Straightening spindles :(

Thought I'd start a new thread on the latest issue to befall my attempts to get my 1924 Relmac lathe operational - how you guys do it with the really big stuff I do not know!

Anyway, having got it all oiled up and running and got the belt tension 'good enough' to turn something I was still concerned about the vibration. I thought this might be a bearing problem, but now I think the spindle is bent.. maybe something fell on it when it was in storage at the previous owner for 6y

I put a MT3 centre in the spindle and put the DTI on it... theres a runout just next to the spindle end of 0.45mm which I'm guessing is rather too much, and I doubt its the centre thats off centre so to speak...

The runout on the main drive pulley is 0.15mm and on the rear shaft out of the headstock (that drives the changewheels) its 0.18mm. I have no idea if these are reasonable but they cant be too far out, can they?

I put the chuck back on and put a short length of 3/4 MS in there and turned it down to 18mm nominal... was actually eccentric 17.99mm - 17.76mm to the best I could measure...

So... the big questions...

Is it likely to be bent?
What else can I do to verify this or not?
Would a worn bearing give the same result?
What should I do next? I suppose getting another spindle made up isn't out of the question, but could this one be trued up and if so, how?

Would a worn bearing give the same result?

Click to expand...

Worn bearing will definitely enable eccentric.

0.45mm is way too much run out. I would equate that to almost .018". 10% of that number would be too much in a good lathe.

I had a bent spindle in a 20" Pratt & Whitney lathe. I took it out and worked on straightening it for about a week. It was not very interested in getting straight again, but I made it happen.

Old plain bearing lathes are very apt to have worn out bearings due to previous owner ignorance about the necessity for all-the-time lubrication of the correct type.

It is easy to tell if bearings are worn - lever up the chuck with a bar and fulcrum and measure lift with dial indicator.

Also easy to tell if the spindle is tweaked. Just put it between two V-blocks on a
surface plate, and put a dial indicator at the center.

These can be straightened if some careful work with a hydraulic press is done.

Jim

Thanks John and Jim... either way it seems there's no option but to take the spindle out... if only I knew how it came out...

There's a rear bush retained by a grub screw that I suspect sets the end float, and the front of the headstock has a collar but no visible means of removing it so I think its integral to the spindle, The bullgear has no visible means of retention so I suspect its keyed to the spindle, while the pulleys/rear gear are free rotating on the spindle. I guess the spindle comes out forwards, so release the grub screw on the rear collar and bash it through with a soft mallet...

Jim:

Why do the two vee blocks have to be on a surface plate? Isn't it sufficient that they just be held firmly, like to a milling machine table? (Or the flat ways of the lathe bed?)

JRR

irving2008 said:

I put the chuck back on and put a short length of 3/4 MS in there and turned it down to 18mm nominal... was actually eccentric 17.99mm - 17.76mm to the best I could measure...

So... the big questions...

Is it likely to be bent?
What else can I do to verify this or not?
Would a worn bearing give the same result?
What should I do next? I suppose getting another spindle made up isn't out of the question, but could this one be trued up and if so, how?

Click to expand...

Your spindle could be bent like a banana, but if all else was OK you would not get an eccentric turning. By which I presume you mean out of round. The only way to get that is through something loose - bearings, chuck on spindle, workpiece in chuck.

John,

I dunno whereas a bent spindle wouldn't be the source of encentric turning. If you put a Morse test bar into a bent spindle, the end of the test bar sort of "scribes out a cone". Surely if you made a cut under that circumstance, it would not be on center with the original center of the test bar. Rather, the new center would be at the center of the cone that the bar was describing. (It's a lot easier to envision this than it is to write it out!)

Irving,

Sometimes you can smack a bent piece of steel shafting back straight by using a strong hardwood block with a vee groove in it. The hardwood block should be about the same length as the spindle. Put the bent shaft in the groove with the high spot up and marked with chalk or magic marker. SMACK the high spot with a heavy soft hammer or a steel hammer cushioned with a flat piece of wood. You must hit the high spot of the bend exactly "normal" to the surface and it must be a HARD smack. Sometimes a shaft seems to have a sort of memory and goes back to straight. I have only done this on shafts up to 1/4" dia and they were not hardened. If it doesn't work, then you can always go to the straightening press......

Good luck.

But, do by all means heed the advice to make SURE the spindle really is bent before you go to the trouble of dismantling it.

