heres how to get 0.0001" precision on your ancient hobby lathe

[spaceman_spiff]

okay heres the latest

I changed the material to 6061-T6 aluminum. More stickout to the cut but more material in the jaws and thicker overall. Not really attempting to do anything particular by making these changes, just experimenting.

The rest of the setup is the same, although I did loosen the gibs a bit on the compound so its now free to jiggle a bit back and forth.

I did my first cut to establish the handwheel to feed relationship, which turned out to be 0.0001065" feed per handwheel thou.

The rest of the cut "targets" were based on that.

I did 4 passes per cut, and at the end lightly placed a very fine file on the workpiece with almost no force for about 4 seconds, just to remove any serious flakes or burrs.

target	result	error
0.0063	0.0062	0.0002
0.0071	0.0068	0.0003
0.0052	0.0050	0.0002
0.0101	0.0095	0.0006
0.0048	0.0044	0.0004

Hmm. At least all the errors are positive. I suppose I should try just increasing the feed by an extra 3 tenths and see what happens.

 

[jtwade]

Used to have to turn some bearing journals to tenths tolerances in a shop I worked at. The way I was taught to do it was fairly simple, rough size to within .005 of finished diameter. Then set up tool post grinder with the compound set at about 84 degrees. At that angle, one thousand on the dial, moved in darn close to one tenth. Forget my trig, but 84 degrees was close enough to get the part close enough:roflmao:. Never tried to get that close with a cutting bit, but see how the math works.

 

[spaceman_spiff]

update

target	result	error	notes
0.0063	0.0062	0.0002
0.0071	0.0068	0.0003
0.0052	0.0050	0.0002
0.0101	0.0095	0.0006
0.0048	0.0044	0.0004
0.0049	0.00515	-0.00025	added 3 tenths
0.0049	0.00465	0.00025	added 2 tenths

Well maybe I've gotten as good as it can get then. The micrometer itself is only accurate to a tenth. And now I've made the error bipolar and its also where I said I wanted it to be so the question now is, can I reproduce that at will and on what materials and locations on the bed?

If we look at this as a practical thing, if you can just INCREASE the cut a 3 tenths at a time, then you can measure and repeat until you are where you want.

So from that perspective, now I need a test.

Perhaps making a spindle block with two bearings, a spindle, and a block with bores, and see what kind of runout I get? Or what would be a realistic test?

 

[benmychree]

Criticism of what I describe here is WELCOME!!! Please, lets dissect this, tear it apart, and make it better.

The good news is: I think I am getting control of diameters, +/- 0.0001" (tenths), on my ancient, worn out Atlas 10F lathe.

I also learned something very important about manual lathes today. That alone is almost more important than being able to cut within tenths.

Heres how I did it, and that special thing I learned:

Notice how the handwheels are graduated in 0.001", and the space between two graduations is about 1/32"? WAY too small to control tenths, accurately anyway. Yes, you can approximate. But seems kind of strange to me that a lathe would have handwheels that coarse. Im sure when even this low grade Atlas lathe was brand new it could cut in tenths reliably. Hmm. Actually..come to think of it the Hardinge HLV-H extreme precision lathe I used at work long ago was also in 0.001". So how the heck is anyone supposed to work in 0.0001"?

Even a DRO or dial indicators wouldn't really help, because you'd still be stuck trying to turn that handwheel just that TEEENY amount!

Well, thats where the compound tool post comes in. It will let you magnify its handwheel by MORE than you would ever want!! Yes, you can turn the handwheel 100 ticks and it will only move the cutter forward by .001" if you want!! So essentially you have made it so that each handwheel graduations of 0.001" actually means 10 MILLIONTHS of an inch! Now tenths become cake!

The idea behind this is to just rotate the compound tool post so its a very slight angle instead of parallel to the spindle axis. To be clear, "parallel" to the spindle axis would mean that turning its handwheel would move it towards or away from the spindle.

Now think about that, if its only moving towards or away from the spindle, its not moving the tool towards or away from the work at all, is it?

So if you put it at a very slight angle, it WILL move the tool towards or away from the work. Turns out that at around 84 degrees, for every 100 thou of its handwheel, it moves towards the work by only 10 THOU! Now each tick is 0.0001"! Tada! And if you want you could theoretically reduce that angle even further and probably multiply it by ten times again!

