quadrature upconversion mixer

I simulated a gilbert cmos quadrature upconversion mixer in ADS. According to 'RF transmitter architectures and circuits' by Razavi, the output frequency should be cos(freq_LO-freq_BB). However, my HB simulation gave both (freq_LO-freq_BB) and (freq_LO+freq_BB) components together with the same amplitude. After that, I plotted transient simulation result in frequency domain, the difference of (freq_LO-freq_BB) and (freq_LO+freq_BB) components is about 50dbc. Obliviously, transient simulation gave me a correct answer, but I wonder why HB simulation not!

could u upload the project to check it
khouly

Please see the figure. HB or transient, which one should be trusted?

hi,

what's the number of harmonics in your HB simulation component?

-Increase the harmonic number and "oversample" ratio ( for instance 8 or 16 ) and try once more..
-Don't trust too much transient simulation of ADS, use spectre instead..