John Hasler wrote:The register at the threaded end is what I called the "back bearing" (Hardinge's term) above. A burr there would angle the collet (which as you note excessive clearence there would do as well).
Correct, but it would manifest itself as a loss of parallelism, not taper. In essence, any machining done to the piece would would be at a slight angle, the amount of the offset created by the burr, but it would NOT be tapered. Only when the piece is offset constantly as it rotates would it manifest itself as taper, which I why I suggested that the tailstock may be part of the problem. It may be offset, coincidence, or not, by the amount that is revealing a taper. The slack of the register (assuming there is any) would permit the collet to "wobble", for lack of a better term, as it rotates, always deflecting away from the cut. It stands to reason that it would diminish to the point of not existing as the cut neared the collet, where there was far more support.
Unless it has slop, it isn't adding any problems, even if the point oscillates a small amount. Because the tool creates a load on the center, it, too, should be offset by a given amount and still support the part as it rotates. It would be my recommendation that a DTI be applied to the point, which should then be rotated to determine if it isn't running true. It would be best done with the center under a slight load, to ensure that the thrust bearing is operating in its expected condition.I think it would be better not to use the live center for these tests.
The OP has clearly stated that he is creating taper. I'm assuming he's not discussing run-out of a part that is installed and a DTI applied. In that case, it might be difficult to determine where the problem lies, as a ding or a mismatch of registers can be the problem. All of my comments relate to a part that is machined in situ.