W5500 stops working

Question

Hey everyone, we’re using the W5500 chip on our own PCBs, so no breakout etc. They’re working quite well most of the time. Nevertheless we got a couple of W5500 that “died” rather unexpected. It’s happening very sporadic so there’s no real reason to identify. Another W5500 decided to stop working recently. The software on the connected MCU hasn’t changed. SPI communication seems to be good as we can read various registers like the version register. But the chip won’t open the socket. None of the status LEDs are on or flashing. Are there any recommended tests to identify the root cause of those problems? Kind regards and a nice weekend, MiD

Eugeny · Answer

What is the value in PHYCFGR register when problem happens? Check 3V3A level. Use scope to see that is going on the lanes (twisted pair) before transformer (on the pins of the network connector) and after transformer (at the input to the RX channel). The problem could be PHY configuration issue (e.g. PHYCFGR[5:3]=110) or electrical problem with W5500 PHY not seeing input signal in the RX channel.

Eugeny · Answer

MiD: I can’t get the register value when the problem appears, because I never know when it’s happening. Just print this register is a loop into the diagnostic output. And when you know it ‘died’ look at the output. MiD: Nevertheless we got a couple of W5500 that “died” rather unexpected. Died completely - you power off the system, power it on, and it does not work any more? Or is it back alive when you power cycle? MiD: The W5500 doesn’t even recognize a plugged-in ethernet-cable. It will only recognize if its core works properly, its PHY works properly, and there’s incoming carrier in the RX channel.

MiD · Answer

Eugeny: Died completely - you power off the system, power it on, and it does not work any more? Or is it back alive when you power cycle? Completely, at least the network part. SPI is still working at the current board. Unfortunately I’m unable to test older defective units as the chips were already replaced. Eugeny: Just print this register is a loop into the diagnostic output. And when you know it ‘died’ look at the output. There is no internal diagnostic output that is running on the MCU all the time. Funny thing is that the W5500 is in charge of providing some sort of diagnostic output for the MCU/system. Eugeny: It will only recognize if its core works properly, its PHY works properly, and there’s incoming carrier in the RX channel. This means that the core/phy of our W5500 ‘died’, since there was no change in either hardware or software?

Eugeny · Answer

Did not see chips dying completely so far. Hard reset /power cycle usually cures the issue. I suspect your circuit must be assessed - I would start with power and network connection. Can you share circuit diagram and board image? How chip is powered - can the issue come from power - what is the circuit diagram of the powering circuit (including filters)?

MiD · Answer

As far as I know we were sticking to the recommended schematic. But I’ll need to ask our electronic engineers about that. I would like to take you up on your offer after I’ve returned from my vacation in the first week of october, if this is okay with you Nevertheless thank you very much for your answers and help so far!