Please help me troubleshoot my lathe problem

[assavalentin]

Hello and best wishes to all watch enthusiasts! I need some help on figuring out what is wrong with my lathe and what can be done to improve things. I've bough vintage American lathe some time ago, classic ww type, and later added faceplate, 3-jaw chucks, cross slide and other things, but never was satisfied with how it worked. As all parts were from different sources it was difficult to find which part contributed the most to bad accuracy. But this time I decided to try. My task is to turn flat rings from copper and silver, from a square sheet blank about 1 mm thick. I made a thicker brass plate to make the workpiece more solid so it doesn't bend when turning and attached both to the faceplate. The problem is I can't turn the metal evenly, it is always that one side of the circle left by the graver in the cross slide digs in considerably while the other side isn't even scratched (see my test copper plate in screenshot). It looks like the faceplate doesn't remain in one plane while rotating as if it had a bent arbor. I measured all three dogs and they are 10.90 mm of height each (could be off about 0.02 mm depending on placing the caliper). Then I tried to measure how much out-of-plane variation I have by putting a graver at the out-most part of the faceplate and rotating by hand. It scratched the most in one place. I marked this place and loosened and rotated the faceplate, so it occupied a different position relating to the headstock bearings (this is possible because the slot pin is broken). This time it scratched the most in a different place. And I found that the scratchy spot changing with each changing of faceplate position relating to the bearings. The scratching is quite subtle though, looking with a naked eye it's hard to notice that the faceplate rotates out of plane. I also made the same experiment with a smaller disk chuck and also by putting some metal rod to the surface of the cone inside the headstock. In all cases it scratched the most in some spot, not evenly around (should it be so?). The headstock itself doesn't show any play in the bearings that I can feel when trying to move the faceplate in different directions, and turns quite freely, though no effortlessly. I didn't take it apart yet. Excuse me for this long message, but has anybody come across a similar case and what is the best way to rectify the problem? I find it difficult to turn flat rings with this setup as the variation across the cutting circle is too big, negative rake graver digs in if I try anything but the most gentle feed and pulls the work out of place if I turn faster. I tried this quite a number of times and filed the edges of my sheet blanks so I don't think it's because they are bent or have burrs. Any thoughts and suggestions are very welcome!

 

[Max Phillips]

Headstock bearings that are worn or out of adjustment could be the problem, have you checked them? Even the tiniest amount of slop caused by bearings that are out of adjustment, can be seen visually if you pay close attention. If you remove the spindle and inspect the bearing surfaces it should also be visually apparent if they are unevenly worn (the races typically wear on the surfaces toward the back of the lathe, since that's typically bearing the pressure from cutting and the belt tension). If the wear is apparent, but not severe, you may be able to improve things by pressing out the bearing races and pressing them back in with the worn surfaces toward the front of the lathe.

If you don't have a good, small dial test indicator, then you should probably try to get one (even if you only borrow one for this testing). You can use it to read the collet taper on the inside of the spindle nose for run-out, and would also be easier to use for testing the faceplate itself, rather than using a graver (though the graver trick works very well if you can't get an indicator).

Ensure that the collet shank on our faceplate is a correct fit for the spindle, as well. I had one lathe with this exact problem, and it turned out to be caused by the tapers being slightly different. Even though it felt like it was engaging properly, I applied some blue marker to the faceplate shank, then carefully installed it into the spindle and tightened the drawbar. I carefully removed it, and saw that one thin ring was all that was making contact, allowing the drawbar to pull the shank off at a barely perceptible angle. Installed into a lathe spindle with the correct taper, it was dead accurate.

 

[motormaker]

I currently have and regularly use 3 old WW lathes and have never had bearing problems. In addition to these, I have also had 3 other WW lathes which were sold. None ever had bearing problems. You will need an indicator and way to securely mount it. Check to see that the bearings are clean, lubed, and adjusted tight enough to prevent play. Indicate the inside taper of the spindle. These lathes are accurate to millionths of an inch. If this runs out too much, the shaft might be bent, bearings worn, or dirt inside the taper. If it runs true:

1. Replace the key (pin to hold chuck in position) - this is explained in the Archie Perkins book. It is an easy job.
2. Check to see if the taper on your face plate matches the lathe spindle taper. The faceplate could have a conoidal profile rather than the standard WW taper.
3. Check runout of installed faceplate. The mounting taper could be bent.
4. Check for play in your cross slide. Tighten gibs as needed. See if slide rocks on lathe bed.

