subtr
Community Member
 
Offline

Analog Enthusiast
Posts: 72
India
|
If you can assume the noise is dominated by white noise, then you can actually make the vco model in verilogAMS by deriving period jitter from it. Now given that the noise at an offset is known, the curve is completely defined in all the points or in other words you can find out the N=H(f) of the noise. Assuming that you need the period jitter, multiply it by a (1-zinv) filter and integrate the noise. This is the sigma of each period. Now you find a random number in your code based on this sigma and pass it to the oscillator. It will generate an integrated noise which corresponds to the PSD of your actual oscillator. Da Dalt's book on understanding jitter has details on retrieving period jitter from spot noise of purely 1/f^2 noise.
You can even model flicker noise by using multirate filters to create 10dB/decade which is explained in Staszewski's book. But it can be a next step.
|