In practice, how much can the 100m ethernet limit be pushed?

Question

I am faced with putting an outdoor WiFi access point about 800 feet from the nearest building. There is no power at the far end. This will require a weatherproof AP that accepts PoE

I've looked at creating a wifi link, but 3/4 of the span is bush and trees.

There are several providers of ethernet extenders, presumably converting the signal to some form of DSL and back. Some provide the ability to power a PoE device at the far end.

It occurs to me that I don't really understand the limitation of 100m on ethernet, and under what conditions it can be pushed. Modern cables are better quality. The original 100m spec figured into the timing for collision detection. Full duplex means collision detection is a non-issue.

The cable will be aerial, suspended from a 12.5 ga high tensile steel wire. (Why not bury it? I don't really want to clear enough of a right of way to run a ditchwitch through 400 feet of poplar swamp.)

I don't need high speed. 10 Mbit/s would be fine.

Option 1: Try it. If it works, fine. If it's flakey put the appropriate additional hardware at the ends.

Option 2: I'm wasting my time. Buy the gadgets.

If option 1 isn't unreasonable, what are my best choices for cable.

I don't need gigabit speeds. Even 10 Mbit/s would be sufficient.

Answer 1

The "100 m Ethernet limit" you refer to exists for twisted-pair copper only. Fiber Ethernet over single-mode fiber (SMF) easily reaches over many kilometers.

800 ft/240 m is far beyond the reach of Ethernet over twisted pair. Not even the best quality, solid-core cable could provide that range. (It might work for half the distance.)

You should be aware that Ethernet speed negotiation does NOT test/train the cable[*1]. Both sides negotiate their best mutual speed/mode and then start linking with that. If the data signal is rather bad the linking fails and the process simply repeats (endlessly). If the signal is borderline the link may come up but the frame checksum (FCS) regularly fails and the effective throughput is low to non-existent.

So, the professional (and on-topic) way of solving your problem would be to deploy fiber and a separate power line for the fiber termination and the WAP. You can get reasonably priced, pre-terminated fiber cable that you can deploy yourself (always use a pipe for protection). If high speed isn't required, 1000BASE-SX with up to 550 m over multi-mode fiber (OM2+) should be the most economical variant. And of course, other powering methods are also possible (solar panel, wind generator, battery, ...).

Also for aerial deployment, the potential impact of lightning should be considered. While twisted-pair Ethernet does provide considerable isolation (per IEEE 802.3 Clause 33.4.1 at least 1500 V), induced voltages might easily exceeded that (depending on the scenario). Fiber has no such susceptibility.

[*1] ... with the notable exception of "smartrate" ports supporting a subset of 2.5/5/10 Gbit/s.

Answer 2

Unlike UL, there isn't any regulatory body that enforces, tests, and certifies ethernet hardware. So one is unfortunately on their own to find these sorts of issues.

"In practice" transmitter power for many commodity network cards is insufficient for a full 100m span. I have seen some (cheapo) switches that can't power a full span as well. You're asking for things to work at almost three times the specified limit. Unless you use unusually large (24awg or larger) cable, I doubt it will work. 802-standard POE absolutely will not work at those distances. Telco line powered gear uses very high voltage (>300vdc) at low current to address line loss due to distance. (good luck finding that sort of gear for a one-off install.)

800ft is nothing to DSL. (a pair of 3 decade on SDSL modems would work.) Your problem is entirely power. "Line power" -- but not ethernet POE -- is a remote possibility. That distance is pretty far to string 120vac residential power as well. (I have to use 8 or 10 awg extension cords @ 100ft. The cord for the RV is 6awg.)