sfriedberg
Diamond
- Joined
- Oct 14, 2010
- Location
- Oregon, USA
So one technique that was used is error mapping, combined with a mechanical cam cut to compensate for the measured error at a particular position along the existing screw. This was not used on lathes used to make arbitrary parts, just on special lathes used to make precision lead screws. Basically, the feed nut was not held rigidly by the apron. It was pinned w.r.t left/right motion, but allowed to rotate on the existing screw. A cam follower was attached to the feed nut. A long linear cam was fastened to the bed. The cam follower (and thus feed nut rotation) was controlled by the linear cam. Because the amount of feed nut rotation was obviously limited, you could not take out large scale pitch errors this way. But you could readily correct for pitch variations by making an appropriate cam.I still don't quite understand how an iterative process allows you to take a lead screw that's out by such and such tolerance and continue running it through a process until it reaches some arbitrary degree of accuracy,
Another technique, which was used with less precise starting screws, was to have the carriage with two separated single-tooth followers, rather than a feed nut. This gives you an averaging behavior, and if you iterate the process, you will end up with a more uniform pitch along the final lead screw. What you won't have is good control over the final pitch, just that it's more uniform. Also you lose effective screw length on each generation, so you start with a substantially longer screw, or at some point in the process you reposition the carriage and pick up a thread on the screw you are cutting to extend it.
[Added in edit] Ah, I see Joe Gwinn described the first technique, and something closely related to the 2nd technique, back in post #5.