The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> 1/f noise up-conversion in oscillators
https://designers-guide.org/forum/YaBB.pl?num=1033535591

Message started by Lieutenant Columbo on Oct 1^st, 2002, 10:13pm

Title: 1/f noise up-conversion in oscillators
Post by Lieutenant Columbo on Oct 1^st, 2002, 10:13pm

I tried to use SpectreRF to simulate the phase noise of a VCO. Since 1/f noise is included in the device model, due to the nonlinearity of the device (MOSFET and varactor), we are supposed to see a 1/f³ slope in the phase noise below a corner frequency. This frequency is normally very low, but it could be >100kHz if 0.18um CMOS technology is used. However, SpectreRF doesn't show this 1/f upconversion effect. In fact, the results do not change regardless of whether I include 1/f noise in the models or not.

Why can't I see the 1/f noise up-conversion?

Title: Re: 1/f noise up-conversion in oscillators
Post by Ken Kundert on Oct 1^st, 2002, 10:33pm

There is a very common user-error that causes this problem. SpectreRF's PNoise analysis provides a parameter maxsideband. This parameter specifies the number of noise sidebands (or how much noise folding) to include in the output noise. For reasons that we still don't understand, both official Cadence documentation and training material mistakenly instructed users to set this parameter to zero. Using maxsideband=0 would prevent any frequency conversion effects from being taken into account; only the noise from sources at the analysis frequency would be included in the output. Thus, you would not be able to see the flicker noise up-conversion, and the result would be a significant under-estimation of the total phase noise of the oscillator.

To see the flicker noise that mixes up from low frequencies in the output, you would need to set maxsideband to at least one. However, in order to account for all significant noise folding, we recommend that you not use a value less than the default value of 7. To be sure you are using a high enough value, try increasing it until the noise results stop changing.

I apologize for this confusion and the trouble it has caused you.

-Ken