We have a debate in my company
on how to calculate the receiver sensitivity with
Sensi = kTB + NF + C/N
is demodulator requirement, and does not vary with
temperature for a digital implementation
is the noise figure, and is referenced to 290K according
IEEE standard (Friis proposed this temperature in
1944). NF varies with temperature of course.
My point is that, when you compute the sensitivity
at different temperature, the "T" of kTB should
remain equal to 290K, otherwise you account twice
for the thermal noise change, once in the NF, and
once in kTB.
What is your opinion on that?