View Full Version : two collar method questions

05-20-2012, 08:24 PM
I have "semi" leveled with my level and can't do anything more until I can borrow my friends Starrett or have him come over.

I've read several post here on the two collar method and tried to find it else where on the internet.

Is there a write up somewhere of how to perform this test and what the results actually tell you? I've seen several post on how it's done, but there isn't much on what the results mean or how to correct it.


05-20-2012, 08:49 PM

The basis is to have a workpiece that has a center section turned smaller so that it leaves two larger diameter rings, one near the headstock and one at the opposite end. You can then take a small cut off one ring, then without moving the crosslide go down and take the same cut off the other. Measuring both rings will tell you something. Most often (?) the test piece is held between centers so this will tell you how much the tailstock must be moved to be centered - that is, to cut a piece with no taper.

Used with the RDM, the piece is not supported by the tailstock but is in free air. In this case what it purports to tell you is whether the bed is twisted one way or the other.

This is a great method for tailstock alignment but I'm a little hesitant to call it good with RDM. To get the best resolution you want as long a workpiece as possible, but with that much hanging out of the chuck the probability of deflection muddying the waters I think is too great to be reliable. YMMV

Forrest Addy
05-20-2012, 09:06 PM
Here,s my rant on the whole topic of lathe leveling that includes the two collar method.

Leveling an engine lathe by Forrest Addy:

Generally it goes like this on the better and larger lathes and import and bench lathes as applicable:

Run the tailstock to the far end to get it out of the way.

Level across just in front of the headstock using the spindle end jacking screws.

Locat a point on the near and far way midway between the spindle ends jacks and the tail stock pedistal jacks as londitudial leveling points. Place the level at these points and adjust the near and far way using the tailstock end leveling crews.

Check across at the headstock and tail stock and adjust. Check longitudinally at the headsstock end. You'll probably find some hump in the bedways at this point because the change gear end of the headstock is unsupported. Lay the level on the near (operator side) way and slowly adjust the near change gear end leveling screw up until you get first movement on the level. Move the level to the far side and repeat with far side.

Start over, refining the leveling this time snugging the locknuts. It's quite possible you'll have a jacking bolt with no weight on it. If the floor is stout (5" or thicker) and intact drill for hold-down bolts working the bolt against the jack screws. You may find your body weight is affecting the leveling. This may indicate the soil has subsided under the concrete. If this is a problem talk to a concrete contractor about slab jacking.

Now to tune up the spindle. For that you need a precision ground test bar unless you’re resourceful. I hate the factory test bars. They're always limber and crooked.

I prefer a home made mandrel in a process I’ve come to call the “two collar method.” You need nothing more than a $4 piece of pipe to accomplish the same test as a $500 test bar. Chuck a piece of Sch 40 pipe that will just clear the cross slide about a foot long. Using HSS and light cuts, turn a couple of short diameters about 1/4" long - one next to the jaws and the other at the far end. Relieve the pipe between the diameters. The pipe is sure to chatter; plunk a piece of plywood against the end of the pipe so it stays there with the center's thrust into the raw wood. This will damp out the ring and chatter without influencing the final straightness of the pipe.

Individually finish the diameters to the same size within 0.0005” and give them a light polish so the indicator doesn’t snag. .Don’t sweat making a perfect cut at the same tool setting. Cut them separately. Check them with a mike and a dial indicator. Find the average point of both diameters (there's always a trifling amount of run-out when machining pipe). These average points are used when checking for spindle parallelism with the carriage motion.

Mount an indicator on the carriage so it registers the diameters in the vertical plane. Traverse the carriage along the ways. Adjust the change gear end to bring the spindle axis parallel with the ways in the vertical plane.

Repeat moving the indicator and differentially adjust the jacks to bring the diameters into parallel in the horizontal plane. A second indicator on the vertical plane will allow you to adjust for the horizontal plane without disturbing the vertical. You may need a helper because this is the point where alignment becomes complicated. Some headstocks on the better lathes are set for parallelism by precision scraping their fit on the bedways. Others have a means of angular adjustment. This can take many forms from opposing jacking screws, to a pulling bolt and a pushing setscrew accessed from the same side. Cheap lathes may use shims. You will have to look if you don’t have the original installation manual

OK assuming the headstock is in alignment, the next step is the tailstock. A precision ground bar between centers is nice if you happen to have a precision ground bar.

When setting up and leveling a used lathe, sooner or later I conduct a short survey of the tailstock’s condition answering to my self the following questions: Is the quill a snug fit in its housing bore? Does the tailstock base fit flat on the bed so I can’t get a 0.001: feeler under it anywhere. Does an indicator scanned along the fully extended quill shows its axis to be parallel to the carriage motion to 0.0005” per ft.? Is the internal taper clean, smooth, unscored etc?