JRR

If I put a MT3 taper in the spindle and turn the spindle using the bullgear the point of the taper definitely describes a circle, small but distinct (esp with a jeweller's loupe to observe it). I've cleaned up the inner bore of the spindle as best as I can and the taper seems to fit cleanly and doesnt have any grime on it when removed and the oil film seems 'disturbed' all over so no obvious high or low spots (I know I should use some blueing but I dont have any)...

I know there is bearing play but I still can't move it more than a few thou by hand with the steel bar in there so I am rapidly coming to the conclusion that the spindle is distorted somehow.... only putting it on v-blocks is going to truely tell... so this weekend I'll have a stab at getting it out.

SouthBendModel34 said:

John,

I dunno whereas a bent spindle wouldn't be the source of encentric turning. If you put a Morse test bar into a bent spindle, the end of the test bar sort of "scribes out a cone". Surely if you made a cut under that circumstance, it would not be on center with the original center of the test bar. Rather, the new center would be at the center of the cone that the bar was describing. (It's a lot easier to envision this than it is to write it out!)


JRR

Click to expand...

I guess the first thing needed is a definition of eccentric. I took Irving to mean that the test workpiece he turned ended up not cylindrical, but oval or otherwise out of round.

He said he held the test piece in a chuck. Doesn't really matter, though - whether held in a chuck on a bent spindle, deliberately offset in a four jaw chuck, or held between centers that are not on center, the test piece he cut should have ended up round. Unless the bearings or spindle were loose or oddly worn, the spindle was flexing, the chuck was flexing on its mount, or the workpiece was moving in the chuck jaws. It won't necessarily be the same diameter over its length, but it should be perfectly round.

Something was loose.

I'd run an indicator in the spindle taper, both near the end and then as deep as he can get it. Providing that the taper is clean and smooth to begin with. The center he is using may have its point concentric with its taper shank, but who knows? He could also run an indicator on his center, remove it and shift it 90 degrees in the spindle and try it again. And again, and again. If he keeps track of the readings with respect to the spindle and the position of the center in the spindle, he should be able to tell whether it's the spindle or the center that's off.

You can put the V-blocks on anything. A bed is pretty
soft and doesn't work too well. A surface plate is a
large, stable flat surface that most shops have handy.

It lets you sweep the length of the spindle to see how
the bend develops and to spin the part in the blocks to
see where the max/min locations are.

It just makes it easy, that's all. You could do it on
a kitchen tabletop but it would take longer and would
be a fussy job.

Have a good dial indicator to the testing with.

But the idea is to have the setup all ready, go to the
press and bring up the pressure to some amount and
see how much it deflects. When you take the pressure
off it will return to the same deflection. Then you
increase the pressure for more deflection. Eventually
will exceed the elastic limit and the part gives. The
trick is to get the pressure right so when it deflects
back after 'giving' it is straight.

Jim

JohnMartin said:

I guess the first thing needed is a definition of eccentric. I took Irving to mean that the test workpiece he turned ended up not cylindrical, but oval or otherwise out of round.

He said he held the test piece in a chuck. Doesn't really matter, though - whether held in a chuck on a bent spindle, deliberately offset in a four jaw chuck, or held between centers that are not on center, the test piece he cut should have ended up round. Unless the bearings or spindle were loose or oddly worn, the spindle was flexing, the chuck was flexing on its mount, or the workpiece was moving in the chuck jaws. It won't necessarily be the same diameter over its length, but it should be perfectly round.

Something was loose.

I'd run an indicator in the spindle taper, both near the end and then as deep as he can get it. Providing that the taper is clean and smooth to begin with. The center he is using may have its point concentric with its taper shank, but who knows? He could also run an indicator on his center, remove it and shift it 90 degrees in the spindle and try it again. And again, and again. If he keeps track of the readings with respect to the spindle and the position of the center in the spindle, he should be able to tell whether it's the spindle or the center that's off.

Click to expand...

John, let me describe what I did in more detail.. firstly when I said eccentric I didn't mean off centre I meant oval as in one diameter bigger than its perpendicular. It may well be eccentric as well but I haven't measured that... it definitely NOT round...

Secondy I put a taper in the spindle, no chuck. I put the DTI on the taper right next to the end of the spindle. I get 0.45mm runout

Thirdly I put the chuck back on and put a 300mm length of 20mm round in there. I can DTi the length of that on the slide and its out by maybe 0.15mm along the length, (but then its not a parallel test bar!). If I bring the tailstock centre up to the bar its offcentre and describes a circle approx 3mm dia on the bar end as the spindle turns. The runout increases as you move along the bar far greater than the .15mm non-parallel of the bar itself 0.6mm at the heastock end to 1.5mm-ish at the tailstock end...