For certain, shorten the overhang or extension of the lathe tool. As setup in picture, the tool will bounce up and down causing chatter. Grind the tool properly for the task at hand. Try running at a slower speed.

Try the above, then let us know your findings.
Jim

 

[StephanG]

Before you take thing apart try something else in the lathe and see if it runs true in the taper.
Try several of your collets with a length of drill rod in them. Does the rod run true.
If your 3 jaw chuck uses the taper try it without the jaws in it and check the face of the chuck body.

If other things run true the problem will be with the face plate arbour.
If you get a similar runout with everything the problem lies with the lathe itself.

Could be a bearing problem or a burr in the taper where the pin once was.

The pin is good to have as it makes sure things always go back in the same place.
Once you get them to run true they should continue to do so each time you fit them.

 

[Discus]

Good info on checking your lathe. Couple of suggestions:
Dial indicators are cheap and even the cheap ones work well.
Then a hint for turning thin flat stock after you get the runout problem solved. Mount a brass rod and face the end flat. Super glue your workpiece on the faced end and machine it. Super glue melts at a relatively low temperature, remove workpiece, and the residue can be dissolved with acetone.

 

[assavalentin]

Thank you all for the useful suggestions! I'm a newbie here and somehow I thought I would get email notifications for responses. It is really great to see this willingness to help in people, I very much appreciate it. I'm still getting along with some special words and figuring out the procedures. It's hard to determine of other chucks run with greater precision as I don't know the acceptable tolerances and they are smaller in size, so less error. I once purchased some Chinese "golden goose" drill chucks and with just one of them (probably the best of the bunch) I can turn 1mm wire satisfactorily. But I'll try to get an indicator head and measure on some hardened drill bit. Of course I didn't do the turning with the tool sticking out that far ) that was just to hear the scratching.
I'll indicate the inside taper, the drill bit and do the marker test with the faceplate and report the results.
Once again, many thanks!
And a question to Discus: am I using the not so good method to turn thin flat stock into rings? If I mount a brass rod which is smaller then the diameter I'm trying to cut, will it not bend my workpiece into a dome?

 

[Discus]

Brass rod should be larger than the finished diameter. Suppose your flat stock is 12 mm square and the finished diameter is 10 mm. Start with a rod at least 12 mm diameter, preferably a little larger. Use your graver to make a continuous cut "inside the corners" so you now have a round piece about 11 mm in diameter and the square corners are discarded.
Finish machine, heat, separate, and clean.
Enco has dial indicators for $10-$15. I bought a cheapie from another company for $8 that works well
http://www.use-enco.com

 

[Max Phillips]

A word of advice on indicators - I've had good luck with the cheap ones as well, but be aware of the different types, regardless of cost. The "plunger" style indicators are usually the least expensive but they're also not very well suited for things like checking runout. Checking runout of a bar protruding from a collet shouldn't be much of a problem with a plunger style indicator, but indicating against the inside taper of the spindle nose would be a lot easier with what's known as a test indicator. Instead of a plunger, it utilizes a pivoting lever so it's a lot easier to set up for indicating this sort of thing. They're also usually much smaller in size, and more sensitive. The drawback to them is they typically have a much smaller range of measure - often less than a quarter of an inch - but for these purposes that is just fine. You want to be measuring a thousandth or less with a test bar chucked into a collet, and probably less than a ten thousandth at the inside taper of the spindle itself.

I'll second Discus' recommendation of Enco, at least if you're in the U.S. I use them often. They currently have one of the import (probably Chinese) brands of dial test indicator on sale for $37: http://www.use-enco.com/CGI/INSRIT?PMAKA=606-4710&PMPXNO=949978&PARTPG=INLMK32

And regarding the $10 plunger-style indicators, this is the one I have for general use: http://www.use-enco.com/CGI/INSRIT?PMAKA=605-4010&PMPXNO=946562&PARTPG=INLMK32

When I need to get serious, I have an old Starrett Last Word test indicator that reads to 0.0001" and an old Federal plunger-style indicator that reads to 0.002mm.

 

[assavalentin]

Max Phillips, thanks for suggestions about indicators! I'll do a bit more research on them. I've searched the ebay and found some interesting digital indicators. Look nice as they probably give runouts automatically without the need to watch the needle. Like this one:
http://www.ebay.com/itm/310575599519?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649
What do you guys think of these indicators?

