RS-232/RS-422 networking experiment

[Bob Fabry--N6EK]

I had many helpful responses to the question I posed a few weeks ago about
how much distance one could cover with RS-232.  The RS-232 standard is
specified to cover 50 feet at 20K baud, but there were several reports of
covering up to 600 feet, sometimes at a lower baud rate and sometimes with
special cable.  Alternatives that were suggested included fiber optics,
ethernet and RS-422 balanced line; the latter is specified to cover 4000
feet at 10M baud.  Unfortunately, K5NA's fiber optics use plastic fiber
limited to 200 foot or shorter runs and the ethernet solution requires
software not currently in CT.

I have a lot of respect for the robustness of RS-232 because when I was in
Computer Science at UC Berkeley, we ran it over a single twisted pair for
TXD and RXD between buildings using the electrical ground as the ground!
This scheme worked well for hundreds of connections for many years in spite
of the transients caused by elevators and so on.  In fact the only real
problem was whenever one of the buildings was hit by lightening!

So it seemed obvious to me that I should compare RS-232 and RS-422.  Based
on the note from Jay O'Brien, W6GO, I contacted R. E. Smith at (513)
874-4796.  They were extremely helpful and shipped me two of their RS-232 to
RS-422 converters for evaluation.  In fact, they have even suggested we take
them to Easter Island and/or Heard Island to continue the evaluation.

I bought a 500 foot box of the cheapest four wire cable I could find:  GC
Electronics 22 gauge solid four conductor telephone station wire, 30-9905,
for $29.  My first test showed that both R-232 and RS-422 worked just fine
at 9600 baud over this length of wire when I was not transmitting.  And the
rule of thumb is if one halves the speed one can double the length, so this
suggests that 1000 feet would work, at least at 4800 baud, or in the worst
case, at 2400 baud, either of which reportedly will support the traffic
requirements of CT.

As Lyndon Nerenberg pointed out, the attainable length may depend on the
RS-232 line drivers in the computer being used.  I was using a clone at one
end and a very old Toshiba T1000 at the other end.  I have no idea how
typical they are.

So I thought to myself that the RS-232 might perform OK with no RF
interference, but that the RS-422 would surely be superior in a hot RF
environment.  I ran a hundred foot serial line out my front door, over my
roof about four feet under the driven element of my forty meter beam when my
tower is lowered, and in the back window.  I fired up at 1500 watts on forty
with the tower down (which I had never done before), and the burglar alarm
went off.  I disconnected the alarm.  Then my wife told me the 60 watt
incandescent ceiling light in her study was going on by itself even though
the switch was off.  I told her not to worry, and figured I had achieved the
proper level of RF heat.  Neither the RS-232 nor the RS-422 network
functioned in this configuration.  Then I added one split ferrite shield
bead at each end of the serial line, and tried again.  Now both the RS-232
and the RS-422 networks functioned fine.

A word about the ferrite beads:  I use Fair-Rite part number 2643164251 and
find it very effective for reducing RFI all over the neighborhood.  It is
about one inch cube with a half inch hole in the middle.  It is number 43
material and has an impedance of 159 ohms per turn squared at 25 megahertz
and I estimate about 60 ohms per turn squared at 7 megahertz.  (To a first
approximation, the impedance of these chokes goes up with the square of the
number of turns.)  I used seven turns of the serial line which should insert
about 3K ohms for common mode RF.  These beads are available from many
sources.  I buy mine from PSC Electronics, 677 Palomar Avenue, Sunnyvale, CA
94086, (408) 737-1333.  Prices go down rapidly with quantity.

Anyway, with everything so RF hot, this seemed like a trivial solution and
my conclusion is that RS-232 ought to be just as good as RS-422 for most CT
networking.  At any rate, don't assume it won't work!  Several people
suggested that shielded wire was important or that special low capacitance
wire was important.  My experiment suggests this may not be necessary.  In
fact, I would stress using a small diameter cable so you can get many turns
on the shield bead.  If, for example, I used a standard computer cable which
carried across all 9 RS-232 wires and had a rugged covering, I might not be
able to get even two turns through the shield bead and I would have to
string together 49 shield beads to get the same RF immunity I can get from a
single bead with the smaller diameter wire.  It seems possible that some of
the bad reputation of RS-232 may be due to people not getting the RF out
rather than from the limitations of RS-232 itself.  Comments welcome!

-Bob Fabry, N6EK


--
Submissions:              ct-user@eng.pko.dec.com
Administrative requests:  ct-user-request@eng.pko.dec.com
Problems:                 reisert@eng.pko.dec.com