From: Ted Dunning
Subject:: Long distance IR detection
-------------------------------------------------------------------------------
In article <1993Apr28.111414.13165@ennews.eas.asu.edu>
gsulliva@enuxha.eas.asu.edu (Glenn A Sullivan) writes:


   Phototransistor-> X100 -> X100 -> NE567
   with 330Kohm              Limiter
   feedback to base
   to control operating point (I expect this really hurts Noise Figure! Any
   comments from Phototransistor experts?)

if you care about noise, you use a photodiode operating as a current
generator.

the circuit can look something like this:





       	 v+
          |
	  |
	  |
	  |
	 ___
	 / \	       	  >1M
	 ---        +----\/\/\/\----+	       	     100K
	  |	    |		    |	        +---\/\/\/---+
	  |	    |		    |	        |            |
	  |	    |    |\	    |	        |   |\       |
	  |	    |    | \	    |     1K    |   | \      +----- output
          +---------+----|  \       +----\/\/\--+---|  \     |
	  |	         |u1 \______|               |u2 \ ___|
	  |	         |   /	                    |   /    |
	  |	    +----|  /	                +---|  /     |
        1 \	    |    | /	                |   | /      |
        0 /	    |    |/	                |   |/       |
	K \	   gnd		               gnd           |
	  /	      		         	             |
	  \			          	             |
	  /			        	             |
	  |     		        	             |
	  |     		        	             |
	  |     		        	             |
	  |              ||             	             |
	  +--------+-----||----+---\/\/\/\/------------------+
	           |     ||    |
	           |           |
	  	   |           |
	      	   |       /|  |
		   |      / |  |
		   |     /  |--+
		   |____/ u3|
		        \   |
			 \  |--+
			  \ |  |
			   \|  |
			       |
			      gnd


u1 should be low current noise (LM11 should be good, op-27 is ok)

u2, u3 should be reasonably high bandwidth (tl085 and similar bifets
are fine, 741 equivalents might be ok if you aren't looking for fast
signals)

pick the loop gain and integrator constants for stability and
bandwidth.

filtering the v+ strenuosly is a *good* thing.

one nice thing is that the current bias on the diode is such that if
you only need 1mV of signal out and have -10V to bias the diode, then
you can see bias away slowly changing lights which are 10^8 larger
than the signal you want to see.  by making the bias loop quick, you
pretty much can even get rid of 60cycle lights.

have fun.  this circuit is posted on an `as is' basis.