1
\$\begingroup\$

I want to connect two phys on pcb through AC coupling (max path 150 mm), one PHY is KSZ9477S embedded PHY, other is DP83849IF PHY. The schematic is attached enter image description here

At first the I made a direct connection and it was 100 Frames per second loss. Now I added 49.9 Ohms and It became 100 times better (1 Frames per second loss, on some channels It 1 packet per our loss. The pcb traces are diff pairs with right impedance and matched length. I only have 200 MHz bandwidth oscilloscope and can't completely evaluate signal integrity, but generally it looks normal (like MLT-3), DC offcets are also fine (at least on DP83849IF side), does DC offsets even metter? KSZ9477S has internal biasing and termination... My question is how to acheave reliable communication in my case? Maybe I should play with capasitors and resistors values? What is the methology? Does it even possible to acheive relaible communication through AC coupling? I just found out that I have pretty bad ripples on 3.3V power supply rails (+- 200 mV, picture is attached) Can it make a difference?

\$\endgroup\$
7
  • \$\begingroup\$ Have you read the TI application note for AC coupling Ethernet? Both sides need termination (be it internal or external) and their own DC bias which is separated by the caps. And you need to set both ends in a mode that supports AC links, like forcing 100Mbps full-duplex links on. \$\endgroup\$
    – Justme
    Commented Apr 18, 2020 at 14:35
  • \$\begingroup\$ What unit is MKϕ? \$\endgroup\$
    – Ariser
    Commented Apr 18, 2020 at 15:07
  • \$\begingroup\$ That's microfarads, in cyrillic alphabets. \$\endgroup\$
    – Justme
    Commented Apr 18, 2020 at 15:11
  • \$\begingroup\$ Thank you for the reply... As I mentioned KSZ9477 has internal biasing and termination (according to datasheet)... So I made exernal termination only on DP83849IF side... MKϕ is 1 microfarad \$\endgroup\$
    – Sergey1101
    Commented Apr 18, 2020 at 16:17
  • \$\begingroup\$ Well how are the PHYs configured? With one PHY it was necessary to change output drivers from default mode to other mode (Class A vs B). Better find some evidence that KSZ PHY supports AC coupled mode. \$\endgroup\$
    – Justme
    Commented Apr 18, 2020 at 17:36

1 Answer 1

Reset to default
1
\$\begingroup\$

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.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.