PSS+Pnoise. Result interpretation

Hi all,

noticed, that resulting phase noise curve I'm geting through PSS
+PNoise simulation starts from positive value if small frequency
offsets are used (like 10..100 Hz). One would expect it to start from
0, since it in dBc/Hz. Is it simulation accuracy (reltol) issue or
something fundamental in how SpectreRF handles this type of
simulations or something wrong can be with setup/circuitry?

Thanks
Dio

 

[Andrew Beckett]

On Wed, 17 Oct 2007 19:36:32 -0700, dionisij@msn.com wrote:

Hi all,

noticed, that resulting phase noise curve I'm geting through PSS
+PNoise simulation starts from positive value if small frequency
offsets are used (like 10..100 Hz). One would expect it to start from
0, since it in dBc/Hz. Is it simulation accuracy (reltol) issue or
something fundamental in how SpectreRF handles this type of
simulations or something wrong can be with setup/circuitry?

Thanks
Dio

Well, you can interpret it in several ways. The real explanation is that noise
models aren't really accurate at very low frequency (say 10 Hz) and so the
numbers aren't meaningful. Adjusting the simulator tolerance won't help...

Another way of thinking about it is that the phase is more than one cycle out -
but in essence that doesn't really help you.

Recent versions of spectre have started outputing a quantity which we've
termed the "line width" (which borrows similar terminology from laser
oscillation I think?) which is the lowest frequency where the phase noise
is meaningful.

Regards,

Andrew.
--
Andrew Beckett
Senior Solution Architect
Cadence Design Systems, UK.