I once wrote:
>> Most RS232 receiver buffers in a pc will accept 3.0V and above as mark
>> and anything below that as space
>
> I sincerely hope that this is not true because it means the receiver has
> zero hysteresis and zero noise margin.

It is true because the receiver chips are made that way. I do not make
receiver chips, only use them. You will see in the spec sheets that the
hysteresis can be (and is) deleted or reduced. I think that they do this
to prevent edge ringing and cross-coupling from unmatched cheap serial
lines at high speed to cause glitches.

>> (rs232 spec says -3.0V and below). So sending 0/5V is barely ok.
>
> So sending 0/5V is out of spec.

Sending 0/5V is out of RS232 spec and within normally used RS232 receiver
chip specs, as used in a PC near you. Ok ?

I am not advocating this for production but it does work for development
and special purpose uses with some caution.

The +/-15V for RS232 is derived from the old serial line specs I think,
where they used current loops with pots wiring at 70, 120, 150 and even
higher volts wired into TTY receivers and sent Baudot at 33 or 45.5 or 50
or such Baud speeds. The negative voltage on the idle line is used because
the idle line has the copper wire at a negative potential wrt ground and
this prevents electrochemical corrosion for a cable exposed to the
weather.  Then they switched to bipolar signalling, which doubled speed
(the receiver was a polarised relay). The forbidden band on +/-3V is to
prevent chatter from ground currents and electrochemical cells formed on
the long wires. So it's a largely historical feature.

And Russell is right, clamping is necessary if the pic is susceptible to
problems from bulk diode currents. Which all pics with a/d are, but those
without a/d do not seem to be affected at all and run fine with
significant current through the bulk diodes. I have used 'my' double
current mirror clamp for this with success.

Peter

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.