[net.periphs] How long can an RS232 line be?

sean@oddjob.UChicago.UUCP (Sean Casey) (10/12/84)

<>

Dear Netland,
I'm planning to run a few rs232 lines around my office and I would like
to know the net's opinion as to

	o	the maximum length a line can be before inductive
		voltages become a problem. Is shielding really
		necessary? (Lines lengths are about 200ft or so)

	o	what length and under what conditions are optical
		isolators (or their equvalent) between the system
		and the terminals required (advisable) for rs232 lines.

Please reply by mail and thanks in advance.

Sean Casey	...ihnp4!oddjob!sean

University of Chicago
Department of Astronomy and Astrophysics

padpowell@wateng.UUCP (PAD Powell) (10/12/84)

Actually, you can run an RS232 line as far as you want.  What is critical is
the signal strengths remaining at the far end.

The standard (ANSI RS-232C), which I don't have beside me just now,
states what the maximum input voltage levels and max/min impedances are.
In addition, it specifies the minimum "received" signal values for a 1 and
0 value, together with the receiver max/min impedance.  The problem is to
figure out what YOU can get away with.

The usual DRIVER for an RS232 signal is an integrated circuit (ARRG!
mem fault- core dump! I can't remeber the number), which generates a
+11 to -9 volt signal.  The matching receiver will detect a 1 volt signal,
and has a hysterisis input, centered on about .5 volts, with about
.2 volts hysterisis.  Note that these values are typical, and can vary
all over the landscape.

The limiting factor on distance is your cable and speed.  The longer the
cable, the more total capacitance, the slower the driver can drive the
signal.  The following combinations have been tried here at Waterloo.
1. Unshielded Bell 4 conductor cable, using (TX,ground) (DX,ground)
	in the pairs.   RTS, DTR, etc. were unused.
	19.2K bps-  150 meters
	9600- 400 meters
	1200- ran out of cable, at least 2 KM.

2. Shielded 50 pair cable:
	TX, RD, with ground, rest jumbled together.
	About the same distances.

By the way, as an interesting experiment, I tried using the new RS422
drivers from Texax Instruments, in a 4 KM loopback mode.  Ran it at
9600 for a week with only a couple of bit errors.  This was in differential
mode.

I made up a small 6 component PC board that has a 422driver/receiver,
	and a small regulator, to convert from 422 to TTL levels
	(which are compatible with the RS232 receivers HINT),
	and use these instead of long distance modems.  The power supply
	is a $3.00 calculator battery replacement.  UGLY? Yerp.
	But it sure is cheaper than $200 for a line driver.  I have
	gotten paranoid lately and added 2 optical isolators for comfort.

	If you want isolation, you have to get a power source from the
	unit;  pin 25 of the RS232 connector is unused, and is usually
	a good place to stick a tap to the +12 or +5 supply.

Patrick Powell

bass@dmsd.UUCP (John Bass) (10/16/84)

At SRI in the late 70's we ran some 2000-3000 ft lines running at 2400/4800
baud with little trouble. Some would work at 9600. The lines were 2 twisted
pairs with foil shield. Some terminals had "marginal" rs232 drivers/receivers
and worked better after using different chips.

Looking at the signals at the computer end the receiver waveforms were nearly
sine waves. Some of the installations also showed 4-15v of 60hz noise ...
much of that was ground noise ... both earth ground variances and inductive
pickup from the lines running along side power lines.

I have heard that some folks have used an RC filter to waveshape the signal and
cut out some of the 60hz noise ... but don't know what was used.

The biggest problems were caused by lines talking to themselves when a remote
terminal was powered off. The floating end of a 2000ft twisted pair lines
passes back all sorts of garbage into the receiver on the computer end. The
logon banner would couple back into the receiver, echo/getty/login output
from the input would keep the line talking for hours on end. A resistive
termination to ground at the terminal end helped on some lines.

Some lines had repeated driver/receiver failure ... ground differential?
inductive kicks?

Have fun .. John Bass