Datacenter
Cat6A copper certification field guide for data centers
What certification proves over a wiremap test, permanent link versus channel, the TIA-568.2 parameters, alien crosstalk, and how to read a fail off the sweep.
Direct answer
Copper certification tests a Cat6A permanent link or channel against the ANSI/TIA-568.2 Category 6A transmission limits across the full frequency sweep to 500 MHz, returning a pass or fail with margin. A wiremap verifier only confirms the pin-to-pin map. Project specifications and the manufacturer warranty govern which limits and model apply.
Key takeaways
- Cat6A certification tests a permanent link or channel against ANSI/TIA-568.2 Category 6A limits across the full sweep to 500 MHz, returning pass/fail with margin.
- A wiremap verifier only confirms the pin-to-pin map; it does not certify or measure insertion loss, NEXT, or return loss.
- Permanent link is fixed cabling plus connectors (90 m), excluding cords; channel adds the cords (100 m). The spec sets which model applies.
- A NEXT fail worsening toward 500 MHz signals too much untwist at a termination; keep untwist under about half an inch.
- PASS* counts as a pass and FAIL* counts as a fail; every parameter must read pass or PASS*, and one FAIL* fails the link.
Copper certification, and what the report actually proves
Certification is the measured proof that an installed copper link meets a category's transmission limits across its whole frequency range, with a recorded margin to the limit. For Cat6A that range runs to 500 MHz, and the limits come from ANSI/TIA-568.2, the balanced twisted-pair standard, or its international counterpart, ISO/IEC 11801 Class EA. The output is a pass or fail per parameter, per pair, with how many decibels of room you had.
That is a different thing from proving the wires are connected to the right pins. A link can be wired perfectly, pass continuity, and still fail certification because the insertion loss is too high or the crosstalk between pairs is too close to the noise floor at 250 MHz. The category limits are a frequency story, and you only see the whole story with an instrument that sweeps the frequency.
On a data center floor the certification report is not paperwork for its own sake. It is the document the manufacturer's warranty hangs on, the thing the owner's commissioning agent checks before acceptance, and the record that settles the argument six months later when a 10G link flaps. No clean report, no warranty, and usually no acceptance.
What does certification prove that a continuity tester does not?
Certification proves the link carries the signal the application needs, across the full band, against a published limit. A continuity tester, or a verifier, proves only that the eight conductors land on the right pins end to end. Those are two different jobs, and confusing them is the single most common mistake on a cabling turnover.
A verifier reads the map: it catches an open, a short, a reversed pair, a split pair, a crossed pair, and the length by time-domain reflectometry. That is useful for a quick is-it-connected check during rough-in. What it cannot tell you is whether the link will run 10GBASE-T, because it does not measure insertion loss, near-end crosstalk, return loss, or any of the transmission parameters that decide the matter at 100, 250, or 500 MHz.
The tell is the price and the adapter. A verifier is a couple hundred dollars and has no permanent-link adapter that references to the standard. A certifier is a calibrated instrument that sweeps frequency, holds an accuracy level under ANSI/TIA-1152, and prints a margin in decibels. If the spec says certify and someone hands you wiremap results from a verifier, the link is not certified, no matter how green the screen looked.
Permanent link or channel: which test model does the spec require?
The permanent link is the fixed cabling: the horizontal run from the patch panel to the work-area outlet, plus the connector at each end, and nothing else. The channel is the whole working path: the same horizontal run plus the equipment cord at the switch, the patch cord at the device, and the mated connections those cords add. The link excludes the cords. The channel includes them.
Most new structured cabling is certified as a permanent link, because that is the part the installer built and warrants, and the test plug at each end is excluded from the result so a worn adapter does not fail good cable. The channel test is what you run when you want proof of the path as it actually operates, with the real cords in place, or when troubleshooting a working link that misbehaves. It is entirely possible to pass the permanent link and fail the channel, because a cheap or overlong patch cord can sink the result the fixed cabling alone would pass.
Set the tester to the model the project specification calls out, and use the matching adapter. Certifying a permanent link with a channel adapter, or the reverse, applies the wrong limits and the wrong reference, and the report is worthless even if the screen says pass. The spec controls which model is required, and the two are not interchangeable.