Basically since there is little movement on the front bearing and little runout on the rear spindle or the pulleys I don't think this is the spindle moving around in the bearing, i think it is bent from the bearing outwards as if something fell on it. Either way its got to come out...

I'll try your suggestion of measuring the runout inside the spindle taper, although the DTIl will have to be angled a little which wont give accuracy. Also of trying the taper in the spindle at various positions and plotting the runout v rotational position. If i do this for the rear spindle and pulley too (although they're both much smaller runouts) and plot them all together it might give clues about the spindle shape if there is a relationship between them.

irving2008 said:

John, let me describe what I did in more detail.. firstly when I said eccentric I didn't mean off centre I meant oval as in one diameter bigger than its perpendicular. It may well be eccentric as well but I haven't measured that... it definitely NOT round...

Secondy I put a taper in the spindle, no chuck. I put the DTI on the taper right next to the end of the spindle. I get 0.45mm runout

Thirdly I put the chuck back on and put a 300mm length of 20mm round in there. I can DTi the length of that on the slide and its out by maybe 0.15mm along the length, (but then its not a parallel test bar!). If I bring the tailstock centre up to the bar its offcentre and describes a circle approx 3mm dia on the bar end as the spindle turns. The runout increases as you move along the bar far greater than the .15mm non-parallel of the bar itself 0.6mm at the heastock end to 1.5mm-ish at the tailstock end...

Basically since there is little movement on the front bearing and little runout on the rear spindle or the pulleys I don't think this is the spindle moving around in the bearing, i think it is bent from the bearing outwards as if something fell on it. Either way its got to come out...

I'll try your suggestion of measuring the runout inside the spindle taper, although the DTIl will have to be angled a little which wont give accuracy. Also of trying the taper in the spindle at various positions and plotting the runout v rotational position. If i do this for the rear spindle and pulley too (although they're both much smaller runouts) and plot them all together it might give clues about the spindle shape if there is a relationship between them.

Click to expand...

Your meaning of eccentric is as I thought it was.

You may have several different problems. You may indeed have a bent spindle. But, once again, a bent spindle would not cause the eccentricity you are measuring. It could be bent like a banana but, unless there was something loose, you should be able to cut a perfectly round test piece. The test cut might not be concentric with your spindle bore, or your chuck mount, but it should be concentric with the axis or rotation through your bearings and it will be perfectly round. Something is loose.

You said that you had cleaned up the tapered bore of your spindle as best you could. It may be that it still has imperfections that are not allowing your center to seat properly. It may even be that you have the wrong center - one could be B&S instead of Morse. Check it with blue - use artist's cobalt blue oil color if you can't find the machinist's blue.

The DTI isn't meant to always be spot on accurate. It's meant to give you qualitative results, not quantitative. If in testing a bore the travel of the indicator tip is not exactly perpendicular to bore you may not know by exactly how much the bore is untrue, but you'll be close enough to know whether it's OK or not.

Your test with the 20mm round bar tells you nothing about a possibly bent spindle. That the tailstock center describes a 3mm circle on the end of the bar tells you that the tailstock is not in line with the axis of rotation through the headstock bearings. That's not your first test, though - it's your final one. Because it may be just a case of adjusting the tailstock.

The important test of the axis of rotation through the bearings is that it be in line with the bed ways and carriage movement. You can check this with your test bar even if it has some runout, by rotating the bar and averaging the readings. If you don't get the same average at the end of the bar as at the middle and at the chuck, you have either a misaligned headstock or a twisted or worn bed. I'd first get a better piece of shafting to use as a test bar, though. You could compensate for the taper of the bar, but it only complicates things.

There are some good writeups available on setting up a lathe, and I'd suggest you look at some. But - although none of your tests have shown it - I think that you have already decided that the spindle is bent and will not be happy until you have pulled it out and tested it, so perhaps that is what you should do first.

JohnMartin said:

Your meaning of eccentric is as I thought it was.

You may have several different problems. You may indeed have a bent spindle. But, once again, a bent spindle would not cause the eccentricity you are measuring. It could be bent like a banana but, unless there was something loose, you should be able to cut a perfectly round test piece. The test cut might not be concentric with your spindle bore, or your chuck mount, but it should be concentric with the axis or rotation through your bearings and it will be perfectly round. Something is loose.

You said that you had cleaned up the tapered bore of your spindle as best you could. It may be that it still has imperfections that are not allowing your center to seat properly. It may even be that you have the wrong center - one could be B&S instead of Morse. Check it with blue - use artist's cobalt blue oil color if you can't find the machinist's blue.

