cadence vco gain

For my LC VCO, the PSS did converge for the voltage vs. frequency curve. But obviously the curve is not right shaped. The gain of the VCO is not linear and the VCO gain jumps around from 1MHz/Volt to 20MHz/volt.

I think should be something wrong in the setup.

The VCO transient is quite good and only nees 50ns to start-up of the oscillation. So in PSS I set Tstab as 100ns. I put max timestep as 10ps (VCO frequency is about 2GHz,i.e., the oscillation period is 500ps). And I made the converge method as gears2only.

What could be wrong to my setup? Thanks

The best way to predict the frequency of oscillation of a VCO is to simulate in transient mode and then make FFT.
Because, sometimes PSS may not converge due to
model properties or some different iteration lacks of simulation methods.

You can build a equation to calculate the frequency and after that you should do "parametric simulation" instead of PSS.

It's quite efficient...

ps. Intialization time is too short.Because a VCO needs at least 100ns to settle it up.

It turns out that the simulation method make a big difference if I switch from gear2only method to trap method. Which one is supposed to be better?