Cat6A, Cat6, Cat5e, or Cat8: which category and why?
Cat6A is specified to 500 MHz and carries 10GBASE-T to the full 100 m channel, which is why data centers standardized on it for access-layer and PoE runs to cameras, access points, and edge devices. Cat6 runs to 250 MHz and will do 10G, but only to roughly 55 m and only when alien crosstalk is controlled, so it is a 1 Gbps cable in practice. Cat5e tops out near 100 MHz and is a gigabit cable, fine for legacy drops and not a 10G option.
Cat8 sits at the other end. It sweeps to 2000 MHz and supports 25GBASE-T and 40GBASE-T, but only over about 30 m, which makes it a short top-of-rack or switch-to-server cable, not a horizontal-distribution cable. The bandwidth buys speed and costs reach. That tradeoff is why no one wires a building in Cat8.
The reason the data center landed on Cat6A is the combination Cat6 cannot give: 10G at 100 m, with the alien crosstalk performance that lets you bundle the cables in tray and conduit without the links degrading each other. Cat6A also carries four-pair PoE with more headroom on heat and resistance than Cat5e, which matters when the same drop powers the device it serves. Verify the category the project specifies and the application it has to run before you pick the cable, because the cable and the certification limits are bound together.
| Category | Bandwidth | Top application | Reach for that application | Data center role |
|---|---|---|---|---|
| Cat5e | 100 MHz | 1000BASE-T (1 Gbps) | 100 m | Legacy drops, not for new 10G |
| Cat6 | 250 MHz | 10GBASE-T, alien-crosstalk limited | Up to about 55 m for 10G | 1 Gbps in practice |
| Cat6A | 500 MHz | 10GBASE-T | 100 m channel | Access and PoE default |
| Cat8 | 2000 MHz | 25G and 40GBASE-T | About 30 m (Class I/II) | Short top-of-rack and server links |
The certification parameter set under TIA-568.2
A Cat6A certification is not one test. It is a battery of transmission measurements run pair by pair across the sweep, each with its own limit line, and the link passes only when every parameter passes on every pair. ANSI/TIA-568.2 sets the limits, and the field tester runs the set defined for field testing under ANSI/TIA-1152. Miss one parameter on one pair and the whole link fails.
Two of the parameters are the simple physical checks. Wiremap confirms the pin-to-pin connections end to end and catches the reversed, crossed, split, and shorted pairs. Length reports the electrical length of each pair, derived from the cable's nominal velocity of propagation, against the 90 m permanent-link and 100 m channel limits. The rest are the frequency-domain parameters where category cable actually earns its rating, and the table below is the working set to expect on the report.
Read the report parameter by parameter, not just the headline pass. A link that passes overall can still carry a worst-case margin of a few tenths of a decibel on NEXT at the top of the band, and that thin margin is the first thing that disappears when the room runs hot or someone re-dresses the bundle. The margin column is where the real condition of the link lives.
| Parameter | What it measures | Common cause of a fail |
|---|---|---|
| Wiremap | Pin-to-pin connections end to end | Reversed, crossed, or split pairs, opens, shorts |
| Length | Electrical length per pair from NVP | Run over the 90 m link or 100 m channel limit |
| Insertion loss | Signal lost along the link, in dB | Run too long, wrong category, hot cable, thin cords |
| NEXT | Near-end crosstalk between pairs, in dB | Excess untwist at the jack, poor termination |
| PSNEXT | Combined near-end crosstalk from all pairs | Same as NEXT, summed across the disturbers |
| ACRF (ELFEXT) | Far-end crosstalk normalized to loss | Pair geometry, mismatched components |
| PSACRF | Power-sum far-end version of ACRF | Same, across all disturbing pairs |
| Return loss | Signal reflected by impedance change, in dB | Kinks, over-cinched ties, impedance mismatch |
| Propagation delay | Time the signal takes end to end | Run length, slow cable |
| Delay skew | Spread in delay across the four pairs | Mixed insulation, different twist rates |
| DC loop resistance | Round-trip resistance of a pair | Long run, thin conductor, bad contact |
| DC resistance unbalance | Resistance difference within and between pairs | Bad termination, broken strand, mixed gauge |
Why did my Cat6A fail insertion loss?
