Here are the files for the AFSK demodulator with current-loop output. 

NOTE: 

rttydemod2.sch and rttydemod2.brd are the same board with a minor modification
suggested by Jim Haynes to 1) add a "snubber" circuit to eliminate the 
possibility of inductive spikes from the selector magnet damaging the 
optoisolator and 2) run the optoisolator outputs in series to eliminate the
need to observe polarity for the loop circuit. I'll try to make sure this
version is the version that goes out to people. 

Standard warnings apply - a 120v/60mA current loop circuit is certainly 
capable of giving you a hell of a shock and possibly injuring you, etc. 
The loop is optically isolated from the rest of the circuit, a bare wire
or tool falling across the board, if it's not enclosed, can easily break
that isolation, which could shock you, fry your laptop, etc... I'm not
responsible for anything bad that happens as a result.  I recommend a fuse 
in series with your loop in any case, for safety. 

While it's working fine for me as built, there could be mistakes in here,
and it's not the most elegant thing in the world. 

Many components can be substituted. Probably any N-Channel FET would be fine
for toggling the optoisolator's LED, I just happened to have a bag of
VN10LMs sitting around.  If you change LEDs you may want to adjust their
series resistors to make sure they don't draw too much current, as loading
down the power supply will also change the demodulator's center frequency, 
which can be frustrating.  The AQV257 and HSR312 are very similar - both 
are "optically coupled solid state relays" and are the most expensive component 
here. AQV257 is good to >250mA at 120VDC and may be usable at slightly higher 
data rates than the HSR312, but neither of them is going to work much past 
mechanical teletype data rates. 75 baud (100 WPM) is fine, and that's as fast 
as I've tested it. 

Note: the loop interface must be connected in the right polarity or it looks 
like it's always a closed circuit.

I've tested this at 45.45 and 50 baud with frequency shifts from 23 Hz to
850 Hz. At really small shifts you have to adjust the center frequency very
precisely. At "normal" shifts like 170 it is pretty forgiving. There is a
"wide/narrow" setting which mostly doesn't matter except at the extremes. 
(ie, 23Hz or 85Hz shift require "narrow" setting) but 170Hz is fine in either
position. 

The center frequency is set by a fixed resistor in series with a pot. You
can replace those with different values to change the usable frequency range.
A 10k resistor plus a 20k, multi-turn pot works well, or a 22k resistor
plus a 5k pot. 

To use it, feed it an audio signal and adjust the "CENTER" pot until the 
LOCK light comes on, then adjust a little bit either direction until you 
get a good flicker of the DATA light and intelligible text on your TTY.

Again I suggest a fuse in series with your loop circuit and putting some
sort of protective cover over the board, as touching the high voltage 
side of the optocoupler could result in a shock. 
eric volpe
1/2/2010
epvgk@limpoc.com