Let me first say that I admire your hack, but for me it brings to the fore various thoughts simmering in the background about IoT. Here are some anecdotes in no particular order. no logical structure, and not directly related to your project.

I don't know how it is where you live, but here Ethernet cabling has to be installed by qualified electricians, which adds to the cost. Of course that 10m Cat6 patch cable between my workstation and the switch in the next room is exempt.

A friend of mine had Ethernet cables and sockets installed in his home a couple of decades ago. He probably wouldn't do it again. I think he used Cat5. It's obsolete now.

The copper in the Ethernet cables in his home is worth more than most IoT devices. And copper is a valuable resource, jokes about copper in Chinese ICs notwithstanding. Also the ports on the switches also cost.

There is a whole generation of homebodies who have never used a wired port, except for the connection between the phone or VDSL socket in the wall to their wireless router.

Given the hassles and cost of setting up wired infrastructure, it's not surprising that wireless has been adopted with a vengance. For years now there have been more wireless devices on my LAN than wired.

Hence ESP8266 or ESP32 in every other MCU project here on HaD or so it seems.

Of course the irony is that now we have a proliferation of devices dependent on wall warts for power. But at least that is more ad-hoc than Ethernet cabling in the wall.

I've come to the conclusion that the next breakthrough in IoT is when devices have the means to harvest power from their surroundings in various ways, and be low powered enough to work with that scant energy. Or maybe tiny fuel cells where you feed the device some crumbs now and then. It would be like maintaining a pet. :) It's actually easier for devices out in the sun, they have to use solar cells. and power cords are not an option.

I don't know where you live, but here you could run any cables in any way inside your property without any authorization from any authorities.

As for cabling - it does not really matter, would you run mains power, some low voltage power or ethernet cable to your device. The price of routing will be the same, and Cat5 is cheaper than nearly any cable sutable for mains or low voltage.

Here we have ~50 sunny days per year, so, solar power is not an option. Wind power is not an option too. Also, we have temperatures below zero nearly half a year. So the batteries is definitely not an option here for outdoors.

And even good batteries usually survive only 500 -1000 cycles even in perfect conditions. So, even you got to live in warm paradise with every day sunny and good constant wind, you have to change them at least once in 3 years.

Also, in our towns WiFi bands is completely full. With huge help of all that WiFi IoT things too. Even 5G band sometimes become unstable.

If there would be small Ethernet capable MCU's, I would definitely prefer them over any wireless solution to get connectivity.

If you go to https://lcsc.com/products/Ethernet-ICs_579.html and filter by Ethernet controller you will find some possible candidates starting at a few dollars and 48 pins.

This is just controllers, not general purpose MCUs with Ethernet MAC and PHY. You can not write and flash your own firmware to this ethernet controllers.

There are no any available small and cheap MCUs with ethernet on market.

Closest thing is PIC18F66J60 in TQFP64. Also, with some imagination, you could count W7100 in QFN64, but it is just two MCU's in one package - one 8051 core you could write code for and some (much more powerful) proprietary core in separate ethernet controller you cannot access.

For hobby applications getting to reliable low-power operation is demanding. I guess that sometimes wired operation is a good compromise.

One option is to have a wired "robust" fall-back, and wireless for bandwidth, mobility, bells, and whistles. Please keep in mind that the WAF (Wife Acceptance Factor) of unreliable wireless opertion is notoriously low.

WAF of wires can be low too. 😉 But I understand the motivation to use what's on hand to solve a specific use case than try for a general solution which will be over-lots-of-things (cost, time, feasibility, etc). After all most if not all of my projects fall in this category. But I do think wired is swimming against the tide of trends.

I'm not even sure you can get 10 Mb switches these days, they start at 100 Mb and 1 Gb is getting common. You can buy Ethernet controller chips, starting at about 40 pins. And it offloads the packet handling.

Did not met any modern switch incapable for 10M on copper port. Few known to me models have few dedicated SFP ports for 1G only modules (even 100M fiber SFP will not work), but all copper ports will work in 10M mode. I think they all can do it, just nobody mention 10M today.

As for controller chips - it is overkill for most simple tasks in IoT or general data acquisitioun and remote control. You don't need some special controller chip for USB or UART. Why should you need it for Ethernet? Ethernet is even much simplier than USB.

Moreover, nearly every "controller" (say, ENC28J60 or W5X00) that could offload packet handling is really an MCU with Ethernet peripheral. However, for completely unimaginable reasons, manufacturer don't want  you to program this MCUs and sell them with firmware in masked ROM as "ethernet controller". So this is some kind of stupidity - use 2 MCUs, one sold as "ethernet controller" and other one as general purpose, just to get value from sensor and send few bytes to server.

Well who cares that the Ethernet controller is a MCU when it's so cheap and you get complete Ethernet compatibility, and of course receive capability. But the real reason why wireless is popular is none of this fiddling with a wired port. It really is a mismatch, a small IoT device connected up to a cable taking much more volume. One justification for a wired connection is if you can use it for PoE.

It's all not about price and possibility. It's about KISS. If you found some small, simple 8-20 pins MCU with Ethernet peripheral, share it with class.

As for wireless ethernet... You can't get any advantages from wireless if your device have to run years constantly without all that battery charging and replacing crap. In most cases you have to power your wireless device with wires. It is just stupid, unless you invent some independent source/storage of energy with capacity/volume value many orders more than all that Lithium garbage we have. The only thing really sutable for powering WiFi comes to mind is RITEG, but I'm afraid that the authorities of your country will become very attended if you will try to buy or build one.

Well yes I agree with you about the source of power and all that. Perhaps PoE will fill your niche. But really I don't think going back to an old technology is a solution for IoT. The mass market will simply not put up with wires so there will be no driver for economy of scale.

This project is not about mass market at all. And it is not even near about that "IoT" hype ( I already had all that "IoT" things 20 years ago, believe me, wall switch is still the best way to turn on the light, and mobile app is definitely the worst :) ).

It's just about connecting nearly any sensor to the network in minutes in cheap and reliable way and without dealing with full TCP/IP stack. May be about playing with Ethernet on lowest possible level (you could change every bit of ethernet packet, from preamble to FCS).

Wireless is not a solution, unfortunately. There are already problems with RF bands used for wireless connectivity, and they will grow very fast. Wired ethernet will be alive for decades, I bet.

Of course wired Ethernet will exist but more and more it will be in data centres rather than the home. Of course it's silly to control lights with a phone. The real promise of IoT is when all the transducers are connected yet untethered. Currenlty they are still hostage to the power source. But I think solutions for power will appear but it won't involve wired data comms, that's just too inflexible. Meanwhile we can continue to hack on interim solutions. They all are interim anyway.

The comment about mass market is simply saying that items not produced on a large scale will become more expensive and scarce. Already we are seeing that an entire ESP32 module is cheaper than the cost of a port and cabling.

I think that @Stanson is right: most network gear can fall back to 10 Mb.

Also I like the idea!

Thanks for providing the source code!

I had a look at it - it's really very skinny. The downside of a receive-only implementation is that the destination MAC address must be "hard coded". This requires a special set-up process or a back channel that works without Ethernet.

If you have display of some sort and buttons, you can add destination MAC setting to firmware. Also, it is not too hard to set MAC on server's network card, or just put it into promiscuous mode.

I just read http://www.microchip.su/showthread.php?t=10418&page=4 - the on-line translation is actually comprehensible. Maybe you should add a link to your homepage (http://www.stanson.ch) to this project page.

I added links to project and profile, however it's in Russian, so most readers have to use translators.

Good - I just love getting a glimpse of famous Russian brand of irony through Google Translate ;-)