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

[george wilson]

Setting the compound at an angle has long been standard operating procedure for working to very fine tolerances. The thing IS,a cut surface will really be too rough to really,truly be accurate to a .0001". You need to grind the surface to truly get the surface smooth enough to be at that tolerance. Otherwise,as soon as the turned surface gets the microscopic "fuzz" worn down,it will no longer be at that accurate a diameter.

So,the answer is,set the compound to an angle and use a toolpost grinder to achieve this close a tolerance.

However,then the problem of how good your spindle bearings are raises its head. You have to have a high quality lathe with high grade bearings to have the piece truly round to that tolerance. When I had an Atlas lathe,my grinding came out looking like the wild grain pattern on plywood. The spindle bearings,and the rigidity of the lathe just were not up to the task.

When I got my first Jet lathe,a 10" x 24",the difference was night and day. And,that was still a low cost lathe. My current lathe,a Hardinge HLVH has high grade spindle bearings accurate to 50 millionths,and a very thick,heavy bed. This lathe can actually work to that tolerance.

I had fun experimenting with tolerances too,when starting out. Keep experimenting,and you will increase your knowledge and figure out new ways to do things.

 

[hermetic]

A 1" long x 1" diameter mild steel bar will expand or contract by .0001" with every 10 degree change in temeprature. You are only holding these claimed tolerances at a fixed temperature, unfortunately we do not live in a fixed temperature environment. That is why engineers work to a tolerance. It is impossible to claim accuracy to 0001" when the actual size varies more than that depending on the surrounding air temperature. It is ok to do it just for fun of course, but don't fool yourself into thinking it will be the same size on a frosty morning. (Thanks to John Stevenson for the above figures)
Phil

 

[spaceman_spiff]

Great info guys!

I tried to reduce variability even further by doing my tests cuts on the same location on the compound, to try and eliminate error from the compounds ways and screw, I did this by:

1- making the first cut using the carriage handwheel.

2- measuring the result of #1

3- making the second cut using the compound handwheel

4- measuring the result of #3 and computing the "handwheel to feed" ratio at that range on the compound

5- moving the compound back to where it was before #3

6- repeating #1 through #3 and comparing this #3 to the previous #3

and I still saw variability around 0.0005".

These cuts are back to back, same workpiece, same cutter, same temperature, same location on the ways, etc..

I think there is at least one more "dumb" thing hiding somewhere thats costing me 3 tenths.

So I'm going to throw the following at it and see if it helps:

1- try a very sharp HSS bit as suggested
2- clean and shield the main lathe ways in case small amounts of crud are getting under them and shifting the whole carriage
3- figure out a way to move the tool in without the compound slide..perhaps a micrometer or something
4- ditch the 4 jaw and use the ER chuck thats coming in the mail

And yes its true, the only thing holding the compound slide against its screw is the cutting force. You can push it forward at any time and modify the cut tremendously. I wonder if somehow sliding the tool forward in its tool post using a micrometer would work better, while keeping the gibs locked down so the compound can't even move. HMMMMM..

I measure the cut without looking at the display on the mic, mostly because I have to focus on getting the anvils centered. Then I use the same number of clicks on the thimble. Then I rotate the workpiece half a turn or so and repeat. It seems like I regularly get zero variability in the mic measurement this way.

 

[John Hasler]

Great info guys!

I tried to reduce variability even further by doing my tests cuts on the same location on the compound, to try and eliminate error from the compounds ways and screw, I did this by:

1- making the first cut using the carriage handwheel.

2- measuring the result of #1

3- making the second cut using the compound handwheel

4- measuring the result of #3 and computing the "handwheel to feed" ratio at that range on the compound

5- moving the compound back to where it was before #3

6- repeating #1 through #3 and comparing this #3 to the previous #3

and I still saw variability around 0.0005".

These cuts are back to back, same workpiece, same cutter, same temperature, same location on the ways, etc..

I think there is at least one more "dumb" thing hiding somewhere thats costing me 3 tenths.

So I'm going to throw the following at it and see if it helps:

1- try a very sharp HSS bit as suggested
2- clean and shield the main lathe ways in case small amounts of crud are getting under them and shifting the whole carriage
3- figure out a way to move the tool in without the compound slide..perhaps a micrometer or something
4- ditch the 4 jaw and use the ER chuck thats coming in the mail

And yes its true, the only thing holding the compound slide against its screw is the cutting force. You can push it forward at any time and modify the cut tremendously. I wonder if somehow sliding the tool forward in its tool post using a micrometer would work better, while keeping the gibs locked down so the compound can't even move. HMMMMM..