If OK I turn a slug gripped in the chuck to the tailstock quill’s diameter. Then I move the tailstock to an unworn portion of the bedways. Setting an indicator zero on top of the slug in the chuck I then crank the carriage to the tailstock and check the top of the extended quill. In a new lathe the indicator will register zero to +0.001 with the quill rising. A worn lathe may have a tailstock whose base is shaped like the runners of a rocking chair. A re-scrape to correct fit and a shim to raise the tailstock quill to the spindle axis is a regular task when re-commissioning an older lathe. . .

Leveling and aligning a lathe is not a 1-2-3 operation. You'll have to go over the machine several times to bring it in level and then in alignment converging eventually on the best results the machine’s condition will allow. If the machine has significant wear you'll most likely have to compromise.

Finally, a worn lathe with significant hollow in the ways and rocking chair on the saddle may not be capable of alignment to turn straight diameters over 6 inches long. There is no golden cure, no trick of alignment or leveling, no special techniques that will eliminate the effect of wear. You can work around wear with skill and technique but it will always be present interfering with efficient and often competitive operation to some extent.

Do not use "Rollie's Dad's method" for lathe alignment. It really doesn't tell you much.

05-20-2012, 10:03 PM
Thanks for the replies.

I've been doing more reading on my search for the "two collar method".

The lathe is a Birmingham ct14X40g Taiwan mfg. It was "lightly" used by a hobbiest. I can still see the original grind marks on the ways from tailstock to spindle. I've checked the run out on the 3 jaw 3 seperate times now using the 1.001" X +/- 30" hydraulic ram. Everytime I've chucked it up the run out is .0005 or less. I think the original owner must have ground the chuck.

I am using "jack bolts" for leveling, but this lathe only has 4 holes for inserting bolts. I've read in other post not to bolt a lathe down, as it may induce twisting and stresses. Not sure I want to start drilling holes and putting anchors down just yet.


J Tiers
05-20-2012, 11:33 PM
There may also be some crazy 'chasing perfection" stuff going on......

I have a machine that, when new , was guaranteed to turn to 0.0001 (a tenth) over 6 inches, which would be basically using the two collars method.

Mine is not new, and I don't expect to get it back to quite that level of perfection.

I don't suggest you do either.

But, with bed wear, and the like, it is possible that merely leveling wil not get you all the way to a reasonable taper. The "TCT" (two collars test) will get you closer than mere leveling, because it tests what you will actually do, which is turning a part.

You might also notice that that manufacturer I mentioned, they didn't spec past 6 inches.

That no doubt is partly because past that you likely need the T/S, and also because even they, with their emphasis on precision, were unwilling to guarantee more than that. yet I read of people trying RDM with 12" or longer test bars..... chasing tenths at the end of those bars......... I believe they are really fooling themselves. I suspect that is about the same as reading the dial of your caliper with a magnifying glass, and interpolating.....you have no right to expect that the measurement is good to that accuracy. ;) ;)

05-21-2012, 12:13 AM

In the two collars method you documented above, it seems you assumed that the headstock on this Taiwan lathe is accurately aligned with the ways so any error must be due to bed twist.

If there happens to be some swarf under the headstock, wouldn't this cause you to twist the bed to compensate for the swarf?

Shouldn't you verify the headstock is aligned to the bed rather than assuming that it is?


J Tiers
05-21-2012, 12:35 AM

In the two collars method you documented above, it seems you assumed that the headstock on this Taiwan lathe is accurately aligned with the ways so any error must be due to bed twist.

If there happens to be some swarf under the headstock, wouldn't this cause you to twist the bed to compensate for the swarf?

Shouldn't you verify the headstock is aligned to the bed rather than assuming that it is?


Not so.....

You level first, then use the test to verify the functional alignment.

If you HAVE headstock adjustment, nobody is stopping you from trying that..... The test is really a test OF effective H/S alignment.

But my suggestion is that unless the error is huge, you CAN assume that it is remaining twist, and adjust it out with very minor adjustments. "Huge" here would be on the order of a thou, not tenths.

The idea of leveling is to ELIMINATE bed twist issues as a major problem.

if you CAN adjust out the error with the H/S adjustments, I see a good reason to DO that..... you already did your best to "un-twist" the bed, so it's the thing you currently have the most confidence in, assuming a good level.

But, for very small errors, probably either works. And if you HAVE no adjustments for the H/S, who yah gonna call other than bed twist?