Hi everybody,
  thanks for reading my post. I'm trying to simulate the IIP2 of a fully differential LNA + transconductor + mixer chain. The BW of my sistem is from 2GHz to 18GHz.

This is the way I set up the simulation.
My input port is sine type, with some power "prf", rf frequency "frf" and "pacmagdb" as the magnitude of PAC analysis in dBm. QPSS analysis run a flexible balance engine on frf and flo (the LO freq.) where frf=2.401G and flo=2.4G, tstab=10n, sweeping prf=pacmagdb from -60 to 10dBm. The I do a QPAC analysis at 2.4GHz, max clock order = 2.

Hence, the fundamental freq is at |flo-fqpac|=0 and the second order harmonic at |flo - (frf-fqpac)|=2.399GHz (flo,frf are (1,-1) in the results window).

The result that I get is in the image and it looks like I'm doing something wrong.

Can somebody help me or pointing me in the right direction?

Thanks a lot,
.mat

The simulation results look suspect below Pin=-40dBm, you should to the measurement (extrapolation point) above -40dBm which will raise your IIP2 estimate, but it still seems a bit low in performance.


One suggestion would be to forget using PAC with QPSS, instead you should run QPSS with two moderate tones with (as vp suggested) 1MHz tone separation and the LO as the large signal. But you need these tones to be offset from the LO by more than 1MHz such that the 1MHz baseband tone will be only due to the IP2 performance of the receiver.

I've another question.

How can I measure IIP2 in presence of mismatches?

Thanks