dog1
|
Hello everyone,
I have a confusion about the basics of phase noise.
I am reading the Razavi's book RF Microelectronics, 2nd Edition. And in Chapter 8.7.3, it says that a for noise around a oscillatory waveform, half of the energy goes into AM side band while the other goes into PM sideband. I think the derivation there also apply to any periodical waveform with dominant 1st harmonic, since it only require a noise around w0. This is confusing. Does this mean that for phase noise simulation of periodical signal, I can always just simulate the total noise spectrum around w0, and divide it by two, regardless of the waveform of the periodical signal and what kind of noise there is? Then why do we need "modulated" type as cadence Spectrum pnoise simulation option? Why don't we just always use "source" type? I also remember that for hard switching circuit, the noise power is mostly PM, but then this contradicts my understanding of the previous statement.
From my understanding, the contradiction seems to arose from the fact that the previous statement is made without consideration for the phase correlation between noise slots in the spectrum. Is that correct?
Thanks
CHEN
|