Hello. I designed a board for ethernet daisy-chaining with a plug-in passive POE board, and would like some help on the next iteration before I order more parts. I know it is niche and RS485 + power would've been more straight forward, but I would like to learn why this works and doesn't work.
The POE board converts the V+ and V- from the Magjack connectors to a clean 5V, and I can interact with the W5500 with a microcontroller, but I've never gotten it to connect to the network properly. Link status is OFF, but the router and cable are fairly new. I think I am not following the W5500 reference designs well, though it is hard to match their designs when I would like to daisy-chain these and need a Magjack connector that can handle POE, and their examples don't do that. My old design (top image) is assembled, and I have not ordered the parts for my new design (bottom image) yet. Any help on this would be greatly appreciated as I am starting to get burned out. Thank you.
EDIT: The new design does work, but does not allow for daisy-chaining.
I don't think it's possible to daisy chain Ethernet. You would have to have a switch (or hub) if you need the signal traverse from one RJ45 to another RJ45
Thank you for the reply. I am looking into daisy-chain ethernet topology and the public info is pretty limited, but a key benefit I have seen listed is the reduced cost and setup time since no switch is required. I will try to find out what kind of custom hub is used in place of that to allow daisy-chaining, but nothing else was connected to the second rj45 pins, so I expect that was not the cause of the cable issues.
Edit: Modbus TCP/IP might be the designs I need to look into. I'm sure the expensive equipment that makes these up wouldn't be able to stay so expensive if they were open source, but I will find what I can.
Hello, and thank you for the reply. My ethernet board had a loose connection to the RJ45 connector, which is why it never connected to the network. I moved some components to a breadboard and used the updated schematic while putting pressure in the right spot of the connector pins and that got it working. However, my design will not allow for daisy-chaining which is what I am hoping to change. I am still working on the issue, and have found a number of potential solutions. Unfortunately, they are solutions I am not very familiar with.
Option 1 is to use a W5500 with an ethernet switch, though I have not found much on W5500s using external PHYs and switches, and this option adds the most cost to the project.
Option 2 is to use an etherCAT chip instead, which seems to be an iteration on ethernet tech that is designed for multidrop.
Option 3 is to use BASE-T1S which is an official iteration on ethernet that is designed for multidrop. I have found nothing on compatibility between W5500 or any MAC and the T1S transceivers because of how new they are.
Option 4 is to ditch ethernet and use CAN FD or CAN XL, though I am worried that they will both be obsolete in a few years.
Option 5 is to use some hybrid of ethernet and CAN so the majority of diagnostics can be done with mini hubs that connect to a central ethernet switch, but each mini hub connects to its sensors and peripherals through GPIO/I2C/CAN.
Using TCP in a daisy fashion seems to be the same difficulty as developing a 2-port switch with the W5500.
The configuration of the device you are trying to implement is understandable, but the circuit is different from the basic w5500 circuit and I do not understand.
It is also necessary to check whether the impedance matching is correct for stable communication.
Of course, I think it would be more convenient to use RS485 as well.
After doing more research, I see that you are absolutely right. The W5500 won't support an external PHY and neither will most modern ethernet controllers, which forces the circuit to be a single port ethernet device, or come up with a janky manual switching system on the MCU.. If I were to make a dual port system for daisy-chaining, then I might as well just get an MCU with MAC and an ethernet switch chip with everything else.
If anyone else finds this thread hoping for a good switch schematic, ignore the images I shared. The ESP32-C3-MINI ($2) + KSZ8863RLL ($5) seems like a decent route for cheapest ethernet switch, but I don't have any schematics for that. You could also set up the ESP32 to switch to wifi if the LAN connection is broken, and there are plenty of ESP32 dev boards to prototype on before committing to the chip.
RS485 seems like the best current option, but I am happy to have learned a lot during this process. I will continue researching T1S+PoDL since I believe we will be seeing a lot more of that or the next iteration of it very soon, but I will also prepare an RS485 system as a safety net.
Thank you for taking the time to reply and explain on this post.
