On the advice of most everybody I converted the smoke detector to a photoLED
( Optek OP999)  in reversed biased mode.  Hook this up to an op-amp, and
compare the result to a constant DC voltage.  Reverse biased Opto LED's leak
more if they are exposed to light, according to theory.

This one has 4.7 megohm in series with the reverse biased LED.  Supposed to
produce about 60 nanoamps dark current (which would drive the whole thing to
about 4.7 volts)  and maybe 1 microamp light current (driving the thing to
about .3 volts)   The op amp compares this with a DC signal produced by a
pot.  Way too high impedances to put into a PIC directly.

The IR emitter and detector is arranged in reflective mode, with the
detector and emitter 90 degrees from each other, so a reflective object
would set them off.

In practice, no such thing happens.  It makes a great solid object sensor
detecting fingers, paper,  etc.  Won't detect smoke.

Had a chat with the OPTEK engineer I'm working with, and this is the
arrangement he recommended.  Said that the reversed biased LED was much more
sensitive than the photologic module I was working with earlier.  So far,
you can't prove it by me.

-- Lawrence "Smokey" Lile

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