Leveling and getting my lathe setup for better preciesion

yendor

Active User
Registered
Joined
Dec 11, 2013
Messages
302
Ok, so LEVEL as we know is a relative thing.
TWIST is the thing to be concerned with right.
I spent some time with a Starett #98 and some shims and have things dialed in as far as the level bubble indicators go front to back at both ends.
NOTES:
It is NOT Bolted to the floor.
The original Atlas legs are mounted to 3" Angle Iron with Castors (but I have them locked in position so it hasn't moved that I can tell)

Now for my (2) Questions:
With what started as a 1" diameter Rod with 3" inside the chuck jaws and 4-1/2" unsupported I have approx .003 of taper over 4".
(it was much --much worse when I started).
Should I continue to chase it down or is that pretty good?
Or is the .003 just tool deflection over the 4" of unsupported stock?

If I check the Lathe Ways end to end both ends I find there seems to be a BOW in the bed as the Left (Tailstock End is (3) Bubble increments lower than the Ways measure End to End near the head stock.

I found an old Atlas leveling document that says the Ways should be leveled end to end is that where my .003 taper comes from?
It shows measuring and leveling out both front to back and end to end - see attached PDF.
 

Attachments

  • Leveling an Atlas Lathe-7401.pdf
    128.9 KB · Views: 25
I would be doing some more shimming.
That is .009 per ft.
 
Yes, more shimming, a Starrett #98 level is not sensitive to do the whole job. You should machine collars on the shaft so that tool wear does not enter into the disparity in sizes, and take very light cuts. I doubt that an Atlas lather is ever going to give perfect results, but is should be better than you are getting.
 
My 1947 SB9C is at .0005 per 6".
 
The first thing that comes to mind is your tool because when doing a leveling process, the tool matters. It must be able to cut at very small depths of cut without deflecting much at all. In most cases, this tool will be a very well ground and honed HSS tool. You cannot use an inserted carbide tool with just a thou or two depth of cut because the nose radius will cause it to deflect. So, tell us about which tool you're using and better yet, show us the tool.

Next, a very short rod held just by the chuck is actually an excellent way to assess headstock alignment with the ways; it is not long enough to accurately detect twist in the bed. Assuming you are using an easily machineable material (like 12L14 or aluminum) and took cuts that were intended to minimize deflection and provide a good finish, the differences in the OD at the chuck and unsupported end on a short rod will tell you more about headstock orientation with the ways.

If you wish to assess for bed twist (more commonly known as "leveling" the lathe) then use a rod with about 10-12" sticking out and do a 2-collar test. I suggest an aluminum rod about 1-2" OD for this and use a really sharp HSS tool to make your cuts. Note that a Starrett 98 is not a really sensitive level but it will work if that is all you have; it will just take a lot more cuts to dial the lathe in.

As for what is an acceptable amount of delta between the two collars, that is up to you. Most folks would accept 0.0002" or less between the two collars as level enough for a hobby shop. I assure you that it is quite possible to dial it in to zero if that is what you want.
 
.0002" taper per foot is good enough for nearly any purpose. generally, .0005" is spec. for an ordinary industrial lathe, half that for a toolroom lathe.
 
Something similar to Rollie's Dad's Method should help you improve beyond one thou per inch taper.

First, figure out a way to apply twist to the bed. Shims under one of the tailstock mounting feet works if your machine is bolted to something stiff in torsion.

Second, get a decent chunk of shafting (https://www.mcmaster.com/1886K42/ is less than $20) that will fit in the best chuck you have. You want it to be round and have a consistent diameter, straight doesn't matter so much. Stiff and not too heavy is nice.

Third, chuck it up, mount a dial indicator on the carriage, and measure runout horizontally on the shaft near the headstock while turning the spindle by hand.

Figure out the *average* displacement. If the indicator moves from plus one thou to minus one thou, the average is zero. Plus 3 to minus 1, the average is plus 1. Because the shaft has a very consistent diameter and is very round, the average of min and max displacement will be the distance to the spindle axis.

Fourth, move the carriage a few inches down the bed at a time, repeating the measurement (average of min and max reading) as you go. Write down the values.

Twist the bed a little and remeasure, repeat until you have things the way you like them.

The nice thing about this approach is that it measures the thing you are trying to improve - distance from the tool mount to the spindle axis as you move along the bed - instead of measuring a bunch of different things that contribute to the end goal.
 
Do you even need a level? I was hoping to get mine straight by measuring two journals that are several inches apart after shimming and turning until the taper is gone or as close to gone as possible. Let me know if I’m wrong
 
It is not tool deflection. That can be expected to be relatively constant from one end of the bed to the other. It doesn't seem to sink in to some people but there is good reason for the factory instructions always assuming that owners will follow them and properly anchor the machine. The second thing that some people ignore is that castors always have give that varies from minimum load all of the way up to failure. If you can not live without the castors because you lack the room to properly anchor a machine and leave it, then use retractable castors, like those that come on most Shopsmith machines. Then you may be able to get away with it. Or you may not. It all depends upon how the bed has warped over the decades.

But anyway, 0.009" per foot is pretty bad.
 
Do you even need a level? I was hoping to get mine straight by measuring two journals that are several inches apart after shimming and turning until the taper is gone or as close to gone as possible. Let me know if I’m wrong

Carpenters level will get you close enough for a 2 collar test.


Sent from my iPhone using Tapatalk
 
Back
Top