Techniques for PLL noise analysis

A Fourier transform will show you the frequency domain noise of the time domain system you've simulated, so in that sense, you can use a DFT. How accurate is your time domain model?

most oft published papers use S domain linaer analysis for phase noise analysis of the PLL

khouly

You should note that for PLL, we're usually interested in phase noise (or time domian jitter). I think that just plotting the FT of the output Transient (time domain) signal would show you output frequency spectrum of the noise (including amplitude noise), while you're interested in phase noise.
For analysis purpose, you can use a linearized s-domain model where the PD is a subtractor, CP=Ip/s, VCO=k/s,......

phase noise value for a VCO at 2.4 Ghz is shown as -116 dbc at 1 Mhz offset

but when carrying out db20(dft) the noise at 1 Mhz offset is only -32 db.

how this could happen

Hi,
Please can you send me a material/tutorial teaching how practically doing that ?

Thanks in advance.

Think about the units. for a DFT, the units are typically power in a spectral bin. Read the documentation on the dft you are using to understand what you are getting out of the calculation

Dave
www.keystoneradio.com

maybe the time u used for doing dft is to short

Here is a good paper on the phase noise of the PLL:
1. VCO phase noise
a. Basic VCO phase noise
b. VCO excess phase
2. Divider and phase detector noise
3. Noise from the resistors and active devices in
the loop filter

Hi all,

Thanks konkon for the paper it was really usefull. Could someone explain how to plot the curves with cadence & Eldo/Spectre ?

You might want to read "Predicting the phase noise and jitter of PLL-based frequency synthesizers", which can be found at http://www.designers-guide.org/Analysis/

-August