Cable testers, toners and loopback plugs

Why this matters

Lesson 17 introduced these tools as part of networking. This lesson is about using them as diagnostic instruments: what each one proves, what it leaves unproven, and how to choose the test that isolates a fault in one step instead of three.

The loopback plug is the addition here, and it is the tool most candidates have never used. It solves a specific and awkward problem: testing a port when there is nothing at the other end of the cable and no second machine to provide one.

Running through all of it is the theme this course keeps returning to. A tool answers one narrow question, and a pass answers only that question. A cable that passes a wire map can still be a split pair. A port that passes a loopback test can still be misconfigured. Knowing the boundary of each result is what separates a measurement from a reassurance.

The lesson

What a wire-map tester proves, and the faults it cannot see

A wire-map tester energises each conductor in turn and reports what arrives at the other end. It proves the physical arrangement of the conductors.

It finds:

• Opens — a conductor that does not reach the far end.
• Shorts — two conductors connected to each other.
• Miswires — a conductor arriving at the wrong pin.
• Reversals — a pair connected the wrong way round.

It cannot see:

• Split pairs, where the pins are right and the conductors were taken from the wrong pairs. This passes perfectly and performs badly, and it is the canonical example of a result that answers its own question and no other.
• Excessive untwist at the termination.
• Attenuation from an over-length run.
• Crosstalk and noise, including interference from nearby equipment.
• Anything about the cable's rated category.

So a passing wire map means "the conductors go where they should". It does not mean "this cable will carry a gigabit". When a link comes up at 100 Mbps on a cable that passed a wire map, the wire map was not wrong — it was answering a different question, and the fault is in the electrical properties it never measured.

The instrument that does measure those is a certifier, which tests against the category specification and produces figures. It is expensive and belongs to installers. A time-domain reflectometer measures distance to a fault, which is what you need when a cable is broken somewhere inside a wall.

Certification against verification, and why the cheap tester is usually enough

Two different activities with different purposes:

• Verification answers "is this cable correctly made and connected". A wire-map tester does it, in seconds, for very little money.
• Certification answers "does this installation meet the standard it was specified to". It measures attenuation, crosstalk, return loss and delay against the category, and produces a report that is a contractual deliverable.

A technician diagnosing one desk needs verification. An installer handing over a building needs certification. Buying the wrong one is buying either an instrument that cannot answer the question or one that costs fifty times what the question is worth.

The practical middle ground when a verified cable still performs badly: replace the patch leads with known-good ones, try a different switch port, and check the negotiated speed. If a known-good patch lead at both ends fixes it, the permanent run is fine. If it does not, the run is suspect and that is the point at which a certifier — or a contractor — is justified.

Tone generator and probe: tracing a cable through a wall and a full rack

The problem: a cable at a desk and a patch panel with forty-eight unlabelled ports.

The method:

1. Disconnect the cable from any equipment at both ends. Injecting tone into a live switch port risks damage.
2. Attach the tone generator at the known end.
3. Sweep the inductive probe along the bundle or across the panel ports.
4. The right cable is loudest, not the only one making a sound — tone bleeds onto neighbours, so work by comparison.
5. Confirm by removing the tone and checking the sound stops on that cable.
6. Label both ends before moving on.

Variations worth knowing: some toners place the signal on a specific pair, which helps in a dense bundle; some kits include a tester so the same visit verifies the cable as well as identifying it; and some generators can drive tone through a connected wall plate where the far end is inaccessible.

The step everybody skips is step 6, and the tool exists because somebody skipped it last time. Labelling is the preventive measure from lesson 34's step six, in the one place where its absence is guaranteed to cost somebody a future hour.

Loopback plugs: testing a port without a second machine at the other end

A loopback plug connects a port's transmit pins to its receive pins, so anything the port sends comes straight back. With appropriate diagnostic software, that tests the port itself.

Types in scope:

• Ethernet loopback, testing a network interface.
• Serial and parallel loopback, on older equipment and industrial gear.
• Fibre loopback, which is a short patch cable from transmit to receive on the same transceiver.

What it proves: the port can transmit and receive, and its electronics work. That is genuinely useful, because it separates a failed adapter from a failed cable, switch or configuration with one part and no second machine.

What it does not prove: that the port is configured correctly, that the driver is right, that the cable beyond it is good, or that anything at the far end exists. A loopback pass on a machine with no connectivity narrows the fault to everything outside the adapter, which is exactly the elimination you wanted.

The examinable distinction is between a loopback plug and the loopback address from lesson 12. Pinging 127.0.0.1 tests the protocol stack in software and does not touch the adapter at all. A loopback plug tests the physical port. A machine can pass one and fail the other, and knowing which you ran is the difference between "the stack is fine" and "the hardware is fine".

For fibre, a loopback also confirms the transceiver is transmitting light, which is otherwise hard to establish safely — and the safety rule from lesson 16 stands: never look into a fibre or a transceiver.

Choosing the test that isolates the fault in one step rather than three

The habit is to ask what the result of each candidate test would eliminate, and to pick the one that eliminates most for least effort.

Worked choices:

• A desk with no link light. Swap the patch lead first — thirty seconds, and it is the most likely single cause. Then move to a known-working port on the same switch, which eliminates the port. Only then tone and test the permanent run.
• A machine that cannot reach the network, link light on. This is not a cable problem; the link proves layer 1. Go to lesson 40's sequence.
• A machine that fails everywhere it is plugged in. The adapter or the machine. A loopback plug isolates the adapter in one step.
• One port in a room that never works while its neighbours do. The permanent run or the termination. Tone and wire-map it.
• A link that negotiates at 100 Mbps instead of 1 Gbps. A broken pair. Wire-map test, which will find it.
• A link that is up and performs badly with no errors on the patch leads. Electrical properties the wire map cannot see; try a known-good run, then escalate to certification.
• A cable that works until somebody walks past. Intermittent, at a termination. Flex the cable at each end during a sustained ping — lesson 36's forcing technique applied to a cable.

The unifying rule: substitute known-good, one thing at a time, and let the change that fixes it name the fault. Every tool in this lesson is a way of doing that when substitution alone is impractical — because the cable is inside a wall, or because there is nothing to substitute at the far end.