• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 34 subscribers
  • Views 1385 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.

How to achieving sub-ns accuracy in ptp applications on dp83640?

Kevin_KDP
Intellectual 260 points
Other Parts Discussed in Thread: DP83640

Hi,

      I have a ptp application, use dp83640 to achieve time precision in point-to-point connections.

      In AN-1730, I see the following words,

      By including the hardware-oriented advantages provided by the DP83640 in a PTP-enabled point-to-point connection, it is possible to achieve precision of                under 10 ns.

       Additionally, by enabling Synchronous Ethernet mode, it is possible to achieve sub-nanosecond precision in point-to-point connections.

     But when i enabling synchronous ethernet mode, the time precision  is 5ns as follows between two pulse per second trigger out.

    

     I have tried many ways, but  no effect.  And i found that the change unit is 8 ns. 

     I think the synchronous ethernet mode is effective. because the error(5ns) is very stable。

    My question ,

     how to revise the time which is less than 8 ns.

    At last, sorry for my bad English...)

  • 0
    TI__Mastermind 23675 points
    Hi Kevin,

    Please review the following application note:

    AN-1730 DP83640 Synchronous Ethernet Mode: Achieving Sub-Nanosecond Accuracy in PTP Applications

    Kind regards,
    Ross
  • 0 in reply to Ross Pimentel
    Intellectual 260 points

    Hi Ross,
    Thank you for reply.
    I have read An-1730 times. And i found that the time difference between two PPS signal is fixed when i use the same cable.
    I think the reason of the difference is phase difference between maser PTP clock and Slave PTP clock. But how to measure them, as we know the unit of PTP counter is 8ns. It's also the unit of Timestamp.
    I was puzzled with this for weeks...

    Best regards,
    Kevin

  • 0 in reply to Kevin_KDP
    TI__Mastermind 23675 points
    Hi Kevin,

    I am not sure what your question is.
    You want to measure the phase difference between the Master and Slave clock?
    or
    You want to get them to align?

    Section #6 in the application note talks about the test measurement setup for viewing the phase difference.

    AN-1729 DP83640 IEEE 1588 PTP Synchronized Clock Output application note (SNLA099) discusses the way to phase align the output clock.

    Depending on what question you are asking, I have listed the way to proceed above.

    Kind regards,
    Ross
  • 0 in reply to Ross Pimentel
    Intellectual 260 points
    Hi Ross,

    Sorry for my bad describe.

    I see that Section #6 in the application note talks about the test measurement setup for viewing the phase difference. My result of PPS phase difference is 4ns ~ 5ns. From the result of Section #6 Figure 11, i think the accuracy can be improved. But i don't know how to do with it...

    Most of the accuracy in Section #6 Table 1 i s less than 1ns when sync ethernet is enabled. So I think there is someting wrong or omission with my work.

    For emamble, If the difference is 5ns, how to adjuest it. I use temp rate adjust and time step adjust on slave, but it didn't work. the difference only from 5ns to -3ns or - 11 ns or 13ns ( i see it on oscilloscope) with unit of 8ns.

    Any other ways to deal with it? Or is there something wrong ?

    I look forward to your valuable suggestions.


    Best regards,
    Kevin
  • 0 in reply to Ross Pimentel
    Intellectual 260 points
    waiting for reply

    Thank you

    Kevin
  • 0 in reply to Kevin_KDP
    TI__Mastermind 23675 points
    Hi Kevin,

    To help explain the results you are seeing, lets break down PTP in both precision and accuracy. Precision being the distribution of repeated events, where the tighter the distribution the more precise it it. Accuracy being the medium based on an ideal reference point. With this in mind, you can have 4 unique states: Not Precise and Not Accurate, Not Precise but Accurate, Precise but Not Accurate, Precise and Accurate.

    The DP83640 uses an internal 125MHz reference clock for its PTP clock network. This means that the cycle time is 8ns, which entails a +/-4ns window essentially for an event to occur.

    The DP83640 allows you to use SyncE, but that Phase will be fixed depending on the cable, link partner and environmental conditions. That Phase can settle anywhere between +/-4ns. That is the maximum accuracy the PTP can be. However, the precision can be sub 1ns as discussed in the application note.

    What you would need to do with your setup is to measure those PPS you are triggering on and see the distribution, that is what is meant by precision.

    When you are adjusting the register you are just delaying the trigger by one clock cycle as you have shown above.

    Kind regards,
    Ross
  • 0 in reply to Ross Pimentel
    Intellectual 260 points
    Hi Ross,

    I think i have get it.
    Thank you for your patience.)

    Best regards,
    Kevin