aaron_do
|
Hi everybody,
i have a question receiver noise figure. Basically I have a system which looks like this:
--------------LNA------|--------------|-------I+ }NF | | Double Bal |_____I- } Port-----Ideal Balun | Quad | | | Mixer |-------Q+ --------------LNA------|--------------|_____Q-
hope that primitive drawing helps. Simple anyway. Anyway I want to calculate the required Input Referred Noise (IRN) of the IF blocks. My LNA/Mixer combination has a gain of 17.4 dB in simulation and a noise figure of 9 dB when simulated across I+ and I- (differential output). The 9 dB is a SSB NF result. Here's my calculation
NFsystem = 19 dB Input Noise (50 ohm) = 4kTR = 8.004 e -19 V2/Hz LNAMixer Output Noise due to source = Input Noise * Gain(LNAMixer) = 4.44 e-17 V2/Hz (A) LNAMixer Output Noise = A * NFLNA = 3.53 e-16 V2/Hz (B) Required IRN at LNA output = NFsystem * A = 3.53 e-15V2/Hz (C) IRN of IF Block = √(C – B) = 56 nV/√Hz
So what i've done is calculated the noise at the output port of the LNA/Mixer with the LNA NF and then the same noise with the system NF and taken the difference. The result is my required IRN of the IF blocks. I decided to double check this result by adding ideal resistors to the output ports (I+, I-, Q+, Q-) with the other end left as an open circuit. I then simulated the system NF again and swept R until the NF was 19 dB. I found this R to be 25 kohm. Since I have 2 Rs for a differential output, the total noise is sqrt(2*4ktR)=sqrt(2*4*1.28e-23*290*25000)=28 nV/√Hz which is obviously not what i calculated. Can anybody offer any insight as to why my calculation is so off? Here's a pic of how i resimulated the NF.
I used the PSS and PNOISE analysis in spectre to simulate the nosie figure.
IF+--------R-------open circuit } NF IF----------R-------open circuit }
thanks, Aaron
|