Hello,
consider a 1st order RC low-pass filter made of 416.7MegOhm switch-cap resistance (2stage, unit capacitor = 100fF, fclk=48kHz, duty cycle=50%, non-overlap time=2ns) loaded by a 10pF cap.

The question I'd like to find an answer for is how much noise (equivalent rms voltage in 20Hz to 20kHz) will I get at the output if noise down-folding (from input signal) is included in the analysis.

I'm using PNOISE in SpectreRF to find an answer. I would expect to get 16.93uV (noise of equivalent RC time-invariant AC model in 20Hz-20kHz band at output node) + contribution of the downd folding effect. The results I'm receiving 5.23uV (maxsideband=300; that is folding from up to 14.4MHz band) is way too optimistic.

I have difficulty interpreting such result..

- I do get matching transfer functions from AC and PAC analyses
- AC noise analysis does match my hand calculation for baseband noise
- increasing "maxsideband" parameter to extreme values produces equivalent noise voltage that exceed sqrt(kT/C)=20.35uV
- 0.18um process; switches are nmoses;
- I'm new to SpectreRF periodic analyses and therefore I may have misunterstood explanations given in "Simulating SC filters with SpectreRF"

Could you help me understand such pnoise result, please?
thanks,
Marek