Insertion loss is the signal you lose walking the length of the link, measured in decibels and rising with frequency, so the limit is hardest to meet at the top of the band near 500 MHz. A fail there is almost always length, temperature, or the cable and cords themselves, not the terminations. Crosstalk lives at the connector. Insertion loss lives in the run.
Length is the first suspect. The permanent link is built around 90 m of horizontal cable, and once you blow past that the loss climbs past the limit no matter how clean the work is. Pull the length result first, because a 95 m run fails insertion loss for a reason no amount of re-termination will fix. Temperature is the quiet second cause: copper gains resistance as it warms, so a link that passes in a cool room can fail the same loss limit in a hot ceiling or a loaded PoE bundle. The standard derates the allowed length as ambient rises for exactly this reason.
The other two causes are cable and cords. A run that is actually Cat6 dressed as Cat6A, or a channel padded out with long thin stranded patch cords, loses more than the limit allows. When insertion loss fails and the length is inside limit, check the ambient, then check whether the cords and the cable are the category the test is set for. The fix is shorter run, cooler path, or the right components, never a re-punch.
NEXT and PSNEXT: the connector's signature
Near-end crosstalk is the signal that leaks from the transmitting pair into the other pairs at the same end of the link, measured as a loss in decibels, where more decibels is better because it means less leak. NEXT is the parameter that reads workmanship at the termination, because the twist in the pair is what cancels the crosstalk, and every millimeter you untwist to land the conductors opens a window for the pairs to couple.
PSNEXT is the power sum: the combined effect of all the other pairs crosstalking into one, which is the number that matters for an application like 10GBASE-T that uses all four pairs at once. A link can pass pair-to-pair NEXT and still sit tight on PSNEXT, because the disturbers add up. On Cat6A at 500 MHz the limits are tight, and the margin you have at the top of the band is usually set by the worst jack on the link.
A NEXT fail points at the termination almost every time. Too much untwist at the punch-down, conductors splayed before the contact, a jack seated crooked, or a plug terminated with the pairs fanned out too far back. The frequency tells you where: a NEXT fail that gets worse as frequency climbs is the classic untwist signature. Re-terminate the end the tester points to, keep the untwist to the minimum the connector allows, and the margin comes back.
ACRF, PSACRF, and the far-end picture
ACRF is far-end crosstalk normalized for the insertion loss the signal already suffered getting down the link, which is why it used to be called ELFEXT, equal-level far-end crosstalk. It measures the leak that shows up at the opposite end from the transmitter, after both signals have traveled the run, so it reads the geometry of the pairs along the whole length rather than the workmanship at one connector.
PSACRF is the power-sum version, the combined far-end leak from all disturbing pairs into the victim, and like PSNEXT it is the number that governs a four-pair application. Far-end crosstalk is driven by how consistently the pair geometry holds across the run and by component matching at the connectors, so a fail here often points at a mismatched jack and plug or cable that is out of spec, not at a single bad punch.
ACRF fails are less common than NEXT or return loss fails on a well-built Cat6A link, and when they show up they tend to ride with another parameter. Treat an isolated far-end fail as a sign to check that the jacks, the cable, and the cords are all the same listed system, because mixing components from different manufacturers is where the far-end numbers drift.
Why did my Cat6A fail return loss?
Return loss is the signal reflected back toward the source by any change in the cable's impedance, measured in decibels, and a fail means too much of the signal is bouncing instead of traveling. Cat6A is built to hold a 100 ohm characteristic impedance, and anything that disturbs that impedance, a kink, a tight bend, a cable tie cinched into the jacket, or a connector mismatch, throws a reflection the tester sees.
The cause is mechanical abuse of the cable far more than it is a bad termination. A cable tie pulled down hard deforms the pairs under the jacket and changes the local impedance, and a bundle dressed with the ties machine-tight is a return loss fail waiting to happen. A bend tighter than the four-times-diameter rule does the same. So does a kink from a cable that got stepped on or yanked through a tight pull. The reflection shows up at a frequency that maps to a distance, which is how you find the spot.
This is the parameter that punishes a bundle that looks beautiful. The crew that dresses a tray with every tie torqued down and every bend crisp is the crew that fails return loss, because the cable wanted to be loose. Hand-snug the ties, hook-and-loop instead of nylon on the dense runs, keep the bend radius honest, and pull the offending cable out of the cinch. Return loss is the workmanship parameter that rewards a gentle hand, not a tight one.
