The cable has been pulled, the computer is connected to the network, the light is on. That doesn’t show the line is within standard limits. A link that carries 1 Gb/s without problems today starts producing CRC errors, dropped connections and unexplained slowness when you move to 2.5G/5G/10G or a high-power PoE load is added. The information that solves the fault is often in the certification report received at handover, provided it’s read correctly. Below we explain the difference between verification, qualification and certification, the permanent link and channel models, and how wiremap, insertion loss, NEXT, PS NEXT and return loss values are interpreted in the field.
The difference between verification, qualification and certification testing
The phrase “the cable was tested” is used to describe three separate jobs. Signing off a report without asking which was done at handover means ending up with a piece of paper that proves nothing later.
| Test type | What does it answer? | Typical instrument | When is it enough? |
|---|---|---|---|
| Verification | Do the conductors go to the right pins, and what is the length? | Wiremap tester, tone and probe | Fault finding, patch cord checks |
| Qualification | Will this link carry a specific application (e.g. 1000BASE-T, 10GBASE-T)? | Application-based tester | Speed upgrade decision on existing infrastructure |
| Certification | Does the link meet the selected category/class limits across the full frequency range? | Certification instrument (e.g. Fluke Networks DSX series) | New installation acceptance, manufacturer system warranty |
Certification measures in the frequency domain against the category/class limits of ANSI/TIA-568.2-D or ISO/IEC 11801-1: up to 100 MHz for Cat5e, 250 MHz for Cat6 and 500 MHz for Cat6A. The instrument’s own measurement accuracy is also defined; according to TIA-1152-A, for Cat6A measurement the instrument must be at least at the Level IIIe accuracy level. Cabling manufacturers’ long-term system warranties usually depend on the report being delivered in the instrument’s native file format. A PDF alone isn’t enough.
Permanent link and channel test models
The test limit changes depending on what is being measured. A link tested with the wrong model either appears to fail unfairly or hides the problematic component.
- Permanent link (PL): The fixed part from the patch panel port in the floor distribution room to the user outlet. It’s at most 90 m; patch cords aren’t included. The PL adapter’s test cord is excluded from the result, but the end connections formed by the adapter plug with the jacks in the panel and the outlet are included in the measurement. This is the model used for new installation acceptance; it measures the components under the installer’s responsibility.
- Channel: The end-to-end path from the switch port to the end device, including patch cords. The total is at most 100 m (90 m of fixed cable + a total of 10 m of patch/equipment cords) and can include up to four connection points. The point where the channel adapter connects to the instrument isn’t included in the measurement.
The practical rule: certify new installations with PL. When diagnosing a line where a network performance problem appears on the user side, do a channel test, because in the field most of the culprits are aged or cheap patch cords. For the channel result to be meaningful, the patch cords that will actually be used must be plugged in. If you change the cord after the test, the result is invalid.
The tips of PL adapters are wear parts. They need to be replaced after a certain number of insertions; a worn adapter produces NEXT and return loss failures on sound links and sends the team to the wrong place. Before starting the test, the reference (set reference) should be set on the instrument and the calibration date should be confirmed as valid. These two checks take five minutes.
Wiremap faults: split pair, reversed and crossed
Wiremap shows whether the eight conductors are connected to the right pins at both ends. These are the most common faults in reports:
- Open / short: A broken conductor or two conductors shorted together. The instrument usually also gives the distance to the fault; if the value is very close to the termination, the IDC connection isn’t fully seated. Open that end and look first.
- Reversed pair: The two conductors of a pair have swapped places (e.g. pin 1 and pin 2). Some end devices keep working with automatic polarity correction, but the link doesn’t pass certification.
- Crossed pairs: Pairs have gone to other pair positions; the typical example is T568A termination at one end and T568B at the other (the 1-2 and 3-6 pairs swap). A single termination scheme (usually T568B) should be used throughout an enterprise network infrastructure project and written into the specification.
- Split pair: Pin continuity is correct, but the signal is split across the conductors of two different physical pairs. Simple continuity testers can’t see this; the certification instrument catches it from the excessive NEXT value. It’s the classic fault that stays silent at 10/100 Mb/s and produces errors at gigabit and above. Sound on paper, problematic in the field.
Insertion loss, NEXT, PS NEXT and return loss
The report’s frequency-domain parameters must be read one by one, because each points to a different physical defect. Glancing at the “All PASS” line and moving on means throwing away half the report.
| Parameter | What does it measure? | Typical cause of a bad result |
|---|---|---|
| Insertion loss | Attenuation of the signal along the link | Excessive length, high temperature, low-quality cable |
| NEXT | Signal leaking from one pair to a neighboring pair at the same end | Twist opened too far at termination, split pair, mismatched component |
| PS NEXT | Combined effect of the other three pairs | Same causes as NEXT; decisive in multi-pair applications |
| ACR-F / PS ACR-F | Ratio of crosstalk measured at the far end to attenuation | Cable and connector quality |
| Return loss | Energy reflected back due to impedance mismatch | Crushed/squeezed cable, sharp bend, mixed components, bad patch cord |
NEXT and return loss are measured from both ends; the report shows “main” and “remote” results separately. If the problem appears only at one end, the fault is most likely in the termination at that end. There’s no need to re-pull the cable. The report also includes propagation delay (limit of 555 ns for channel, 498 ns for permanent link), delay skew (channel 50 ns, PL 44 ns) and length. Length is calculated from the NVP (nominal velocity of propagation) value entered into the instrument; if the wrong cable type is selected, the length is also reported wrongly.
