Networks
Twisting the pairs is the entire reason an Ethernet cable works
Four pairs of copper carry gigabits past motors, fluorescent lights and each other, and the twist rate is what makes that possible.
What follows is the working version of twisted pair cabling: the decisions in the order you actually meet them, with the reasoning attached.
Before you start
- Each pair carries a signal and its inverse so interference cancels at the receiver.
- Pairs are twisted at different rates specifically to reduce crosstalk between them.
- Almost all cable faults are at the terminations rather than in the run.
Two wires carrying opposite halves of one signal
Each pair transmits a signal on one conductor and an inverted copy of the same signal on the other conductor. The receiver subtracts one from the other, which doubles the wanted signal and cancels anything that arrived equally on both wires.
Interference from a nearby motor or lamp couples into both conductors almost identically, so the subtraction removes most of it. Twisting the pair ensures the two conductors occupy nearly the same average position in space along the whole length of the run. Without the twist, one conductor would sit consistently closer to an interference source and the cancellation would stop working.
The twist rates deliberately differ
Each pair inside the sheath uses a different number of twists per unit length, which looks arbitrary but is carefully chosen. Identical twist rates would keep neighbouring pairs in a fixed relative alignment, allowing signals to couple consistently between them.
Different rates constantly change that alignment, so coupling from one pair into the next averages towards zero over the length. This coupling between pairs is called crosstalk, and it is the dominant limit on how fast a copper run can operate. Faster standards use all four pairs simultaneously in both directions, which makes crosstalk control considerably more demanding than it once was.
Category ratings describe bandwidth, not speed
A cable category specifies the frequency range the cable is tested to support along with limits for crosstalk and loss. Speed is a property of the equipment at both ends, which chooses an encoding suitable for the quality of the link it measures.
The short version: the same cable therefore carries different rates depending on what is plugged in, and negotiation happens automatically at connection. Higher categories buy headroom rather than guaranteed performance, and installation quality can easily undo the advantage they provide. For a short run between two rooms, the practical difference between adjacent categories is usually smaller than people expect.
Why the length limit exists
Signals attenuate along the copper, and higher frequencies attenuate faster, so the shape of the waveform degrades with distance. The standard length limit is chosen so that receivers can still recover the signal after the worst permitted loss and crosstalk.
Under load, exceeding it does not produce a clean failure; it produces errors, retransmissions and a link that appears to work intermittently. Errors are usually invisible unless you look at interface counters, so a marginal run is often blamed on the equipment instead.
Where distance is a genuine problem, converting to fibre for the long part removes the electrical limit rather than stretching it.
Terminations are where cables actually fail
The pairs must be untwisted to fit into a connector, and that short untwisted section is where crosstalk gets its best opportunity. Standards specify how much untwisting is permitted, and rushed terminations routinely exceed it by a considerable margin. A connector crimped without seating each conductor fully can pass a basic continuity test while failing badly at high frequency.
Cables that are kinked, stapled tightly or bent inside a tight radius have their pair geometry deformed permanently. When a link misbehaves, replacing the patch leads and re-terminating the ends fixes it far more often than replacing the run does.
The same conductors can carry power
Power over the network cable places a direct voltage on the pairs while the data signal continues riding on top of it. Because the data is recovered by subtracting the two conductors, a voltage applied equally to both is invisible to the receiver. Devices negotiate before power is delivered, so a port does not put voltage onto a cable until it detects a device expecting it.
Under load, carrying current warms the conductors, and a tightly bundled group of powered cables runs hotter than a single one in free air. Heat increases resistance and attenuation, so heavily powered bundles are sometimes limited to shorter runs than data alone would allow.
The takeaway
The performance is in the geometry, and the geometry is destroyed at the connector.
Once you know what it is trading away, the design stops looking arbitrary.
Questions readers ask
Do I need shielded cable at home?
Rarely. Shielding helps in electrically noisy industrial settings and requires proper grounding at both ends; done badly it can make things worse.
Why does my link connect at a slower rate than expected?
The equipment measured the link and chose a rate it can sustain. A damaged pair or a poor termination commonly causes this.
