I recently splurged on cable separators just to keep Ethernet away from my power cords. I also hid cables under furniture to avoid the tangled mess behind my TV. After a full day spent on cable management, I wanted to know whether I'd made a difference. So I pulled out a 75-foot Cat 5e cable and tested four common household cabling scenarios. I expected the results to tell me I'd wasted my time, but as it turned out, I couldn't have been further from the truth.
One Cable, Two Computers, No Internet Access
To set up for the test, I connected my Windows laptop (10.77.77.1) directly to a spare Zotac Mini PC (10.77.77.2) using the 75-foot cable. I then confirmed the two network adapters had negotiated a gigabit link. There wasn’t a switch between them, and my internet connection didn’t factor into the results. I used the network tool iperf3, which generates traffic between two computers and measures network throughput. To keep the comparisons consistent, I used the same cable and computers for every layout. Three 6-second tests in either direction, comparing the median speed reported by the computer on the receiving end. Network error counters are checked before and after each batch run. With iperf3 running in server mode on the mini PC, I ran a PowerShell test loop. The -R flag reverses the traffic direction, while -O 3 omits the first three transmissions to provide a warm-up period before measuring. That setup gave me repeatable measurements instead of eyeballing ICMP ping results.
Coiling and Crossing
I started with the network cable stretched across the floor, far away from any power cords. It delivered roughly 949 Mbps from my laptop to the mini PC, and 929 Mbps in reverse. That was the benchmark I would use to compare the next set of test results against. Next, I wound nearly the entire cable into a tight coil about 1 foot across, with enough cable at each end to reach each computer. The results of the PowerShell and iperf3 tests barely changed, so I crossed neat Ethernet coils off my list of layouts that could affect network speed. For the crossing test, I placed the Ethernet cable over the power cord of my room heater at a 90-degree angle. The heater itself was rated at 2,400 W and stayed turned on throughout the testing. As with the coiled cable, the test results showed no meaningful impact on throughput. Then I ran 15 feet of network cable alongside the extension cord supplying electricity to the heater. Along the distance, the two cables were parallel, with their jackets touching for most of the run. This time, I observed a measurable drop in network speed. These are the median speed results reported by the receiving computer, in Mbps:
Since one result isn’t necessarily conclusive, I reran the test after moving the Ethernet cable away, but leaving the heater running. This restored normal throughput, and moving the cable back to the power cord brought the slowdown straight back. Finally, switching the heater off restored the network performance even though the cables were still running in parallel. During all tests, neither computer moved, the laptop’s charger connection stayed unchanged, and I didn’t unplug any Ethernet cable during any of the comparisons. Only the power cord’s position and the heater’s operating state changed. The mini PC’s receive-error counter was already at one before the first parallel test runs and never increased during or after them. There just wasn’t anything else that could explain the slowdown. I wasn’t able to directly measure electrical interference, so I can’t claim this was the cause of the drop in network speed. But the repeatable drop and subsequent recovery gave me more than enough reason to keep these cables apart. My dodgy plug still connected at gigabit speed.