Propagation delay and delay skew
Propagation delay is the time the signal takes to cross the link, and it tracks length, so it rarely fails on a run that already passed the length test. Delay skew is the interesting one: it is the spread in arrival time between the four pairs, and it matters because 10GBASE-T splits the signal across all four pairs and reassembles it, so if one pair runs much slower than the others the receiver has trouble lining the streams back up.
Skew comes from the cable, not the install. The four pairs in a cable are twisted at different rates so they do not couple, and they can be insulated with slightly different materials, both of which make the pairs different electrical lengths even though they are the same physical length. Good cable holds the skew well inside the limit. Cheap or counterfeit cable, or a channel that mixes two different cables in one path, is where skew creeps toward the line.
A delay skew fail is a cable problem. If it shows up, you are not going to fix it by re-terminating, so check whether the run is a single continuous cable of one type or a Frankenstein of two, and check the cable is what the box claims to be.
DC loop resistance and resistance unbalance for PoE
The DC parameters do not matter for data the way the frequency parameters do, but they decide whether a link carries four-pair PoE cleanly, which on a data center floor is most of the access-layer drops. DC loop resistance is the round-trip resistance of a pair, and it sets the voltage drop and heat a powered device sees at the far end. DC resistance unbalance is the difference in resistance, both between the two conductors of a pair and between the pairs, and it is the one that bites four-pair power.
Four-pair PoE splits the current across all four pairs and across the two conductors in each pair, and it counts on that sharing being even. When one conductor or pair has lower resistance it hogs more than its share of the current, runs hotter, and drops more voltage than the design assumed. The standard caps the unbalance to keep the sharing even, and a fail usually means a bad termination, a partially broken strand, or a channel that mixed conductor gauges. For the heat and delivered-power side of the same problem, see the PoE voltage drop and heat guide.
Run the DC resistance unbalance test on any link that has to carry Type 3 or Type 4 power. It is cheap insurance, and it catches the run that passes length and wiremap but cannot carry four-pair power evenly. When a powered device misbehaves on a link that certified clean for data, resistance unbalance is the next thing to check.
What is alien crosstalk, and why is it unique to Cat6A?
Alien crosstalk is the interference that leaks from one cable into a neighboring cable in the same bundle, as opposed to the pair-to-pair crosstalk inside a single cable. It splits into ANEXT, alien near-end crosstalk, and AFEXT, alien far-end crosstalk. Cat6A is the first category that has to control it, because at 10GBASE-T the cable-to-cable leak in a tight bundle, not the internal crosstalk, becomes the parameter that limits the link.
This is why Cat6A cable is physically bigger, often with a separator and sometimes a shield, and why bundle density and cable separation matter on a Cat6A pull in a way they never did on Cat6. Pack the cables tight and parallel for a long run and the alien crosstalk climbs as more of them go active. Loosen the bundle, vary the lay, or use a shielded construction and it drops. The cable design and the pathway both drive it.
Alien crosstalk is tested differently from the in-cable parameters, and that is the part crews miss. It is not a single-link measurement the certifier runs in one shot like NEXT. ANEXT is characterized from both ends and AFEXT from one, against a victim cable surrounded by disturbers, often the six-around-one configuration the cable makers use to qualify a system in the lab. In the field it is a slow, sampled measurement, so most projects rely on a manufacturer-qualified system and good bundling practice rather than testing every link for alien crosstalk. Confirm what the spec and the warranty require, because some demand sample alien crosstalk testing and most do not.
Setting up the field certifier
The certifier is a two-unit instrument, a main and a remote, that you set to the standard, fit with the right adapters, and reference before the first link. Set the test limit to the category and model the spec calls for, Cat6A permanent link or Cat6A channel under TIA-568.2 or the ISO Class EA equivalent, and set the cable type so the tester uses the right nominal velocity of propagation for the length calculation. The NVP is how the instrument turns time into distance, and a wrong NVP gives a wrong length.
