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.

DS250DF230: CDR Lock Time

Part Number: DS250DF230

Hi,

1. Only the Typ value is listed, do you have the Max value data? Is it possible to exceed 0.5s or 1s in some cases? Is there a formula for CDR Lock Time?

2. It says Typ <100ms when Normal Lock Mode is 25.78Gbps, but does this value change when it is 9.8304Gbps? If it does, is it longer or shorter?

3. Can you tell me the max value of CDR Lock Time at 9.8304Gbps? Could you tell me the value which is definitely not exceeded, such as max=250ms, 300ms, 400ms?

Best Regards,

Nishie

  • Hi Nishie-san,

    1) I don't have a formula available to share regarding CDR lock time.  Since this metric is not evaluated in production testing, there is not a max value.  TI empirical testing shows that for 25.78125 Gbps, adapt mode 2, CDR lock time remains well under 100ms.  Depending on channel loss, I would expect CDR lock times from 30-70ms.  I would not expect a case where 0.5s would be observed.

    2) For 9.8304 Gbps, the CDR lock time would increase.  At most, it would be double the lock time for 9.8304*2 Gbps.  In practice, since I don't have CDR lock time data for 9.8304*2 Gbps, assuming a CDR lock time of <250 ms for 9.8304 Gbps data would be fine.

    3) Note that this is not a production tested max value, but based on <100ms for 25.78125 Gbps data, <250ms for 9.3804 Gbps is a good expectation.

    Thanks,

    Drew

  • Hi Drew-san,

    Thank you for your reply.

    Let me ask you an additional question. Is it correct to recognize that CDR Lock Time<500ms when the data rate is 4.9152Gbps?

    Best Regards,

    Nishie

  • Hi Nishie,

    Yes, this is a good value to recognize.

    In practice, you may be able to improve CDR lock time if needed.  There are two methods that come to mind:

    1) A component of CDR lock time is time spend identifying the best CTLE and DFE values.  The retimer accomplishes this by incrementing through values and measuring the eye using the internal eye monitor.  It then uses a figure of merit based on HEO/VEO to select optimal CTLE/DFE values.  The time spent on each eye measurement can be reduced, which would in turn reduce CDR Lock + Adaptation time.  See "Set EOM Timer" in programming guide.

    2) Since a notable component of CDR lock time is adaptation, simplifying the adaptation process can reduce lock time.  Adapt mode 1 would result in faster lock time than adapt mode 2, and adapt mode 0 would be the fastest.

    Thanks,

    Drew

  • Hi Drew-san,

    I would like to try how to improve CDR Lock Time.

    Also, let me check if my understanding is correct from your past answers.

    1) Is the relationship between data rate and CDR Lock Time inverse proportion?

    2) Does 9.8304*2Gbps mean 19.6608Gbps?

    3) If yes, is the 9.8Gbps lock time exactly 262ms instead of 250ms?

    Best Regards,

    Nishie

  • Hi Nishie-san,

    1)  I believe there are some parts of the CDR lock time that are not dependent on data rate.  For example, I expect the time required for the device to detect a signal and enable functional blocks within the device to be independent of data rate.  However, I think a conservative estimate for CDR lock time is to just treat lock time as inversely proportional to data rate, as you mentioned.

    2) Yes

    3) Yes, my apologies. I had estimated the 250ms, but 262ms is more accurate.  Please keep in mind that CDR lock time is expected to be <262ms.  Based on this, for 4.9Gbps, please consider CDR lock time to be <524ms.

    Thanks,

    Drew