I measure the cut without looking at the display on the mic, mostly because I have to focus on getting the anvils centered. Then I use the same number of clicks on the thimble. Then I rotate the workpiece half a turn or so and repeat. It seems like I regularly get zero variability in the mic measurement this way.

You might want to try combining these techniques with a shear tool. It is supposed to provide a very good finish while consistently removing 1 or 2 thousandths accurately (I've never tried one myself).

http://www.gadgetbuilder.com/VerticalShearBit.html

 

[spaceman_spiff]

A 1" long x 1" diameter mild steel bar will expand or contract by .0001" with every 10 degree change in temeprature.

F or C? If its F, the expansion ratio is actually 0.0000645in/in/F, if its C, its 0.000120in/in/C. (Thanks to Machinery's Handbook for the above figures)

You are only holding these claimed tolerances at a fixed temperature, unfortunately we do not live in a fixed temperature environment.

Ring gauge standards accurate to millionths of an inch are specified at a certain temperature. The machining process is not less accurate because its result are a certain value at a certain temperature.

That is why engineers work to a tolerance.

No its not. Tolerances exist for economical reasons, not because of thermal expansion. Thermal expansion increases the need for tighter tolerances, it doesn't reduce it. Tolerances are not a design spec, they are a manufacturing spec to reach the design spec.

It is impossible to claim accuracy to 0001" when the actual size varies more than that depending on the surrounding air temperature.

Hrm... master gages, interference fits, ball bearing diameters, precision 123 blocks, etc..etc..all have specifications equal or tighter than 1 tenth. The machining can be that accurate. Thermal expansion is irrelevant. Nobody is claiming they are machining something and that it suddenly becomes immune to thermal expansion.

It is ok to do it just for fun of course, but don't fool yourself into thinking it will be the same size on a frosty morning. (Thanks to John Stevenson for the above figures)
Phil

So higher precision work is only okay for fun? Is that really your argument? Think about that...

 

[spaceman_spiff]

Have you checked the finish cut for taper?

The test cuts I've been making are only about 1/2" long, I measure in the middle. The ways definitely have significant wear and there probably is a problematic taper that needs to be dealt with beyond the range I'm doing these tests in, but I'm trying to solve one problem at a time and just focus on cut accuracy at a very limited carriage travel. I figure if I can get repeatability and accuracy with 2 or maybe 3 tenths, then I'll call it a day and move on to the zillion other inaccuracies this worn out hunk of iron has.

 

[spaceman_spiff]

I swing the compound as you describe when I'm using a tool post grinder. The one thou graduations on the cross feed are pretty heavy cuts for the grinder. When trying to take the light final cuts you might have some luck using a HSS bit ground as a shear tool, they're great for shaving those really light passes.

Greg

Something I dont understand about the grinders is, they are still subject to all the inaccuracies of the ways and slop in the screws and backlashes, so the only improvement to accuracy they would provide is by virtue of the lighter but more consistent cut forces right?

 

[Bill Gruby]

No, the accuracy in grinding comes from the finish. The smoother you can get the part the closer it will be. Ground finishes with the right set up will get you much closer.

"Billy G"

 

[george wilson]

Exactly what I said about ground finishes,Bill. Turned finishes are just not smooth enough to base really small tolerances on.

 

[spaceman_spiff]

No, the accuracy in grinding comes from the finish. The smoother you can get the part the closer it will be. Ground finishes with the right set up will get you much closer.

"Billy G"

Could you explain this further? I have never been able to get a grip on grinders and why they do a better job as far as accuracy, without the machine they are attached to also being better.

For instance, lets say the finish using the turning tool are relatively rough, but consistent. Then as long as the increase in cut is made relative to the measured finish/diameter combo, the quality of the finish falls out and isn't part of the result anymore.

Another way to look at it is this, I'm seeing errors in the range of 0.0005" to 0.001". The finish is much smoother than that, even just eyeballing it. A 1000 microinch finish is what you get with a saw or flame cut apparently. So as long as the finish remains consistent, I dont see how improving it would increase accuracy.