• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 33 subscribers
  • Views 257 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

DP83822I: Process of measuring clock jitter

David Huang
Intellectual 2285 points
Part Number: DP83822I

Hi 

The customer is using our DP83822I device for the Ethernet PHY, but have problem measuring the clock jitter of the crystal oscillator. 

They tried to order a 25MHz crystal oscillator to match our device, and sent the following spec in our DP83822I device to the crystal oscillator maker,

Jitter（Long Term）Accumulative over 10ms： 1ns max

and the maker doesn't know how we measured this value.

Could you make an answer about how we measured this long term jitter (Accumulative over 10ms) to solve the customer's problem?

And if possible, giving some recommendation of what crystal oscillator should they be choosing, that would be very helpful.

Thank you

David H.

  • 0
    TI__Expert 7040 points

    Hi David,

    The DP83822 EVM is a good reference for a crystal that will meet the datasheet specifications: ABM8AIG-25.000MHZ-12-2Z-T3

    The long term jitter specification comes is determined by providing an external clock to the PHY and increasing the accumulated jitter until we start to see link instability in the DUT. The jitter of external clock is recorded and this is what all clock sources must meet, including crystal oscillators. To record the jitter of a crystal oscillator, you should be able to measure using a low capacitance probe from XI to GND and measure the jitter over time. 

    Regards,

    Justin

  • 0 in reply to Justin Lazaruk
    TI__Intellectual 2285 points

    Hi Justin

    Thank you for your reply.

    That provides the idea for me when explaining to the customer.

    Can you elaborate a little about this 'Jitter（Long Term）Accumulative over 10ms' as well? The customer seemed to be confused about the 'accumulative over 10ms'. Is it safe to assume that the TIE is 1ns when the measurement time is at 10ms?

    Thanks

    David H.

  • 0 in reply to Justin Lazaruk
    TI__Intellectual 2285 points

    Hi Justin

    Adding one more question here.

    The customer also mentioned that it's difficult to do this accumulative over 10ms measurement with the equipment they have, so they are asking if this measurement can be done with reduced duration but multiple tests, e.g. 100us for over 100 times.

    Regards

    Huang

  • 0 in reply to David Huang
    TI__Expert 7040 points

    Hi David,

    Yes, that can be done to calculate the accumulated jitter.

    Regards,

    Justin

  • 0 in reply to Justin Lazaruk
    TI__Intellectual 2285 points

    Hi Justin

    Thank you for the answer, it's good to know it works.

    Just one last thing to confirm. You mentioned that 100 times of 100us measurement works for the 10ms accumulated jitter test, does it mean that it's likely to obtain 1ns max jitter on every test?

    (My understanding of testing accumulated jitter is like for the picture below, assume the red rising edge is the ideal clock and the black is the real clock with jitter, the jitter shows below should be ranging in a certain range - in this case, ±40ps. And since you said that it's okay to do the measurement of 100us for 100 times, it's safe to expect seeing 1ns max of jitter in every test.)

    Sorry to ask so many times... I just want to make sure my understanding is correct before explaining to the customer.

    Thanks!

    David

  • 0 in reply to David Huang
    TI__Expert 7040 points

    Hi David,

    I would expect the cumulative total to be 1ns accumulated jitter over all tests. I believe it is okay to break the 10ms down into 100 tests and that the results can be combined to calculate the jitter as if the test was 10ms.

    Regards,
    Justin