Lathe leveling and the SB Two Collar Method

It depends on what you are happy with. Yoc could get it a lot better. I get less taper than that(.001) on a ten foot bed American Pacemaker that is as old as that. Its all in how you adjust the leveling pads , if you have the lathe anchored to the floor and how strong the floor is.

It depends on the bed condition - if there is little wear it could be quite a bit better.

If the zero twist setting lifts part of the cabinet off the floor just stick a shim under the offending corner. The thermal expansion, thermal time constant and shift with humidity of almost all floor materials are so different to that of steel that it's usually better to avoid anchoring the lathe too solidly to the floor.

If you're chasing tenths, even a 0.002" cut on widely separated parts of a bar is not totally reliable. It's safer to carefully turn and accurately measure the diameter of two collars. Preferable but not essential that the two diameters be equal.

Clock the first diameter near the chuck with a tenths clock, then repeat it on the second collar, correcting for any measured difference in the two diameters. The effect of any residual eccentricity can be cancelled by making the measurement at the 90 deg point half way between the maximum and minimum clock reading.

With the clock bearing on the collar furthest from the chuck,the beauty of this system is that it can directly observe the effect of adjusting the tailstock end hold downs.

Jim

Steve, Setting up a lathe is a 3 step process. First the bed, then the head stock and last the tailstock. In order to do this, two tools are required. First is the precision level and next a truth test bar. There are no short cuts, especially if there are wear and misalignment issues. First, let's get rid of the wives tail that the lathe must be level. That's a crock of @#$%. We use the level state only for a common reference.

First level the bed in all directions. This may require floor bolting in order to generate enough torque. If you can't get it level, resurface the bed.

Second, install the truth bar into the headstock spindle taper with the flat on top. With a dial indicator mounted on the carraige, slide the carraige back and forth observing parallel error, testing both on the test bar flat (verticle plane) and on the front centerline (horizontal plane). If error is present, try to shim the headstock at the bed mount to compensate or resurface the bed because of saddle wear. Last is the tailstock. To observe tailcenter height error, use the truth bar flat again along with the test bar center.

Cutting a test piece is the acceptance test, not an alignment process. Steve

Sounds interesting but I'm not sure I follow what you are saying. I think you are saying, turn two spots on opposite ends of the bar and carefully measure the diameters. Preferably make sure that the measured diameters are equal.

"Clock" is a term I don't know. I think you are saying clamp a tenths indicator in the cross slide and compare it's reading on the two ends of the bar (maybe adjusting readings if the two diameters were not the same.)

I guess that may make sense. I'm just trying to be sure I follow the description.

How would things get elliptical? That would have to be stuff you can't fix, like headstock bearings or bearings in a live center in the tailstock, no?

So, if I have this right, to correct any difference in readings, how do you twist the tailstock end to fix it? I'd think if the tailstock end reads high (bigger) you would twist the tailstock end of the base clockwise (looking toward the headstock) and vice versa. Correct?

How do you be sure that you have the tailstock centered absolutely right, and you are not turning a taper because the tailstock centering adjustment is off? I think I have the general idea, but it seems like these two issues (twist and tailstock center) are hard to separate. Actually, tailstock height can factor in too.

Is there a scientific way to measure and balance all these things other than trial and error? I'm sure it's been figured out, but it ain't completely clear in my brain.

From your description, I think I can figure what a truth bar is, but this is the first I've heard of one. How long is the projection? Do they come up on eBay?

xray sez: ""> How do you be sure that you have the tailstock centered absolutely

IMO, you should minimize taper from the tailstock first - bearing in mind you can't absolutely "zero" it out if there is bed twist. But, get it as close as possible before attacking the bed. As Jim said, "The effect of any residual eccentricity can be cancelled by making the measurement at the 90 deg point half way between the maximum and minimum clock reading." You can't do much about headstock height, but fortunately for us, an error in height has a great deal less effect than a similar "sideways" error.

Bob Swinney

My two cents: A lot has been said about its not being too important for the lathe to be (perfectly) level, so long as it's true. Well, yeah... but there are some strong advantages to getting it all dead-on level.

The first one is this -- If you use a decent level, you can get virtually all of the twist out of the bed before you ever start worrying about tweaking it to zero.

