Hi,
I am now testing the IQ down-conversion mixer with post-simulation layout.This IQ down-conversion uses Gilbert-cell with independent gm stages.So,I and Q path may not effect with each other. At the same time, two LO ports are used to provide LO frequency in I and Q paths respectively. LO port in Q path(LOQ port) is set 90° phase difference from LO port in I path(LOI port).
Then, I find that,when LOQ port is set with different initial phase shift, different IIP2 results appear.
Here is my setting:
HB: 1 tone @13GHz,  Harmonics=5, oversample=5,errpreset=conservative
Hbac：rapid IP2, f1=12.3GHz,f2=12.7GHz,IM output=400MHz,Linearity output=700MHz
Results:
when LOI_phase=0, LOQ_phase=0, I_IIP2=59.96dBm,Q_IIP2=68.86dBm
when LOI_phase=0, LOQ_phase=-90°, I_IIP2=61.31dBm,Q_IIP2=66.46dBm
when LOI_phase=0, LOQ_phase=90°, I_IIP2=58.28dBm,Q_IIP2=71.85dBm
Why does IIP2 in I and Q path have such differences when LOQ_phase differs?