To my lathe problem, what I managed to do without the indicator (yet) is I cleaned both headstocks in an ultrasonic cleaner and took pictures of the bearings. They do show some abrasion races, which are uneven (don't go around a full circle). One bearing shows what looks like a tiny crack (I marked it in red in the picture).
I would really appreciate if someone could tell me based on experience whether these headstocks look unusable or acceptable to oil and put back together. I've read somewhere about the suggestion to use abrasive paste to lap the bearings - have anybody tried it? To me it looks like a better idea to buy a new chinese lathe if the condition of these headstocks is too bad to work without restoring with abrasive paste.

Here are the images of the first headstock parts:


And here are the images of the second headstock parts:

 

[Max Phillips]

There is evidence of at least some uneven wear but from the pictures doesn't look too bad. The "crack" looks more like a nick or scratch to me. I would expect a crack to extend all the way out to an edge. I don't see shiny spots around the scratch so if it ever was a problem it appears to have worn itself back in. You could press the bearing races out of the headstock, and press them back in rotated 180 degrees but from the look of the wear it probably won't make any immediate difference (it could prolong their lives though). The belt tension will cause the back surfaces of those races to eventually wear more than the rest of the surface. Lapping them is an option but you have to do it right. Other than obviously making sure everything is 100% clean afterward, be aware that you cannot do this by driving the spindle with a belt. Lapping the bearings using a motor and belt drive will result in the bearings wearing one side down (as I mentioned earlier, due to belt tension) even more rapidly. If you have a lot of patience you could use a similar method to hand-lapping automotive engine valves - basically a suction cup on a nifty handle that lets you spin it rapidly by hand while still applying some consistent downward pressure. Alternatively, you could rig up something driven by a power drill, allowing you to directly spin the spindle while applying some pressure. You'd have to constantly stop and check/adjust the bearing load (adjustable via the rear bearing) to make sure both front and rear bearings are being lapped. It could be tedious and nerve-wracking and I'm not sure you'd see any benefit from it. Probably best off to clean, oil, and reassemble then take some measurements/comparisons to determine for sure where your problem is coming from.

I'm not sure about that particular electronic indicator... they claim a pretty ridiculous resolution (0.000005") but the repeatability is only 0.0005". The biggest problem is that it's not a test indicator, it has a plunger. It would be incredibly awkward to use for your particular issue here. A test indicator is really what you want... and honestly I would go with a mechanical one with a dial. I'm not even sure you'll find an inexpensive digital test indicator, but if you did I bet it'd be way too big to actually be useful. I'll recommend a Starrett Last Word all day long for this kind of thing, not only are they nice but they're incredibly small which makes them much less of a headache to try and use on a small lathe like this. The cheap-o test indicator I linked to above has a 32mm diameter dial, to give you an idea of size. Any bigger than that is going to be pretty awkward. The dial on my Starrett is only 22.5mm. They're not cheap, but they are awesome. I got mine used for cheap, that's always a gamble... but if they're taken care of they'll last several lifetimes. Having a test indicator that reads down to 0.0001" (many are 0.001" or 0.0005") would really be ideal for what you're doing.

 

[Neuron]

I agree with Max that your cone headstock bearings look OK. The spindle bearings also look fine. You have the so-called hard (steel) rather than bronze (B) cone bearings. H bearings are more durable than B, but they are more sensitive to grit and are a bit more finicky to adjust. Soft bearings tend to self-heal and seat evenly with wear and the slack is easy to tighten out. Anyhow, your bearings appear to have ordinary wear. Put it back together, oil it, and adjust it carefully, and it should work OK. Unless you are confident, I wouldn't press out the bearings and rotate them, as you could create new problems.

I suppose that you could try to "polish" the bearings if you find that despite careful lubrication and adjustment the spindle still doesn't rotate smoothly. Since these are tapered (cone) bearings, it's possible to apply a very fine abrasive compound and carefully turn the spindle to polish the bearing surfaces, then clean out the parts and reassemble.

