Mark wrote:
Note: For many LNA designs, the best signal/noise ratio occurs at an
impedance that is close to, but not really, a perfect conjugate match.
The signal is coupled to the amplifier best at the conjugate match
impedance, but sometimes the noise is enhanced even more.


That brings up an intersting question I never did get a good answer
to...

It is my assertion that an LNA that is physically at room temperature
(290K) can have a noise figure no better than 3 dB (i.e. its effective
noise temperature is 290K) IF it is also conjugatly matched i.e.
looks like 50 Ohms.

Yes you can make the noise figure better than 3 dB, but then you must
either cool the device or MISMATCH it to the line.


I make amplifiers with 50 Ohms input impedance and 300pV/rtHz input-
referred noise. By your assertion, that should have been impossible.
The trick is that the input impedance obtained by feedback: The
cooled resistor trick. It works.

Jeroen Belleman