Close encounters of the CADR kind

Noel Chiappa jnc at mercury.lcs.mit.edu
Mon Jan 16 11:41:23 CST 2017


    > From: Al Kossow

    > I assume that CONS ww panel we had on display in Boston was there
    > because it was wired by a robot they built.

I don't know for sure about the CONS, but I'm pretty sure no CADR's were wired
at the AI Lab; ISTR that some outside fabricator did them. (Probably a place
with tape-controlled wire-wrap machines.)


The robot that the AI Lab _did_ build, that was used in the production of
CADR's, was a wire-wrap testing machine, used to verify the huge CPU panels
immediately upon arrival. It was this large frame (made out of Dexion, IIRC)
in which the CPU panel was laid horizontally.

It had two 'test heads' (not sure of their exact name) which were moveable
(details below), each of which had a little arm with a little round solid test
probe sticking down. The test head would roll up to where the pin it wanted to
continuity check was, and lower the little arm (it moved about 1/2", or so,
IIRC). If the probe wasn't centered exactly on the pin, the test probe would
slide off the pin, and go further down, and the machine could sense that. So
it would raise the probe, and hunt around a bit in a trial-and-error process
until it landed on the pin. Then the other test head did the same thing, and
then it could do the test (dunno if it checked for shorts, as well as pin-pin
continuity per the wire list); repeat ad nauseam.

The two test heads were supported on beams (supported at both ends, IIRC)
which could run back and forth on tracks, with the heads moving up and down
the beams, all driven by a pair of plastic chains (which looked like miniature
rope ladders), one each for X and Y axis motion. Those were driven off a
couple of stepper motors attached to the frame, IIRC. I don't recall how it
was done so that the two heads and their support beams didn't interfere with
each other - possibly there was one on each end, and each had half of the
panel to itself? (I think probably that, now that I think about it.)

It was the most amazing Rube Goldberg device, and a blast to watch running,
but it did the job.

	Noel


More information about the cctech mailing list