Fit the adapter that matches the model. A permanent link adapter has a high-quality test cord and a plug that mates with the installed jack, and its known characteristics are excluded from the result so the adapter does not fail good cable. A channel adapter accepts the actual patch cords. Do not mix them, and watch the wear count on the permanent link adapter, because the test plug is a consumable that drifts as it ages.
Run the self-test and set the reference before you certify anything. The self-test connects the two adapters together and confirms the instrument and the adapters are healthy, which is how you catch a worn or damaged permanent link adapter before it fails a hundred good links. Setting the reference establishes the baseline the measurements are taken against. Both units also need to be inside their calibration cycle, typically an annual factory calibration, because a certifier out of cal is producing numbers nobody has to honor. Keep the cal date and the adapter test counts in the record.
Reading a fail off the frequency sweep
Every transmission parameter is a curve plotted against frequency with a limit line over it, and the shape and the location of a fail tells you what went wrong before you ever touch the cable. Pull up the plot for the failing parameter and read where on the band it crosses the line, because that points at the cause.
A NEXT fail that grows worse as frequency rises, peaking near the top of the band, is the untwist signature at a connector. A return loss fail shows as a spike at a specific frequency, and the tester can convert that frequency to a distance down the link, which walks you to the kink or the over-cinched tie. An insertion loss fail that runs high across the whole band is length, temperature, or the wrong cable, not a connector. A fail on one pair points you at that pair's termination, while a fail on all pairs points at the cable or the length.
The fault-locate function on a good certifier does most of this for you, naming the worst parameter, the worst pair, and the distance to the largest reflection. Use it, then go to the spot it names instead of re-terminating both ends on a guess. The plot is the diagnosis. The asterisk and the margin tell you how close you were, and the distance marker tells you where to put your hands.
Termination workmanship that passes versus fails
The work that passes certification is the work that keeps the cable as close to its made condition as the connector allows. The pair stays twisted right up to the contact, the jacket stays on as far as it can, the bend stays loose, and nothing crushes the cable. Every departure from that is a parameter giving up margin.
Untwist is the one that fails NEXT. Each conductor you peel back from its twist to land it opens the pairs to each other, and Cat6A at 500 MHz has little tolerance for it. Keep the untwist to the minimum the jack or plug needs, commonly held to about half an inch or less, and follow the connector maker's termination diagram rather than fanning the pairs out for a tidy look. Manage the pairs into the jack in their lay, do not splay them flat.
The cinch and the bend fail return loss. A cable tie torqued down with a gun deforms the pairs and throws a reflection, so dress bundles hand-snug with hook-and-loop on the dense runs, and never machine-tight nylon. Hold the bend radius to at least four times the cable diameter at every turn and at the back of the jack. And remember the PoE cable runs warm: a bundle that certifies clean cold can drift on insertion loss once it is loaded and the ambient climbs, so the derate for hot, loaded bundles is part of the workmanship, not an afterthought. The crew that dresses gently and keeps the geometry honest is the crew whose links pass on the first sweep.
The PASS* and FAIL* asterisk, and measurement uncertainty
An asterisk on a result means the margin is smaller than the tester's own measurement accuracy, so the instrument cannot tell with certainty which side of the line the link truly sits on. A PASS* is a marginal pass: it passed, but by less than the accuracy figure, so a more accurate instrument might read it either way. A FAIL* is a marginal fail by the same logic. The asterisk is the instrument being honest about its own uncertainty.
The rule for the report is set by the standard, and it leans the safe way each time. A PASS* counts as a pass and is compliant. A FAIL* counts as a fail. To earn an overall pass, every parameter has to read pass or PASS*, and a single FAIL* drops the whole link to fail. So the asterisk does not change the verdict, but it tells you how thin the result is.
Treat a link full of PASS* margins as a warning even though it passed. The margin that vanished into the measurement uncertainty is margin the link does not really have, and it is the first thing to go when the room heats up or someone re-dresses the bundle. A higher-accuracy tester, set to a tighter PASS* threshold, narrows the gray zone, which is one reason the spec sometimes calls out a tester accuracy level. When a job is full of asterisks, go find out why before you hand it over.
The certification report and turnover
The deliverable is not the green screen, it is the report: a record per link of the test standard, the model, the cable type, the length, every parameter with its worst-case margin and the pair it occurred on, the overall pass or fail, and the tester with its calibration date. The certifier's software rolls these into one file for the whole job, and that file is what the owner, the commissioning agent, and the manufacturer accept against.
