Scraping a lathe to the limits of mensuration

I'm a retired research scientist. But long before I became a scientist I was hopelessly perverted by C.L. Stong's volume of "Amateur Scientist" columns from "Scientific American".

In the late 40's John Strong, a premier experimental physicist, constructed a new ruling engine at Johns Hopkins. IIRC this was the first successful build since the 3 engines Rowland built 50 years earlier. A.A. Michelson failed in his attempt which says a lot about how hard a job it is. Though in fairness to Michelson, he had *very* ambitious goals.

The person selected to machine the pair of screws which positioned the carriage and stylus was David Broadhead, an amateur. After eliminating all spindle errors, Broadhead rescraped his 13" Southbend so that the ways were flat and parallel to less than a tenth over the entire length.

I have a 10x20 Clausing 4902 and a 6x24 8520, both of which are large enough for anything I expect to do. I bought the 8520 in "as new" condition with the original inspection sheet which it still met 30 years later. The 4902 has seen quite a bit more use, but is still in good shape. But it does need TLC.

I want to "recondition" these machines to the limits of my ability to measure the errors. I bought and read Connelly cover to cover 20 years ago, I read "Foundations of Mechanical Accuracy" by Wayne Moore long ago, but only got a personal copy a few weeks ago. Any book I come across that discusses high precision and accuracy machines I buy. Suggestions welcome.

In preparation for rescraping the 4902, I bought a 7x14 Chinese machined casting set to practice on. I intend many alterations to the Chinese casting set design, but in this forum I'd like to use it as a case study for discussing the nuances of scraping the alignment of a lathe to the limits of mensuration. I chose that machine for a warm up exercise because they are cheap and, tuned to high accuracy, a convenient size for making gauges or precision RF connectors which tend to be *rather* pricey.

I "reconditioned" a new Chinese 4" x 6" bandsaw from ENCO to cut square and precisely about 25 years ago by traditional benchwork. It was a lot of work, but well worth it. But I've not done much such work since. So I'm rusty. Hence my desire to practice on something less valuable than my 4902.

I hope by Monday to have a 3 point mount with differential thread leveling screws on which to mount the 7x14 lathe bed. However, the rest will not be quick. To compensate for the deflection of the floor, mount, etc I plan to attach a pair of levels to the bed and then use another pair of levels and a small microcontroller (e.g. Arduino) to compensate for the movement of the bed as I move around it and report the deviation of the surface I'm working from the reference plane. I wander between electronics, metal work and programming as suits my needs and interests. I am well prepared.

At present I'm trying to get pricing on commercial sensors. I can build suitable sensors from Chinese level vials for about $15 each which will provide better than 1 millionth per inch accuracy. For a few bucks more I'd rather buy them.

My hope is that others, even if not interested in pursuing the absolute limits of accuracy will be interested enough to contribute their insights into the problems associated.

Scraping is time consuming, but easy to do. Mensuration is another matter.

Have Fun!
Reg

Check out these guys. They also publish books on the subject. I will be teaching a class on Rebuilding and alignment in August in Rockford IL inside Bourn & Koch. Noncontact Displacement Sensors: Capacitive and Eddy-Current
Lion Precision's Technical Library

I always say...Scraping is easy, but knowing where to scrape and how much to take off is the secret of the trade. I also say using 3- Points is also a "lost art" It is so simple but no one uses it. Thanks for your help in teaching.

temperature difference and weight deflection can easily distort parts. like a dog chasing its tail. modern machines often have automatic electronic compensation for temperature changes
.
extremely common for larger parts to need shims to compensate for sag by the parts own weight. sort of like you got a 1/2" dia steel rod thats 10 feet long. it bends when you try to pick up.
.
bigger parts very common for coolant to touch one end of a part for that one end of the part to get .0003" smaller. when coolant stops after a few hours it grows back to normal size.
.
same with a big building the steel outside building columns can be growing .030" everyday when the heat of the sun after 8 hours warms up the side of the building where the columns on the middle of the building that are temperature stable do not change in size that much. thus the building distort or changes shape hourly, daily , etc
.
then there is bimetalic effect where a steel part bolted to a massive concrete floor (more stable slower changing temperature)the steel can warm up from sunlight coming through windows and the steel expands and constrained becomes curved or snaked wavy and when on a cloudy day the steel shrinks and is straight again
.
sort of like holding a inside micrometer or a .0001" indicator bore gage it changes calibration just by holding it in your warm hands. literally every 5 minutes it can be changing length
.
literally a 10" dia aluminum boring bar when coolant hits it it can grow smaller so the back end of a 12" deep bore is smaller than the front cause it was changing size diameter over a few minutes of temperature exposure from coolant evaporation having a chilling effect
.
and many a time i have had a optical level setup and fork truck drive by and the whole floor tilts like being on a trampoline or a spring sofa bed. or visualize being on a boat. if many people go to one end of the boat that end goes down or the boat tilts or leans. literally if heavy weight in center of the boat the center of the boat is lower or boat is curved cause ends of the boat are higher
.
then there is assembling machine parts where thickness of fingerprints can distort part .0001" or more and need to work in temperature controlled hepa air filtered dust free room with face masks. skin cells and spit when you exhale and touch stuff has effects. and obviously bolt torque can distort parts. soft foot effect can effect alignment just by changing bolt torque. often changed part alignment more than .0002" that way varying torque on a few bolts. and of coarse its often not stable after a few hours alignment and part distortion can change. we used to say its got to think about it
.
dog chasing its tail
.
Harrison clock was a masterpiece of automatic temperature compensated parts. the mechanical clock was accurate to like a second of time per month.

