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.

DP83867CS: DP83867CS application inquiry

Part Number: DP83867CS
Other Parts Discussed in Thread: DP83869

FCI_10131762-101LF-1 (3).pdf AMPHENOL_10124558-101LF_Product SPEC.pdf

Hi TI,

We'd like to use DP83867CSRGZT both on RGMII>SGMII, SGMII>MDI transceiver, and use HX5008NL for magnetics.

Could you help to check the topology as attached? we concern about trace length, connector(spec as attached) and impedance.

please give us any suggestion on it. 

  • Hello,

    Here are my comments:

    1. The traces from RJ45 to PHY should be less than 2 inches.
    2. The RGMII trace should be less than 6 inches, which is good on your block diagram.
    3. The SGMII trace (including the connector) overall should be less than 18 inches.
    4. For the MDI traces, ensure a controlled differential impedance of 100 ohms.
    5. For the MII traces, ensure a controlled single ended impedance of 50 ohms.

    As for the connector, this is a connector we have never used/tested, so we cannot comment on it.

    Regards,

    Adrian Kam

  • Hi Adrian, 

    Thanks for your assistance,

    There are some questions need to clarify:

    Do you have any reference document for MDI/SGMII/RGMII maximum trace length?

    If we can not shorten the trace length, does it has any redriver device that we can use?

    Have you ever seen the application before?

    Thank you very much.

  • Hello,

    1. For SGMII/RGMII, the maximum trace length can be found in the specifications of the interfaces. For MDI, it is a bit more complicated. While it is recommended to keep the traces short, it can be longer if the signal losses on the trace is kept to a minimum. If the long traces cause noticeable signal loss, it also affects the cable reach of your application. In the end, the MDI trace length varies based on your application/PCB design.
    2. The first link below is a link to TI's ethernet redriver product folder. However, I have never used/tested redrivers with the ethernet PHYs, so I cannot comment on a specific redriver device you can use. The second link below is an E2E thread confirming that there are redrivers that support SGMII.

    https://www.ti.com/interface/ethernet/retimers-redrivers-mux-buffers/overview.html

    https://e2e.ti.com/support/interface/f/interface-forum/429472/redriver-for-sgmii

    Regards,

    Adrian Kam

  • Hi Adrian,

    Thanks for your reply, and there are some questions since we have never implemented the application before.

    1. Do you have any experience for this application (RGMII>SGMII>MDI)? Is the purpose for trace length requirement?

    2. Any other solution for such long trace length in our application?

    3. The purpose of RGMII>SGMII>MDI is BMC dedicated LAN for debug. does it has any effect except lower speed if we exceed the trace length?

    Thanks.

  • Hello,

    1. Yes, we have experience going from RGMII to SGMII to MDI. Another one of our devices, DP83869, has the ability to act as a media converter, which allows RGMII signals to be converted to SGMII signals or vice versa. The purpose of this is to connect two devices with different interfaces, like an adapter. The converted RGMII or SGMII signal can then be routed to another PHY, which is connected to the MDI. As stated before, the trace length requirement is in the specification of SGMII/RGMII.
    2. Outside of using redrivers, I do not think there is any other workaround for your long trace lengths.
    3. If you exceed the trace length, there is a possibility that the signal loss would be too high, so the device would not work properly. If the using lower speeds results in lower signal loss on the trace, then the lower speeds would work, but not the higher speeds. In the end, it has to do with signal loss, not speed.

    Regards,

    Adrian Kam