[Laser] Photodiodes in series

[F1AVYopto at aol.com]

John 
>It  seems like putting photodetectors in series would be a  pretty
>unpredictable configuration, unless they were being used in a  photovoltaic
>mode or some such.  I usually think of photodetectors  as operating as
>current sources, having some dark current, and producing  some current from
>light photons striking their surface.  In which  case, I would consider
>connecting them in parallel, where their dark  currents simply add, and their
>internal capacitances simply add, their  active areas just add, but their
>responsivity in mA per mW is  unchanged.  You just get a bigger detector and
>lose a little  sensitivity and bandwidth.  But putting them in series ... I
>don't  know ... Csh are in series, Rsh are in series, which doesn't change
>the  bandwidth much, unless the load is much less than Rsh.  Yves may  have
>the best available analysis, having tried it.
>John Matz  KB9II
I wrote   
>The  Rshunt increasing should reduce the Johnson noise as the square root of 
the  diodes >number. 
It  was exactly "The Rshunt increasing should reduce the noise as the  square 
root of the diodes number." 
The Johnson  noise INCREASES like the square root of the diode numbers 
because the  Rshunt INCREASES like the diodes numbers. In= (4KTf /  Rshunt)^1/2 
But  as the Rshunt INCREASES like the diodes numbers, the useful  signal 
voltage (Rshunt x IL on infinite load) goes up like  the diodes number. 
So the final  S/N improves like the square root of the diodes number on an 
infinite  load. 
This  reasoning is only right in photovoltaic mode. 
In  photoconductive mode It seems to me the shot noise each diode not 
correlates and  adds also as a square root of the photodiodes number? 
73  Yves.