 

[assavalentin]

Hi and many thanks to Neuron and Max Phillips!
Hope everybody is doing well and having good time.
(BTW I read your article about finishing a Chinese lathe and it is just great)
Awkward to bump up this thread after a bit of a break but I finally managed to acquire a test indicator and would like to share my results of measuring runouts. Maybe this could be useful to other people for justification. The main question is: are the results acceptable to use these tools to turn disks from flat stock?
Here are the measurements.
My indicator is Starrett 711-t1, it says 0.0001" and has marks on the dial 0-1-2-3-4-3-2-1. Each mark is divided in 10, so I concluded these numbers are thousands of an inch (0.001") and measured everything in them.
For each measurement there is a photo of my test setup. Maybe they are very incorrect so any criticism is welcome! To my mind they are not precise as the ball pin has to be applied at slightly different angles to clear the obstacles, but they give the idea of approximate figures.

1) headstock inside taper, reads 0.2 (thousands of an inch) on several full rotations



2) 3-jaw chuck drill bit radial, reads 2.5



3) 3-jaw chuck base lateral, reads 0.5



4) 3-jaw chuck base radial, reads 0.5 too (not pictured, same setup but ball pin is applied to the outer surface of the smaller cylinder shape

5) faceplate base lateral at the outer edge, reads 0.7


6) faceplate with mounted brass support base plate, reads 1.5


What of these figures look like acceptable (if any) and what is too much?
My own observations are that my chuck bases are roughly on a par with the headstock, but assembling the whole setup drops down the precision. Wonder why... The 3-jaw chuck is mounted on 3 screws and I noticed that the final precision depends on how it is put together, it has some freedom of side movement before the screws are tightened. Maybe it takes some special procedures to achieve better precision after assembling it? Concerning my brass plate - looks like the brass plate itself is responsible, I just measured the thickness and it reads from 3.19 to 3.21 mm. Can't figure out how to sand it for consistent thickness...

 

[bkerr]

I am just now looking at this thread, it would make sense that the further away from the spindle the more run out you would have. I would think that the pic one would or should be near zero run out. You might check the tension, load, on the bearings. .002 is quite a bit that close to center. you could possibly lap the bearings in and re check your work.

The indicator is fine, make sure to zero it out each time.

Good luck

 

[Neuron]

First of all, congratulations on your dial gauge run-out testing jig! I don't know how to evaluate your posted results but maybe Jerry Kiefer, who is very expert in micro lathe and micro millwork will be able to comment. I can say that if you want to test the truth of your headstock, the best way to check for run out is to secure a round piece in a collet and measure run out while that is spinning in the headstock. You'll want the piece to be perfectly round, rigid, and big enough (diameter-wise) to be reliably measured by the dial gauge feeler. I like to use hard steel stock. 25 to 50 is large enough to give a reliable reading, and stiff enough so that you can measure run out on the stock near the collet and 1 to 2 cm out. You could also check different collets, as they can be a source of off-axis (i.e. angled) alignment errors. In any case, a truer picture of your headstock performance will be obtained by using the collets for which it was designed. Measuring run out on the inside of the HS collet bearings is not particularly useful, mainly because the collet is gripped by a fairly large cone (compared to the small segment you are measuring). Your point measurements can seem excessive when in practice the workpiece held in a collet runs true.


As regards the eccentricity errors in work pieces held in size-adjustable faceplate, 3-4 jaw, and Jacob's style chucks, these are by their mechanical design prone to more eccentricity error than precisely sized wire chucks. That's OK because they are typically used for machining larger parts where their errors are tolerable. For really fine work (like turning a balance staff), the wire collet comes to the fore.

 

[Firegriff]

I am currently cleaning and setting up a Boley watchmakers lathe so reading this chat is helpful. For my next lathe rather than a Chinese lathe I am going to purchase a Shereline lathe when they get the DRO(Digital Read Out) version back online both lathe types seem to have their pluses and minuses. There have been other chats on the Chinese lathes that have some issues with tolerances and proper tempering of the metals used so I would not purchase one over the Shereline lathe.

 

[assavalentin]

Thank you very much for suggestions!
Experience is immense value. I see the point in using a high precision rod, but I don't have one, so I use what is available to evaluate.

 

[Max Phillips]

Thank you very much for suggestions!
Experience is immense value. I see the point in using a high precision rod, but I don't have one, so I use what is available to evaluate.

The way you were measuring, in the spindle taper, is more fundamental than checking a test bar in a collet, so you're certainly starting with the most important aspect. Measuring a test bar held in a collet combines the error of the collet, which we don't care about until after the spindle itself is running true. As has already been mentioned, double check your bearing adjustment. 0.0002" is too much runout there in the spindle taper. There really shouldn't be any runout there (maybe the slightest wobble of the indicator needle, at worst). The whole purpose of this type of lathe is to be deadly accurate. Any error inside that spindle taper will only be amplified as you read further away (both along the lathe axis and radially outward from it). That spindle error alone, without compounding error of the faceplate and jaws involved, could easily result in 2 or 3 thousandths runout turning a 2.5" diameter flat disc.

