I heard from a buddy that he heard from an ISP tech that he heard from some source that cable splitters can degrade one’s cable signal, which I assume means decreasing Internet speed, bogging down uploads, increasing latency, things like that. Perhaps a splitter makes it more difficult to slaughter noobs in Warzone with your rootkit aimbot. Since I can’t (or at least, haven’t yet) stick my modem on the outside of the house to get the upstream cable straight into my equipment and eliminate splitters or couplers altogether, I decided to replace the splitter with a coupler. Maybe the dead end of the splitter causes losses on the line?
Here’s where I started — I took this photo before I move cables so that I could undo any unwitting mistakes.
The cable coming into the splitter from the bottom is the upstream ISP cable (I didn’t color-code that one). The left side of the splitter (blue) goes to the bedroom coax jack, which was live when we moved in. The red cable goes to the basement, where a cable dangles from a heating vent like a 75mbps iguana in a Florida double-wide. Without a coax tester handy, I guessed (correctly, on my first try) that this was the basement cable; and I was thankful, since I had decided to play with coax cables when it was 12 degrees (F) outside… The cable you see on the right side of the splitter is irrelevant; I just used that cable to test the upstream cable and splitters themselves with ISP tech support (again, no coax tester handy).
I decided to replace the splitter. I figured that any substantial, meaningful signal improvements would show up on the wire. And since we all live on the wire, it wouldn’t mean anything to point to a tester and say, “Look, it’s better,” when someone inside the house, sitting on the couch watching Netflix, never noticed a difference. I decided simply to run speed test after speed test using different tools. This isn’t the most scientific or exact method.
I ran fast.com (Netflix’s speed test), speedtest.net (Ookla, the one everyone uses), the M-Lab test (Google’s default “you searched for ‘speed test'” result), and the Windows 10 network tester (free from the Microsoft Store), all on 5ghz Wifi from my 7 year-old laptop. I ran each test three times and averaged their results. For each test, I noticed a few differences in methodology. Ookla had all the metrics. The Microsoft test really flattered my download speeds, and I think it somehow averaged idle, upload and download latency metrics, because its latency numbers seemed very different than the other tests. For my first three runs on the M-Lab test, I hadn’t noticed that they did provide latency metrics, so when I noted the latencies for my second round of tests, I deduced that they only provide the unloaded latency, which is by far my lowest ping across all tests, and I discarded that second round of latency from my averages — the averages you see here should reflect loaded latency. I learned something new as a result of playing with these numbers: “Unloaded latency measures the round-trip time of a request when there is no other traffic present on a user’s network, while loaded latency measures the round-trip time when data-heavy applications are being used on the network” (about.netflix.com).
Here are the numbers with the splitter still in place:
Ookla 1 2 3 avg down 87.84 89.81 89.84 89.1633333333333 up 4.51 4.75 4.72 4.66 latency 92 96 86 91.3333333333333 MS 1 2 3 avg down 93.54 96.65 104.46 98.2166666666667 up 2.03 3.69 2.24 2.65333333333333 latency 39 38 28 35 Google 1 2 3 avg down 86.8 85.4 84 85.4 up 4.29 4.3 4.22 4.27 latency Fast 1 2 3 avg down 86 95 87 89.3333333333333 up 4.3 4.2 3.7 4.06666666666667 latency 82 113 88 94.3333333333333 Up Down Latency Ookla 4.66 89.163 91.333 MS avg 2.653 98.21 35 Fast 4.066 89.33 94.333 Google 4.27 85.4 avg 3.91225 90.526 73.5553
Here are the numbers with the new coupler in place:
Ookla 1 2 3 avg down 88.86 88.77 87.1 88.2433333333333 up 4.47 4.67 4.73 4.62333333333333 latency 98 89 88 91.6666666666667 MS 1 2 3 avg down 105.1 92.45 97.38 98.31 up 1.91 2.45 2.37 2.24333333333333 latency 38 39 45 40.6666666666667 Google 1 2 3 avg down 87.8 87.7 87.5 87.6666666666667 up 4 4.36 4.19 4.18333333333333 latency 10 9 10 9.66666666666667 Fast 1 2 3 avg down 91 87 90 89.3333333333333 up 4.1 4.3 4.1 4.16666666666667 latency 92 97 103 97.3333333333333 Up Down Latency Ookla 4.623 88.243 91.667 MS 2.243 98.31 40.667 Fast 4.167 89.333 97.333 Google 4.183 87.667 avg 3.804 90.8883 76.5567
So, at least in my case, with the tools I’m using… no appreciable, noticeable difference in performance. If anything, it’s slightly, unnoticeably worse. Quite likely, the splitter wasn’t causing a problem at all, or maybe the quality of materials differs between the coupler and the splitter — maybe a poor-quality coupler is the same (or worse) than a high quality splitter — but I don’t know enough about coax standards and materials to say one way or the other. The coupler was $3.64 for a two-pack at the local Walmart. The splitter is from Commscope, probably installed by an ISP tech. I see a date of 10/6/21, which I figure is a manufacturing date.
Anyway, this was way more fun for me to write than it was for you to read, so, uh, thank you! Hope you learned something, and feel free to comment with some knowledge or some ignorance below! I welcome both.