then there is dynamic alignment. on a printing press had a alignment distortion happening every minute of many parts moving in relation to one another. when machine stopped alignment was ok.
.
i bolted on a temporary precision level and saw machine distortion happening every minute periodically when machine running. had to shim a small part .001" to compensate for part bending as the force slid down the length a a alignment bar. sort of like when you mill with a end mill close to collet it doesnt bend as much that if you mill at end of end mill sticking out 6" out of the collet
.
the computer CAD model didnt show part bending under varying load.
.
and of course on a big cnc mill with 10 ton table. when hydrostatic way oil pressure failed intermittently one end of the 12 foot table was going up or down .002" with varying way oil pressure. some machines you can measure 10 ton machine breathing or see .0001" movement every time way oil is pumped and when pump stops. or see servo oscillation where machine moves back and forth never stopping all day long or sometimes it does stop after 10 seconds. dynamic alignment changes are the hardest to see and measure
.
same with a machine slide tilting on direction reversal. very common to see .0002" just changing direction

Hi Reg,

What do you plan on doing with this "ultimate" machine?

If you plan on touching this machine you will be distorting
the outcome. If you are going to machine with it, how will you control
the temperature and the distortion of the machining process?

Some grinding machines that are capable of working in the millionths,
do their final position adjustments by warping the whole machine.

If you want to operate on the edge of meteorology you might be
better served by using fully a characterized and calibrated control
system(s). Dynamically measure all the variables and adjust accordingly.

I have owned a machine shop for 50 years, and was a flight controls engineer
for Boeing for 13 years, and I scrape. I am not trying to be a "know-it-all",
I'm just putting in my 2 cents worth.

Paul

Richard,

Thank you for helping to keep the art alive. BTW I read your thread on 3 point support. Very good.

On the question about where to place work supports, as close to the three centers of mass which are farthest apart. The location of the supports stipulated for a surface plate are based on the geometry of a surface plate. Actually determining those points accurately for a machine part is more trouble than it's worth. A lathe bed will still be in the same position relative to the supports, but nothing else will be.

I'm amazed that 3 point support is ignored. It's vital to evaluating the surface accurately. If a machine was installed with a twist, the wear pattern will preserve that twist. So letting the part untwist is critical to doing a good job.

Thanks for the link to Lion.

There is a long thread on capacitive sensing of spirit vials here:

suggestions for high-resolution tiltmeter (inclinometer) sensor? - Page 1

From reading some bubble vial literature from Fredericks Company, it appears that the repeatability of 2"/2 mm vials is 2.5 millionths per inch. However, resolution using the NASA Tech Brief circuit mentioned in the thread is at least 10x better.

It's very nice to have people to discuss the project with that understand that what I am proposing is doable, just difficult. When I get close to the end I'll have to pay attention to temperature gradients and to achieve best accuracy the lathe will have to warm up by running the spindle for several hours before making the final adjustments to the headstock alignment. And will need a similar warm up for use if the best accuracy is needed. Determining how long a warm up is needed will be a very interesting problem. The practice machine has a flat top headstock, but my Clausing does not. I'm hoping that with an electronic level sensitive to 1/4 millionth per inch I can level the lathe and monitor the level of the headstock until it stabilizes. I suspect I'll need a level on the headstock and another level or two referenced to the bed at one end and the headstock at the other. At those deflections, shifting your weight from one foot to the other will move the floor. The purpose of the 3 point support frame I'm making is so I can measure the deflection of the supporting surface and correct the level readings.

Have Fun!
Reg

Measuring cross slide dovetail alignment with side

After considerable study, I concluded that the cross slide is my best choice of datum. Once it is true on all faces and the dovetails, I can use it to spot the saddle and square the saddle to the long axis of the bed.

In looking through Connelly I don't find any mention of aligning the dovetail to the side of the slide. The closest I found was a discussion of fitting a tapered gibb.

Initially I had though to use a set of precision ground rods to measure the dovetail alignment, but on consideration that seems not wise as the clearance surface is not a precision surface and one side of the rod would be in contact with that.

That seems to leave placing the slide on a surface plate and using a tenth indicator to trace a series of lines along the face of the dovetail followed by some trig and algebra.

I'd like to avoid scraping the dovetail such that I need to do extra scraping on the side to align it to the dovetail. I particularly want to have all the slide faces properly plane and aligned so I can reliably indicate off them. I also plan on quite a range of attachments for doing things such as align boring between centers, etc.

I still need to consult the books by Wayne and Richard Moore on the subject. I'll edit this later after I go play with the slide for a bit. I realized in writing this I had not yet spotted all the faces to see how flat they were or were not.

