The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Question about discrepancy between PNOISE and NOISE analysis
https://designers-guide.org/forum/YaBB.pl?num=1679936730

Message started by eye_see_design on Mar 27^th, 2023, 10:05am

Title: Question about discrepancy between PNOISE and NOISE analysis
Post by eye_see_design on Mar 27^th, 2023, 10:05am

First time posting here so please bear with me if my formatting looks terrible.

A little bit of background…I have a circuit that has an autonomous output waveform as it is the result of periodically charging and discharging a capacitor. There are no driven sources in the schematic. I have a transient testbench where I use acnames to break and run noise analysis. I have utilized the “@sigoutCross” function to define the threshold crossing at 0.5V that the noise analysis is to be performed. That is done using acnames. At exactly 0.5V on the first rising edge crossing noise is to be performed. Which is the yellow curve in the plot I have attached, this looks like what I expected.

Since I need to evaluate the impact of noise around the trip point, I have set up the PSS/PNOISE analysis as well. However, this is in a separate test within that same testbench so I can make use of the calcVal() function. I have verified that the PSS transient waveform looks correct, it is one cycle of operation. The “VAR(“period_calc”) function is defined in a separate test, I run transient to extract the period of the signal since it is not a known parameter to me and it is called using the calcVal() function. The PNOISE analysis (red plot) does not look correct to me as it is 500x higher than what I calculated in NOISE analysis. Is there something obvious that I am doing wrong?

Rather than fail spectacularly at formating my images, I decided to upload them as a PDF. Hope that is alright.

Thanks in advance!

Title: Re: Question about discrepancy between PNOISE and NOISE analysis
Post by smlogan on Jun 8^th, 2023, 6:02pm

Dear eye_see_design,

> I have a transient testbench where I use acnames to break and run noise
> analysis. I have utilized the “@sigoutCross” function to define the threshold
> crossing at 0.5V that the noise analysis is to be performed. That is done
> using acnames.

If I understand your methodology correctly, it appears you are running an AC noise analysis at the time at which your oscillator is crossing its threshold voltage.

> ...The PNOISE analysis (red plot) does not look correct to me as it is 500x higher
> than what I calculated in NOISE analysis. Is there something obvious that I am
> doing wrong?

First, I am not sure how you calculated the AC noise result. From your ViVA plot, its units are shown as nV/root Hz. An AC noise analysis output has the units of V/root Hz or V^2/Hz. Secondly, an AC noise analysis is a small-signal simulation only and creates a linear model of devices based on the operating point at which which you specify the analysis. As a result, there is no folding of signal/noise harmonics due to the non-linear nature of the limiting mechanism of the oscillator. a large-signal PSS analysis followed by a pnoise analysis does include the non-linear behavior of the limiting mechanism and will include noise due to folding of signal and noise. Therefore, I would expect the total noise over some bandwidth of a PSS/pnoise set of simulations to be much larger than that predicted by an AC noise analysis.

Hence, I am not surprised at your results.

Shawn