on the table, and taking measurements around the upper face.


st

ps, cold rolled, don't know what type. Machined really nice. I've heard of the stresses before that's why I took of a couple od passes, and then finished the faces, which were already rough faced. I did drill and tap it after facing. Could that screw things up?

Tony Ennis
05-04-2010, 02:34 PM
I did drill and tap it after facing. Could that screw things up?

You betcha. Are you sure you didn't raise a .0001 burr?

softtail
05-04-2010, 03:10 PM
You betcha. Are you sure you didn't raise a .0001 burr?
Should have noted, I also faced off around the hole (counter faced?), to avoid any burs, etc coming in contact with the table. On the top face, it doesn't matter.

st

Tony Pratt
05-04-2010, 03:10 PM
4" high on a surface grinder is not a problem to someone who knows what they are doing. Just throwing it on a mag table will not get a truly parallel job unless you are very lucky, you will need to tell the machinist how good you want it and he can then decide how much time and your money will be required to achieve this.
Tony

softtail
05-04-2010, 03:54 PM
I'll go out when the rugrat takes a nap and take proper measurements.

It seems to me, turning all at once in a 5c spindle mounted collet with a dead center on the other end should have produced two parallel faces, and perpendicular sides. My gut says using a 3 or 4 jaw and swapping ends would get me worse results. The faceplate idea sounds like it would work, but conceptually it doesn't strike me as any more accurate or foolproof than a 5c collet. But hey, I reserve the right to wrong!


st

Tony Ennis
05-04-2010, 04:22 PM
I reserve the right to [be] wrong!

That attitude is critical for achieving long term success.

softtail
05-04-2010, 06:25 PM
Ok here's the measurements, taken with a B&S dti reading .0005

Wish I had a 4" mic.

-There's a taper to the od of 0.0015 along it's 4" length, when placed in a v block and a height gauge used.. wider at the tailstock end. Again this doesn't matter in and of itself for my purposes, but it probably speaks to other problems I guess it would technically be half that number.

-Both faces dip towards the center .00025 to .0005. This would not be a huge deal either as long as any 'dip' was the same across the face and didn't affect parallelism. Would like to know why it's there though.

-The max difference I got around both faces was a bit over .0005. Of course the post was resting on one or the other of the faces, so if one is 'good' and the other 'bad' I have no way of knowing.

Would the tailstock mis-alignment throw of the facing?

I feel like these tolerances aren't that bad, what do you guys think? No good for this project though.. maybe over my head.

st

clutch
05-04-2010, 06:34 PM
Taper might be tailstock set over or ways that are droopy.

What is the condition of your lathes ways?

Can you turn a 6" long round supported by a tail stock center and get same diameter at both ends? I'd use at least 1" dia stock.

Clutch

softtail
05-04-2010, 06:40 PM
Taper might be tailstock set over or ways that are droopy.

What is the condition of your lathes ways?

Can you turn a 6" long round supported by a tail stock center and get same diameter at both ends? I'd use at least 1" dia stock.

Clutch

Would that affect getting parallel faces?


No obvious signs of wear....Would that amount of taper be considered high? Is zero taper attainable?

st

Carld
05-04-2010, 07:43 PM
The "dip to the center" your seeing is correct for a lathe face cut. A lathe is set to face convex rather than flat or convex. That is how they are designed.

What you can do is set the shaft on one end on a surface plate or something you know is very flat and use a dial indicator on a mag base with the mag turned off and slide it around the top end. That should tell you how much out of parallel they are.

softtail
05-04-2010, 08:04 PM
The "dip to the center" your seeing is correct for a lathe face cut. A lathe is set to face convex rather than flat or convex. That is how they are designed.

What you can do is set the shaft on one end on a surface plate or something you know is very flat and use a dial indicator on a mag base with the mag turned off and slide it around the top end. That should tell you how much out of parallel they are.

That's what I did if I'm understanding you. One end of shaft on the surface plate, and then swept the other with a dti on a stand. Curious if I nailed one face and the other one is suspect or if they are equally to blame. If the sides were straight, I'd clamp it in a v block, stand it on one end so the shaft is suspended, and sweep each end to figure it out. The side taper kills that idea. It's here nor there I suppose.. the thing could be parallelogram for all I care.

