Hi, I am trying to find out whether the jitter result simulation from pnoise analysis matches with the result given by transient noise analysis.

Before that, I have verified that the transient noise result matches with the noise reported on a small signal ac analysis by trying it out on a simple opamp.

Now comes the interesting part. First, I used pnoise analysis to find the edge-to-edge jitter (Jee) of a chain of inverters driven by a 100MHz clock. My integrated noise bandwidth is from 100kHz to 50MHz. I then increased the number of sideband until there is not much change in Jee. This is what I got as follows:

For no. of harmonics=1, Jee=30fs
For no. of harmonics=640, Jee=60fs

I then ran a transient noise analysis with the following settings:
noisefmin=100kHz, noisefmax=50MHz, stop time=100us. The Jee I got from the waveform is 29.5fs. Does that mean transient noise analysis does not take into account of noise folding effect? I have repeated the experiment with different noise bandwidth and all yielded similar conclusion.

Being a curious little kitty, I then repeated the experiment on VCO and I am even more confused. The oscillating frequency is at 800MHz and I have chosen the noise bandwidth from 100kHz to 10MHz to avoid integrating noise too closed to the carrier. From pnoise analysis, the cycle jitter (Jc) with different no. of cycles (k) is as follows:

k=1, Jc=415fs
k=8000, Jc=50ps

On the other hand, transient noise analysis (with noisefmin=100kHz, noisefmax=10MHz, stop time=100us) gives the following:

k=1, Jc=583fs
k=8000, Jc=278ps

The jitter from transient noise analysis is much more worse than the one reported by pnoise analysis.

Has anyone of you tried the experiment before and please share some of your findings. Thanks!

Count

P.S. I am talking about rms jitter all this while.