• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 33 subscribers
  • Views 1180 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.

DP83867CR: Connection between two PHYs through a transfomer.

Elder Costa17
Expert 2825 points
Part Number: DP83867CR

May I connect two PHYs (one of them a DP83867CR) through a single 100BASE-TX transformer? I remember seeing a question where the PHYs were directly connected but in my case the DP83867CR is electrically isolated from the other end through the 100BASE-TX transformer. I want to avoid placing a second transformer on board if I can due to space constraints.

  • 0
    TI__Mastermind 19780 points
    Hi Elder,

    You can connect 2 PHYs through a single transformer on a PCB.

    As you selected a DP83867, I suspect you're trying to use 1000M speeds which requires all 4 channels. A 100base-tx transformer may only have 2 channels inside.

    Best Regards,
  • 0 in reply to Rob Rodrigues
    Expert 2825 points
    Hi, Rob.

    I will be using two channels. I chose this part because of Linux and U-Boot support, not speed. A 100base-tx part that was suggested to me in another thread (DP83822) lacks U-Boot support whereas DP83867 has support (it may require small tweaks though). For my application, 100Mbits/s suffices and magnetics with required isolation is available for this speed only.

    About the implementation, is there an application note or desig reference I could take as a basis? Or I should just connect them and that is it?

    Should I cross the connections (RX-TX and TX-RX) or it is not necessary?

    The transformer has the common mode chokes built-in on the media side, thus connected to the end opposite to DP83867. Is there any problem that they could cause?

    Finally, I have read in a topic about magnetics-less connection between PHYs that Ethernet assumes a minimum cable length and, therefore, some load capacitance was added because of that. The person linked the app note below:
    www.intel.com/.../8255x-fast-ethernet-controllers-without-magnetics-appl-note.pdf

    On the other hand, TI's slna088a/AN-1519 app note has none.

    So, in my case, would it be advisable adding some capacitance? If so, in which point?

    I apologyze for so many questions in a row.

    Thank you very much for your support.

    Elder.
  • 0 in reply to Elder Costa17
    TI__Mastermind 19780 points

    Hi Elder,

    I will be using two channels. I chose this part because of Linux and U-Boot support, not speed. A 100base-tx part that was suggested to me in another thread (DP83822) lacks U-Boot support whereas DP83867 has support (it may require small tweaks though). For my application, 100Mbits/s suffices and magnetics with required isolation is available for this speed only.

    [Rob] Great.  Then both DP83867s will have to be configured to 100M mode, or have the 1000M autonegotiation capability removed. 

     About the implementation, is there an application note or desig reference I could take as a basis? Or I should just connect them and that is it?

    [Rob] No, the only reference would be for the TLK series of devices which shows the use of a single transformer only.  This has also been tested in our lab an works. 

    Should I cross the connections (RX-TX and TX-RX) or it is not necessary?

    [Rob] This is not necessary as auto-mdix will resolve the connection.  Also

    The transformer has the common mode chokes built-in on the media side, thus connected to the end opposite to DP83867. Is there any problem that they could cause?

    [Rob] No, the common mode choke placement will not be an issue.

    Finally, I have read in a topic about magnetics-less connection between PHYs that Ethernet assumes a minimum cable length and, therefore, some load capacitance was added because of that. The person linked the app note below:
    www.intel.com/.../8255x-fast-ethernet-controllers-without-magnetics-appl-note.pdf

    On the other hand, TI's slna088a/AN-1519 app note has none.

    So, in my case, would it be advisable adding some capacitance? If so, in which point?

    [Rob] The need for capacitive loading on the front end of a PHY will vary based on the PHY in question.  We have not seen short cable issues with the DP83867, yet I would strongly recommend leaving a place holder for some load to be added.  I would ensure the load is at the output of the driver as shown in the 8255x app note.

    Best Regards,

  • 0 in reply to Rob Rodrigues
    Expert 2825 points

    Hi, Rob,

    Thank you for the answers. 

    Regarding this:

    [Rob] No, the only reference would be for the TLK series of devices which shows the use of a single transformer only.  This has also been tested in our lab an works.

     

    Do you have a link (data sheet or app note) for the above?

     

    BR

     

    Elder.

  • 0 in reply to Elder Costa17
    TI__Mastermind 19780 points
    Elder,

    This is the app note I was thinking of: www.ti.com/.../slla327.pdf

    It doesn't have the single transformer configuration I was thinking of, however. My apologies.

    Best,
  • 0 in reply to Rob Rodrigues
    Expert 2825 points
    Hello, Rob,

    I had to look twice to see what you meant (option 2 on Fig. 9) as at first sight I thought it was a conventional connection. AN-1519 I mentioned above has a similar diagram that went unnoticed by me. Anyway, your hint is very helpful as it shows the path I must follow.

    Thank you very much for your support.

    BR.

    Elder.
  • 0 in reply to Rob Rodrigues
    Expert 2825 points

    Hi, Rob.

    [Rob] It doesn't have the single transformer configuration I was thinking of, however. My apologies.

    There is no need to apologize. The top diagram on Fig. 9 provides enough info to allow me to create an electric diagram. I assume that was one of the scenarios you mentioned were tested. In my case the connectors are not necessary, as well as coupling capacitors.

    BR

    Elder.

  • 0 in reply to Elder Costa17
    TI__Mastermind 19780 points
    Great. I am glad it helped, Elder.

    Best Regards,