Concave is the way it should face then? I can live with that as long as it's concentric and parallel.

st

JCHannum
05-04-2010, 08:37 PM
I am not sure I follow what you are doing altogether, but when making a cylinder square, the center portion of the face is undercut leaving only a narrow section actually in contact with the surface plate, say about a 1/4" wide raised area on the OD of the face. This minimizes the effect of the lathe's tendency to produce a concave face.

reggie_obe
05-04-2010, 08:38 PM
You're trying to make a cylindrical square, are you not? Something like this: https://www.subtool.com/tp/9141-43_taft-peirce_cylinder_squares.html You don't need or want it to faced completely flat across both ends. Only a outer ring of metal needs to be left on either end to contact the surface plate. This was described completely in an old issue of Project in Metal. Maybe someone here can find an online copy of the article.

Arcane
05-04-2010, 09:08 PM
Why don't you try lapping the ends flat and then work at lapping them parallel to each other?

softtail
05-04-2010, 09:48 PM
reggie_obe, yeah, that's basically it, with all thread out each end.

JCHannum, got it. I did take away some material towards the center, but not that much. If it's concave, it seems like there would only be the outer edges touching anyways.

I'm still not be where I need to be, because it's out of parallel a smidge over .0005 measured at the outer 1/4", which gets magnified greatly over the 3 ft span I need to cover.

st

Paul Alciatore
05-04-2010, 10:57 PM
I'll go out when the rugrat takes a nap and take proper measurements.

It seems to me, turning all at once in a 5c spindle mounted collet with a dead center on the other end should have produced two parallel faces, and perpendicular sides. My gut says using a 3 or 4 jaw and swapping ends would get me worse results. The faceplate idea sounds like it would work, but conceptually it doesn't strike me as any more accurate or foolproof than a 5c collet. But hey, I reserve the right to wrong!


st

Well, the collet and the collet closer will EACH have some runout and there is little you can do about it. It may or may not make a difference in your method above as you did turn all three surfaces in one setup. However, if you face the face plate in place on the spindle, then it should be dead nuts. Or at least as good as you will ever get.

What it does is allow you to face both ends concave. And I just thought of another advantage, you will be facing both ends at just about the same location on the ways so if there is any wear there, it will be the same for both. When you are going for tenths, everything counts.

Actually, it is hard to see why either procedure would produce unparallel faces.

You didn't say what kind of steel. Could be there are internal stresses that are being released as you machine it, causing distortions as it settles. You might try putting it in the oven for an hour or so at the highese setting and then turning the temp down in degrees over a two or three hour period until it cools to room temp. Then try again.

softtail
05-04-2010, 11:06 PM
Well, the collet and the collet closer will EACH have some runout and there is little you can do about it. It may or may not make a difference in your method above as you did turn all three surfaces in one setup. However, if you face the face plate in place on the spindle, then it should be dead nuts. Or at least as good as you will ever get.

What it does is allow you to face both ends concave. And I just thought of another advantage, you will be facing both ends at just about the same location on the ways so if there is any wear there, it will be the same for both. When you are going for tenths, everything counts.

Actually, it is hard to see why either procedure would produce unparallel faces.

You didn't say what kind of steel. Could be there are internal stresses that are being released as you machine it, causing distortions as it settles. You might try putting it in the oven for an hour or so at the highese setting and then turning the temp down in degrees over a two or three hour period until it cools to room temp. Then try again.

I drilled and taped it after facing... 3/4-16. Maybe that made it wonky. It's cold rolled 1018 or something.. I'll dig up my invoice...

st

J Tiers
05-04-2010, 11:31 PM
Whoa, there............!

3" diameter, to set a 3 foot piece on...... I see some issues

1) the multiplication effect from that is obvious, depending on whether the piece is centered of offset...

2) the piece will presumably sag somewhat (everything being made of jello, as we know), so that may affect apparent alignment

3) the reason you MAY want more than a narrow ring on top is that otherwise you may pull the part down into the recess, warping it a bit.

As for the taper..... that shows the centers are not aligned, OR the lathe cuts a taper for some other reason.

IF it cuts a taper, then one end should be coned convex, the other coned concave. The natural tendency of the lathe to cut the tailstock face of the work concave will exaggerate one end, and increase the coning at the other.

