Re: My Logan Clunker

Jack,

Doug gives some good info in his response. What he didn't mention is
that EMC has the sort of leadscrew calibration (error correction
tables) he describes. So it is/can be more than a mental exercise.

It might be worth a call to Fred Proctor at NIST to ask about your
specific application. His phone and email is online at the nist/emc
site.

Hope this helps.

Ballendo

P.S. One other way to simplify this problem could be to use a
mechanical system of correction. This allows standard software to be
used. This was described here by Les Watts awhile back in relation to
his gantry mill, but Lindsay books has some booklets on leadscrew
accuracy which will be pertinent.

Assuming there is only a few thou to deal with, what you would do is
to use a "straight cam"(in line with the bed) which will ACT ON the
crosslide leadscrew. You'd make a helical holder for the end mount of
the leadscrew, and an arm riding on/against the "cam" would provide
the compensation, advancing and retarding the crosslide l/s rotation
as necessary. If the lathe has a telescoping taper attachment, you
could put a compensation "cam" back there instead. Downside would be
the limited length of travel for the compensation.

P.P.S. It IS an interesting thought that you have presented here;
i.e., fixing clunker lathes with software!

--- In CAD_CAM_EDM_DRO@y..., jmw@c... wrote:
> If I were to convert my old 12 x 36 Logan to CNC, is it reasonable
to
> expect that the system could correct for the machine's worn ways?
>
> Presumably the system 1) would know the carriage position along the
> long axis and 2) could be given a map of the way wear. While the
> machine ran, a demon in the cross slide could be checking 1) and 2)
> and making little in and out adjustments to correct for the wear.
> Maybe this could be relatively transparent to the user, and maybe
the
> Logan could turn out a cylinder again?
>
> If this is a standard feature in CNC lathe systems, pls excuse the
> bandwidth--I did check the archives.
>
> Thanks in advance.
>
> --Jack