Ethernet Cable Categories and Terminations

Why this matters

More trades now pull network cable than ever did: a camera on the garage, an access point in the attic, a wired connection to a pool controller or a solar gateway because the Wi-Fi out there is hopeless. A run that tests "good" on a cheap tester and then drops the camera every afternoon is a callback you will struggle to diagnose, because the cable is inside a wall and looks fine. Nearly every one of those failures traces back to a choice made before the cable was pulled: the wrong category, the wrong construction, the wrong jacket, or a termination that passed a continuity check and failed at speed. This card is the set of choices. The hands-on crimp and test procedure is its sibling. See related: Terminating and Testing an Ethernet Cable.

The category tells you the bandwidth, not the speed you will get

A cable category is a performance grade set by the TIA-568 standard (the North American structured cabling standard; ISO/IEC 11801 is the international equivalent). The grade is defined by the frequency the cable is tested to, in megahertz, which in turn limits the data rate it can carry over distance.

The 100 m figure is the channel: up to 90 m of permanent cable in the wall plus up to 10 m of patch cords at both ends combined. A run longer than that is outside the standard at every category, and the usual symptom is a link that comes up at a slower speed or drops intermittently.

Read the table correctly. A higher category does not make a camera faster; a camera needs a small fraction of what Cat5e carries. What Cat6 buys on a residential run is margin: more tolerance for a slightly long run, a tight bend or a mediocre termination. That margin is why it is the sensible default for any new run. Cat6A is worth its stiffness and bulk only where a long run or a 10 Gb link is actually planned.

Solid core for the wall, stranded for the cord

Each of the eight conductors in a cable is either a single solid copper wire or a bundle of fine stranded wires.

• Solid core is for permanent runs: in the wall, through the attic, to a jack or a patch panel. It carries signal further with less loss and it is what punch-down jacks are designed to grip. It cracks if flexed repeatedly, so it does not belong in a cord that gets moved.
• Stranded is for patch cords, the short cables between a jack and a device. It survives flexing, but it loses more signal per meter, which is why the standard caps total patch length at 10 m within the channel.

The connector must match the conductor. A plug made for stranded wire uses contact blades that pierce between the strands; on solid wire the same blades can skate along the surface and make a contact that tests fine and later goes intermittent. Plugs labeled for solid wire, or for both, exist for exactly this case. Better still, terminate the solid run on a jack and use a factory patch cord to the device.

Watch for copper-clad aluminum. Some low-priced bulk cable is aluminum wire with a thin copper coat, sold as "Cat6" without meeting the standard, which requires solid copper conductors. It has higher resistance, breaks more easily at a termination, and runs hotter under power over Ethernet. If a spool is unusually light or the conductor shows silver at a fresh cut, do not install it.

The jacket rating is a code question, not a performance one

The jacket on the outside determines where the cable may legally be installed, under the communications articles of the National Electrical Code in whatever edition your jurisdiction has adopted. The common markings, from most to least restrictive in where they may go:

• CMP (plenum): permitted in air-handling spaces. Low smoke when burning.
• CMR (riser): permitted in vertical runs between floors.
• CM (general): general-purpose runs in walls and ceilings that are not air-handling spaces.
• CMX (limited): limited use, typically in one- and two-family dwellings and in short exposed lengths.

A higher rating can always substitute for a lower one. Most residential runs inside walls are fine with CM or CMR, but a run through a return-air space used as a duct, or any commercial plenum ceiling, is a different matter. Check with the inspector or the local code if a run passes through a space that moves air. Outdoor runs need cable rated for sunlight and moisture; buried runs need direct-burial cable, which carries a gel or a water-blocking tape. Indoor cable outside degrades within a few seasons.

T568A versus T568B: pick one and never mix

The eight conductors are four twisted pairs: orange, green, blue and brown, each a solid color twisted with a white-striped partner. The standard defines two wiring patterns, T568A and T568B, which differ only in swapping the orange and green pairs. Either works. What does not work is one pattern at one end and the other at the other end, by accident.

