• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 21 subscribers
  • Views 2729 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Period jitter in calculation

Mahesh KC03
Intellectual 1360 points
Other Parts Discussed in Thread: ADS5463-SP, CDCM7005-SP

Hi Team,

I was calculating the clock jitter at the input of ADC, ADS5463-SP. The customer is using an XO with period jitter (1 sigma) of 25 ps rms and 12K to 20M jitter of 0.3ps rms.

They are using the CDCM7005-SP in buffer mode to fan out this clock to other end points also. The CDCM7005-SP adds 0.1 ps rms jitter from its output buffers.

Should we consider the period jitter from the XO as well in the calculation or is it just the RMS sum of 0.1 ps and 0.3 ps?

Can you please explain with reason?   

Thanks,

Mahesh

  • 0
    TI__Mastermind 24205 points

    Hi Mahesh,

    Please consider the RMS sum of 0.1ps and 0.3ps for your jitter estimation. Peak-Peak jitter is unbounded and grows with the number of samples taken for the measurement. The one sigma (RMS) jitter does not change significantly with samples.

    However, the bandwidth of integration could be different from 12kHz-20MHz (this needs to be verified). A phase noise analyzer is appropriate for the measurement of this parameter as it has a lower noise floor than most real time oscilloscopes.

    Regards

    Arvind Sridhar 

  • 0 in reply to Arvind Sridhar
    TI__Intellectual 1360 points
    Hi Sridhar,

    Thank you for the answer.
    The period jitter is more of a statistical (histogram) value and it increases with the number of samples thus we can't rely upon this value for calculation? The band limited jitter on the other hand is a function of phase noise integrated between an offset of 12K to 20M from the carrier. The phase noise function does not depend on number of samples.
    Is my understanding correct?

    Thanks,
    Mahesh
  • 0 in reply to Mahesh KC03
    TI__Mastermind 24205 points

    Hi Mahesh,

    Correct. The time domain period jitter peak-peak measurement increases in theory with the number of samples. You could however measure the 1 sigma period jitter that does not increase significantly with sample size. The integrated phase noise (frequency domain representation of jitter) does not really grow with the number of samples since it is not a sampled system. 

    The sampled nature of the period peak-peak jitter measurement on a real time oscilloscope implies that this measured parameter is subject to aliasing, which results in noise components at frequency offsets above half the sampling rate, i.e., fc/2 to alias to frequency range between 0 and fc/2 . The high frequency noise components which may not contribute to the RMS phase jitter when using a phase noise analyzer, will contribute to period jitter when measured using a real time oscilloscope. Integrating the area under the phase noise curve over an application specific passband (12kHz - 20 MHz) does eliminate the contribution due to jitter components above and below the high/low-pass frequencies.

    Regards

    Arvind Sridhar