Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!wuarchive!uunet!mcsun!ukc!slxsys!jpp
From: jpp@specialix.co.uk (John Pettitt)
Newsgroups: comp.sys.transputer
Subject: Re: Control signals on optic link
Message-ID: <1991May8.154509.21817@specialix.co.uk>
Date: 8 May 91 15:45:09 GMT
References: <9105041115.AAnac14693@nac.no>
Organization: Specialix International, London
Lines: 42

@autro1.UUCP ("teig_y", yvind Teig) writes:
>---------------------------------------------------------------------
>Transputer link via optic cable. 

>We are thinking about using optic links between transputer embedded-
>controller boxes. I know about the Inmos application note "Connecting
>INMOS links". I also know about a board from TTT "Transputer Technologie
>Transfer" that has Tx-links to optic, with all control lines encoded
>into fibeglass optic.

>Does anyone out there have any experience?

We have been running T222's on the end of up to 1Km of glass fibre.
We have an effective bandwidth at that distance of 1mbit (we run
at 10 mbit but the speed of light delays the data by 6 micro seconds
at a km).  We took the notes in the application note and the HP apps
notes on fibre and away we went.  We are using 62.5 glass (plastic is
not up to much more than 30M).

We are seeing error rates close to 0 - I.E. better than the 1 in 10^14
that we were expecting.  As it happens we do run a checksummed transport
protocol over a conectionles network layer to provide robustness - but then
we are in a hostile environment (office use ! :-).

[ the product is a distributed, fault tolerant I/O system for UNIX
systems - up to 128 ports at 38K baud with 17 T222's spread over up
to 4 Km of cable.  Topology is free form and will tolerate loss of
links between units by routing over alternate links ]

We don't send the control signals down the line  - as the remote is 
expected to work.  If it does not poke a watchdog every 250ms it
gets reset by a one shot in the hardware design.  This removes the need
for reset in our application.   Multiplexing two control signals is possible
with varying degrees of dificulty depending on the delay in propagation you
will tolerate.  If delay is not a factor then you can use a subcarrier on
the fiber to carry the control signals.

Hope this helps - if you want more info please mail me.
-- 
John Pettitt, Specialix International, 
Email: jpp@specialix.com Tel +44 (0) 9323 54254 Fax +44 (0) 9323 52781
Disclaimer: Me, say that ?  Never, it's a forged posting !