On Tuesday, January 28, 2014 12:28:26 PM UTC-5, haitic...@gmail.com wrote:   
   > >    
   >    
   > >    
   >    
   > > A lock-in is exactly equivalent to a symmetrical bandpass filter    
   >    
   > >    
   >    
   > > followed by phase-sensitive downconversion to DC.  Most of the time you    
   >    
   > >    
   >    
   > > want it to track the signal phase, but once in awhile it's useful in    
   >    
   > >    
   >    
   > > measuring the noise floor, or asynchronous signals.  (You usually want a    
   >    
   > >    
   >    
   > > 2-channel lock-in for that, so you can get the total amplitude, as in    
   >    
   > >    
   >    
   > > the aforementioned AM radio. ;) )   
   >    
   > >    
   >    
   > >    
   >    
   >    
   >    
   > snip snip   
   >    
   >    
   >    
   > Phil, Here is the way I see lock-in amplifiers. They are ways to 'mark' a   
   signal so that the noise cannot sneak into the part you want to measure.   
   Consider a square wave which has a complex 16 bit pattern on it, there are   
   various codes which are low    
   probability of appearing in noise. These codes are used in CDMA, spread   
   spectrum, pseudo noise, Huffmann codes, Barker codes, etc. In this case, a 16   
   bit binary code will out-perform a simple square wave because its a rare   
   pattern and the information    
   gained, the SNR boost, is directly related to the Szilard formula for   
   information and entropy. So if you demodulate a rare long bit pattern, it is   
   like a demodulator for spread spectrum sequences, and it will reject noise   
   well.    
   >    
   > Lock-in amps have been built with "Barker" codes of 13 bits, but according   
   to the tenets of information theory, a longer sequence of bits might even do   
   better - A practical advantage of this approach is it can be done by   
   microprocessor with minimal    
   part count - See Cappels.org.   
   >    
   Hi jb,  this is wierd.  In one post you say there is no 'return' from   
   information theory.  And above is perfect use.     
      
   I don't know anything about these longer bit sequences. (but I can imagine.)   
   There's certainly a bandwidth/ averaging time trade-off.  (it takes longer to   
   average a long bit sequence.)   
      
   George H.   
   >    
   > jb   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)