Hi All,

I am trying to run a pnoise simulation on an inverter. My understanding is that, when we plot "output noise" after a pnoise "sources" run, it plots time-averaged PSD of the periodic output noise. Here, time-averaging is one period of the beat frequency. In my simulation beat frequency is 4.14 GHz, so for each offset frequency voltage noise at the output is averaged over 1/4.14GHz = 241.5 ps. Using output-noise (from "sources") plot, I integrated the output noise from 1 KHz to 2.07 GHz (the limits of frequency sweep) to obtain -71.7 dBV (3dB already added to account for two sides of the spectrum).

Now for comparison with pnoise "timedomain" simulation, I ran the simulation [img][/img]with 68 points making sure that sufficient points around the transition region are included. From the directplot form, I plotted the integrated output noise over time. The attached file has the plot. The average value of this noise plot is -63.4 dBV. Hence, there is a discrepancy of 8.3 dB between the integrated noise numbers obtained from timedomain and sources noise simulation results.

It would be great is someone can take a look and tell me what I'm doing wrong here?
Thanks.