I was working on doing phase noise simulation for a ring VCO using EldoRF.
I found a very strange output result where the phase noise curve goes above the 0dBc/Hz.
Anyone has a physical explanation for that? or it is just simulation error.
That is the way the simulator sees phase noise. It is due to the small signal approximation the curve goes up as 1/f^3 without limits. But even for small frequency mismatch, the phase error builds so heavily that this model does not hold and displays large discrepancy at low frequency offsets. The phase noise curve really is lorentzian., which tapers the noise at low frequencies. There is nothing to worry.