W5500 False Reset due to AC switching noise

Hello,

We are using W5500 module for connectivity. Whenever we are switching AC Contactor, W5500 chip resets and all ip settings become 0. We get -3 error from socket function. We tried using multiple decoupling capacitors (100pF, 1nF, 0.1uF, 4,7uF) on Vcc and Rst pin, but the problem didn’t solve.

The waveform observed on the rst pin is as follow -

1 Like

This is not that straightforward. When you measure anything with scope its wires may take up the noise and add it to what you see on the screen. First of all, I believe problem with supply is more important factor than something on the reset pin - you should ensure power supply and wires going to W5500 are not affected and continuously supply required current to the chip. Second, system as whole is properly grounded. You also say nothing about what that contactor does - what load does it turn on? Why there’s so much EMI? Maybe problem somewhere else (not in EMI at all), and any other electronic device like W5500 will fail in your conditions?

  1. Power to PCB is through 12V, 7AH SMF Battery. It can easily source the current. There is no AC supply given to our PCB.

  2. No load is connected to contactor. We are just powering contactor coil (230VAC). Contactor coils are powered on-off manually through a switch. So contactor is electrically and physical not connected to our PCB.

  3. We have multiple chips on board such as stm32 microcontroller, RTC, EEPROM, LCD, etc But none of the chip are getting influenced by contactor switching.

  4. We get -3 return error from socket function which is meant for ip configuration is 0. In runtime IP configuration can become 0 only in 1 case when w5500 resets. So we are assuming that w5500 is getting reset.

  5. Apart from contactor switching, PCB and connectivity is very stable.

Thanks. There was similar case in the past, when we suspected chip’s circuit is influenced by the EMI, but we did not finish troubleshooting due to priorities change. See here. First of all I believe we must ensure that problem is in EMI. How close contactor is? If you move it away physically, does it help? If it is proven, then there’s a number of techniques to suppress differential mode noise, which, IMHO, most possible cause of the problems with contactor influencing transformer in the jack or TX/RX lines on the board - in other words, if reset happens, the first source of the issue I would consider is PHY. However I would first prove that W5500 actually resets, and in which its part - digital part of/and PHY part. For this you need to attach some kind of diagnostic output to the MCU driving the chip and cause it dumping all the registers on request (then you will be able to identify states of registers and thus what may have happened).

  1. If the contactor switching connection is away from PCB then problem doesn’t occur. If the switching point is within 20cm radius of PCB then problem occurs. Frequency of problem increases as the distance reduces. So radiated EMI must be the problem.

  2. We shall share diagnostic output of all common registers tomorrow for further analyses.

We have read all control register after getting -3 error. It can be clearly seen that all register values are set to default. So the entire chip is resetting.


Seems you are right. Take a look here. Here’s what I found on the internet (no ad intended, just read the theory). Two things are on my mind - grounding and shielding, to either devices - contactor or W5500 assembly. It would be wise to do them to W5500 because then it will be secured from other EMI sources. As I said before IMHO most susceptible parts of W5500 circuit are (analog) jack with magnetics and RX/TX wires.

Combination of 0.1uF and Ferrite Bead on reset line has considerably improved the situation. It is not solved 100% but frequency of occurrence has reduced majorly.

As per my understanding, it is a ground plane coupled EMI.

1 Like

Is “Reset Line” the W5500’s “Reset Line”?
Every time the inductive load you mentioned is loaded, the surge power splashes too hard… It doesn’t look normal. I think we need to turn it on properly first. Isn’t the GND line firm?

Did you find a solution to this?

No Omkar. I think w5500 has a problem. It is too much sensitive to EMI.

Are you aware of any other alternatives? I am thinking to consider ENC28J60.

Let me know if you find any success. I will too.

Thanks :wink:

We haven’t used ENC28J60. I think someone from Wiznet team should seriously address this issue.

The EMI issue is an important issue that continues to be addressed.

If there is a problem with the relay, there will be two cases.

  1. If there is a problem with the supply of the W5500 because the relay is being used for the power supply → this can be solved by L,C design at the power supply end. This is not a problem with the W5500, so you have to review the power design yourself.

  2. MDI line is affected by noise flowing into GND
    → This can be resolved by separating Analyst GND and Digital GND. However, if this is the case, the area of Analyst GND should be enlarged as much as possible. # Of course, the impedance design of the MDI line should be done.

If you are suffering from EMI issue, please send us circuit and Gerber file for review. And if you let me know which frequency band is the problem, I will review it.

We usually have a lot of problems with multiplied frequency, which can be solved by circuit changes and Artwork.

Hii Alan,
Can you share your email ID so that we can send you our design for review.

Thanks,
Omkar I.

Hello @Alan
I am using the W5500 Lite - T-Lite W5500 – LILYGO®

I am facing the same EMI issue. Is the implementation from Lilygo board causing the problem?
I have implemented a sophisticated setup with line choke and buffer capacitor with a good power supply arrangement.

Can I solve this by using WIZ850io Module or some other module from Wiznet?

I’m sorry for the late response. If you send me the details by e-mail, I can review it quickly.
Alan@wiznet.io

Has a solution for this problem been found?

No. I think w5500 is having unstable performance and wiznet team is not interested in addressing the concern.