I don't "get" how BOTH ends are cut concave...... Not if both faced without taking the piece out of the centers...
What SHOULD have happened is that even if perfectly aligned, the tailstock end should have been cut slightly concave and the headstock end should have been cut convex the same amount.

winchman
05-05-2010, 07:17 AM
Here's a fixture I made to get both ends of 4" lengths of 2" thin-wall tubing parallel:
http://img.photobucket.com/albums/1003/winchman/000_13112.jpg

http://img.photobucket.com/albums/1003/winchman/000_13122.jpg

The plastic liners were turned with the fixture in the chuck, so the bore is concentric with the spindle axis. The fixture is marked to always go back in the chuck in the same position to eliminate chucking error. I use a piece of all-thread through the spindle to hold the fixture solidly against the chuck.

You might need something a little heavier to do your job on a piece of steel, but the concept will work.

softtail
05-05-2010, 09:21 AM
Whoa, there............!

3" diameter, to set a 3 foot piece on...... I see some issues

1) the multiplication effect from that is obvious, depending on whether the piece is centered of offset...

2) the piece will presumably sag somewhat (everything being made of jello, as we know), so that may affect apparent alignment

3) the reason you MAY want more than a narrow ring on top is that otherwise you may pull the part down into the recess, warping it a bit.

As for the taper..... that shows the centers are not aligned, OR the lathe cuts a taper for some other reason.

IF it cuts a taper, then one end should be coned convex, the other coned concave. The natural tendency of the lathe to cut the tailstock face of the work concave will exaggerate one end, and increase the coning at the other.

I don't "get" how BOTH ends are cut concave...... Not if both faced without taking the piece out of the centers...
What SHOULD have happened is that even if perfectly aligned, the tailstock end should have been cut slightly concave and the headstock end should have been cut convex the same amount.

Yes, agreed. I will ring only the face that sits on the table. In practice the part (bike frame) gets supported by a small bottle jack deal with a somewhat softer tip (hard palstic, ect) at it's farthest point from the post mount on the alignment table.

But for checking how the post came out, letting the part sag is fine as it will sag to the same spot within reason. The readings were still all over the board.


Both ends being concave seems wrong to me too (if done in one setup), but one face, the upper face has a shoulder in the middle (shoulder specs don't matter, only the bottom face), as the part that lays on it is a tube about 1.5", so getting a handle on flatness is more difficult on the upper face than the lower.

st

metalmagpie
05-05-2010, 10:48 AM
If you do wind up borrowing access to a surface grinder, here's a trick for grinding 90 angles. (If you just grind both ends you will get ends flat within .0001" or so but will they be perpendicular to the sides?) For a cylindrical square this will only work on something that's truly cylindrical, but it's a good trick anyway.

Mount a DTI with a plunge-type actuator on a base that needn't be magnetic. Turn off the mag chuck. Slide your DTI base so when the base touches your workpiece you can read the indicator gage. You should verify your setup by sliding the DTI base away and then returning it solidly against the base, to see that your measurement is completely repeatable.

Now rotate your workpiece 180 on the mag chuck and repeat your measurement on the other side. If it reads the same, then your part is perpendicular. If not, then you will see the difference in readings. Shim your part appropriately and regrind top & bottom. Repeat the tests on both sides, regrinding as necessary until both sides measure the same. Now that part is ground with the sides truly square to the top & bottom, and also your DTI is set up for measuring perpendicularity.

metalmagpie

softtail
05-05-2010, 12:12 PM
Glad were on the web 'cuz I'd get a swift kick in the a** for this one.

I got a some .020 feeler gauges out and noticed that the face of the tube I was loading into the post wasn't sitting flush on the upper post face. At first I thought maybe is was just binding tight ascew.

Then it hit me.. I made a top washer that goes between the nut and top of part clamped onto the post.. *both* the id of the washer that slips over the all thread and the shoulder on the washer that goes into the mounted part were a hand slip fit. The hole for the all thread needs to be opened up, b/c it was binding on the all thread and keeping the face of the part from seating flat on the upper suface of my post. Flipped the washer upside down (to get the shoulder out of the picture), and sure enough, I can sweep the whole table within .02 which I'm calling good enough for now.

Thanks for the tips.. good stuff in here. It did force me to look at the part closely and discover a number of areas that could be improved for better accuracy. In the end, the faces are not 100% parallel, but they are pretty damn good. I was looking at shouldered locating cylinders last night and they listed tolerances to .0002, so I knew getting too much better with my setup wasn't likley.


Thanks,

st