Eccentric turning - not intentional

[CalM]

Be sure to use a very short rigid spotting drill or centerdrill. The point shape of the spotting drill is important. A conventional grind on the drill leaves a small flat at the end of the web. That can cause the drill to walk when it contacts the work. If that happens you can see the drill wobbling when you start the hole.
You might also want to check the tailstock alignment. If the spotting drill is not on center when it contacts the work there will be a greater tendency for drill to walk.

You didn't say how far the part projects from the chuck when you start drilling. If it projects a long way that will exacerbate the problems related to runout. Also, with ER collets the workpiece needs to held for the full length of the collet. If the work is chucked only part way in the collet the back end of the collet will collapse more than the front, resulting in a tapered bore. That will allow the work to wobble in the chuck resulting in an eccentric hole.

I suggest the following steps to find the cause of the problem:
Verify that the collet is running true.
Make sure the part is held full length of the collet.
Check tailstock alignment.
Use a rigid spotting drill

If I may criticize...

A collet running true means nothing to the first side, but has a lot to do with eccentricity when the part is flipped end to end for repeat operations . CHECK That!

Part held full length of collet is not an issue IF the part runs true while the center drill is applied. Bumping the part to run true is not unheard of!

Tailstock alignment matters if the tail stock quill is not stiff enough to Cut a center rather than just follow the center of the drill. Some machines are stiff enough, some are not. AS mentioned, easily observed.

WTF is a "rigid spotting drill", If the drilling bit flexes in the cut, it is NOT a spotting drill. ;-)

[fixerdave]

Image is not working but to me it looks like more runout than what collet chuck would cause.
Here's the image if I had better luck:

Wouldn't you get just that if your tailstock was not aligned?

The hole would be on center but the OD would, if you used the center while turning, end up a taper, smaller at the end, and not concentric around the hole.

Have you measured the diameter of your part after turning to see if it holds along the length?

David...

[Paul Alciatore]

One of three things, perhaps four:

1. Bad collet

2. Bad collet chuck

3. A chip or other foreign item between things

4. The spindle itself is off

Remove everything. Carefully clean everything: use a flashlight to look into the bore of the spindle and of the collet chuck.

Check the spindle for runout.

Mount the collet chuck and check it for runout.

Install the collet and tighten it on a piece of drill rod or precision shaft. Check for runout on the drill rod or shaft. Do this close to the collet and an inch or two out from it and compare the readings.

One of those should reveal the problem.

[754]

Drill it all from one end, then finish the od . How long is it?

[CalM]

Wouldn't you get just that if your tailstock was not aligned?

The hole would be on center but the OD would, if you used the center while turning, end up a taper, smaller at the end, and not concentric around the hole.

Have you measured the diameter of your part after turning to see if it holds along the length?

David...

NO!

If the tail stock is out, the starting drill will first find an eccentric "center, but based on stiffness, will soon enough find the center of gyration of the spindle.

The animation should be obvious to all but the most callous observer.

[754]

FWIW, I think those are usually made from tubing, and drawn shells.
You may be able to buy something close.

[fixerdave]

NO!

If the tail stock is out, the starting drill will first find an eccentric "center, but based on stiffness, will soon enough find the center of gyration of the spindle.

The animation should be obvious to all but the most callous observer.

Yeah... I miss-read the OP. It's off at the collet end, not at the center. And, yes, a center would, well, center on the hole, so it would have to be a mucked live center to be off like that. But, wrong end anyway.

So.... (much snipped)... what Peter said,

David...

[darryl]

I would chuck the part then turn just enough off the od to true it up. Then turn the part over and true up the rest. Remove the part and roll it on a surface plate, or anything flat. If the two turned sections don't align perfectly, your collet is probably not able to keep the part on axis, even if the exposed portion of the part seems to center well near the end of the collet. Another test would be to mount a length of tgp rod, say about 6 inches long, and insert it only as far as you would your parts. With the collet lightly drawn up, see if you can induce wobble in the rod, and see if you can easily knock it into running true. Repeat this test with the collet drawn up more tightly.

I suspect that the collet isn't capable of holding the part true when the part is inserted only a short ways.

[photomankc]

Yes. One of the tests i did for my SetTru 3 jaw was to do just what is mentioned by darryl and CalM: Turn one side true, flip the part keeping the alignment to the jaws and then turn the other side to the same diameter. If the chuck is really on those two should be REALLY close to being perfectly matched. On mine the difference was under 0.0005" so the set true stuff was working well.

