After viewing a lot of posts, I feel it is necessary to open a specific post to discuss why Cadence put this mode in the tool.

I noticed a few things:

1. FM jitter mode doesn't require setting the strobe time or threshold. Which indicates that it is time averaged noise. So is source mode and modulated mode.

2. I tried with a simple VCO and found that FM jitter mode gives identical result as source mode. The only difference is that in FM jitter mode the result is DSB format.

3. This mode is accessible only in oscillator mode.

4. Based on one of Ken's papers, FM jitter is more to modeling the jitter accumulating behavior. This seems to indicate that we should always use FM jitter mode for oscillators. However, I also found that PM jitter will give the identical result (if we convert Jee back to phase noise) without losing the accuracy (assuming no AM noise appearing).

5. It is stated that FM jitter mode is based on modulated mode. But I'm not sure and have no idea how it works.

6. This appears to me that FM jitter mode is almost a repeat of source mode. I need to further confirm that the phase noise result in FM jitter mode is real phase noise (PM) or just the sum of total noise (AM + PM).

Thanks for pointing to that post.

It looks like that FM jitter is the same as modulated mode. It just present the result in a different way.

Cadence really should write a document "Noise Type De-mystified".

Thank you Frank.

Frank, Do you think they have it?

I tried a couple of times to get more understanding on this. But it turns out to be much more effective to just go here.

Mentor's ELDORF goes a more straight-forward way. They naturally split AM/FM noise  in their result .sstnoise. However, EldoRF has terrible convergence performance on those highly non-linear circuits. It also has no strobed analysis yet. (Maybe I'm a bit out-of-date on this point.)