RE: [CAD_CAM_EDM_DRO] Ball-screws

-----Original Message-----
From: James Owens [mailto:wotisname@...]

>... The screws claim a repeatability of 50 micron,
>correct me if I am wrong but 0.000005" is quite small.

I believe 50 microns is actually .00197 inches, which is quite big. I suspect I
could get that repeatability out of standard threaded rod.
Even 50 micro-inches would be 0.00005" not 0.000005"

It seems to me that there should be three measurements of interest here:
1) The repeatability, that is, the error you get on multiple positioning
operations to the same theoretical position (always approaching from the same
side, to avoid backlash). This would be increased by roughness of the thread
running surface, the balls, and asphericity of the balls themselves, but in
general would tend to be a pretty small number. Even regular threads with a
regular nut should be able to reposition to within better than 0.001".
2) The absolute positioning error (sometimes called cumulative or linear error),
that is, the error you get between the actual and theoretical distances between
two positions (again, both approached from the same side). This can only really
be measured as accurately as the repeatability. Errors here can (in theory) be
compensated for by the CNC software or by tweaking dimensions. These errors are
caused by poor dimensioning of the overall thread, and cyclic errors can also
occur if the thread is not perfectly helical. This is where standard threaded
rod really loses.
3) Backlash, that is, the error you get in actual position when you move to the
same theoretical position from opposite directions. This is caused by clearance
and material elasticity. Again, this can only be measured as accurately as the
repeatability allows.

If the ballscrews have .004"/foot linear error and .002" repeatability, that
pretty much seems that the linear error is really "about as small as can be
measured given the repeatability".
-Kevin Martin