Since you have limited range (a plus), I would recommend you start again
with a more tradiitonal approach. Use a 20 ma IR LED (.25 from BGMicro
and up to .50 from others, 1.69 at Radio Shack), modulate it with about
40khz (tons of schematics around, but you can use the one at:
http://www.seattlerobotics.org/guide/infrared.html It is a simple
74HCT04 circuit with very few components and it is tunable. Use an
Everlight receiver (3.69 at Radio Shack, as low as 1.95 if you search
around), or a Sharp GP1U52X or similar (harder to find). Put the output
of the receiver directly into the port on the PIC. Drive the 74HCT04
inputs with the PIC with the modulation you want (I've used everything
from home brew codes to Sony IR codes to just a regular frequency). The
total cost could still be less than $4 - and probably a lot less.

You can actually modulate the IR LED without the 74HCT04 if you want to
get a little more involved with the PIC code.  See:
http://www.verinet.com/~dlc/projects/botproj.htm#irprox for Dennis
Clark's implementation using a 12C508. He gives you the code and it
should work with other PICs with no problem.

Dream up what you want to send, and squirt it out.  You can decode the
received signal using a pulsin (similar to Basic Stamp code) routine.
See http://www3.sympatico.ca/donroy and follow the links to the news
letter for a sample of code that will decode (or, with mods, encode) the
Sony IR remote signals (in PBasic - but hey, it's a challenge! ).

Dan


>Ok, I am new to this infrared stuff, so if someone could point me to a
>simple design for an asynchronus transceiver over IR I would be grateful.
>
>Insofar I am using the following equipment:
>
>Matched Lite-ON IR LED (940nm Vf=1.2V), it is in a clear plastic T-1
>package with
>3mm diameter package.  I was going to write a software UART into a 16c715
>with an inverted output (compared to normal UART) i.e. active high.  I was
>going to take that TX pin and tie it to the base of a 2n3904 through a
>3.3k resistor.  Then tie the NPNs emitter to ground and the NPNs collector
>to the cathode on the IR LED (Liteon LTE-4206). Then I was going to tie
>the anode of the IR LED through a 100ohm resistor to +5V.  The NPN should
>get saturated when turned on which will pull the collector down to about
>0.3 volts which would let a current of (5-(1.2+0.3))/100 = 35mA flow
>through the IR LED, maximum forward current for the IR LED is 60mA and at
>35mA (59%) power the output is about (57%) of that at full power, a nice
>linear relationship between radiant intensity and forward current.  I
>could tweak the 100ohm resistor to give let 60mA through the LED but
>unfortunately I am limited to 90-100mA for the entire circuit and I have
>another 2 visible LEDs that suck 20mA a piece.  Ok the entire transmit
>circuit that I just outlined will cost 36 cents in parts.
>
>Now I need an inexpensive receive circuit using the matched
>phototransistor, which conveinently has a darkened lens to cut visible
>light (LiteOn LTR-4206E), it is also small in size.
>
>BTW I only need 6-12" of transmission/reception range on this system (it
>functions as an optical calibration port), and is very cost sensitive.
>
>Primarily I need help getting the raw signal cleaned up from the
>phototransistor, so that it is ready to be fed to an input pin on the PIC.