• State Resolved
  • Locked Locked
  • Replies 2 replies
  • Subscribers 33 subscribers
  • Views 255 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.

DP83867IR: Confirmation of layout rules

John Ford
Intellectual 530 points
Part Number: DP83867IR

Hi all.

Hi.

The DP83867 datasheet contains a small section on layout, but leaves many questions unanswered.

I've read a lot of the posts on this forum about layout rules, and I have a question that I would like to ask about layout between the PHY and the magnetics.  Specifically, about length matching. 

1) As I understand it, the most critical rule is to match the lengths of the 2 signals in a pair.  I can't find any TI recommended maximum mismatch.  Is there one?

2) From reading the 802.3 spec, section 40.1.3.2, the receivers are independent of each other, and so each pair's length is somewhat independent of each other?  Is the clock recovery system duplicated on each receiver in the DP83867?

40.1.3.2 Physical Medium Attachment (PMA) sublayer

...

The PMA Receive function comprises four independent receivers for five-level pulse-amplitude modulated signals on each of the four pairs BI_DA, BI_DB, BI_DC, and BI_DD,"

 

  • 0
    TI__Genius 9920 points

    Hi John,

    1. We have not characterized nor finalized the maximum mismatch but our recommendation would be for the pair to mismatch between 50 mil 

    2. Again we do not have it characterized in our datasheet but each pair's length are independent with reason. It is recommended to have the pairs be at most 250mil of length mismatch. And yes, there is a clock recovery system on each receiver. 

  • 0 in reply to Cecilia Reyes
    Intellectual 530 points

    Thanks.  We are seeing some communication errors with our current board layout, which has some mismatch in the pair length, in the worst case about 60 mils.   We are searching for the cause.  It seems maybe this might be the problem or not.

    John