Digitizing arm - CMM

>I've got an Immersion Microscribe digitizing arm, and although it was an
>expensive piece of equipment ($3k) it doesn't seem very complex
>mechanically. The whole thing swivels on its base, then rises about 10
>inches and supports the main arm, which pivots in a plane perpendicular to
>the table it sits on. At one end of this section is a counterweight, at the
>other another pivot with a shorter section of arm with another joint from
>which the stylus depends. All the joints, with the exception of the
>swiveling base, are parallel, operating in the same plane of movement.
>These are also very heavy-duty joints, in what appears to be die-cast
>metal. If someone could figure out the angular encoders and software
>interface, the rest could probably be done in a simpler manner if someone
>had a reasonably well-equipped machine shop.

The way you describe it, it only uses 4 encoders? That shouldn't
be too difficult to do, I'd think. Using 2000cpr encoders you
could probably get pretty darn accurate too. It would be direct
drive to the encoders, so no backlash to worry about either. The
only critical thing would be to get the dimensions of the arms
dead accurate or your readings will be off. Using the index pulse
off the encoders would also be a must, and the initial dialing in
to determine where each arm is in space is also very critical.
Luckily, this is all one-time work, and once it was done, you
could just get down to using it.

As far as the software goes, I could handle something like that
pretty well I think. Well, let me get a little specific on that
one: gathering the point cloud would be really easy, but what do
I do with it then? Output into some kind of CAD program or
something? You'd need to stitch all those points together into
surfaces or you wouldn't be able to do much with it.

Give me a few ideas about how to use the output data, and we could
really go somewhere with this. I've got a few rotary encoders
sitting around here, and I could probably lash up something that
would be really inaccurate but good enough to test the concept.

-Tom Kulaga