Designer’s Guide Community

pcardoso73

Junior Member

Offline

Posts: 25
Portugal

Phase-Noise results not consistent with Jitter results
May 11^th, 2011, 12:27am

Hi all,

I am simulating a VCXO at 80 MHz, to be later integrated into a PLL. I am using PSS+PNOISE to extract Phase-noise and then Jitter.
As I am trying to decide which inversion region performs better (I need low bandwith so Weak Inversion is not an issue). I have simulated this circuit for strong and weak inversions. From my understanding, the phase-noise in Weak inversion should be better than in Strong inversion, once it uses exponential law (like BJT's). But what I get is a better phase noise and worse Jitter. This is contradictory.

1 - I am using Jcc. Is this correct ?
2 - Where the first jarmonic should be I get a very small "bump"
3 - I have set the PSS as follows: from 1 to fo/2; beat frequency 80 MHz; Number of harmonics 10; Conservative analysis.
4 - PNOISE: Output Frequency Range 1-80 MHz (or 40 MHz); Maximum sidebands 30; noise type FM (I am not sure about this one)
5 - Direct Plot Form: First I plot phase noise, than jitter (Jcc); number of cycles 1; rms set; Freq. Multiplier 1; Integration Limits 1 - 40 MHz.

Please tell me if this procedure is incorrect for autonomous circuits.

Best regards,
pedro

Back to top

phase_noise_WI2_small_001.jpg

IP Logged

rajkumar palwai

Community Member

Offline

Posts: 46
Sunnyvale, California

Re: Phase-Noise results not consistent with Jitter results
Reply #1 - May 15^th, 2011, 8:50am

Hi Pedro,

The Jitter number which we have to consider depends on the application.

JCC gives u the jitter between the two succesive clock cycles from a VCO. But, in most of the serial link cases you should look into the absolute rms jitter on a edge. This jitter is also called as accumulating jitter. PSS+PNOISE does not do this absolute jitter measurement automatically.

You can obtain the absolute jitter number �by integrating the total phase noise power and dividing it by the 'w0' (fund frequency in rad/sec). You can do this integration in MATLAB.

regards
rajkumar

Back to top

IP Logged

rajkumar palwai

Community Member

Offline

Posts: 46
Sunnyvale, California

Re: Phase-Noise results not consistent with Jitter results
Reply #2 - May 15^th, 2011, 8:58am

Hi Pedro,

You are saying that you got a better phase noise after biasing VCO in weak inversion. Is this trend same for entire freq range? Can you give the plots of phase noise for weak & strong inversion?

Because, as JCC is the correlation between 2 succesive clock cyles, most of the low freq phase noise will not give significant contribuition towards the JCC jitter number. It can happen that your weak inversion low freq phase noise is better, but high freq phase noise might be bad than the strong inversion case. And hence the weak inversion case was showing higher JCC number.

Regards
rajkumar

Back to top

IP Logged

pcardoso73

Junior Member

Offline

Posts: 25
Portugal

Re: Phase-Noise results not consistent with Jitter results
Reply #3 - May 16^th, 2011, 12:43am

Hi rajkumar,,

Thank you for your answers. Well, I cannot do Jee measurements, because it would be needed a reference and that's not the goal.
Regarding Jitter in VCO's I have read some papers and almost all of Ken, and everything points to Jc or Jcc not Jee.

Anyhow, the contribution from phase noise to jitter until 1st harmonic is almost zero, all the jitter comes from the frequencies close to the 1st. harmonic. As I am integrating from 1 to fo/2, I think I am catching the skirt of the 1st harmonic having, thus a variation in Jitter. Maybe, in this case, the integration period should be less than fo/2 in order to avoid double counting.

Attached you can find the plots.

Best regards,
Pedro

Back to top

Forums.zip

IP Logged