Oz: When you control both the emitter (eg led) and sensor (eg PD with   
   transimpedance amp) with same reference signal.    
      
   I have seen some very simple designs and more complex. The more complex   
   generally have phase lock and other features.   
      
   Going for the simple, I have seen circuits which just use a square wave to   
   excite an led and the same reference signal is read by an ADC on a mcu. In the   
   simplest case, the "off" reading is subtracted from the "on" reading.   
   Alternatively, this can be    
   done in the analog realm by switching an op amp with a cmos switch and a low   
   pass filter on the output.   
      
   Anyone know any drawbacks to the simple? (ISTM you won't get the full SNR   
   benefits w/o phase lock, and also a square wave has odd harmonics.)   
      
   An interesting twist on this lock-in is the algorithm described by Horowitz in   
   his Art of Electronics, ca. p. 1027. The algorithm is so simple it is   
   mind-bending: Just add many measurements, and the "grass" on top of the   
   mountain will become visible. No    
   averaging involved in its execution. I know this has been used in the Hubble   
   telescope, and wonder if anyone has used this in more mundane apps.   
      
   Horowitz explains this by noting that the noise increases as the sqrt of   
   number of observations, but the buried signal increases in a linear fashion.   
   So, if the measurement value is 100 with noise fluctuation 5, and the signal   
   is say .01, then after    
   10exp6 measurements, the signal is now 10,000 and the noise fringe 5,000.    
      
   The signal value will be 100 million from the addition, so a microprocessor   
   seems a good way to do this.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)