VCO jitter measurement in transient analysis and pnoise analysis

As for the proper jitter to be measure, you need to understand which type of jitter is important for your system.

ffrom what I understand, Jc(K=1) results in what would be called Period Jitter in the time domain.  That is important for digital timing type applications. ie: What is the shortest clock period that will be produced?

Jcc would be cycle-to-cycle jitter, which has less applications, generally.

Jee is another, and it would be called Phase jitter, or Absolute jitter (Jabs).   That is important in SERDES and ADC/DAC type applications, where the frequency spectrum of the jitter needs to be bounded.

Dear analog design,

If it is of any use to you, I have a few comments on your choiice of analysis and settings (in addition to those of Mr. SWurzer).

analogdesign said:

am simulating  active inductor VCO with free running frequency of 5GHz in 65nm technology.

          version IC 6.1.3

          using transient analysis (enabled transient noise) the measurement of period jitter is 400ps

          In tran analysis the noise setttings,  noiseseed 1, noisefmax 2.5G and didn't set noisefmin

If you are using an active inductor in a VCO whose fundamental frequency is 5 GHz, that suggest that the technology you are using can support frequencies far in excess of 5 GHz. In essence, your inductor's impedance is created using a device's gm and hence its frequency response must exceed 5 GHz. As a result, your upper limit for noisefmax in your transient noise analysis appears too low. You are limiting the maximum noise frequency to 2.5 GHz - but your technology supports a much greater frequency than that. Hence, your simulation will not include any noise folding of noise frequencies over about 2.5 GHz. I would not expect the transient noise simulation noise results to correlate with the pnoise results you are simulating as I noticed you have pss and pnoise simulation sidebands set to 10.

analogdesign said:

3) how to interpret phase noise of -75dBC/Hz in terms of jitter

You need to integrate the phase noise over some frequency range to estimate the random jitter. The frequency limits ti use in the integration, and potentially any filtering to apply, are dependent on the standard your VCO application is intended.  The Jcc and period jitter are not estimates of the random jitter - which is what the integrated noise result will provide. They are related - but the the same.

Shawn

Hi Shawn!

Are you Shawn Logan of the former Agere/LSI days?  We were colleagues in the old days.  Good to see you're still kicking around!

By the way, after I replied to this thread, I noticed that it's a 6 year old question  ..... 

Not sure if the original questioner is still monitoring, but good info none-the-less.

Hi Steve,

I am afraid I am the guilty party Steve - I noticed the last name right away, but was not sure if it was you - good to hear form you too!

Shawn