The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Help with VCO PN simulation using SpectreRF
https://designers-guide.org/forum/YaBB.pl?num=1096758225

Message started by Frank F on Oct 2^nd, 2004, 4:03pm

Title: Help with VCO PN simulation using SpectreRF
Post by Frank F on Oct 2^nd, 2004, 4:03pm

Hello,

I am currently simulating the phase noise of a VCO wih cadence spectreRF AA simulation tool. I have read the help files of AA concerning the phase noise simulation of oscillators and cadence's whitepaper "oscnoise.pdf" available on the net.

The problem which I am having is the following:

What exactly is the meaning of a "sideband" (referred by the "maxsidebands" field located in the PNOISE form)?

In the help file,a sideband Ki is defined by the following equation:
f(noisesource)= f(out)+ Ki*fund(pss)
Does this mean that a sideband defines the noise frequencies which are used in the computation of the PN around the desired harmonic of the output?

According to the help files, one should set the maxsideband field to 0 (at least for PN of oscillators). Shouldn't one set the maxsidebands field to a large number as to evaluate the PN due to all sidebands (whatever they are)?

My problem is that my results improve by 40dBs when I shitch from a "maxsidebands" of 0 to a "maxsidebands" of 1 ?
I would like to know which of the the results is the real one.

Any help will be appreciated.
Regards
-Frank

Title: Re: Help with VCO PN simulation using SpectreRF
Post by Mighty Mouse on Oct 2^nd, 2004, 5:33pm

When you specify maxsideband you specify how much noise folding to account for. Ideally you would want to set it to infinity, but then the simulation would never finish. So you must set it to some finite value that is low enough to simulate quickly but high enough to accout for all significant noise folding.

I can pretty much assure you that setting maxsideband to either 0 or 1 is way too low. By setting it to 0 your are not accouting for noise folding at all. In particular, besides missing all the white noise that is folding both up and down to the output frequency, you are also missing the flicker noise mixing up to the carrier.

I recommend starting at the default value (around 7 or 8 if I remember correctly) and increasing it (8, 16, 32, etc.) until the output noise stops increasing significantly.

You might want to take a look at the paper on simulating noise in switch capacitor filters (under Analysis on this website) for a discussion on how to choose maxsideband.

- MM -

Title: Re: Help with VCO PN simulation using SpectreRF
Post by Andrew Beckett on Oct 4^th, 2004, 2:57pm

Some of the confusion arises from the fact that there was some incorrect information in the SpectreRF documentation up to and including IC446, which suggested setting maxsidebands to 0. Some of this came about because in the spectreS interface in the past, setting maxsidebands to
"none" (or something like that) in the UI, actually didn't
set the maxsidebands parameter, allowing spectre to use its built-in default of 7, giving reasonably OK results for most circuits.

However, setting it to 0 makes no sense, because it means that you are only including the noise at the oscillator frequency itself, and not the upconverted noise around DC (mainly the flicker noise), and downconverted higher frequency noise. The lack of upconverted noise would be a major error in the noise output.

Andrew.

Title: Re: Help with VCO PN simulation using SpectreRF
Post by frank F on Oct 4^th, 2004, 3:15pm

Thanks.
I incremented the maxsidebands to 5.
Consequently, I obtained -5dBc/Hz at 1kHz offset with ~30dB/decade slope.
I guess the flicker noise has a pretty strong impact on the phase noise.

-Frank