• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 32 subscribers
  • Views 608 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.

DP83849IF: DP83849IF to KSZ9477S MDI connection

Andrey Pavlov
Intellectual 800 points
Part Number: DP83849IF
Other Parts Discussed in Thread: DP83849I,

I have a custom board with DP83849IF's TX/RX diff pairs connected to KSZ9477S diff pairs in 100BASE-TX mode wothout any pull-ups, pull-downs. So MDI interface  of DP83849I directly connected to MDI interface of KSZ9477S chip. Do I need in this case 49.9 Om resistor and 0.1 uF capacitor like in the reference design?  I'm experiencing some problems with packets loss, which I think occurs somewhere in  DP83849IF...

  • 0
    Intellectual 800 points

    Also I want to add that the problem looks similar  to the following thread.

    The MLT-3 encoding whitch goes from PHY to switch  doesn't look good, but link is always established but there are some packets loss...

  • 0 in reply to Andrey Pavlov
    TI__Genius 16030 points

    Hi Andrey,

    I am not clear about the connection between phy and the KSZ9477S. Are you connecting them through AC coupling cap instead of magnetics or is it a direct connection?

    Can you send me following snapshots :

    1. Snapshot of  schematic showing the connection of phy and KSZ9477S.

    2. Snapshot of the front end of KSZ9577S (if possible) that you are referring to.

    I can review these and let you know the minimum required components.

    --

    Regards,

    Vikram

  • 0 in reply to Vikram Sharma
    Intellectual 800 points

    Thank you for the reply. Now KSZ9477S chip and DP83849IF connected directly without any AC coupling, biasing, termination or magnetics, just traces on the PCB. The schematic snapshort is attached (schematic_snapshort.jpg). Also I've made the following experiment: I cut off solder mask on pcb and solder 49.9 Ohm resistors on 24 and 23 pins (TPRPD_A, TPRPD_B, I beleive it's a receive diffpair, not sure though because of MDI-X) and connect them ro 3.3V like in reference design for DP83849IF (pull ups to 3.3V through 49.9 resistors). The performance became 100 times better. If 100 packets per minute were lost then now it's about 1 packet per minute (I use special 4000 frames/s stream for testing).  Also I can provide waveform from oscilloscope before and after modification. Also I asked microchip for KSZ9477S front-end, maybe they'll respond...

  • 0 in reply to Andrey Pavlov
    TI__Genius 16030 points

    Hi Andrey,

    Though we recommend AC coupling and termination resistor but you have an interesting use-case. It all depends upon the front end of KSZ to find out :

    1. Whether there is an electrical conflict on these lines or not because of direct short.

    2. Whether the impedance offered by KSZ is sufficient and with correct common mode potential connected for good signal integrity.

    Let me know when you get more information about KSZ and I will also check the same. 

    But just curious why will you have a phy connected on board on MDI lines directly to switch? MAC extension? What is the reason for not going for AC coupling and resistors? Let me know and I may help you with optimization further.

    --

    Regards,

    Vikram

  • 0 in reply to Andrey Pavlov
    Intellectual 800 points

    So I've managed to solve the problem by doing three things: I've placed capacitive coupling (0.1 uF) between DP83849IF and KSZ9477S, also I've placed 49.9 pull ups on DP83849IF side, and the last thing I did was turning off autonegotiation on KSZ9477S chip. So now there is no packet loss.