Need Advice - 140m Ethernet Run Between Buildings

Question

I'm facing a technical challenge and could really use your expertise. I need to establish a wired network connection between two points 140 meters apart - well beyond the standard 100m limit for Ethernet.

Key Details:

Outdoor installation (connecting two separate buildings) Requires stable, reliable bandwidth Must be wired (WiFi isn't suitable for this application) Looking for real-world experience:

Has anyone successfully implemented a similar long-distance run? What solutions actually worked in practice? Any pitfalls or lessons learned I should know about? I'm especially interested in hearing about what's worked (or failed) for others in field deployments.

Thanks in advance for sharing your knowledge!

PS: I'm aware fiber is an option, but curious about other approaches that might work with just rj45 cables.

Answer 1

Ethernet beyond 100 m reach is only possible with fiber variants.

Depending on the required link speed, 1000BASE-SX or 10GBASE-SR over multi-mode fiber are probably the most common options.

If you require more bandwidth now or in the future I'd seriously recommend starting with single-mode fiber, which doesn't add much cost, if at all (using 1000BASE-LX, 10GBASE-LR, 40GBASE-LR4, ...).

Of course, there are several non-IEEE options for transmitting Ethernet over copper over more than 100 m, but many are proprietary or hard to get - keep in mind that you might need to replace components after a couple of years and exotic hardware might be hard to get then.

Answer 2

While you might find a combination of gear that will work beyond spec, you're just asking for trouble in the future. In fact, there's a surprising amount of gear that won't drive 100m - takes a lot of power, and more sensitive receiver. There are DSL (VDSL these days) "ethernet bridges" that can easily span that distance, but they're going to be slow (~100Mbps), and increasingly hard to source.

As Zac has already said, fiber is always the answer. Future proof your installation with single-mode, and more than one pair. I highly recommend installing conduit between the buildings; it gives more options for the cable, and options into the future. (be mindful of elevation... we created a water spout in one of our offices.)

Fiber is immune to lightning (and to a degree, water), and will not create electrical/ground loops. However, direct burial stuff often contains a steal tracing wire.

(I've used SDSL devices in a pinch. But that was decades ago, and it tops out at 2Mbps.)

Answer 3

Ethernet has 100 m reach per segment. They make PoE-powered repeaters or extenders which are essentially two-port switches. Of course, you can use a normal switch in the middle.

Answer 4

My personal experience is very similar to yours. I had a building about 150m from the base building to which I ran Cat 6e and fibre underground, with the expectation that I would need the fibre connection to get any decent throughput. However thanks to delays in getting the fibre terminated, I started with just the Cat 6e. 18 months later I'm still using just the Cat 6e (as it happens because I'm in a rural area it's been hard to get someone to come out and terminate the fibre for me) and I'm having no problems transferring files to and from computers at speeds close to 1000BASE-T (typically around 800Mb/s). The cable runs between two switches (one Dell and one Ubiquiti).

Having said that I have been pleasantly surprised that it works, and wouldn't recommend using copper cable over that distance, especially for mission critical applications. If you're running cables from scratch, run fibre.

Answer 5

Long ago, when 10BASE-T and 10BASE-2 were a thing, every packet was broadcast over the whole network. Hubs (in 10BASE-T) were only repeaters and in 10BASE-2 the cable was the hub itself. Switching hubs (now only called "switches") happened later.

In order for a broadcast-only network to work properly, one needs an upper limit of the propagation delay in order to deal with the packet collisions. This was arbitrary decided at 100m and the physical layer was engineered around this.

Well, it is not that everything worked at 100m - low quality cables and gear usually constrained the networks at lower size.

When the switching hubs started to pop up here and there, the propagation time was no longer an issue. The limiting factors became the signal attenuation and the noise pickup over the length of the cable. With high quality hardware and some luck one could run almost 200m segments. I have some experience in neighborhood-sized LANs where segments of up to 160m running reliably at 100MBit/s were a common practice.

While I never did myself this at 1000MBit/s, there is no reason why this should not be at least tried.

---

Well, the propagation delay assumption is not completely abandoned these days. You may have to play with the flow control settings.