Hello all,
I am trying to simulate VCO pulling. I am using Spectre  mmsim-71.

My simulation setup includes VCO which drives a transmitter supplied by BB signal. The transmitter than is coupled back to the VCO. Then the VCO is pulled by a harmonic of the TX signal at the VCO frequency. The result is that spurs are created in the VCO spectrum at offset equal to twice the BB signal. The amplitude of the spur depends on the frequency offset from the VCO frequency, and for a free-running VCO it must decrease by 20deb/dec. This is what I also see when I run PSS+PXF, but this is a small-signal simulation.

Now I would like to do the same simulation, but when I use real BB CW signal at given BB frequency. For this I use HB with 2 tones. The first tone is my VCO signal, while the second tone is the signal I inject at BB. For this simulation I use 10 number of harmonics for the osc! signal with oversample factor of 1. For the injected signal I use 5 as number of harmonics and also oversample factor of 1. Then I also use tstab as harmonic balance homotopy method. And finally, I enable the oscillator option.

Likely the HB converges and when I probe my VCO signal, I can see the spurs at the correct offset. The problem now is that first the spur amplitude does not follow the one predicted by the pxf analysis, and second that the spur amplitude does not scale by 20dB/dec with the BB signal.

My question is whether somebody have already some experience with such kind of simulation. According some Cadence application notes, this is possible, but I wounder why I do not see the 20dB/dec slope.

cheers,
Ilian