REALITY CHECK

-You are still limited by the accuracy of the threads, ways, etc.. You cannot precisely move the tool by 10 atoms just because you have the slide at an angle. But where that limit of accuracy is is now something you can move on to dealing with, instead of trying to turn a handwheel by some teeny amount.

-The angle readings on the compound tool post are not very fine. You should get the angle to approximately where you want and then do some test cuts to confirm the new tool post handwheel versus actual increase in cut distance, and then do some arithmetic and calculate how many thou you need to turn the handwheel to get what you want. If you dont have a tenths micrometer, now you have an excuse to buy one!!

-You will likely need to rotate your toolpost so the tool is again perpendicular (or whatever you need it to be) to the work, since you are rotating the tool post.

RESULTS!!!

So...does this work or not? The answer appears to be YES!

I spent the last couple hours experimenting to see if I could really make reliable cuts in one shot based solely on the handwheel.

Material: 303 stainless
RPM: about 800
Cutter: carbide TPG322 in a very rigid insert holder and tool post
Lube: none
Diameter of workpiece: about 1/2"
Workholding: 4 jaw chuck. Stickout about 2".
Feed: automatic, about 5 thou per rev
Machine: Atlas 10F from 1944

Cutting notes:

All done in three passes. First two were easily seen. The last pass was BARELY visible taking place..I'm not sure if it was doing anything or not. I have very good vision up close and could just barely see the "line" moving by getting up real close. And to be clear, I made NO adjustments to the handwheels between passes. I just turned the motor off, used the carriage handwheel to move the tool back to the beginning of the cut, and then powered it up and repeated the cut.

I made many test cuts. Some of them came out right, some did not. A few came out about 1 thou off!! But I think I have it figured out now.

The final two cuts I made came out within about 0.0001" of where I intended. I think what was happening with the cuts that didn't work out was that I was not taking deep enough cuts. And I also screwed up the math a few times since my calculator was dying and I couldn't read it. I would get your notepad and calculator handy and keep a log of what you do so you can figure out how deep you need to cut and what RPM/stickout you need to do, etc..

FINAL CUT #1:

Target radial reduction: 0.0117"
Actual radial reduction: 0.01175"

FINAL CUT #2

Target radial reduction: 0.0107"
Actual radial reduction: 0.0108"

(the pic shows the incremental change in diameter)

Please join in the fun and report your results!!! You will need an accurate tenths micrometer to play along. The one I'm using was $30 from horrible freight!

Pretty much a big waste of time, especially on a notably imprecise machine, ways, geometry, and unmentioned, spindle bearings. Cutting tools are a large part of the accuracy equation as well; we do not need to have such complication in the (machine's) ability to measure, all that is necessary is an accurate finish cut, leaving a very small amount of stock to (lightly) file and subsequent polishing with abrasive cloth. It may be another matter with ultra precision machines, but turning and grinding is the real answer for more complicated precision parts that turning alone is not likely to give.

 

[SG51Buss]

That 'shear' tool article was fascinating, and so counterintuitive that I had to re-read it 3 times before it sunk in. Gotta try that someday, including the 'notched' parting blade.

Thanx...

On the thermal expansion issue, I've always wondered if hi-quality micrometers expand uniformly with shop temperatures, yielding the same measurements on same-temperature workpieces, or if the bows are designed to mitigate that expansion, including the 'warmth' of the operator's palm.

It's interesting that your 0.0002" - 0.0003" errors are consistent. I wonder if that's caused by the rounded edge of the carbide insert. When I first started using insert tooling, I also experienced a similar reduction in depth-of-cut to bit-advance ratio.

Oh, wait. That consistent error may also be caused by not being at a precise 84.26° compound angle. Look at the error to target-feed ratios. For example, in your second test, to get the 0.0071" target, the compound was advanced 0.071"? The 0.0003" error over that 0.071" travel is about a 0.24° angle...

 

[John Hasler]

That 'shear' tool article was fascinating, and so counterintuitive that I had to re-read it 3 times before it sunk in. Gotta try that someday, including the 'notched' parting blade.

Thanx...

