[GreenKeys] Stop bits and all

[jhhaynes at earthlink.net]

In the 1950s Teletype was making electronic time-division multiplex.  
First the vacuum tube AN/FGC-5 and then the transistor AN/UGC-1 and
AN/UGC-3.  In these systems the input was normally 7.42 code.  The
multiplex distributor ran a little faster than the input character rate
so as to assure that it would never fall behind, since there was no
buffer.  This meant that characters came out the receiving side a
little fast: the correct baud rate but with a shortened stop pulse.
About every 50 characters the multiplex distributor would get ahead
of the input so there would be no character to send, and then a
multiplex blank would be inserted.  These didn't go to the output, so
a printer would pause for one character time when one happened.

I recall the printer sounded a little bit fast, but it never got so
fast as to be 7.00 unit code.  The multiplex ran about 2% fast.
7.00 code is about 10% fast compared to 7.42.

Apparently this did cause some problem with crypto equipment, which
I never understood.  Western Union got a contract to make a "Printing
Telegraph Signal Normalizer" which had the job of slowing things down
to squeeze out the multiplex blanks.

It was something of an annoyance to Teletype to have to make gears and
cams for Western Union that were different from those for everybody
else.  I believe this applied only to the transmitting equipment, as
late-model printers ran fast enough that it didn't matter to them whether
the input was 7.42 or 7.00.  There is a patent 3,351,711 that addresses
the problem by making a transmitter-distributor have a variable length
stop pulse without changing any parts (but by making it awfully
complicated).  I don't know if any of these were ever built or sold.
About the same time W.U. decided to quit buying 7.00 unit equipment.
By then they were mostly buying for Telex or for government contracts
and not for their own systems.  Maybe it was by mutual agreement of
W.U. and Teletype; I don't know.

It's certainly possible to design a printer selector that does not need
a whole unit of time to latch up the clutch.  Thus even when receiving
7.00 code the selector does not rotate continually; it stops between
characters.

Once upon a time I posted to the list a chart of speeds of various
things - I don't have time to look for it right now.  If you can find
it I think you'll see that the receiver shaft speed is faster than one
revolution per 154 milliseconds at 60 speed.  Or you could work it out
on an actual machine by counting gear teeth and knowing the speed of
the motor.  The keyboard is geared down to run slower than the main
shaft.


-- 

jhhaynes at earthlink dot net