[mod.telecom] RS-232 problem

S.PAE@DEEP-THOUGHT.MIT.EDU (Philip A. Earnhardt) (03/04/86)

We have a noise problem with our RS-232 lines at work. We run pins 1,2, and 3
(with 7 shorted to 1)
for 2 lines in every office. Unfortunately, the 6 wires are run on a single
cable; the 2 lines are not shielded from each other. This still works OK
if someone has 2 terminals in their office or on terminals with multiple
ports. However, folks who use a T-switch to switch between their 2 lines
in their office will reliably send garbage down the unselected line.

Has anyone solved a similar problem? I'm guessing the what will 
is to put a load between pins 1 and 3 on the unselected line. To do this
automatically, we'd use a matrix switch instead of the t-switch:

Computer A ====== Terminal  *   Computer A ==  == Terminal
                            O                \/
                            R                /\
Computer B ====== Resistor  *   Computer B ==  == Resistor

Will this work? Am I in any danger of damaging our computer terminal
boards? Any estimate of an appropriate value for the resistor?
-------

rab@WELL.UUCP (Bob Bickford) (03/09/86)

   Methinks you should not tie pins 1 and 7 together.  Just leave pin 1
completely out of the connection; it's only a frame ground and if you're
in the same building you don't need it.  We don't use it!

   Most T-switches do *not* switch the pin #1 connection, so there may
be the source of the trouble:  they're always tied together.

   The resistors should go to a 12v reference, not to ground, although
ground will work with many receiver IC's.  I would expect there is
already a resistor in the interface PCB, going to 12v, and if you
added one to ground at the switch you'd get a tendency to float to 6v,
maybe leading to more noise than you have now.  (assuming you use the
same value, namely 10K ohms)

embick%tetra@NOSC.ARPA (Edward M. Embick) (03/12/86)

We have a similar setup here in CSC's complex in San Diego.  We
are running 3 wire communications, but unlike you we are running
pins 2 (TD), 3 (RD), and 7 (SG) only between the computer and the
terminals.  The RS-232 25 pin connectors have pins 4 and 5 and 
pins 6, 8 and 20 shorted at the terminal end and at the host.

We run our lines several hundred feet in a very noisy environment
without serious problems.  The terminals are operating at 9600 baud.

I believe the problem you are experiencing is due to the transmit line (pin 2)
not being terminated at the data terminal end.  It acts like a very
long antenna.  Any electrical noise, or something like a login
prompt from the host, will be transmitted to the host.  The host,
upon receiving the garbage, will respond with something if a terminating
character is detected.  This response goes out over the receive line (pin 3)
and is picked up by the transmit line (pin 2), etc., etc. etc.

This also can occur when a terminal is powered off and the UART completes
the connection between pins 2 and 3, as well as with an AB switch.  

We solved the problem by having the AB switch short pin 2 to pin 7
on the line to the host that is not connected to the terminal. 
This does no damage to the host port, and keeps it from talking to itself.
------------------------------------------------------------------------------
Ed Embick    (the more paths I make, the more paths they break! waaaaaaa....)
Computer Sciences Corp.                embick@noscvax.UUCP  or
4045 Hancock St.      {akgua,allegra,decvax,ihnp4,ucbvax}!sdcsvax!noscvax!embick
San Diego, CA 92110                MILNET:  EMBICK@NOSC
(619) 225-8401 x516