On the thermal expansion issue, I've always wondered if hi-quality micrometers expand uniformly with shop temperatures, yielding the same measurements on same-temperature workpieces, or if the bows are designed to mitigate that expansion, including the 'warmth' of the operator's palm.

As far as I know they are made of invar

http://en.wikipedia.org/wiki/Invar

to minimize thermal effects.

 

[dsh1001]

First of all, congratulations on figuring out that trick. That is something a person would normally learn at the elbow of a master machinist and you came up with it yourself. Nice

Grinders definitely create a very fine finish. I don't beleive that, in itself, gives a ground finish any better dimensional accuracy though. Think of the finish as a peaks and valley on the surface of the part. Whether you are talking about a lathe finish of 125 microinches (pretty darn rough) or a ground finish of 10 microinches (pretty darn nice finish) you are still doing your measuring on the peaks of that finish. The very fine tolerances come from the very fine cutting tools (grit of the grinding wheel), the wide contact with the part for even pressure dispersal, the rapidly rotating cutting tools (which makes the grinding wheel essentially one wide tool instead of thousands of tiny tools) and the very fine advance of the grinding wheel usually in the range of ten millionths to a hundred millionths (.000010 to .000100).

Just my two cents,
David H.

 

[Andre]

I think MrPete222 showed this in one of his videos. Nonetheless it is a great trick! I'd play along but my lathe doesn't have dials :lmao:

 

[george wilson]

Ground finishes that are done properly definitely leave smoother finishes than turned finishes. That is why all manner of parts that need to be accurate are ground. The valves and valve stems on your car are ground,for 1 simple example. They need to fit accurately into their holes without the "peaks and valleys" that turned parts have,becoming soon worn off,resulting in undersize valve shafts that leak oil.

If you look around at
what things are turned and what are ground,or even precision lapped,you will begin to understand why ground surfaces are used. The surfaces of ball bearing races are ground. Even the outside surfaces are ground. This to again,prevent the "peaks" of a rougher surface from wearing off,leaving the bearing loose in its mounting. Really precision things like gage blocks are first precision ground down to very close limits. Then,to make their surfaces even smoother,and more long lasting of tolerances,they are lapped until they actually stick together,so flat and smooth are their surfaces.

So,grinding not only makes the surface smoother,it makes a longer lasting surface that will remain at that diameter.

One final,but very important reason for grinding is that many,if not most real precision surfaces have been made longer lasting by hardening them. This applies to all kinds of moving and sliding parts as well as gages. They have to be ground since they are hardened.

By the way: If you look at a precision grinding machine while the wheel is grinding a surface,you will notice that the grinding is done at the leading edge of the wheel,not over the entire surface. Plunge grinding operations ,like crankshaft journal grinding,do involve the whole surface of the wheel. Most grinding operations do not involve the surface of the whole width of the wheel. The wheel has to be dressed very often in cases where the whole wheel is used,as on crankshafts,where there's no room to move the wheel sideways. Possibly for every grind. I don't do automotive work.

You really do not need an experienced master to tell you these things. You can look at mechanisms and see for yourselves what is ground and what is not. However,you must always enter into these investigations with an open mind,and not with your mind already made up. Open mind is needed to really learn.

 

[Bill Gruby]

First of all, congratulations on figuring out that trick. That is something a person would normally learn at the elbow of a master machinist and you came up with it yourself. Nice

Grinders definitely create a very fine finish. I don't beleive that, in itself, gives a ground finish any better dimensional accuracy though. Think of the finish as a peaks and valley on the surface of the part. Whether you are talking about a lathe finish of 125 microinches (pretty darn rough) or a ground finish of 10 microinches (pretty darn nice finish) you are still doing your measuring on the peaks of that finish. The very fine tolerances come from the very fine cutting tools (grit of the grinding wheel), the wide contact with the part for even pressure dispersal, the rapidly rotating cutting tools (which makes the grinding wheel essentially one wide tool instead of thousands of tiny tools) and the very fine advance of the grinding wheel usually in the range of ten millionths to a hundred millionths (.000010 to .000100).

Just my two cents,
David H.

The peaks and valleys are not as pronounced on the Grind Finish, hence the super accuracy. I do not grind for a living. I use it on rare occasions in the hobby. My use of it and the results I get are what I stand by.

"Billy G"