I understand that an Ethernet bus design with hubs has largely been superseded by star topologies with switches and that is largely due to the issues of collisions. However, I am interested in networking a large number (say up to 1000 for now, possible 10 times more in the future) soil moisture sensors and latching solenoids that have very low bandwidth needs. These would be located along rows in an orchard.

Currently, people are generally using LoRaWan or other radio networks, and fewer sensor locations, but these have disadvantages such as of lack of handshaking to ensure message delivery, the need for lots of batteries, and regulatory limits on transmission volumes.

What are the practical considerations for using a PoE Ethernet bus design or are there reasons this would not be a good idea? Is the hardware available to link multiple controllers together this way?

  • \$\begingroup\$ I can see why a wireless technology is preferrable to a wired one, especially if you have to run PoE since you need to run one wire to each device and back to a switch. For 1000-10,000 that is a lot of power hungry network equipment. A solar powered battery device would probably be a lot better. You could have a wired hub device that reports back on its "local" sensors, banks of 100 or so. That way you don't have to worry about running thousands of Ethernet cables. Plus outdoor direct bury CAT6 is expensive, on the order of 5-10 USD per foot. \$\endgroup\$ – Ron Beyer Sep 23 '19 at 1:14
  • \$\begingroup\$ Agree that star topology isn't practical so was asking about a bus. Solar has issues under a tree canopy and you still have to run a lot of wire for the power and signal. I have only seen sensor hubs for maybe 10 sensors, but it may be possible to do more. \$\endgroup\$ – haresfur Sep 23 '19 at 1:25
  • \$\begingroup\$ When you say "bus" do you mean a ring or daisy-chain? Or do you mean something that is multi-drop more along the lines of CAN? Ring networks usually have limitations especially if you want single break recovery, plus the devices need to support ring protocols and not just a device-level switch. If you are hell-bent on Ethernet, you may find a combination of ring/star with logical network segments to be the best. I think though that this is more network engineering than electrical at that point. \$\endgroup\$ – Ron Beyer Sep 23 '19 at 1:42
  • 1
    \$\begingroup\$ @RonBeyer cat6 6 is not needed, and lat time I purches birect byry cable it was less than $1000 for 500m \$\endgroup\$ – Jasen Sep 23 '19 at 6:01
  • 3
    \$\begingroup\$ as You need to connect many nearbly nodes with a low signal rate, RS485 might be a better match, with high impedance transceivers a 4 wire bus (2 signal 2 power) can service 256 devices. \$\endgroup\$ – Jasen Sep 23 '19 at 6:11

These are some considerations that come to my mind:

1) Rats - they will chew down on any wires they will find tasty enough. Laying down metal shielded cables might be costly and difficult.

2) Range - Ethernet is limited to 100 meters (328 feet). That might be a challenge.

3) Cost - Ethernet might be more expensive than other technologies.

4) Failure - If you are using a common power supply for creating the POE, shorting of one sensor might bring down the entire network.

5) Modification to the network is complex - Once you have laid down the network, moving anything around is difficult due to the wires.

6) Damage due to machinery being used in the orchard - Some worker might accidentally damage the cabling while working in the field.

Just for the sake of comparison, RF based sensors might be able to last for few years. Lora claims 10 years or so. Maintenance might be easier for these and the casing can be made strong enough so that rats don't destroy these sensors. Even if the battery lasts for 1 year, changing the battery once an year might be less time consuming as compared to maintaining the Ethernet network.

| improve this answer | |
  • \$\begingroup\$ These are all valid concerns. My major concerns with batteries are cost, eWaste, and the labour to replace them. Not a big deal for 100, but when you start scaling up by a factor of 10 or 100... Damage may be an issue but many orchards have drip lines along the rows already and there is no traffic there. At current prices RF sensors and battery systems become cost-prohibitive, which is why I want to explore all options to make a step-change. Also thinking about wired 24-48V power with radio comms. \$\endgroup\$ – haresfur Sep 24 '19 at 0:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.