Ethernet Cable Loss Calculator Cat5e / Cat6 / Cat6A Insertion Loss
Check TIA-568 channel insertion loss for a copper twisted-pair run, see pass/marginal/fail margin, find the fastest Ethernet standard your cable category and length actually support — and look up T568A/T568B wiring, color codes, and crossover cables.
TIA-568 channel insertion loss · Pass/fail margin · Max supported speed
Cable & run
A standard channel is a 100m link: up to 90m of permanent-link (solid) cable plus up to 10m total of stranded patch cords at each end, through up to 4 connections.
TIA-568 channel comparison
Reference values only — TIA-568-C.2 channel insertion loss limits at 100m, scaled linearly to your entered length. Cable construction, connector quality, and installation (bends, untwisting at terminations) all affect real-world loss. For a certification-grade result, verify with a calibrated cable tester.
PoE voltage drop
Will a powered device at the end of this cable get the power it needs? Copper resistance eats into the voltage on long runs. Pick the PoE class, the cable, and the run length.
Source voltage and device power start at the class's worst-case values from the standard. Change them to match your actual switch and device. Uses the unit toggle above for length.
Worst-case PoE class limits
Class
Power at switch
Power at device
Min. switch voltage
Max. loop resistance
Pairs used
3 (802.3af)
15.4 W
12.95 W
44 V
20 Ω
2
4 (802.3at)
30 W
25.5 W
50 V
12.5 Ω
2
5 (802.3bt)
45 W
39.94 W
50 V
6.25 Ω
4
6 (802.3bt)
60 W
51 W
50 V
6.25 Ω
4
7 (802.3bt)
75 W
62 W
52 V
6.25 Ω
4
8 (802.3bt)
90 W
71.28 W
52 V
6.25 Ω
4
Values from the Ethernet Alliance's "Power over Ethernet – Cable Losses" whitepaper (2017) and its 802.3bt overview (2018). The Class 7–8 minimum switch voltage is shown as 52 V in that source; the final IEEE 802.3bt text may differ slightly (some references list 52.7 V), so check your switch's datasheet. Cable resistance here is calculated from wire size for typical solid copper at 20 °C. Real cable is often a little higher (twist, temperature, stranded patch cords), and copper-clad aluminum cable is much higher. This is a planning estimate, not a substitute for testing with a PoE tester.
Max supported Ethernet speed
Standard-defined maximum channel length by category and Ethernet speed. These are protocol-defined limits (IEEE 802.3), not just insertion loss — 10GBASE-T over Cat6 is limited by alien crosstalk in bundled cable runs, not raw attenuation, which is why its range varies with installation conditions.
Ethernet standard
Cat5e
Cat6
Cat6A
10BASE-T / 100BASE-TX
100m
100m
100m
1000BASE-T (1G)
100m
100m
100m
2.5GBASE-T
100m
100m
100m
5GBASE-T
Not officially rated (may work at reduced distance)
100m
100m
10GBASE-T
Not supported
37–55m (alien-crosstalk dependent)
100m
Per IEEE 802.3an (10GBASE-T) and 802.3bz (2.5G/5GBASE-T). 5GBASE-T was formally specified for Cat6, not Cat5e — a Cat5e run may pass 5G traffic in practice but isn't standard-guaranteed. 10GBASE-T over Cat6 UTP ranges from ~37m (worst-case, bundled cables) to ~55m (better-case) per the standard's alien crosstalk allowance; shielded Cat6 (F/UTP) can do better.
T568A vs. T568B pinout
Same RJ45 connector, same 8-conductor cable — T568A and T568B are just two different orders for which color pair lands on which pin. Only the green and orange pairs swap position between the two standards; blue and brown stay put.
Pin
T568B color
T568A color
Pair
1
White/Orange
White/Green
2
2
Orange
Green
2
3
White/Green
White/Orange
3
4
Blue
Blue
1
5
White/Blue
White/Blue
1
6
Green
Orange
3
7
White/Brown
White/Brown
4
8
Brown
Brown
4
Per TIA/EIA-568. Pin 1 is the leftmost pin with the RJ45 clip facing away from you and the cable pointing down.
Why two standards, and which one to use
T568A is the older standard — it's backward-compatible with single-pair USOC phone wiring, and some government specs (and older residential wiring) still require it. T568B became the de facto standard in most US commercial structured cabling, originating from an older AT&T wiring spec that got widely adopted. Electrically, neither is "better" — a channel wired end-to-end in either standard performs identically. The only rule that actually matters: both ends of a given cable run must use the same standard. Mixing them on a single cable is what creates a crossover, intentionally or by mistake.
Straight-through vs. crossover cables
A straight-through (patch) cable has both ends wired to the same standard — T568B/T568B or T568A/T568A. That's the cable in virtually every patch panel, wall jack, and PC-to-switch connection. A crossover cable has one end T568A and the other T568B, which swaps the transmit pair with the receive pair.
