Aigneryu
|
Dear Ken,
In your PLL phase noise+jitter paper (version 4b), the result of Demir is adopted to model a free-running oscillator. The perturbation including white and flicker noise is described as eq.(20): Su(df)=1+fc/df. And the resulted Sphi is described as eq(20): Sphi(df)=n(1/df^2+fc/fd^3). According to these two equations, the phase noise spectrum should introduction a 1/f^3 and 1/f^2 corner frequency fc,phi, which is exactly the same as the flicker noise corner frequency. However, in the well-know phase noise tutorial written by Ali Hajimiri, who introduced the impulse sensitivity function, the 1/f^3 corner frequency should not be the same as the flicker noise corner frequency, except some special situations. Can Spectre simulator be utilized to examine the inconsistency? I mean can we use an ideal MOS, which incorporates basic non-linear I/V characteristics, white noise and other colored noise sources to make a oscillator to find out the 1/f^3 corner frequency which is caused by flicker noise.
I remember that the flicker noise corner frequency of MOS transistor is usually very large as well as several mega Hertz, but when I measure or simulate oscillators, I find the corner frequency in phase noise is one or two order lower than the flicker corner frequency.
Sincerely,
|