One thing that has helped me track down problems has come from some good hard slaps to the face online. I was told to stop "knowing" what was already good and start measuring everything from the ground up. In this case I would say: Measure the naked spindle for run-out and mating surface squareness. Then measure the collet chuck's mounting plate for squareness to the spindle and the boss for any run-out. Then measure the collet chuck's ground surfaces. Finally measure a pin, ground rod, or something you turned in a single operation in the collet. This should give you an idea where the issue starts OR verify that those things are providing the expected accuracy.

I own a CDCO collet chuck boat anchor. After tearing my hair out trying to find the issue I eventually discovered that the chuck was ground with conical run-out. When clamped to a 1" bar of material that was turned true, the back of the chuck wobbled...visibly. Complete garbage. This testing even showed a problem in my 4 jaw chuck as well. It also suffered a similar issue that seemed to come from the jaws because everything else measured out damn near perfect. The jaws were clamping things at a slight angle to the axis which caused similar conical run-out. You could of course turn something true in the chuck, but flipping the part and centering at the chuck could not center the part at any other distance from the chuck.

Had I not gotten methodical about measuring everything I would have spent a lot of time not understanding the problems I was running into.

[Kleinpm]

One of three things, perhaps four:

1. Bad collet

2. Bad collet chuck

3. A chip or other foreign item between things

4. The spindle itself is off

Remove everything. Carefully clean everything: use a flashlight to look into the bore of the spindle and of the collet chuck.

Check the spindle for runout.

Mount the collet chuck and check it for runout.

Install the collet and tighten it on a piece of drill rod or precision shaft. Check for runout on the drill rod or shaft. Do this close to the collet and an inch or two out from it and compare the readings.

One of those should reveal the problem.

It ended up being the clamping collar that clamps down on the collet. Run out was acceptable inside the collet holder but when I put in a collet and snugged up a drill rod run out was .008 with about 1 inch exposed and there was a noticeable wobble with about 3 inches exposed. I have two er-32 collet holders - MT4 and MT3. I swapped the clamping collar/nut from one to the other and the problem went away. I swapped them back and the problem came back. I swapped a few more things around just to confirm but the problem followed the collar. It seems like the clamping collar was clamping the collet in a little cockeyed.

Thanks for all the great suggestions and help. I really appreciate it.

Patrick

[wheeltapper]

a stupid question I know, but, are you putting the collet in the collar properly?

just a thought.

Roy.

[Jaakko Fagerlund]

And just to confirm as you are using ER collets, do you have material the whole length of the collet? If not, it will tighten whatever way it wants to. To chuck shorter than collet lengths you have to insert a dummy rod in the back of the collet so it squeezes closed parallel to the work.

[tmarks11]

I suspect, as other have said, that the collet chuck isn't running true. when you flip the part over, run a dial indicator on the surface of the part and see how much run-out.

Failing that, maybe your drill is walking and cutting off center?

Did you use a spotting drill first (not a center drill)? If not, that explains a lot.

If you need that second hole dead-accurate, consider drilling undersize and boring if possible (although it might be kind of small for that).

[RB211]

Collet runout, stock not round, or tailstock is off when drilling. I have yet to see a chinese small lathe where the tailstock was true in any axis.

[BCRider]

When I want a part to be truly concentric I do all the operations to the outside and inside with the raw material held in one complete setting. When you do things from one end and start flipping them around if the setup is not truly very accurate you will find that any error typically doubles up unless you take care to find the "null error" position if there is one.

So I'd suggest that you alter your methodology to allow for cutting the inside and outside of each collar all in one holding setup. That ensures that the parts are in fact true. If you're doing enough of these in batches then setting up stops and cutting jigs would save a lot of time. For example a tail stock tool head with a boring bar and an external cutter to both bore the pilot hole out to proper size as well as trim the outside to the right wall thickness in one go. A ruler or dial gauge would measure the depth of this dual cut. Or figure out some sort of stop to control the cut in a more "production" like manner.

If the outsides need some decorative cuts then do the "sizing cuts" using a tail stock tool head and do the outer decorative shaping cuts with cutters in the tool post including the parting off cut. Then pull the stock forward using a positioning stop and repeat. This would avoid any accumulative or other repositioning errors.

It also sounds like it would be highly advantageous to get back to basics and check your tail stock alignment. A good way to do this is to cut a piece in the chuck so it is very close to the same diameter as the tail stock ram. Then with a dial gauge on the carriage check for a match between the turned test stub and the tail stock ram. You can do this both horizontally and vertically and even traverse the extended and locked ram to see if it is sagging or pointed up slightly WRT the bedways. Side to side error can be zeroed out with the offset screws. Vertical error and sag might need some shimming between the base and upper parts of the tail stock to correct such an error unless some serious amount of wear is present.