Cadence Customer Support Solution 11175734

PHASE NOISE IS NOT CONSISTENT FOR PNOISE SOURCES AND JITTER IN ADE

Problem statement:
You are running a PSS/PNOISE analysis on an autonomous circuit in Analog Design Environment. You first set a pnoise=sources and then a pnoise= jitter in the analysis form. After running both simulations and plotting the results from the Direct Plot form, you notice a 3db difference between the phase noise from Pnoise->sources and Pnoise-> Jitter . Why?

Solution:
When you run pnoise=sources, spectreRF just calculates the noise as the time-average power spectral density and therefore does not provide correlations in the noise and so the cyclostationary nature of the noise. The output noise is the voltage noise, just as it would be measured by an oscilloscope. The pnoise=source analysis does not separate Voltage noise into AM and PM components. The phase noise is therefore being calculated as the output noise scaled back by the power in the carrier.Indeed, for autonomous case and close to the carrier, entire noise is phase noise, so we can substitute S_v for S_phi in dBc units. This assumption is not valid away from the carrier.

When running pnoise-> Jitter for autonomous circuit, spectreRF invokes in the background pnoise modulated analysis that captures the full cyclostationary nature of the noise. Pnoise modulated analysis uses the spectrum along with information about the correlations in the noise between sidebands. This analysis figures out correlation between upper and lower sideband. Based on relation between USB and LSB, spectreRF assesses the level of AM and PM noise.

The difference observed hence occurred because Pnoise-> Jitter does compute PM/AM components and uses correct definition of the phase noise whereas Pnoise->Sources is using the fact that in special case (for autonomous circuit close to the carrier, see above) the whole noise is due to the PM component. Assuming your circuit is fully matched, you should therefore see a 3 dB difference between phase noise (3dB is the result of is the scale of two was due to usage of one sided PSD in PNOISE).

Source:http://sourcelink.cadence.com/docs/db/kdb/2005/Apr/11175734.html