Why long cables misbehave
A cable that works perfectly at two metres can fail mysteriously at twenty, and the reason is not a fault in the cable but physics accumulating with length. Every metre of copper resists the signal a little, absorbs a little energy, and admits a little interference, and those small losses add up linearly until the signal arriving at the far end is too weak or too distorted for the receiver to decode reliably. The result is not usually a clean failure: a long HDMI cable shows sparkles, dropouts, or a blank screen that comes and goes, while a long Ethernet run negotiates at a slower speed or drops packets intermittently. The calculator on this page estimates that accumulated loss in decibels from cable length and type, so a run can be judged before it is buried in a wall or ceiling where fixing it is expensive.
Decibels are the right unit for this because signal loss is multiplicative in nature but additive in decibels, which makes long runs easy to reason about. A loss of three decibels means roughly half the signal power is gone, six decibels means three quarters, and ten decibels means ninety percent. Different cable types lose signal at different rates, because conductor thickness, shielding, and construction differ: the model here uses about 0.18 decibels per metre for Cat6 Ethernet, 0.4 for copper HDMI, 0.12 for coaxial cable, and 0.3 for copper USB. These are planning figures, not laboratory measurements for a specific cable, and real loss varies with frequency, cable quality, and connectors, but they are accurate enough to separate safe runs from marginal ones, which is the decision that matters.
Ethernet: the hundred-metre rule and its fine print
Ethernet's hundred-metre limit is the best-known cable rule in home networking, and it is often quoted without its fine print. The limit covers the whole channel: the fixed cable in the wall, plus patch leads at both ends, plus connectors. A ninety-metre fixed run with five-metre patch leads at each end is already at the limit, and a run that measures ninety-five metres of fixed cable alone is over it once patching is counted. Loss also rises with frequency, so a cable that passes at one gigabit may fail at ten gigabits over the same length, because higher frequencies lose more per metre. When planning a run near the limit, measure the actual path including rises, bends, and service loops, not the straight-line distance on a floor plan, and keep a margin of several metres rather than aiming exactly at one hundred.
Cable quality changes the margin considerably. Solid-core cable is the right choice for fixed runs, because its thicker conductors lose less signal and terminate more reliably in wall sockets, while stranded cable suits patch leads that flex. Shielding helps in electrically noisy environments, such as runs parallel to mains cables or near motors, but shielded cable only helps if it is terminated and grounded correctly; a badly installed shielded run can perform worse than a good unshielded one. For runs that must exceed the limit, the honest solutions are a network switch as a repeater at the midpoint, a fibre run with media converters, or moving the equipment closer together. Passive wishful thinking is not a solution, and neither is a cheap extender that adds latency and unreliability.
HDMI, coax, and USB: different cables, different breaking points
HDMI over copper is the most fragile of the common home cables, because its high-frequency video signals lose energy fast: at the model's 0.4 decibels per metre, a fifteen-metre run has already lost six decibels, and many passive cables struggle beyond ten to fifteen metres at full 4K resolution. Lower resolutions survive longer runs, which is why a cable that works fine at 1080p can fail when the same system is upgraded to 4K: the signal got faster, not the cable longer. For runs beyond fifteen metres, active HDMI cables, which amplify the signal along the run, or fibre HDMI cables, which convert to light, are the reliable choices. They cost more, but they cost less than opening a finished wall to replace a passive cable that never worked.
Coaxial cable is more forgiving per metre, at about 0.12 decibels per metre in this model, but antenna and satellite runs are often very long, so the total still matters: a hundred-metre coax run loses about twelve decibels, which is enough to push a weak broadcast signal below the receiver's threshold. Amplifiers help, but they amplify noise as well as signal, so an amplifier placed at the far end of a long, lossy run amplifies a degraded signal, while an amplifier near the antenna boosts a clean signal before loss occurs. USB over copper is the shortest-range of all in practice: the specification limits passive USB runs to about five metres, and longer runs need active extension cables or conversion to another medium. The calculator's loss figure for USB is a planning guide, not a substitute for the specification's length limits.
| Cable type | Loss per metre (model) | Typical safe run |
|---|---|---|
| Cat6 Ethernet | 0.18 dB | Up to 100 m channel |
| HDMI copper | 0.4 dB | Up to 10 to 15 m passive |
| Coaxial | 0.12 dB | Depends on signal strength |
| USB copper | 0.3 dB | About 5 m passive |
Planning a run that works first time
A reliable cable run is planned on paper before a single hole is drilled. Measure the route, including vertical rises and bends, add ten percent for service loops and routing around obstacles, and run the calculator with that total. If the estimate exceeds the ten-decibel budget used here, or approaches a specification limit, change the plan now: choose a better cable, add an active device, split the run, or move the equipment. Running the cable first and testing afterwards reverses the cheap and expensive steps, because testing is cheap and re-running a finished installation is not. Test every run before walls are closed, with the actual equipment that will use it, at the actual resolution and speed it must carry, because a test with easier equipment proves nothing.
Installation details protect the margin the calculator predicts. Keep data cables away from mains cables where possible, crossing at right angles rather than running parallel, because parallel runs invite interference that adds to loss in practice. Do not kink cables or bend them tighter than the manufacturer's minimum radius, because a kink changes the cable's electrical properties at that point and creates a reflection that degrades the signal. Leave service loops at both ends, a little extra cable coiled neatly, so a damaged connector can be re-terminated without re-running the whole cable. These details cost nothing during installation and save a great deal afterwards, which is why professionals insist on them even when the customer cannot see them.
- Measure the real route, including rises and bends, not the floor-plan distance.
- Add ten percent for service loops and routing.
- Test with actual equipment at actual resolution before closing walls.
- Keep data cables away from mains runs where possible.
- Never kink a cable or bend it tighter than its rated radius.
Use the calculator above on every long run before you commit to it. A minute of arithmetic now is worth far more than an afternoon of fault-finding later, when the cable is already hidden and the screen is already blank.