Some results are shown for information only at low-loss frequencies. At frequencies where insertion loss stays below 3 dB, the return loss result isn’t included in the evaluation under the “3 dB rule.” That’s why you see values marked informational on short links; it isn’t an error.
In PoE projects, also look at DC loop resistance and DC resistance unbalance within a pair. Unbalance causes PoE current to be distributed unevenly across the two conductors of a pair and degrades the data signal. For the effect of bundle heating with high-power PoE, see the heating in PoE and cable bundles article.
Interpreting marginal and failed results
The report shows two values for each parameter: the worst case margin (the margin at the frequency closest to the limit) and the worst case value (the worst value measured). A positive margin is a pass, a negative margin is a failure. First look at the margin, then at which frequency and which pair combination that margin occurred.
When a measurement falls within the instrument’s accuracy uncertainty, the result is marked with an asterisk:
- PASS: All parameters are better than the limit by more than the uncertainty margin.
- PASS*: At least one parameter passes the limit, but the margin is within the instrument’s accuracy uncertainty. In the TIA and ISO approach, it’s generally accepted as a pass.
- FAIL*: The limit has been exceeded, but the difference is within the uncertainty. It’s considered a failure.
- FAIL: A clear failure.
If you see a low margin on the same parameter across multiple links, the problem is systematic: the same team, the same termination habit or the same component batch. A low margin seen on a single link is usually physical damage specific to that link. The pattern gives you the fault’s address.
Common causes of test failures and how to fix them
The diagnostic functions of certification instruments (HDTDR and HDTDX analyses on the Fluke Networks DSX) show the fault’s position on the link in meters. A team that reads the position and the symptom together doesn’t blindly re-pull the cable. The order we follow in the field is roughly this:
- High NEXT at the end: The twist has been opened too far at the termination, or the jacket has been stripped back too much. Re-terminate the relevant end. With Cat6A, tolerance on this is even lower.
- Return loss problem in the middle of the link: A cable tie pulled tight in the tray, crushing, a sharp bend or a point that has been stepped on. Loosen the cable ties or replace them with hook-and-loop straps; if the damage is permanent, replace the cable.
- High insertion loss: The link exceeds 90 m, the cable runs through a hot route, or non-standard cable (e.g. copper-clad aluminum, CCA) was used. If the same failure shows up on many links at once, first check the test limit and cable type selected on the instrument; often the problem isn’t in the cable but in the menu.
- NEXT/return loss failures at random locations: A worn PL adapter or a dirty/damaged test interface. Retest on a known good reference link.
- Mixed components: A Cat6 outlet on Cat6A cable, or modules from a different manufacturer. The category is only as high as the weakest component end to end.
- Damage during pulling: Pulling tension for 4-pair cable should be limited to 110 N (25 lbf), and the manufacturer’s bend radius value should be respected. Exceeded tension disturbs the pair twist, and it doesn’t come back.
Handover checklist for the test report
When receiving the report, don’t settle for the “All PASS” statement on the summary page. The items below are the minimum framework for acceptance; for the wider process, the structured cabling specification and handover acceptance checklist article gives details.
- The test instrument model, serial number and calibration date are in the report.
- The selected test limit (e.g. TIA Cat6A Perm. Link) is the same as in the specification.
- The cable type and NVP value have been selected to match the cable used.
- There is a single result for each link; repeated records for the same link ID are explained.
- Link IDs exactly match the labels in the field (consistent with the TIA-606 labeling scheme).
- PASS* and FAIL results are listed separately, and the failed ones have been fixed and retested.
- The report has been delivered both as a PDF and in the instrument’s native file format (a LinkWare file for Fluke).
- As a sample, a few links have been retested in front of the receiving party.
If the cable category hasn’t been settled yet, start with the choosing Cat6, Cat6A and fiber article. The measurement method for fiber lines is different; details are in the fiber testing with OLTS and OTDR article. For projects where design, installation and certification should run under a single responsibility, see our structured cabling and low-voltage infrastructure service page. The primary source for the definition of test limits is TIA’s standards publications (tiaonline.org).
Frequently asked questions
Why can a line that passes the permanent link test fail the channel test?
The channel test also includes patch cords and equipment connections. A low-quality, worn or category-mismatched patch cord produces NEXT or return loss failures even if the fixed part is sound. Replace the patch cord first and repeat the test.
Should a link with a PASS* result be accepted?
In the TIA and ISO approach, PASS* is generally counted as a pass; the measurement passed the limit, and the difference is only within the instrument’s uncertainty. However, manufacturer warranty terms or the project specification may be stricter. Write the evaluation rule into the contract in advance.
Can a simple device that does wiremap testing replace certification?
No. A wiremap device shows whether the conductors go to the right pins; it doesn’t measure frequency-domain parameters such as split pair, NEXT, return loss or insertion loss. A certification instrument report is required for installation acceptance and the manufacturer warranty.
How can you tell whether an existing Cat6 infrastructure will carry 10GBASE-T?
IEEE 802.3an specifies for 10GBASE-T over Cat6 (Class E) cabling not 100 m but at most 55 m, and even this distance depends on alien crosstalk within the bundle. Measure the links with a certification instrument, assess alien crosstalk on long, densely bundled routes, and plan a move to Cat6A on critical lines.