A switch will happily negotiate 1 Gbps over a cable that fails every category test. The link works until it carries real traffic, at which point errors appear that look like a switch problem.
Testing is how you find that cable before it becomes a ticket. Certification is how you prove it.
Three levels of testing
| Level | Question | Tool |
|---|---|---|
| Verification | Is it wired correctly? | Basic tester |
| Qualification | Will it carry this speed? | Qualifier |
| Certification | Does it meet the standard? | Certifier (Fluke, etc.) |
Only certification produces a report you can hand to a contractor or an auditor.
What copper tests measure
| Test | Detects |
|---|---|
| Wiremap | Miswires, opens, shorts, split pairs |
| Length | Runs over 90 m |
| Insertion loss | Signal attenuation, often heat or length |
| NEXT | Crosstalk at the near end |
| Return loss | Impedance mismatches |
| TDR | Where the fault is, in meters |
The last one is the useful one during troubleshooting: a TDR trace says "the fault is 32 m from this end," which turns a hunt into a repair.
Fiber tests
| Test | Measures |
|---|---|
| Loss budget | Total dB against the optic's allowance |
| OTDR | Loss per event, and where |
| Visual inspection | Contamination on the ferrule |
The visual inspection is not optional. A single speck of dust on a single-mode ferrule can exceed the entire loss budget.
Test the channel, not the permanent link
The channel includes both patch cords. Contractors often certify the permanent link and then install whatever cords are in the van. Test the channel, or test the cords separately and record them.
Reading a report
A certification report that matters contains:
- Cable ID and location
- Category and test standard used
- Length, loss, NEXT, return loss, and pass/fail per pair
- The tester's calibration date
- A date and technician
A report with only "PASS" and no per-pair data is a claim, not evidence.
Worked example: a new wing
- Contractor certifies 120 runs to Cat6A.
- Report shows 118 pass, 2 fail on return loss at 78 m.
- TDR localizes both to a bend behind a wall plate.
- Re-terminate and retest.
- Store the report with the as-built drawings, keyed by cable ID.
Six months later, a flaky AP is a five-minute lookup instead of a cable pull.
What breaks in production
- No report, so every fault starts from zero.
- Permanent link only, with unknown patch cords.
- Certifying to Cat5e because that was the tester's default.
- Dirty fiber ends, producing intermittent loss that moves when the cable is touched.
- Reports filed nowhere, so the data exists and cannot be found.
Checklist
- Every new run is certified to its intended category.
- The channel is tested, including patch cords.
- Reports are stored with the as-builts, keyed by cable ID.
- Fiber is inspected and cleaned before every mating.
- Testers are calibrated, and the date is in the report.
- Failures are re-tested after repair, not assumed fixed.
Next
Cabling has to carry power as well as data. Continue with Cabling for PoE.