W5500 stuck on socket close

TCP/IP Chip W5500
21

Sorry for the late answer.
So do you mean it is not working on Ethernet Linking status?
Unfortunately, there is no way to check the chip condition without PHY Link on.
Could I know why you need to check chip status before PHY link?
And if you share your h/w schematic, we can also check if there is any issues or not.
Thank you

1 Like
22

Hello @lawrence,

Iā€™ve a faulty W5500 which cannot establish the Ethernet link, as Irina pointed out probably the PHY is not working correctly, but the chip is not entirely fried because the SPI works, the register RAM works and the 25 MHz clock is oscillating correctly.
Iā€™m trying to understand what happened to this chip to prevent this kind of problem.
Schematic attached.

schematic
schematic1261Ɨ678 16.4 KB

Thank you.

23

I have one question and an opinion, by not sure if it is appropriate to say anything.

24

Eugeny, seriously, no hard feelings, ask away. I was just that me repeating the same things and you dismissing my questions was a waste of your and my time, and I had little time. Water under the bridge, I hope you understand.

25

@gama sure, also no problem from my side. I was waiting for answers for questions I was asking, you were asking more questions without answers.
So hereā€™re questions (I broke one big question up into several):

  • does this failure occur after you power off device, and then power on, or may appear spontaneously during operation? As I see you initially said that you close idle sockets just after power up/during initialization, you noticed the problem after power on, right? Or you identified problem after power on, but issue occur during normal operation?
  • what is powering the W5500 - is circuit available? Is it battery?
  • what is the rise time for the power during power on? What is the slope of the power voltage rise?
  • would be great if you have scope and sample the waveform for the voltage on the VDDA rail several times to see the actual power rise cycle (and at best power fall cycle during power off). But you must measure these on the normal W5500, not defective one (as defective may have different active and reactive resistance and impact the readings).

And my opinion is that problem should not relate to network cable, it must be somewhere on the board, or outside the board, and the first candidate is power. While you can see filtering circuit on the reference schematic and you have just copied it (however seems ferrite bead is 181, not 121), the VDDA power rise profile heavily depends on the power rise profile at the entry to the filter. If you happen to have battery, closing contacts (powering the filter) will cause very fast voltage rise, and it may cause heavy overshoot after the filter. Digital circuits may have different sensitivity to these overshoots and survive them. If you are interested in theory and simulation, then use LTSpice and see how different power rise profiles affect the filter output voltage shape.

Edit: the first question which actually must be asked - seems you have your own board, how do you solder W5500 chips?

26

@eugeny:

  • the problem happened only on one chip and I donā€™t know when it happened, I only realized there was a problem after a power up, but before the board was powered without ethernet cable for 2 o 3 days straight, the problem could have happened then and I didnā€™t realize

  • the 5500 is powered by a step-down converter from 12-24 VDC, the converter is a MC34063 with a standard circuit, it is a bit noisy but nothing alarming, the ripple is 200mVpp max at 24V input

  • the max slope is about 1V/10ms on VDD and AVDD

  • the slope is quite consistent on multiple power-ups, it is really soft without any overshoot, I measured it on a working W5500

  • the chip was originally soldered in a oven, after the problem I removed and re-soldered it manually

I agree that the power could be the cause, probably Iā€™ve fried it myself somehow. However I find strange that something on the chip still works, I would expect a short-circuit on one of the power rails, also on the same 3.3V there is a STM32F103 which was not damaged, and Iā€™ve found the STM32F to be quite weak.

Anyway, now I have 10 boards working for two weeks without problems, so I tend to think it was a sporadic incident.

1 Like
27

So this research did not give us definite answer to what have happenedā€¦
The WIZnet chips are very sensitive to soldering temperature, when soldering using reflow please ensure you fully comply with the profile in the chapter 6 of the datasheet, and when soldering with iron ensure you set temperature below 325 C (I solder at 315). Soldering at 350 may damage the chip and soldering at 375 will kill the chipā€™s functionality in weird way.