The DTI isn't meant to always be spot on accurate. It's meant to give you qualitative results, not quantitative. If in testing a bore the travel of the indicator tip is not exactly perpendicular to bore you may not know by exactly how much the bore is untrue, but you'll be close enough to know whether it's OK or not.

Your test with the 20mm round bar tells you nothing about a possibly bent spindle. That the tailstock center describes a 3mm circle on the end of the bar tells you that the tailstock is not in line with the axis of rotation through the headstock bearings. That's not your first test, though - it's your final one. Because it may be just a case of adjusting the tailstock.

The important test of the axis of rotation through the bearings is that it be in line with the bed ways and carriage movement. You can check this with your test bar even if it has some runout, by rotating the bar and averaging the readings. If you don't get the same average at the end of the bar as at the middle and at the chuck, you have either a misaligned headstock or a twisted or worn bed. I'd first get a better piece of shafting to use as a test bar, though. You could compensate for the taper of the bar, but it only complicates things.

There are some good writeups available on setting up a lathe, and I'd suggest you look at some. But - although none of your tests have shown it - I think that you have already decided that the spindle is bent and will not be happy until you have pulled it out and tested it, so perhaps that is what you should do first.

Click to expand...

John, once again, thanks for your response. After sketching an imaginary spindle/workpiece in different rotational positions I have come to agree... a bent spindle will still cut round but offcentre... the depth of cut will vary according to angular rotation but the average depth of cut remains constant unless, as you say, something else shifts with rotation and/or load... which to get the result I saw means either the workpiece was shifting in the chuck, the chuck was shifting on the spindle or the spindle was moving in the bearings (or all three!). I shall repeat the cutting exercise but try to make sure everything is tight.. my thought is that I would put a non-cutting tool in the post, offer it up to the bar, lock everything down as tight as it will go then DTI the contact point, the toolpost, the slide, etc. as I rotate the countershaft by hand (to avoid adding movement/stress to the spindle) - and see what moves... I presume movement of the crossslide on its screw or the saddle on the bed would have the same effect so will check those too... however they wouldnt produce the DTI'd runout at the spindle I'd measured, but that could be, as you say, to the taper not being centred.

I'm pretty sure this is an MT3 taper. I measured the bore at various depths and plotted the result and its spot on an MT3, nothing like a B&S or any other taper I have data for.

My gut feel is the movement is probably in the front bearing (despite my attempts to show it by DTI and bar) as the pinch bolt for this is locked down as tight as it will go and I cant pinch the bearing on the journal (I can with the rear one to the point of locking the spindle). It may be that my attempts to show bearing play aren't in the right direction - maybe it needs to be 45degrees rather than vertical/horizontal....

Before I strip it down I will certainly do more testing... I'm not wedded to the idea of taking it apart unnecessarily, just resigned to the fact

it's all come to bits...

Well I went back and did some more measurements, tests, etc. and convinced myself that the spindle wasn't bent. I DTI'd the inside of the spindle and discovered a lump which I previously hadn't noticed when I cleaned it up. Took some removing but eventually it was gone. Retried the taper and the runout was now a much more reasonable 0.04mm (2 thou) average (rotating the taper in the spindle), max .06mm, min 0.02mm. DTI'd the inside of the spindle and got .03mm (1.2thou) runout.

So then went back to the issue of the eccentric cut. Set it all up again, tried bar test for bearing movement and put more effort into it and was getting .2mm (8 thou) movement lifting vertically, but .4mm (16 thou) pressing down. I also discovered 0.4mm end play on the cross-slide (at certain positions of the handle you can physically move the slide - with a clunk!) that will have to be dealt with as well but still felt the problem was in the bearing.

Decided to take the plunge and get the shaft out. In the end it came out surprisingly easily... here are some pics of what I found...




can you see whats wrong yet

I think I know whats happened... the bullgear is keyed onto the spindle but to get the shaft out you have to turn it so that the mark on the spindle lines up with a mark on the headstock casting, that puts the key at the 3-o-clock position looking into the headstock allowing it to move through the bearing (you can see it out of focus on the left in 1st pic). If you don't line it up and try bashing it out with the bullwheel at the 'normal' position (i.e. the locking pin at 12-o-clock) the key tries to cut through the bearing - hence the partial slot at the 6-o-clock position in the 2nd pic. Sometime in the past someone has tried to remove the spindle and gave up, smashing the bearing in the process... this also probably explains why the back end of the spindle is burred over and wont admit a 3/4" bar (fortunately the thread for the changewheel retainer doesnt run right to the end).

