definition of noise
raymond_luo2003 said:
Hi,all,
1) Phase Noise
2) Spectral Density
3) Power Spectral Density
4) Spectrum
5) Single side band power spectral density
6) Auto-correlation function
It seems that they all different but related!! Who has very clear reference on that? Very appreciate if any can share your understanding and reference!!
Raymond
Phase noise is in fact the random frequency fluctuation observed in real oscillators. As the frequency fluctuation is normally small and centered around 0, we may take its integration, which is phase, as a random process to study.
When phase noise is treated as a (typically wide-sense stationary) random process, it is best characterized by its time-domain self-correlation function. If you take the Fourier transformation of the correlation function, you get a power spectrum, which is what you see in publications.
As a spectrum plot is hard to describe verbally, typically phase noise is quantized by xxx dBc at xxx offset. Such as -110 dBc at 1 MHz offset, with center frequency at 5.2 GHz, it means that at 5.201 or 5.199 GHz, the power spectrum density of phase noise is 110 dB BELOW the power of fundamental (desired) signal.
By default the phase noise (power) spectrum is expressed in double-sideband format, as F0 >> 0, say = 5.2 GHz. However, in some cases, such as direct down conversion reception, one of the two sidebands is folded back and overlaps the other, so the power adds up. Hence, sometimes single-sideband phase noise is more convenient.
Anybody has some thought over the calculation of phase noise for simple (analog, not ring) oscillators?