Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!swrinde!elroy.jpl.nasa.gov!jarthur!nntp-server.caltech.edu!woolstar
From: woolstar@nntp-server.caltech.edu (John D. Woolverton)
Newsgroups: comp.arch
Subject: Optical Interconnect could be rather easy!
Message-ID: <1991Apr10.050149.21587@nntp-server.caltech.edu>
Date: 10 Apr 91 05:01:49 GMT
References: <12618@pt.cs.cmu.edu>
Organization: California Institute of Technology, Pasadena
Lines: 44

lindsay@gandalf.cs.cmu.edu (Donald Lindsay) writes:
>But how to attach a fiber to a chip?

>There are several answers. I think one group made a notch in the edge
>of a chip, so that a fiber in the notch would be properly aligned
>with an edge-emitting diode.  Now that we can make arrays of
>face-emitting lasers, there's a simpler answer: just epoxy the fiber
>end onto the chip. One of the lasers will happen to be on the fiber's
>optical axis, and one is enough.
>
> [more interesting ideas]
>
> D.C.Lindsay .. temporarily at Carnegie Mellon Robotics

   If lasers are fairly cheap (ie 1,000 cost the same as one),
have the chips self aligning, then we can program the chip
to configure itself.

   This idea comes from my reading on the JTEG proposal.
Under the JTEG proposal, each chip will have four testing pins:
a clock, directional signal, pass in, and pass out.  (I think)
The idea, is to connect up every internal pad through a shift 
register, and connect up all the chips on a board, so that you 
can check the connections between every chip on the board
from one simple connector.  This also makes testing easier
than a bed of nails.

   For optical chips will still need a few metal pins 
for power and ground, so we could keep the JTEG functionality
in metal as well.  Then we put about 10 lasers in a row for
each output, and about ten receptors for inputs, connect up the
chips (with lots of tolerance), and use the JTEG to test which
output/receptors are connected.  The have some fuse mechinism
to enable the proper laser(s), and the board is wired.
   You could even use a scheme like this to modify the pin
out of chips to suit your need, or have extra lines to be
used incase a fiber is bad.

   Doing board-to-board connections could be a little 
trickier.  Maybe we'll have dual-port laser drivers on
each side of a connector.

	woolstar@cobalt.caltech.edu
	John D. Woolverton, Engineer