1
\$\begingroup\$

I have used W5500 hardware for my gateway device. It has been implemented in 80-100 locations. But in 4-5 locations I'm getting an unusual issue related to the network connection.

The ethernet mode is static and it is connected to a cisco network switch.

  1. It gets connected to network in 100mbps full duplex mode for an uncertain period and then gets disconnected for which the network configuration is needed to be changed to 10mbps full duplex to connect it again.

  2. The second observation is alternating the port connection. When the network gets disconnected again it gets connected when port is changed in the cisco switch. Also if from the gateway device side the ethernet cable is reinserted it starts connecting to the network.

  3. When configuration of switch is set to the auto mode, it starts working. But, after 2-3 days stops connecting and again configuration changes to 100mbps and full-duplex it starts working fine.

As per the reference design the network mode is set to auto negotiation. I have this much clue on this issue, help me to find out if there is any limitation in firmware or hardware.

enter image description here

Top Layer enter image description here

Bottom Layer enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ Maybe a schematic design or PCB layout issue (assuming you're using a 'raw' W550 and not a module). Edit your question to show us the relevant pieces of the schematic & PCB layout. \$\endgroup\$
    – brhans
    Commented Oct 22, 2020 at 13:32
  • \$\begingroup\$ @brhans Yes, I'm not using module. I have integrated into our gateway device. Here I have attached the Schematic and board views. \$\endgroup\$ Commented Oct 22, 2020 at 13:53
  • \$\begingroup\$ Consider this (as the author was also using W5500's): superuser.com/questions/1527294/gigabit-ethernet-switch-issues \$\endgroup\$ Commented Oct 22, 2020 at 15:35

1 Answer 1

1
\$\begingroup\$

Let me express my thoughts on the case. It is not complete answer as details are not enough, but hopefully will give a direction for further investigation.

  1. Not clear if the problem is PHY circuitry or digital circuitry. Look into the W5500 driver logs to see if chip resets or having issues during disconnect time. Behavior will depend on the driver code behavior; it must reinitialize the chip if there's major problem with functionality.

  2. Commenters are right, layout is not the best. There's link to HW design guide on the W5100 page, this guide also applies to W5500. What I would change - separated 49R9 resistors into pairs, maximally decreased their path to power. I do not see how ETH_PWR is designed. Damping 33R resistors may actually do harm, try removing them/replacing with 0R. I do not see any rationale using TVS diode array because PHY is decoupled by magnetics, and at the time of diode breakdown W5500's PHY must be dead already; I would consider using TVS diodes as unsupported configuration because they are not present on WIZnet reference circuit.

  3. Having internal PHY chip may heat considerably. Does your board and chassis design assume proper cooling?

  4. The behavior you observe may be caused by the Cisco switch, with W5500 and Cisco port driver screwing up each other (there were cases). Dig into the switch port configuration. One of the settings I advise to check is port fast start option with fixed speed setting.

  5. Last but not least. Your board is custom, you must strictly follow soldering temperature profile in the datasheet because these chips do not like overheating and may be defected in strange ways.

\$\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.