Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!ucsd!ames!eos!shelby!neon!kaufman
From: kaufman@Neon.Stanford.EDU (Marc T. Kaufman)
Newsgroups: comp.dcom.modems
Subject: Re: Why 300 baud?
Message-ID: <1990May31.025608.18545@Neon.Stanford.EDU>
Date: 31 May 90 02:56:08 GMT
References: <9686@discus.technion.ac.il>
Organization: Computer Science Department, Stanford University
Lines: 37

In article <9686@discus.technion.ac.il> joel%techunix.bitnet@jade.berkeley.edu (Yossi (Joel) Hoffman) writes:
>I was wondering:  why is it that all the common modem speeds are
>multiples of 300 baud?  Is there anything special about that number?

Why, yes, there is.  Long ago in the olden days the Teletype corporation
invented a terminal (models 33 and 35) that would wack along at 10 characters
per second, which was 110 baud (because there were 2 stop bits those days).
Later, Teletype tried to see just how fast a mechanical device could go, so
the model 37 was invented - which had upper and LOWER case characters, black
and RED ribbon, and in general was always on the hairy edge of working.  At
that time, they decided that the mechanical clutch used to decode characters
would drop out in 1 stop bit time.  The mechanism could push 15 characters
per second, so it ran at 150 baud.  That's probably close to the limit, as
IBM only got 14.8 cps on its Selectric terminals (2740 and 2741).  The clutch
dropout time was the reason old Baudot (5-level) teletypes required a stop
bit that was 1.42 data bits long (since rounded to 1.5 in VLSI chips).

The Bell 103 modem, which was current in those days, would run reliably only
up to 300 baud, because of bandwidth constraints -- so the next logical
baud rate was 300.  The Bell 212 actually runs at 600 baud, coding 2 bits
per symbol.  The natural progression was then, 1200, 2400, 4800, 9600, 19200.
[Why powers of 2?  Because you have to synthesize the baud rate from some
standard crystal frequency, and divide-by-2 circuits are easier than other
divisors -- before LSI, we had to do it with flip-flops and gates].

ANSI was trying to standardize baud rates during the same time frame, and so
there is an ANSI standard somewhere setting standard baud rates at 200, 400,
800, 1600, etc.  I actually build an interface to a 1600 baud banking terminal
once.

Not to be outdone, IBM -- who always want proprietary hardware, started using
1800, 3600, 7200, and 14400 bps as their synchronous bit rates.

Tannenbaum was right.  The nice thing about standards is that there are so
many of them.

Marc Kaufman (kaufman@Neon.stanford.edu)