I have a receiver architecture built up in Genesys with two different analysis for the chain. The first analysis is a System sim at the center frequency of the band which is accurately predicting Gain/P1dB/NF/etc. The other analysis is a Linear analysis to predict S-params.
The Linear analysis matches the System analysis within 0.3 dB when it comes to predicting the forward gain of the receiver, but the noise figure is considerably different. The System analysis, reports a NF of 5 dB, which is the value I expected at Fc. Using the Linear analysis, NF is reported as 2.8 dB, which is considerably different and unexpected.
Why is that? I would think that the linear analysis would be equivalent to doing a parameter sweep of Fc in the System analysis. What is it about the Linear anlysis NF (CS Noise correlation matrix in S-Parameter form) that is so different from the Spectrasys CNF function?
1. I have been looking at other NF questions and saw this CNF vs NF noise fgure calculations
1a. I am using a 180deg splitter as the first element in the chain which then uses another 180deg splitter later to recombine. This may not be the correct 'physical' implementation of my HW, but still doesn't change the question.
3. I have my system BW set to 1 Hz to eliminate BW considerations and force a calculation analogous to a spreadsheet.
If I change my first element to a 2:1 transformer to implement the balun/splitter, the System analysis reports the same value for CNF (2.8 dB) as the Linear analysis NF (2.8 dB). Curiously though, the Linear analysis reports a NF of 2.8 dB regardless of which element (2:1 transformer or 180deg splitter) is used.