Every day it seems a new looney pops up. Good luck with your endeavor, I highly doubt you'll come to an end game.

Reg....One can see your a scientist and your over thinking this. Some of the hardest people for me to teach sometimes are the brightest. In Taiwan I had a genius who had 3 PHD's and for the life of me and him he could not master scraping in the first 45 hour class while the other students whizzed by him. His boss came in and told him or ordered him to quit trying to figure it out, just do it!

Everyone always quotes about this and that from the Connelly Book. My Father met Connelly and he said he was not a scraper or machine builder. He was a technical trade school teacher and he organized the book. Look in the first few pages of the book and see where he credits all the contributors. My Dad never used the book, other professional rebuilders do not use the book as a guide. We were taught by Master Journeymen Rebuilders and were not self taught from the book like many on here do. As I have always said on here "there are many ways to skin a cat" and as long as you can skin the cat as fast as a Pro can, go right ahead. You have to start at the bed and work up, referencing from original surfaces that are not worn and not trying to re-inventing the wheel. So do the bed, match fit the saddle to the bed following the original surfaces, then do the saddle ways, squaring to the bed, cross-slide, tail stock and the headstock. Compound can be done anytime. That's the way the new machine builders do it. Many rebuilders do the tails stock last. Many never remove the headstock. Being a detective is the key and not over thinking it.

Warming up the spindle on a lathe before aligning it is nonsense...how many lathe operators warm up the spindle to work on it? This is not rocket science or a space ship headed to the moon. How many shops have temp control.. The Moore book is also a book that 90% of all shops do not need. A conventional lathe is set up to be .0002" / 12" Moore uses .00005" / 12". Enjoy..

plastikdreams said:

Every day it seems a new looney pops up.

Click to expand...

I think this thread has potential

Demon73 said:

I think this thread has potential

Click to expand...

Yeah being as the 2 lathes we have at work are past menstruation

I spotted the slide and then measured it with an indicator and surface plate. It appears to be within a half thou except that the sides are not parallel in the vertical direction. I did not attempt to check the ends.

During the indicator work I noticed that the friction moving the slide across the surface plate varied quite a bit and in a consistent fashion. That leads me to think that the B grade surface plate I have is not adequate for this project.

At present I do not have a King Way. Nor do I have a large enough surface plate to test my 24" straightedge from "Other Brother". All I have is the parts of the machine, a 9" x 12" surface plate, 5"/2 mm level and dial indicators. I have stuff on order, but I'm trying to do what I can with what I have while I wait.

My normal practice for leveling my Clausing is to place my level on the cross slide and adjust the legs for the least deviation of the bubble as I move it from one end of the bed to the other.

Yes, that will impart some twist to the ways if they are more worn at the headstock end than at the tailstock end which would be the normal case. But as a practical matter in setting up a lathe it seems to me a reasonable method. I'd be interested in alternative approaches.

In any case, the cross slide must be true. If the slide is true, then with the slide on the saddle, any error is either the saddle or the bed. My plan is to scrape the flat way of the bed and then scrape the vee of the saddle.

With that done, a level on the cross slide will reveal the condition of the vee on the bed though not the straightness. That requires a measurement from the back edge to both faces of the vee.

The comment about Connelly is interesting. It certainly explains the poor presentation. I reference it because it's all I've got other than the Moore Tool books. A better book would be very nice to have.

There are a lot of PhDs who got the degrees by memorization or because their supervisor just wanted to get rid of them. People with the real thing put us on the moon and only got one. If you have a real PhD you don't need anyone to teach you. A teacher is a convenience, but not a requirement. I had a classmate who did not know Snell's law, but he got a PhD. I can't imagine someone with a functional PhD scraping machines except as a hobby. Everything needed to scrape a machine is part of the high school curriculum in the US, though not many seem to learn it these days.

As to bearing heating. My Clausing is a plain bearing and I intend to replace the ball bearings in the practice lathe with a plain bearing. All bearings warm up in operation. This is why they have excess clearance versions. The heat conducted from the bearing to the headstock will cause the headstock to expand. But the tailstock won't experience that heating, so it will not. This will result in a taper. Whether that matters depends upon how demanding the work is. I'm pushing for the limits of possible. It's purely entertainment.

After 50 years of playing guitar, I finally reached the point I can put a random CD on and play along in key within a few bars even if I've never heard the tune. There is hard and there is *really* hard. I'm more interested in the latter. And if you don't understand what is meant by the thread heading, I don't care what you have to say. Scraping is easy. Mensuration is the hard part. If you pick up a class 1 gauge block with your fingers you just changed the dimensions by more than the block set tolerance.

Have Fun!
Reg

correct me if I'm wrong,but don't 4900 series Clausings have flame hardened bedways?

Your on your own Reg as you seem bent on "teaching the forum" how to, when your guessing at it and have never scraped a machine yourself. I was willing to help until your last speech.

I'd take up golf instead if I were you. It's just as frustrating but at least you get a walk in the fresh air.

Regards Tyrone.

Little bit of OCD ehh? I guess if its cheaper than therapy, go for it!