How I Ran Gigabit Ethernet Over Cat3 Cable and Sinned Against the Ethergods

*This is not how we could do it nowadays, but this is what I got…*

When we moved into our first townhouse in 2023, I attained something I’d never had before, even after starting to permanently work from home in 2020: a Home Office.

This had a lot of great perks: a full desk without bumping my chair into my bed when I backed up, ample storage, more privacy for calls. But one thing was lacking: a great internet connection.

One day after a lot of stuttering on a Teams call for work, I did a speed test and discovered that I was only getting 60-80 Mbps down, and a paltry 10 Mbps up. Initially, I was certain my ISP was to blame, and readied myself for a frustrating support call. That was until I took my laptop downstairs and re-did the speedtest next to the router. Sure enough, the WAN connection was fine. The problem was my Wi-Fi.

Chances are, like me, you have a WiFi router that works on the 802.11a/c/x standard that uses the 5GHz frequency band. These modern standards can (theoretically) deliver speeds in the ballpark of gigabits/second. Unfortunately – unlike the slower, 2.4GHz standards of yesteryear – their effective range leaves more to be desired. At 40ft, reliability and speed drop off rapidly – and that’s not accounting for factors such as interference, number of clients, and the amount of physical clutter between sight lines.

While my router is in the central part of my house (and thus great for my wife’s office setup, our living room, our bedroom, and a myriad of other devices) my office is a bit of an outlier. It’s about 40ft away from the router, with a lot of house structure in-between.

I’ve had bad luck in the past with wireless repeaters, so my thoughts immediately turned to how I might bridge the distance with an ethernet cable.

Our house was built in the 1990s, which means it was mainly wired for telephones. Future-looking support for a networked household was present, albeit a bit nearsighted. Checking inside the phone jack boxes I found Category 3 cable (“Cat3”). It was 4-pair, so the right number of wires, but Cat3 is an old standard originally specced for 10Base-T (10 Mbps ethernet).

Thinking back to my days in downtown offices, I knew I’d seen situations where gigabit ethernet ran successfully and reliably over Cat5 cables (it’s supposed to require Cat6). So I decided to do some Googling.

Officially, if you want your ethernet to work reliably at the speeds your hardware supports, you need to use the cable standard that the speed was designed for. The increasing standards (Cat3, Cat5, Cat5e, Cat6, etc) have increasingly rigorous requirements, such as more twists in the wire pairs. These standards allow for decreased EM interference generated by said cable’s own traffic, and that of its neighbours when it’s in bundles.

Unofficially? A lot of the time you can plug in any old cable and it will just work. I found tons of references online of Cat3 working reliably as a medium for gigabit ethernet. For short runs of cable, with minimal neighbouring cables and traffic saturation, it should work just fine.

This assertion seems to make telco cabling guys absolutely livid. “Standards matter” I was told when I asked for opinions on Mastodon. I can certainly sympathize: when you cut corners in IT, what works reliably in situation A may not be viable in situation B. We pay extra for that reliability.

However, I was just looking to use one cable: the one going to my office. If it worked, for a minimal investment I would have a solution. The alternatives were to live with crappy Wi-Fi, buy a Wi-Fi repeater that might not work, or lastly, pay probably thousands of dollars to have a cabling guy cut open my walls to remove Cat3 (which would be attached to studs at random intervals inside the wall with metal staples) and replace it with Cat6 in conduits.

Compared with those alternatives, this was a low-risk experiment.

That said, I was aware that I was testing the EtherGods with my hubris, and if I wanted to succeed, I should give my experiment the best chance for success. I already had some Cat6 cable in my ethernet tool kit (a relic from my days as a jack-of-all-trades IT guy). My employer’s recently closed-offices meant that a lot of networking surplus had been free for the taking. So tools and equipment were already on-hand to get the job done.

One thing I did purchase was an 8-port patch panel. I decided I wanted this to look relatively clean when finished. I also bought some additional Cat6 patch cables as needed (something I would have needed to buy anyway even if I had gotten new wiring installed).

I noticed something fortunate: the jack a few feet away from my modem/router was directly above the basement’s utility room where the wires all met (and where the patch panel would go). Furthermore, the jack faceplate opened to a mostly-empty, columnar space that housed a furnace exhaust duct and not much else. This allowed me to easily run a new Cat6 cable from the existing jack down to the utility room. This way, the “trunk line” from my modem/router to any switch I installed in the basement would be a high-quality wire.

*The old Cat3 wire and the new Cat6 wire with keystone jack*

Each step of the way, I would test runs with my cable tester to make sure that I’d crimped tightly, punched down sufficiently, and not mixed between TIA 568 A and B. When the first outlet was patched back to the router, I tested the line from the utility room – as expected, just shy of 1Gbps down, 100Mbps up.

*A $30 Trendnet 8-port patch panel I got off amazon joins Cat6 (blue wire joining port 2) to Cat3 (grey wire joining port 8)*

Next, I identified the wire going to my office jack using my cable tester. I punched a Cat6 keystone jack on one end of the Cat3 cable, and my patch panel on the other. Last, I used a short patch cable to connect the two ports on the panel. It was time for the first real test.

From the port in my office, the results were exactly the same as before: just under 1Gbps down, 100Mbps up.

Here I decided to use a D-Link DGS-105 gigabit switch before sending another Cat6 patch cable to the other side of the room where my desk was. The switch would have two functions: First, if I wanted to plug something else in at some point, I’d have that ability. Second, my hope was that any difficulties were had with the cable, the switch would have sufficient error correction to accommodate them.

My first work day using the cable was a busy one. It started with a lengthy call, involved lots of work over RDP, and even a few YouTube videos (during breaks, of course). It worked flawlessly. In fact, it worked better and more reliably than the WiFi connection had.

So where are the dropped packets? The endless renegotiations back and forth between 100Base-T and Gigabit? Strange recurrent errors that left me screaming madly for the sweet release of death? All the horrible things the telco cabling guys would have you believe?

This seems to confirm the sysadmins side of the story: for light work, cabling not meeting the minimum spec of the protocol will, in fact, work just fine. While I don’t have firsthand knowledge of low-level hardware, there seems to be a consensus that modern networking controllers contain lots of error correction features to make these interfaces very robust in the face of interference or some substandard wiring.

Numerous times, folks online have relayed the story of chipset salesmen running ethernet over barbed wire as a demonstration of how robust their products were, knowing full well that their competitors’ chips were capable of the same efficient performance – thanks to those error correction features.

And I get it: the telco guys have probably shown up to buildings where dozens or even hundreds of long, traffic-saturated bundles of Cat3 were running side-by-side, interfering and causing crosstalk – where a mid-level manager tells them they want to make the network run faster without replacing the cabling. But my home is not that scenario.

I have a friend who works in enterprise IT and has participated in a couple of big cable builds. When I explained to him how former telco types had framed the perils of running gigabit over Cat3 and asked for his take, he responded thusly:

“It’s a wire.
the internet comes out of it.
you don’t work at or live in CERN so you’re fine.
make sure you tell the cranky purists to find a decent dumb switch…”

So that’s how I made gigabit ethernet work over the Cat3 installed in my home. As long as your needs are modest, and you already own some ethernet cable tools, you should try it too! Your results may vary, but if it works, it can save you a costly cable install.