Here are the spindle journals...

Front


Rear


I think these need regrinding (I would if this was a car crankshaft, but maybe a lathe is different), at least the front one does, its well pitted.

Here's a view of the broken up bearing , note the small pieces of my carefully tapped oiler hole , the bearing material went right up through the headstock casting. Does this suggest they might have been poured in situ??



and a view of the headstock with front bearing removed. Is that an oil well in the bottom? If so I dont see how it works as theres no hole in the bearing and it wasn't full of oil. Also I notice the bearing is offset in the headstock - there's considerably more metal to the right where the pinch bolt is than to the left. I have no idea how to specify this or replace it...



Finally, the rear bearing. Its intact, but do I need to replace this? Its a bit scored but otherwise doesn't look too bad.



The last pic shows clearly how offset these bearings are... would these have been manufactured/bored in place? How would I go about specifying or making new ones? have I bitten off more than I can chew? All answers gratefully received...

Faces with a similar problem on a (sorry to mention it) SB, Here is what one enterprizing member did.

http://www.practicalmachinist.com/vb/showthread.php?t=146690&highlight=headstock+bearing

Jim B.

The paint on the inside of the bearings tells a bit of a story. Was it 'tidied up' for sale?

billmac said:

The paint on the inside of the bearings tells a bit of a story. Was it 'tidied up' for sale?

Click to expand...

Not by the previous owner. He owned it 6y but only used it a couple of times and the state of the paint suggests this was done some considerable time ago... but I did wonder how paint got inside the bearings!

Jim B. said:

Faces with a similar problem on a (sorry to mention it) SB, Here is what one enterprizing member did.

http://www.practicalmachinist.com/vb/showthread.php?t=146690&highlight=headstock+bearing

Jim B.

Click to expand...

Jim, thanks. I have already read that thread. Unfortunately the headstock is not removeable so has to be done in-situ and I dont have the tools or resources to do that so I will have to get this done professionally and that may make it a non-starter price wise...

Bearings & shaft

If you can't buy the bearings, you should be able to make them, or have them made. A good dead center in the tailstock should give you the measurements for the offset. This type bearing is made so slack can be taken up w/the hold down bolt. Just make sure the hold down is backed off when you measure the I.D. of the headstock, and bore the bearing .002 larger than the shaft O.D. The shaft you have looks good overall, but I would make sure there aren't any high spots where the dings are. You don't have bearing caps on this lathe, so grinding the shaft and making undersized bearings is unwise if you have anything that rides between them(gears, pulleys, seals, or wear rings). If the surface is round, and straight, and no high spots, a little polishing should do. I've cleaned up and polished far worse ones than that, in babbit lined ones, and they did fine. Good luck on this project. Looks like you'll have a good one without too much hastle(at least it hasn't been in the woods for 20 years like my first one).

lazarus said:

If you can't buy the bearings, you should be able to make them, or have them made. A good dead center in the tailstock should give you the measurements for the offset. This type bearing is made so slack can be taken up w/the hold down bolt. Just make sure the hold down is backed off when you measure the I.D. of the headstock, and bore the bearing .002 larger than the shaft O.D. The shaft you have looks good overall, but I would make sure there aren't any high spots where the dings are. You don't have bearing caps on this lathe, so grinding the shaft and making undersized bearings is unwise if you have anything that rides between them(gears, pulleys, seals, or wear rings). If the surface is round, and straight, and no high spots, a little polishing should do. I've cleaned up and polished far worse ones than that, in babbit lined ones, and they did fine. Good luck on this project. Looks like you'll have a good one without too much hastle(at least it hasn't been in the woods for 20 years like my first one).

Click to expand...

Thanks for this. I'm on a steep learning curve here. Would the outside of the bearing take up the shape/markings of the casting under pressure from the pinch-bolt or would that indicate they were poured in situ (but if that was the case I cant see what would stop the bearing material filling the well in the bottom of the casting).

I've read up about making babbit bearings and it wouldnt seem too hard to make up a mould with an oversized outer case and a undersized mandrel off-centre and then get them turned down and bored out then slit and the oil hole drilled (or maybe the last 2 before boring). I just don't have the tools for the job (yet).

What would you suggest as regards polishing? I could easily rig up a way to spin the spindle using a powerdrill at one end and the MT3 dead centre running in a countersunk hole at the other but what do I use for the polishing agent, or is this something again to get done outside?