Connecting
Cable type
PC / server ↔ switch, router, or wall jack
Straight-through
PC ↔ PC (direct, no switch)
Crossover*
Switch ↔ switch (uplink, no MDI-X uplink port)
Crossover*
Switch ↔ router (most modern gear)
Straight-through
*Mostly historical. Virtually every NIC and switch port made since the mid-2000s supports Auto-MDI-X, which automatically detects and corrects for straight-through vs. crossover wiring — so a straight-through cable works for a direct PC-to-PC or switch-to-switch link on modern hardware too. Crossover cables are still worth understanding (and keeping one on hand) for older equipment that lacks Auto-MDI-X. Note also that a true Gigabit crossover needs all 4 pairs swapped, not just the 2 that a basic T568A/T568B crossover swaps — one more reason Auto-MDI-X made dedicated crossover cables mostly obsolete rather than just less common.
Recommended tools
TREND Networks SignalTEK QT Pro (R166001) An all-in-one tester for copper, fiber and Wi-Fi: 10G qualification, PoE load testing and network diagnostics, in one handheld.
As an Amazon Associate we earn from qualifying purchases.
What is Ethernet channel insertion loss?
Insertion loss is the reduction in signal strength (in dB) as a signal travels through a cabling channel — the cable itself, plus every connector and patch cord along the way. TIA-568-C.2 defines a maximum allowable insertion loss for each cable category at a 100-meter channel length, tested at that category's rated frequency.
Too much insertion loss means the receiving equipment can't reliably recover the signal, causing packet loss, retransmissions, or link speed downshifts (e.g. a 10GBASE-T port falling back to 1G).
TIA-568-C.2 channel insertion loss limits
Category
Test freq.
Max loss (100m channel)
Cat5e
100 MHz
24.0 dB
Cat6
250 MHz
21.3 dB
Cat6A
500 MHz
20.9 dB
Reference values — confirm against the current published TIA-568-C.2/D standard for certification work.
Why does insertion loss go down as frequency goes up?
It doesn't — higher category cable is actually rated to higher frequencies (Cat6A up to 500 MHz vs. Cat5e's 100 MHz) because better shielding, tighter twist rates, and improved dielectric materials let it carry more bandwidth before loss becomes a problem. The max allowed loss numbers above look similar across categories because each category is tested at a different, harder frequency — a fair comparison would ask "how much loss at the same frequency," where better cable always wins.
Worked example: is a 75m Cat6 run good for 10GBASE-T?
Say you've got a 75m Cat6 permanent link run from an IDF to a desk, and someone wants to plug in a 10GBASE-T switch port at both ends. Two separate checks matter here, and it's easy to only run one of them:
Check 1 — insertion loss. Cat6's TIA-568-C.2 limit is 21.3 dB over a 100m channel at 250 MHz. Scaled to 75m, the budget is roughly 21.3 × (75/100) = 16.0 dB, and a well-installed 75m Cat6 run should land close to that estimate. On pure attenuation, this run is fine — there's no reason a 10GBASE-T signal can't physically travel 75m of Cat6 without excessive loss.
Check 2 — the real limiter. 10GBASE-T over Cat6 isn't actually loss-limited in most real installs; it's limited by alien crosstalk from adjacent cables in the same bundle, which is a separate spec (ANEXT/AXT) this calculator's insertion-loss numbers don't cover. IEEE 802.3an guarantees only 37m worst-case for unshielded Cat6 in a bundled, high-density run — up to about 55m in a lighter-density or shielded install. A 75m Cat6 run is past even the optimistic end of that range, so it's a real risk for 10G even though the raw insertion-loss math looks comfortable. The fix, if 10G is required, is either Cat6A cable for that run or accepting the link may negotiate down to 1G or 2.5GBASE-T instead (both of which are solidly within spec at 75m on Cat6).
This is the single most common mistake installers make with this kind of math: insertion loss passing doesn't mean a higher-speed standard is supported — check the category/speed/distance table on the "Max speed by category" tab too.
Frequently asked questions
Does a shorter cable always mean lower loss? Yes for a fixed frequency and cable type — insertion loss scales roughly linearly with length. But going from Cat6 to Cat6A at the same length lowers loss more than shortening a Cat6 run by a few meters would, because Cat6A's tighter twist rates and thicker conductors reduce loss-per-meter at any given frequency.
My cable tester reports a different number than this calculator — which one is right? Trust the tester. This calculator scales the TIA-568-C.2 standard's maximum-allowed loss linearly by length as a field estimate; a certified tester measures your actual installed cable, connectors, and terminations, which is always the authoritative number for certification or troubleshooting a specific run.
Why does the "connections in channel" field not change the result? The TIA-568-C.2 channel limit already assumes up to 4 connections (2 permanent-link ends plus 2 patch cord ends) as part of the standard 100m channel definition — it's shown here for reference since exceeding 4 connections is itself an out-of-spec condition worth flagging, not because each connector's individual loss is broken out in this version of the tool.
Can I use this for a permanent link test instead of a full channel? The 90m permanent-link limit is roughly 0.9× the 100m channel limit shown here (TIA-568-C.2 defines it as a slightly tighter per-meter budget since it excludes patch cords) — close enough for field estimation, but use the channel numbers above if you're testing with patch cords attached at both ends, which is the more common real-world test configuration.