Phase Noise and SpectreRF

[hrkhari]

pss saveinit

Hai Folks:

I had simulated the phase noise of a LC-QVCO using cadence spectreRF, with a pulse source perturbation. But I'm getting positive phase noise for offset frequency less than 100Hz. I thought the phase noise plot is a normalized plot of the carrier. It should start from 0 at 1Hz of offset and falloff negative. Why does this happen?

Rgds

 

[ieropsaltic]

phase noise spectrerf

I think that you should check that you VCO started properly. You can use: results->direct plot->main form->PSS->voltage and check time option then plot the output net and check its amplitude, or you can also check saveinit option at PSS and follow the same previous steps but choose startup transient V except for voltage. If the oscillator doesn't start, you should first check that you have enough gm, then you should try using a different initial condition.
You should also check the output log and see if there's a warning at PNOISE. If there's one, try increasing tstab or changing the accuracy.
Also, make sure that you plot PHASE NOISE of the PNOISE plots.

 

[Mazz]

spectrerf positive phase noise

I think this is not a problem of simulation.
The assuption that at 1Hz is 0dBc is correct?
Take as example the measurement of a VCO+divider done in different conditions (attached). At freq of VCO the phase noise is higher! and if you extend plots to lower freqs you'll get positive phase noise at 1Hz.
Mazz

 

[hrkhari]

why positive phase noise?

Hai Folks:

I had jumped across a similiar inquiry in the Designer's Guide forum. According to Ken Kundert, the author of Designer's Guide to Spice and Spectre, the low fequency corner in the phase noise response is caused by large signal effects and the simulator cannot predict it accurately because they are using small signal analysis.

I hope you could comment more on this statement and direct me to any materials proving this statement. Thanks in advance

Rgds

 

[ieropsaltic]

positive phase noise

I don't think so. PSS is not a small signal analysis and it takes in account circuits non-linearities and that's why it is always used in oscillators' simulations. To increase the accuracy at low phase noise offsets, try using larger number of side bands at PNOISE analysis.