• Same pattern both ends produces a straight-through cable, which is what every run you install should be.
• A at one end, B at the other produces a crossover cable. Modern equipment detects and corrects for it automatically, so the link usually still comes up, which is exactly why a mixed run slips through and confuses the next person who reads the jacks.

Pick one for your shop and write it down. In North American residential and light commercial work T568B is the more common choice; T568A is the one some government and older specifications call for. Consistency matters more than which.

Installation practice that keeps the category you paid for

A cable keeps its rating only if it is installed to the standard's handling limits. Under TIA-568 for four-pair cable:

• Untwist no more than 13 mm (half an inch) at a termination. The twist is what rejects interference; untwisting more is the single most common way a Cat6 run tests as something less.
• Bend radius at least 4 times the cable's outside diameter. A cable kinked around a joist corner has a deformed pair geometry at that point for the life of the run.
• Pull tension no more than 110 N (about 25 lbf). Pulling harder stretches the conductors and changes their spacing.
• No tight cable ties. A tie cinched until it dents the jacket crushes the pairs; use hook-and-loop straps or leave ties loose enough to slide.
• Separation from power wiring. Do not run alongside line-voltage cable for long distances. Where a low-voltage cable must cross a power cable, cross at a right angle. Local code may require a specific separation or a barrier; follow it.

What a test actually proves

There are three levels of tester, and each proves something different:

The faults a wiremap test reports: an open (a conductor not connected), a short (two conductors touching), a reversed pair (the two wires of one pair swapped at one end), and crossed pairs (two pairs swapped, often the A/B mix-up above). The one it often does not report is the most deceptive.

The split pair, the fault that passes the cheap test

A split pair happens when the conductors land on the right pin numbers at both ends, but the two wires on pins 3 and 6 (or 1 and 2) come from different twisted pairs. Pin for pin, continuity is perfect. Electrically, the signal on those pins is no longer traveling on a twisted pair, so it picks up crosstalk from its neighbors. It usually happens when someone wires both ends in the same wrong color sequence, for example in simple color order rather than to T568A or T568B.

A worked case. A tech pulls a 48 m run of Cat6 from the basement to a garage camera, terminates both ends in the same sequence, and the wiremap tester shows all eight pins green. The camera powers up. Over the following week it drops offline most afternoons and the recording shows gaps.

Reading it: 48 m is well inside the 100 m channel and Cat6 has plenty of headroom, so neither the length nor the category is the suspect. The pattern (works, then drops) points at a link that is marginal rather than broken. Checking the switch shows the camera's port negotiating at 100 Mb/s instead of 1 Gb/s and logging receive errors, which is a link fighting interference. Looking at the plug through its clear body shows the color order: white-orange, orange, white-green, green, white-blue, blue, white-brown, brown. That is color order, not T568B. Pins 3 and 6 carry white-green and blue, from two different pairs. Both ends match each other, so continuity passed.

Re-terminating both ends to T568B fixes it, and a tester with split-pair detection then shows a clean result. Because the camera does not need much bandwidth, the run might well have survived at 100 Mb/s on a quiet day. The afternoon drops line up with the garage door opener and a shop compressor on the same wall running, both of which add the interference the split pair could not reject.

What would have changed the answer. If the camera had dropped at random times with no link speed change and no errors, the cable is probably fine and the fault is upstream: power, the switch, or the address. See related: IP Addresses and Why a Device Disappears. If the run had been 95 m to the camera plus long patch cords, the channel limit becomes the first suspect.

References

• TIA-568 structured cabling standard (category performance, channel length, installation limits)
• National Electrical Code, communications articles, in the edition your jurisdiction has adopted (cable jacket ratings and separation)
• See related: Terminating and Testing an Ethernet Cable
• See related: Power over Ethernet for Cameras and Access Points
• See related: Low-Voltage Wiring Basics for Non-Electricians