• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 33 subscribers
  • Views 3589 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.

Ethernet PHY and PCB design according to SLVA531A

anduss
Prodigy 80 points
Other Parts Discussed in Thread: DP83822I

We are designing in a new product that should be using an Ethernet interface. We are trying to choose a PHY device. A simple one with RMII interface.

We are worried about the design of the PCB. We had been told that we should have DISTANCES between the PHY and the magnetics of AT LEAST 1 inch and DISTANCES of 1/2 inch between the magnetics an the RJ45 connector. That seems surprising...

Yet we know that we should have good trace layout for the high-speed signals and differential pairs. We have consulted your SLVA531A documentation and there on page 15 it is written that we should try and make all distances as short as possible,

Would those LONG distance requirements we cited above depend on a certain type of PHY chip or would that be a GENBERAL requirment for Ethernet interfaces?

Thank you in advance for your response.

André Dussault

GEXELEC inc.

418 842 2198

  • 0
    TI__Mastermind 18425 points
    Hello Andre,

    The distance requirement is a general guideline that we recommend to customers and not tied to a specific PHY.

    -Regards,
    Aniruddha
  • 0 in reply to Aniruddha Khadye
    Prodigy 80 points

    But are the very short distances recomended in all cases of connection?

    - magnetics to the RJ45 jack?

    - PHY to the magnetics?

    - COU (MAC) to the magnetics?

  • 0 in reply to anduss
    TI__Mastermind 18425 points
    Hello Andre,

    The recommendation was for both PHY-to-Magnetics and Magnetics-to-RJ45. I am not sure what you mean by Mac to Magnetics, did you mean MAC to PHY?

    Leaving 1 inch between PHY and magnetics and 1/2 inch between magnetics and RJ45 should be ok as long as the traces are differential with 100ohm differential impedance. Are you anticipating that PCB space will be a constraint and want to reduce the length?

    -Regards,
    Aniruddha
  • 0 in reply to Aniruddha Khadye
    Prodigy 80 points

    We had had recommendations that the distance between PHY and magnetics should be AT LEAST 1 inch up to 2 inches and that the distance between the magnetics and the RJ45 jack should be at least 1/2 inch.

    That seems to be bizarre, since with correct PCB layout the differential lines could be laid out correctly almost any distance...

    BUT what we are after is keepingthe distances AS SHORT AS POSSIBLE.

    Is there any objection in having those distances as short as possible while keepingthe lines correctly laid out?

    Are there PHY chips that would oblige a design to have those long lengths of traces (or DISTANCES) between the components? We had designed a board with an SMSC LAN8720A and we have a lot of unwanted emissions and someone told us that we should have respected the distances I mentioned above.

    In such a case are those recmmendations for ALL the PHY cips or for SOME? Could you please recommend a PHY chip that could work in the RMII mode and be placed in a tight configuration on the PCB.

    Our concern also lies with ellectromagnetic emissions in order to pass the FCC A-level

    I hope this response is clearer for you, thanking you in advance,

    André Dussault

  • 0 in reply to anduss
    Prodigy 80 points
    NOTE: the recommendations for the MINIMUM legths and/or distances did not come from TI
    AD
  • 0 in reply to anduss
    TI__Mastermind 18425 points
    Hello Andre,

    Thanks for the additional information. As long as the traces are designed correctly, short distances should not be a problem. Is RMII the only primary requirement for your system? If yes, then I would recommend DP83822I. It is our latest 10/100M Ethernet PHY.

    -Regards,
    Aniruddha
  • 0 in reply to Aniruddha Khadye
    Prodigy 80 points
    Thank you, we will look at tyour recommendation tomorrow morning.
    AD
  • 0 in reply to anduss
    Prodigy 80 points
    Hello,
    Concerning the distance "problem" I want to solve, I found a document by SMSC that I suggest you read because it contains some information that goes agains the "short if possible" approach (the one that I would adhere to and prefer...)
    www.google.ca/url
    Look closely at its Section 2.3 and its section on the distances between the PHY and the magnetics. In that document they also refer to "other PHY vendors", saying "Among PHY vendors, the 25mm rule is considered good practice..."
    Thanking you in advance for your comments and suggestions.
  • 0 in reply to anduss
    TI__Mastermind 18425 points
    Hello,

    We have not specifically tested systems with less than 1 inch distance between PHY and magnetics since it would need multiple board designs and multiple EMI test runs. However 1 inch distance between PHY and magnetics should be good for DP83822. Since EMI is a key point for your application, sticking to the 1 inch distance would be better idea. We also recommend not to have any metal beneath the magnetics on any layer.

    -Regards,
    Aniruddha