The manufacturer warranty is the reason the report has to be clean and complete. A system warranty on a Cat6A plant is granted on the strength of the certification records, run on a calibrated tester of the required accuracy, against the right limits, with no unresolved fails. A report with FAIL* links, a stale calibration date, or the wrong test model can void the warranty the owner paid for, so the report is checked, not just filed.
Tie the report to the link IDs and the as-built so the next person can find a given drop. Certification is one layer of the structured cabling turnover package that the data center cabling overview covers end to end, alongside the labeling, the pathway records, and the as-built. The report proves the copper. The package proves the system.
What to document
A Cat6A result earns its keep only when it points back to a specific drop on the as-built. The report has to map to the link ID and the as-built, and it has to carry enough that a reviewer can reproduce the verdict and the warranty can stand on it.
Capture the link ID, the test standard and limit, the test model, the measured length, the worst-case margins on the parameters that matter most, NEXT, insertion loss, and return loss, the overall pass or fail, and the tester serial number with its calibration date. If a link reads PASS* anywhere, note which parameter, because the marginal ones are the links that come back. Keep the record in the format the warranty submission requires, because reformatting a thousand links after the fact is its own job.
| Field to record | Why it matters |
|---|---|
| Link ID | Ties the result to a physical drop and the as-built |
| Test standard and limit | Confirms the right Cat6A limits were applied |
| Test model (link or channel) | The wrong model invalidates the result |
| Measured length | Catches over-length runs and bad NVP settings |
| Worst margin NEXT / IL / RL | Shows how close the link is to failing |
| Overall pass or fail | The verdict the warranty hangs on |
| Tester serial and cal date | An out-of-cal tester voids the report |
Common mistakes
- Handing over wiremap results from a verifier and calling the link certified.
- Certifying with the wrong adapter or the wrong test model for what the spec requires.
- Untwisting the pairs too far back at the jack and failing NEXT at the top of the band.
- Dressing bundles machine-tight with nylon ties and failing return loss from the cinch.
- Bending tighter than four times the cable diameter at the jack or the tray turn.
- Ignoring alien crosstalk on dense Cat6A bundles when the spec or warranty calls for it.
- Accepting a report full of PASS* margins without finding out why the link is marginal.
- Certifying on a tester past its calibration date or with a worn permanent-link adapter.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
The category limits live in ANSI/TIA-568.2, the balanced twisted-pair cabling standard, which defines the Category 6A transmission requirements the link is measured against, with ISO/IEC 11801 Class EA as the international counterpart. The field-test side, the accuracy levels a certifier has to meet and how the parameters are measured in the field, comes from ANSI/TIA-1152, the standard for field test instruments for twisted-pair cabling. These are the two documents the report answers to.
The application limits come from IEEE. The 10GBASE-T specification in IEEE 802.3 is what drives the 500 MHz, 100 m target Cat6A is built to meet, and the higher-rate specifications for 25GBASE-T and 40GBASE-T are what push Cat8 to 2000 MHz over its shorter reach. The exact clause, table, and revision letters of all of these shift as the standards are updated, so name them by topic and confirm the edition the project specifies before you cite a specific section on a submittal.
Above all of it, the project specification and the manufacturer's system warranty govern. The spec sets the category, the test model, and the tester accuracy required, and the warranty sets what the certification records have to show to be honored. Where a manufacturer or the contract demands something tighter than the general standard, that requirement controls. Cite the standard that governs the point, and let the spec and the warranty override a rule of thumb.
Units, terms, and conversions
Copper certification mixes decibels, frequency, distance, and a few parameter acronyms that read differently across a tester screen, a standard, and a spec, so the same result can look unfamiliar depending on where you meet it.
Transmission parameters are in decibels, where for a loss-style parameter like NEXT or return loss a larger number is better, meaning less leak or less reflection. Bandwidth is in megahertz, the frequency range the cable is swept across, 500 MHz for Cat6A. Length is in meters in the standards, where 90 m is the permanent link and 100 m the channel, and a US drawing may give the same run in feet, roughly 295 ft and 328 ft. ACRF is the current name for what older testers and documents call ELFEXT. AXT, alien crosstalk, is the cable-to-cable family that splits into ANEXT and AFEXT.
