On Tue, 7 Jun 2005 19:47:54 -0700, "Joel Kolstad"
wrote:

Hi Al,

"Anonymous" wrote in message
.. .
Actually, RS-232 has always baffled me, and from some measurements I made it
doesn't seem to be a very strict standard.

There really is an RS-232 standard, it's just commonly abused. Here's what I
have from a "cheat sheet" I have at hand:

-- Transition levels are between +/-3V (this is what you have to detect)
-- Maximum open circuit voltage is 25V (!)
-- Maximum load capacitance is 2.5nF
-- Load resistance is 3-7kohm (this is ignored all the times -- often the load
is much, much higher than this)
-- Maximum slew rate is 30V/us (probably violated occasionally but no one
checks :-) )
-- Maximum transition time is 4% of bit time
-- Maximum data rate is 20,000 bits/second (obviously ignored with good
reason...)
-- The maximum distance is 15 m (or was it 50 ft originally). This is
often ignored and usually the problems are related to ground potential
differences.

The 20 kbit/s makes sense, when you look at the other requirements. At
this speed, the bit time is 50 us and a 4 % transition is 2 us. At 30
V/us, that would be a 60 V transition from -30 V to +30 V. Compare
this to the 25 V open circuit voltage.

To charge a 2.5 nF capacitor with the 30 V/us slew rate, a 75 mA
current is required. IIRC, the standard also contained a short circuit
current limit about that magnitude. The power dissipation can be quite
high in the transmitter.

These calculations clearly indicate the fundamental problems with
RS-232.

The RS-422/485 and later LVDS solved the problem by radically reducing
the driver voltage swing to a few volts or even below 1 V. Even with a
modest 30 V/us slew rate, the transition from one state to an other
does not take a long time. With the termination resistance of the
order of 100 ohms, the connection can also be analyzed as a bipolar
current loop. Since the characteristic impedance of a twisted pair
cable is typically around 100 ohms, this is also a well matched (low
SWR :-) transmission line system, so there are not much reflections
that would mess with the bit transitions. Also the differential
construction avoids most of the ground potential difference problems.
RS-422/485 is usually implemented with transceivers running from a
single +5V supply

It is sad that RS-422/485 is not used in common PCs, but instead all
kinds of hacks violating the RS-232 are used (some even claim TTL
signal levels RS-232 "compatible").

Paul OH3LWR