With any luck it is just your bearings at fault - either the adjustment, or they are just worn. The nasty alternative is that the spindle itself is at fault. A new spindle could be made but that may not be the best option (unless, like me, you just enjoy a challenge). A note on lapping cone bearings, if you are tempted to try: Other than the issue of cleanliness, keep in mind that you cannot lap the bearings using the belt and pulley drive currently mounted to the machine. The belt tension will ruin your results. You must set up some sort of fixture to drive the spindle via the nose, while applying pressure on the spindle that is ONLY directly in line with its axis.

 

[Neuron]

The way you were measuring, in the spindle taper, is more fundamental than checking a test bar in a collet, so you're certainly starting with the most important aspect. Measuring a test bar held in a collet combines the error of the collet, which we don't care about until after the spindle itself is running true...

What you say is true, but I was suggesting this collet test measurement because, as I said in my post, it is difficult to use the OP's dial gauge set up to measure the inside of the bearing cone. Yes, the collet itself will add potential eccentricity error, even if the HS cones and spindle are perfectly aligned, but the lathe is meant to be used with compression collets and that error is what you need to know when you are using the lathe. Of course you need to use collets that are in spec for the HS spindle/cone geometry.

 

[Dushan Grujich]

As has already been mentioned, double check your bearing adjustment. 0.0002" is too much runout there in the spindle taper.

G'Day Max!

I believe that You are going too far as the measured runout at the inner spindle taper of 0.0002" or 0.2/1000" being equal to ~5/1000 mm, is according to the Swiss precision lathe manufacturer the maximal allowed runout (applicable to cone bearing watchmakers' lathes) measured at the inner spindle taper. In practice it ends up being 2-3 micrometers.

Talking of tolerances better than five micrometers, in this context, we are entering the area of uncertainty caused directly by temperature changes, in both, measuring and operating the spindle.

Cheers

Dushan

 

[Max Phillips]

G'Day Max!

I believe that You are going too far as the measured runout at the inner spindle taper of 0.0002" or 0.2/1000" being equal to ~5/1000 mm, is according to the Swiss precision lathe manufacturer the maximal allowed runout (applicable to cone bearing watchmakers' lathes) measured at the inner spindle taper. In practice it ends up being 2-3 micrometers.

Talking of tolerances better than five micrometers, in this context, we are entering the area of uncertainty caused directly by temperature changes, in both, measuring and operating the spindle.

Cheers

Dushan

Hi Dushan, I hope you are well!

If that 0.0002" happens to be pure axial runout then I suppose I would agree... but it seems to me that the OP might be having a problem with distorted/damaged (out-of-round) spindle taper, in which case that 0.0002" could be making a significant contribution to the total error he is seeing with his flat discs held in the faceplate (the jaws do also seem to be partially at fault, from what I read). Given that the surface of the faceplate will already be protruding from the spindle a bit, that error could easily end up being several thousandths at the outside diameter of the disc (depending, of course, upon the diameter of the disc).

 

[Dushan Grujich]

G'Day!

I keep coming back to this topic as I deem understanding of watchmaker's lathe capabilities and limitations as very important, as well as the ability to maintain the lathe in order to preserve its accuracy for unlimited future use.

To help the OP to properly estimate errors, of both, axial and radial run-outs I am attaching a control slip showing how and where the measurements of the Favorite II were made. The control slip gives allowed departures from norm, tolerances, and it shows the actual measured values of this, cone bearing, watchmaker's lathe.

Favorite II lathe was made in Switzerland in 1951, a smaller version of the Schaublin 65 and 70, produced to more stringent requirements of accuracy than its bigger brothers. To open the full sized image hover the cursor over the thumbnail and click the mouse. The language is Swiss French, however it is quite easy to understand when used with related drawings.

Basically, I follow the same protocol in checking my Favorite II, once each year, to verify that there are no changes, I can also say that the lathe still performs as it did when it left the factory in 1950's.

If anyone may want it, I can also provide similar protocol for testing the Schaublin 70 lathe.

Cheers

Dushan

 

[Max Phillips]

Thanks for posting that, Dushan! I have filed it away for my own reference and hopefully it has helped the OP as well.