>It depends on the axis that your looking at, and the type and date of
>manufacture of the Lathe. If it is the tool bed i.e. The bit that moves
>along the X axis, then a shaft encoder connected to the ACME thread is all
>that's required to give that sort of measurement (With reference points
>along the bed of course).

Backlash would kill you with this setup.  All the CNC machine tools I've
seen used optical encoders on the motors or ballscrews.  They could get
away with this for two reasons--ballscrews have less backlash, generally,
than an Acme screw and nut, and they can compensate for the residual
backlash in software.  Ballscrews usually aren't practical for manual
machine tools (too easy to backdrive).


Most of the DROs for manual machines I've seen have used a simple linear
quadrature optical encoder.  With good seals, the connector or cables give
out before the glass.  (I seem to recall replacing an IR led on one unit,
but that's unusual.)

I've seen the Sony ads, and they really push the durability aspect of their
units.  I've seen one, but I didn't get a chance to open it up.  I suspect
it's inductive rather than magnetic.

I've seen references to "Inductosyn" linear sensors in some old CNC books,
but I've never seen one.  It certainly sounded rugged, but I think the
drive requirements were a problem.  (I vaguely recall it being the linear
version of an analog resolver, but I could be way off here.)

The common handheld digital calipers use a capacitive sensor, I think.
(Well, except for the old Brown and Sharpe I've got, which uses a optical
glass scale.  I've only seen one example though, so it must not have been
very popular.)  All the handhelds  I've used have been _very_ sensitive to
moisture.  Mitutoyo has a cool absolute caliper that won't lose zero, but
there's a small price premium involved...

Long stroke linear pots do exist, but they're uncommon.  LVDTs would
probably work well if short distances were involved.  Of course, both of
these sensors are inherently analog...


newell