• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 2396 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.

DP83640: PTP Function Requestion DP83640

SHH
Mastermind 32196 points
Part Number: DP83640

Tool/software:

Hi Team,

If customer's architecture is as shown below, can customer use the DP83640 to implement and support PTP?
SOC (Support PTP) --SGMII--> DP83640 --PHY--> Connector --PHY--> DP83640 --RMII--> SOC (No Support PTP)

In addition, the app note mentioned the software approach. Using a standard MAC and PHY can also achieve PTP.
Do you have any related information for this part that you can share?

0726.TI_IEEE1588_PTP.pdf

BR,

Scott

  • 0
    TI__Mastermind 30610 points

    Hi Scott,

    DP83640 does not support SGMII, can the SOC supporting PTP also support MII or RMII?

    SHH said:
    Using a standard MAC and PHY can also achieve PTP.

    DP838640 has the capability to support PTP using timestamping. The purpose is to synchronize a master clock across multiple devices. A standard MAC may be able to use the PPS or clock signals to synchronize to the master clock through software as described in this app note.

    Best,

    Shane

  • 0 in reply to Shane Hauser
    Prodigy 20 points

    Hi Shane, Can PPS alone achieve synchronization accuracy better than 1 ms in your setup, or would hardware timestamping be required to reach that level?

  • 0 in reply to Chung Jack
    TI__Mastermind 30610 points

    Hi Mr. Jack,

    From the linked app note section 5.3, the average PPS synchronization between the two connected PHYs is ~869ps with a standard deviation of 7.87ns. From this, the PPS signals of the PHYs can be synchronized within 1ms.

    Are you asking whether the PPS from the PHYs can synchronize the other clocks in your system to within 1ms? On further inspection the PPS alone likely will not synchronize the clocks as it is mainly a reference signal.

    You can use the 1588 output clock from the PHY as a reference PTP clock, but if you want true PTP synchronization you should send the timestamp data from the PHY to a PTP software stack. 

    Best,

    Shane

  • 0 in reply to Shane Hauser
    Prodigy 20 points

    Hi Shane,
    1.Our objective is to synchronize the system time of a Master and a Slave, with a synchronization error of less than 1 millisecond (ms). We are considering two methods: pure software implementation of IEEE 1588 and hardware-assisted IEEE 1588. Can both of these methods achieve our required accuracy?
    2.Do both the master system and the slave system need to support PTP, or is it sufficient for only the two DP83640 devices to support PTP?

  • 0 in reply to Chung Jack
    TI__Mastermind 30610 points

    Hi Mr. Jack,

    1. Yes for <1ms accuracy a pure software approach will likely work. Hardware-assisted timestamping can improve the accuracy down to the nanosecond range, but may not be necessary here.

    2. Both Master and Slave systems need to support the IEEE1588 PTP software stack. A standard MAC/SoC should support this stack as it is software based, but check with the SoC manufacturer to be sure. The PTP hardware support from DP83640 is primarily to improve the accuracy of the PTP synchronization. 

    A good resource for software vs hardware PTP implementations can be seen in this YouTube video from TI. Specifically minutes 12:26 - 20:30 covers the different PTP implementations.

    Best,

    Shane