Confusion about conversion from phase noise to jitter

[lunren]

phase noise to jitter conversion

Dear all,

Recently, I read the paper by Ken"Predicting the phase noise and jitter of PLL-Based frequency synthesizers",(h**p://www.designers-guide.org/Analysis) It is really a wonderful work for pll disigners. Afer reading the paper, however, I am confused by some questions. I hope experienced experts who read the paper can give me some hints.
1) With pss+pnoise analysis, I get the curve of L, acorrding to Ken's paper "Predicting the phase noise ...", page 33, we can calculate the jitter of the circuit for example a VCO. But I am confused by the fact that: when I used L=-71dBc/Hz at 100KHz from the carrier (f0=409MHz), the jitter is 3.39ps; when I used L=-42dBc at 10KHz from the carrier (f0=409MHz), the jitter is 9.59ps. I am confused by the way to calculate jitter according to Ken's paper. Which one can I accept as my jitter? What's the exact meaning of the jitter culculated from equation(74)? It is the period jitter just at the offset frequency "point" or some thing else?

2) I find another paper about conversion from phase noise to jitter. h**p://www.maxim-ic.com/appnotes.cfm/appnote_number/3359. From this paper, the jitter is calculated by integrating L from for example f1 to f2. What's the difference between Ken's method and maxim's?

Thanks in advance.

Best regards,
Lunren

 

[Vamsi Mocherla]

phase noise conversion

The answer to your question lays in the offset frequency, which is the channel resolution of a frequency synthesizer. Hence, you can make a measurement at the channel frequency

Jitter can be random, average and deterministic. Average jitter can be obtained by integrating across the interval.

 

[windman]

phase noise jitter conversion

Probably you haven't fully understood the method described in Ken's paper. I think you calculated the close-in noise as the wide-band noise which can be predicted by the simple equation Peff=noise power density * Bandwidth.
In fact, for close in phase noise, the noise power density is depend on the frequency offset, it can only be calculated by integrating the noise power density over offset from 0Hz to the bandwidth of interest. That's the method described in Maxim's application note.

 

[needforspeed]

maxim phase noise jitter conversion

Probably you haven't fully understood the method described in Ken's paper. I think you calculated the close-in noise as the wide-band noise which can be predicted by the simple equation Peff=noise power density * Bandwidth.
In fact, for close in phase noise, the noise power density is depend on the frequency offset, it can only be calculated by integrating the noise power density over offset from 0Hz to the bandwidth of interest. That's the method described in Maxim's application note.

What's the meaning of the bandwidth of interest?
when we get jitter by integrating Phasenoise, how can we choose an integration interval?

 

[zizu]

Hi all,
After reading h**p://www.maxim-ic.com/appnotes.cfm/appnote_number/3359, I can't understand how the equation 8 can be derived, anybody has a idea or hint? any assumption here?

 

[biff44]

I have not read the paper, but to convert from the frequency domain (phase noise) to the time domain (time jitter) you must INTEGRATE the continuous phase noise curve over the entire applicable frequency range. That typically means integreating the phase noise curve from say 10 hz offset from the carrier to perhaps 5 MHz offset from the carrier.

You can not just take the phase noise at one data point, say 10 KHz offset frequency, and then deduce the time jitter from it alone. You need to also add in what happens at 10 Hz, 100 Hz, 400 Hz....1 Mhz...

Rich