Schematic review - W5500 not working

Question

Hi, I have a W5500 on my PCB but it doesn’t appear to be working (I have tested code on a different W5500 board before). I used the reference schematic from here - https://docs.wiznet.io/Product/iEthernet/W5500/ref-schematic Here is my PCB’s schematic: image1920×568 140 KB Is there anything I have missed because it looks exactly the same to me? The only thing I can think of it that the EVB has the reserved pins tied to GND through a resistor even though the reference schematic hasn’t? Thanks in advance!

Eugeny · Answer

How it does not work, and why 82 Ohm resistors? Edit: not sure what external transformer circuit has 82 Ohm resistors instead of 49.9 Ohm.

Eugeny · Answer

Matt: Is this schematic wrong? Thanks! This is a good question, but this must not be related to issues with SPI, if there’re any. Does it turn link LED on when you plug live cable in? OMG, your design does not have it!

Matt · Answer

Yeah, unfortunately I did just take the reference schematic as blind faith and if it wasn’t on there it isn’t included in mine. Is there a newer than 2015 reference schematic that does have the correct layout? Thank you!

Eugeny · Answer

As long as it is on the official website we can not deem it as incorrect but looks strange, and probably not a reason for thing not to work. You must identify what exactly does not work and keep working towards the root cause. For now SPI communication for digital part and link LED for analog part.

Matt · Answer

Unfortunately some other testing has blown my micro so I need to wait for a repllacement or a new revision to test it properly. Does this schematic look better? image1920×697 170 KB Thanks for the help!

Eugeny · Answer

Do not make changes until we identify what exactly is wrong with the device to see the direction to dig into. Hope you’ll fix it quickly and issue is not that bad.

Matt · Answer

Sure, just the schematic for now. I won’t send it to get made untill I can prove it out. My thinking was that all of the changes were pull downs/ups instead of floating so I can just DNP if needed. Also added the LED’s and changed the resistors to 49.9ohms to match every other schematic I have seen. I am sending the board off tomorrow to get a new uC to get testing with this. It’s a BGA so need it done properly and X-rated. Will report back what I find

Alan · Answer

Hello. I ask because I can’t see the whole circuit. Is the MCU controlling the RST? I don’t think that’s the case with the drawings you showed me.

Matt · Answer

Hi, nope I tied it to 3.3V to always be activated. Should I keep this 10k but put it to the micro aswell to pull low in startup?

Alan · Answer

Yes, the MCU must control the RST. All examples contain that part.

Eugeny · Answer

Agree with Alan, if you make new board design MCU to be able driving reset pin to perform hardware reset after power up. At worst you must have at least some capacitor-based reset circuit, but it is not the best practice.

Matt · Answer

Thanks both. I have added the reset signal from the MCU. The only thing I am not 100% on now is post transformer but hopefully as that is the same as the reference design then that bit is ok! Just need to add a voltage level translator as my MCU outputs 5V. I now have this: image1920×687 171 KB

Eugeny · Answer

Matt: Just need to add a voltage level translator as my MCU outputs 5V. This is another point of careful consideration. While datasheet says pins are 5V-tolerant, my experience (with W5100) showed incorrect behavior when driving with 5V signals - and I dropped 5V design completely, using 3.3V MCU (actually FPGA). I do not say it will not work, I show you a possible risk I have removed in my pervious designs.

Matt · Answer

Yeah that is a good point. I have just thrown in a voltage divider for the 3.3V reset pin but don’t want to do that for SPI. Should I drop those down to 3.3V SPI as well? I can’t change MCU unfortunately.

Eugeny · Answer

The level translation (from 5V to 3V) could be done using simple series resistor or divider as you seem have done already - its nominal is calculated by the desired current of the protection diodes at both sides and desired input voltage. You are lucky all the SPI lines are unidirectional, so you do not need bidirectional translators. What you can do right now is preserve seats for resistors, solder 0R nominals, but in case of communication problems play with other nominals.

Alan · Answer

I answered your inquiry through Channel Talk.

Alan · Answer

Oh… I don’t think it should be organized like this. I don’t think RST will work normally… Combined with pull-up resistance, there can be a lot of problems… For example, when the voltage is applied from the Pull-up power source, the power is divided, so it’s only about 1.98V. There could be a lot of problems besides this. Not recommended. (There is also a risk of damage to the parts due to reverse voltage.)

Alan · Answer

I think it would be better if you use a buffer that can distinguish power or use TR to down the level!

Alan · Answer

The W5500 is particularly sensitive to ESD. You have to pay a lot of attention to Artwork. If you send Gerber, I can check it for you.

zzwwss · Answer

Hi Matt Is your circuit board running normally now? I used TXB0108 bidirectional level conversion chip. The debugging is not smooth at present. I want to ask if you have successfully debugged it.

zzwwss · Answer

image1415×926 162 KB This is my circuit diagram, which is currently undergoing debugging. Could you please help to see if there are any other issues? The previous issue was that the LINKLED was always high and not lit. Previously, SPI communication was normal, but LINKLED remained at a high level. After a long investigation, it was re welded and the welding temperature was lowered to below 300. Now, by plugging in the internet cable, LINKLED can light up. I found the solution from the current website. Thank you all.

Alan · Answer

I’m glad the problem is solved. Thank you.