Re: gEDA-user: ngspice shot noise

[Dan McMahill <dan@xxxxxxxxxxxx>]

Karel Kulhavy wrote:
> > > > thermal noise due to base resistance were your limiting factor.
> > >
> > >
> > > Because only RF transistor has transition frequency of 25GHz.
> >
> > Fight, but you said you were going to use this where a 2N3904 would have 
> > enough speed so why do you need a 25 GHz Ft?  Also you said you were 
> > going to run at a current level low enough to drop the Ft down to 300 MHz.
>
>
> Because 2N3904 will have enough speed at 5mA collector current which
> generates huge shot noise. 25GHz Ft will have 300MHz speed at I don't
> know 1uA of current which will generate substantially less noise?
>
> CL<


lets see, if you have a low impedance source then your input refered 
noise at 5 mA will be on the order of 0.4 nV/sqrt(Hz) ignoring base 
resistance which you really can't.  At 1uA of collector current, your 
input refered noise will be around 290 nV/sqrt(Hz).  Now you probably 
can ignore the base resistance!  So you're 37 dB worse in terms of SNR. 
 Now, if you have a high impedance source, than this is a different 
story.  In that case, as John has mentioned, you're better off with a 
FET input.

For a given input voltage (signal), your output current in a common 
emitter amplifier drops linearly with bias current while your noise 
drops with the sqrt() of bias current so you lose.


-Dan