- dB
- Decibel; the unit for loss-style parameters, where more dB of NEXT or return loss is better
- MHz
- Megahertz; the frequency the link is swept across, 500 MHz for Cat6A
- NEXT / PSNEXT
- Near-end crosstalk between pairs, and the power-sum from all pairs at the near end
- ACRF (ELFEXT)
- Far-end crosstalk normalized to insertion loss; the older name was ELFEXT
- Return loss
- Signal reflected by an impedance change; a kink or an over-cinched tie throws it
- Delay skew
- The spread in propagation delay between the four pairs, a cable property
- AXT (ANEXT / AFEXT)
- Alien crosstalk between adjacent cables in a bundle, near-end and far-end
- NVP
- Nominal velocity of propagation; how the tester converts signal time into length
FAQ
What is the difference between a permanent link and a channel test?
The permanent link is the fixed horizontal cabling plus the connector at each end, excluding patch cords. The channel adds the equipment and patch cords and the connections they bring. Most new installs certify the permanent link for warranty; the channel proves the path as it actually runs. The spec sets which model applies.
Why did my Cat6A fail NEXT certification?
A NEXT fail almost always means too much untwist at a termination, since the pair twist is what cancels near-end crosstalk. A fail that worsens toward 500 MHz is the untwist signature. Re-terminate the end the tester names, keep the untwist under about half an inch, and follow the connector's termination diagram instead of fanning the pairs.
Why did my Cat6A fail return loss?
Return loss fails come from impedance changes, not bad punches. A cable tie cinched machine-tight, a bend tighter than four times the cable diameter, or a kink deforms the pairs and reflects signal. The tester maps the reflection to a distance. Loosen the ties, fix the bend radius, and pull the offending cable out of the cinch.
Cat6 vs Cat6A vs Cat8: what is the difference?
Cat6 runs to 250 MHz and does 10G only to about 55 m. Cat6A runs to 500 MHz and carries 10GBASE-T the full 100 m, which is why data centers standardized on it. Cat8 reaches 2000 MHz for 25G and 40G but only about 30 m, making it a short top-of-rack cable.
What is alien crosstalk in Cat6A cabling?
Alien crosstalk is interference leaking between adjacent cables in a bundle, split into ANEXT at the near end and AFEXT at the far end. Cat6A is the first category to control it, because at 10G the cable-to-cable leak in a tight bundle limits the link. Loose bundling and shielded cable reduce it; dense parallel runs raise it.
Does a wiremap verifier certify a Cat6A link?
No. A verifier only confirms the pin-to-pin map, catching opens, shorts, and reversed or split pairs. Certification measures insertion loss, NEXT, return loss, and the full parameter set against the TIA-568.2 limits across the frequency sweep with a recorded margin. If the spec says certify, wiremap results from a verifier do not satisfy it.
What does a PASS* asterisk mean on a certification report?
An asterisk means the margin is smaller than the tester's measurement accuracy, so the instrument cannot be certain which side of the limit the link sits on. A PASS* counts as a pass and is compliant; a FAIL* counts as a fail. A report full of PASS* results is marginal, and worth investigating before turnover.
How do I read which parameter caused a certification fail?
Open the plot for the failing parameter and read where it crosses the limit. NEXT worsening toward the top of the band is connector untwist. A return loss spike maps to a distance, pointing at a kink or tie. Insertion loss high across the band is length, heat, or the wrong cable, not a termination.
Why does Cat6A need DC resistance unbalance tested for PoE?
Four-pair PoE splits current across the pairs and counts on even sharing. DC resistance unbalance means one conductor or pair carries more than its share, runs hotter, and drops more voltage. The standard caps it, and a fail signals a bad termination or broken strand. Test it on links carrying Type 3 or Type 4 power.
What standards govern Cat6A copper certification?
ANSI/TIA-568.2 sets the Category 6A transmission limits, with ISO/IEC 11801 Class EA as the international equivalent. ANSI/TIA-1152 covers the field test instrument accuracy. IEEE 802.3 defines the 10GBASE-T application. Confirm the edition against the project specification, since the manufacturer warranty and the contract documents ultimately govern which limits and model apply.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.