gjm (03/25/83)
This is the summary of the responses on the question of autobaud via CR
(in:--in:) vs BREAK (in:-BREAK-in:) to switch to the proper speed for low
speed modems.
There were only 5 responses, but all of the input was helpful.
Important points include:
1) Some systems have fixed speed lines (datasets) where neither
autobauding nor BREAK applies.
2) Some systems might require long BREAK's to set the baud rate as
desired, especially from 300 up to 1200.
3) Autobauding is based on framing errors, and works better when
setting to a low speed from a high speed default answer.
4) Standard (212A type) modems should be able to break from 300 up
to 1200.
5) CR should be used to autobaud from (known) 1200 answer down to 300.
BREAK's should be used to switch from (known) 300 answer up to 1200.
If you don't know the answer speed, use BREAK's.
Thanks to those who sent in there helpful comments to enlighten us.
Gary Murakami
Bell Labs - Indian Hill
ihnp4!gjm
----------------
Messages follow:
>From sdo Wed Mar 9 08:31:50 1983 remote from u1100a
Re: Autobauding
The u1100? systems don't do auto baud changing. An incoming line has
a fixed speed. Also, BREAK doesn't work at all here.
Scott Orshan
>From rochester!Liudvikas.Bukys Wed Mar 9 17:08:50 1983 remote from allegra
From: Liudvikas.Bukys
Subject: BREAK really used
ritcv (a neighbor of ours) defaults to 300 baud and needs a BREAK9 (no
mere BREAK will do, and I had to read the uucp source to find BREAKn)
to switch to 1200 baud. Don't ask me why, but it's true.
>From we13!rjk Wed Mar 9 21:09:48 1983 remote from lime
To: lime!houxz!ihnp4!gjm
Subject: UUCP
Our modems autobaud 1200-300-110-150-1200... In going to a lower speed
you always have the advantage of the framing error because your loooonnnggggg
data frame at a lower baud will always overrun the speed of a higher
data set.
Perhaps we could settle on the autobaud convention of changing speeds in
uucico until a newline is recognized.
Randy King
..we13!rjk
>From vax135!floyd!cmcl2!lanl-a!unm-ivax!bob Thu Mar 10 22:20:36 1983 remote from ariel
Subject: autobauding vs. break
Granted we are not part of Bell Labs, but our phone lines answer at 300 and
break to 1200.
-- bob
>From watmath!watcgl!dmmartindale Tue Mar 15 02:16:47 1983 remote from decvax
Subject: switching speeds on dialin lines
By "autobaud" I assume that you mean switch speeds when a carriage
return is received? This isn't what I call autobauding, but is probably
what applies here. All of the 1200 baud modems on the three VAXes in the
math faculty - watmath, watcgl, and watdaisy - are VADIC triples. They
determine the speed of the calling modem automatically, but getty (the
process that prints login:) doesn't have the information available. It
is set up to cycle between 1200 and 300 each time a break (framing error)
is received, and this does work quite reliably. A carriage return sent
at 300 baud to a DH set to 1200 baud also generates a framing error, so
you get the 1200->300 switch, but 1200 baud characters received at 300
will be ignored or appear as a character of some sort, depending on the
actual bit pattern transmitted. Thus BREAK is a more reliable way to
switch speeds if you don't know what the other end is doing, but carriage
return is a bit faster.jim (03/29/83)
If you have a Vadic triple, there is a rs232 line that gives the speed the modem is set at. It should be possible to jumper this line to one of the lines that your DZ (or whatever) can sense, like maybe CTS, and have the driver choose the right speed. Of course the driver would have to know which ports to do this on. I intend to try that one of these days. We have a Perkin-Elmer Unix system here with Vadic triples which have been fixed up so that the speed-control line from the modem actually controls the hardware speed of the computer port. This isn't too hard on P-E computers, since the speeds are set by wire-wrapping (!) but might be difficult on a DZ-11.