[GreenKeys] Telephone Men

[Jim Haynes]

When I was working at UCSC we ran a lot of terminals at 1200 or 2400 baud
using RS-232 signals over twisted pairs - completely ignoring the supposed
length restrictions, but it worked fine.  People who were more serious
about it bought "modem eliminators" that translated RS-232 into something
else to send over wire pairs.  But we did what we got away with.

One time we sent out a request for information to a bunch of computer
companies, to see what they would propose for our future time sharing
requirements.  We specified speeds up to 2400 baud or so asychronous 
for the terminals.  The IBM guy said they couldn't meet that - they
only provided up to 1200 baud asynchronous.  Eventually had me talk to
one of their data communication experts, and he kept saying he didn't
understand our requirement - where were we getting the clock?  (Obviously
he was thinking of modems that were only available up to 1200 baud
async and to go faster than that you had to use synchronous and the
modem supplied a clock.)

Another time one of the faculty told me they were running a connection
to Stanford, using 2400 baud synchronous modems, with ordinary ASCII
async terminals.  This kinda floored me - with sync modems you are 
supposed to put the bits in when the clock pulse from the modem tell
you to.  But it turned out they were getting more-or-less good 
transmission with async terminals - the modem just sent whatever it
saw on the data line when it was clock pulse time.  So what came out the
other end was more or less what went in, with occasional errors where
the async bit happened to be changing just at clock pulse time.  And
with some distortion caused by the clock pulse not falling in the middle
of the input bits.

jhhaynes at earthlink dot net