Say the level is good to 1.5 minutes (90 seconds) per mark. That's about 5 mils per foot. Say the bed is 6" wide. Assume you get both front and back as level longitudinally as you can. Then if you get both the head and tailstock transverse to read splat-on to the same mark, you're reasonably less than 0.001" out for twist.

That would be with a 'basic' machinist's level, like a Starrett Model 98. With an inspection-grade level, you could pull that down to two-tenths on the first pass.

That's a pretty good starting place for the last tweaking, and it's a bit more intuitive than working out a twist by examining the results of a spindle test turning.

LLoyd

In answer to your first question I'm simply assuming that you'll be taking the 1" dia bar that you've already been using and producing two accurately turned and measured turned collars on it, one close to the headstock and the other close to the far end.

To help getting a good finish on the far end collar it is permissible to temporarily improve the rigidity by supporting the end of the bar by the tailstock center while it is being machined. This end of the bar must of course be free and not supported by the tailstock when using the bar for checking lathe alignment

"Clocking" is exactly as you have described it. The probe should be accurately located at centre height so that it duplicates the cutting edge of a lathe tool. A lever type clock can be used but the plunger type is usually more convenient. The collar width should be small and the probe repeatably placed at the middle of the collar width so that any residual taper does not have a significant effect on the measurement accuracy. The bar should be rotated thru a full turn to check for the presence of eccentricity.

The method assumes that the lathe produces cylindrical (i.e. not elliptical) collars but is tolerant of minor eccentricity. For instance if you turned the far collar when supported by a badly aligned tailstock the collar could run slightly eccentric when released from the tailstock centre. However the true axis of the collar can still be accurately measured by averaging the maximum and minimum clock reading or by simply making the measurement at the 90 deg positions, half way between the minimum and maximum points.

With a typical lathe with two hold downs under the headstock end and another pair near the tailstock end, a good strategy is to fully tighten the headstock pair and differentialy tighten or shim the tailstock pair. Clockwise twist increases the diameter that will produced by a cutter mounted close to the tailstock end - the the effect can be seen directly by watching the clock as you tighten the hold downs.

The nice thing about this method is that it doesn't need accurate test bars or precision levels - you only need decent clock and this has a multitude of other uses in the workshop.

It of course ONLY checks bed/headstock alignment. Tailstock alignment is a different matter. There are numerous methods for checking this - my favourite uses a lever type clock supported by a 4 jaw chuck and arranged so that the tip of the lever bears on the inside (or if you prefer it - the outside of a tailstock centre) of the tailstock barrel. With the chuck gently rotated by hand it directly reads out tailstock horizontal and vertical alignment.

Jim

Yes, I have bought 3 different sizes on eBay. I now have one for every lathe I own. Typically they will range from 8" projection on a MT3 bar to over 12" on an MT6 taper. Steve

It seems to me that you need to do some more research on aligning a lathe. From what you have said, you're going about it backwards.

The first step is to get the twist out of the bed. For that, forget you even have a tailstock. A precision level at various places along the bed will tell you if it's straight or twisted. When you've got it straight, then try a test bar in the chuck. In the chuck only - no tailstock. Put a test indicator on the carriage, and move it along the test bar to see if you get the same readings. If you do, the headstock is aligned with the bed.

If the test bar shows different readings, it means that the headstock is not aligned with the carriage travel. Since you've already got the bed straight, there are two possibilities. One: the headstock is misaligned with the bed. Two: wear on the bed is causing the carriage travel to vary. If you can realign the headstock, do it - but keep in mind that some are not adjustable and that this is not the most likely scenario. You can now induce some twist into the bed to compensate for wear, although it would of course be better to rescrape the ways.

Again, the test bar is held in a chuck and is read with a test indicator. Trying to accomplish the same thing by taking test cuts on a bar may not give you the same results because of deflection due to tool pressure. The deflection has two sources - worn chuck jaws and the bending of the test bar. Try pushing and pulling on the test bar and see what happens to the test indicator reading - that will tell you just how rigid your setup is.

Now you can take some cuts on that test bar, supported by the tailstock. Move the tailstock until you get the same readings.

John Martin

Steve Lusardi wrote: snippage

Hell No! Mmmmm.. wives tail, that's sometimes the best kind

WB ...........