• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 32 subscribers
  • Views 616 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

MDI signals on a 40-50cm cable

Giuseppe Modugno
Prodigy 70 points
Other Parts Discussed in Thread: DP83848C

Dear forum,

I want to use a Ethernet PHY DP83848C from National. The PHY should be placed near the CPU on a circuit. This board will be installed at the center of an enclosure. Naturally I'd like to have the RJ45 Ethernet port located on the panel that is 40-50cm far from the circuit.

One solution is to put the CPU, PHY and RJ45 with magnetics on the central board and connect a patch cable (RJ45/RJ45) from the central board to a female/female adapter on the panel.

But I don't have too space on the CPU board and I'd like to don't lose the LED signals for RJ45 connector, so I was thinking if it could be possible to leave CPU and PHY on the central board. I'm going to deliver the MDI signals from PHY to a simple plastic RJ11 6-poles connector (no magnetics) where I can connect the two 10/100 pairs and the two LED signals. With a patch CAT5 cable (RJ11/RJ11), I go on a small board on the panel where I'll have the RJ45 connector with magnetics and LEDs.

What do you think about this approach? How should I care the two MDI signals pairs? Is CAT5 suitable to transport MDI signals from a Ethernet PHY?

Thank you for your answers and suggestions.

  • 0
    TI__Mastermind 31600 points

    Application note AN-1469 (Literature Number:  SNLA079C) is the design and layout guide for the DP83xxx Ethernet devices.  It provides some general guidance and some considerations with respect to engineering tradeoffs.  It sounds like your application exceeds the bounds of the guide, but it should provide a starting point for discussion.  The guide is available on-line at:

    http://www.ti.com/litv/pdf/snla079c

    Regarding your specific application, I would like to understand your constraints and flexibilities a bit better.  40-50cm from the Phy to the connector is a longer than usual run, but may be feasible.  Based on the text above though, I am not sure if that is even a possibility. 

    Will the main board abut the enclosure such that the traces could be run from the Phy to the connector without using an intermediary cable? 

    What interface are you planning to use between the CPU (MAC) and the Phy, RMII or MII?  If you cannot move the CPU (MAC) closer to the connector, could you move the Phy and increase the distance between the CPU (MAC) and the Phy? 

    Are you planning to use an RJ-45 with integrated magnetics or are you plannig to use discrete magnetics (separate from the RJ-45 connector)? 

    Patrick

  • 0 in reply to Patrick OFarrell
    Prodigy 70 points

    Patrick O'Farrell said:

    Application note AN-1469 (Literature Number:  SNLA079C) is the design and layout guide for the DP83xxx Ethernet devices.  It provides some general guidance and some considerations with respect to engineering tradeoffs.  It sounds like your application exceeds the bounds of the guide, but it should provide a starting point for discussion.  The guide is available on-line at:

    http://www.ti.com/litv/pdf/snla079c

    Regarding your specific application, I would like to understand your constraints and flexibilities a bit better.  40-50cm from the Phy to the connector is a longer than usual run, but may be feasible.  Based on the text above though, I am not sure if that is even a possibility. 

    Will the main board abut the enclosure such that the traces could be run from the Phy to the connector without using an intermediary cable? 

    What interface are you planning to use between the CPU (MAC) and the Phy, RMII or MII?  If you cannot move the CPU (MAC) closer to the connector, could you move the Phy and increase the distance between the CPU (MAC) and the Phy? 

    Are you planning to use an RJ-45 with integrated magnetics or are you plannig to use discrete magnetics (separate from the RJ-45 connector)? 

    Thank you for point me to the AN-1469 document, its reading was very interesting for me.

    Unfortunately the CPU (with embedded Ethernet MAC compatible with RMII) will be placed on a central circuit, but I want the RJ45 connector on the panel. The magnetics will be integrated in a RJ45 connector from ERNI. Do you think is it better to have longer RMII signals instead of longer MDI signals? After reading AN-1469, it seems to me all the signals are sensible and should be traced with very short traces.

  • 0 in reply to Giuseppe Modugno
    TI__Mastermind 31600 points

    There is no hard limit on the spacing between components.  It is a matter of performance.

    Extending the distance between the Phy and the magnetics / RJ-45 will reduce performance in two ways:

       1. Signal integrity into the Phy will get worse.  This could impact cable reach.
       2. Radiated emissions could get worse.  Longer traces on the high speed MDI lines will act like antennas and radiate.

    We designed the device with margin to the specifications, but deviating from best practices will reduce that margin.  In general, for spacing between components, I try to prioritize them in this order:

       1. Distance between the Phy and the 50 Ohm termination resistors.
       2. Distance between the magnetics and the RJ-45.
       3. Distance between the 50 Ohm termination resistors and the magnetics.
       4. Distance between the MAC and the Phy.

    Due to the fact that the RMII signals are synchronous and lower speed than the MDI signals, we have the most flexibility with respect to the distance between the MAC and the Phy. 

    Given the distance that you want to cover, I think that the best implementation would be to split the difference between the RMII and MDI traces.  If you could reduce both the RMII traces and the MDI traces to 15-20 cm, that would give you the best opportunity for success.  The shorter you can get those traces, the better the performance will be.