Introduction to Ethernet attenuation and channel distance
Ethernet signals become weaker as they travel through balanced copper pairs. That reduction is called attenuation; when a completed link is tested, the measured quantity is normally called insertion loss. A longer cable loses more signal than a short one, and a high-frequency signal loses more than a low-frequency signal on the same cable. Connectors and stranded patch cords also consume part of the available signal budget.
This calculator estimates loss for a complete copper Ethernet channel rather than for an ideal reel of cable. It combines solid-conductor horizontal cable, stranded patch and equipment cords, and mated connections such as patch panels, outlets, cross-connects and consolidation points. The result shows total insertion loss in decibels, remaining margin against the budget you enter, a standards-oriented reference limit and the maximum horizontal distance that fits the selected loss budget.
The calculation is useful during design and troubleshooting. It can reveal that shortening two long patch cords may recover more margin than replacing the horizontal cable, or that an apparently generous loss budget still cannot override the physical channel-length limit. It does not replace a calibrated field certification tester, because insertion loss is only one of several required measurements.
The Ethernet insertion-loss formulas used here
The category curves use the familiar three-term cabling model. Frequency f is entered in megahertz, and the result is the worst-case cable loss in decibels per 100 metres. The coefficients differ by category.
The square-root term primarily represents increasing conductor resistance caused by skin effect. The linear term represents dielectric loss, while the final correction improves the curve at low frequencies. The calculator scales that 100 m value to the installed cable length and then adds the cord and connection allowances.
In this expression, Lh is horizontal solid cable in metres, Lp is the combined length of all stranded cords, and N is the number of mated connections. The factor 1.2 models the approximately 20% higher loss of stranded cords. Each connection is assigned 0.04√f dB, which equals 0.4 dB at 100 MHz and about 0.63 dB at 250 MHz.
Solving the same equation for horizontal length gives the loss-limited distance. Connector and cord losses are deducted before the remaining budget is assigned to horizontal cable.
How to use the Ethernet distance calculator
Begin by selecting the category actually installed, not the category printed on one isolated component. A channel is only as capable as its lowest-rated cable, jack, panel or cord. Enter the horizontal cable length between the patch panel and work-area outlet. Do not include patch cords in that field, because the calculator applies a separate stranded-conductor allowance to them.
Next, total the cords at both ends and count the mated connections. A simple permanent link commonly has two connections: one at the patch panel and one at the outlet. A consolidation point or cross-connect raises the count. Select the highest test frequency relevant to the class or application being assessed. Frequency is not the same as the advertised data rate; for example, Class E is characterised to 250 MHz while Class EA is characterised to 500 MHz.
Finally, enter a dB budget and calculate. The result distinguishes a loss-only distance from the physical limit. Cat5e through Cat7 channels are normally limited to 100 m in total, with no more than 90 m of horizontal cable. Category 8 uses a much shorter 30 m channel and 24 m permanent link. A calculated loss distance above those values is spare electrical margin, not permission to exceed the standard.
Category coefficients and practical interpretation
| Category | a / b / c | Rated frequency | Typical channel limit |
|---|
| Cat5e, Class D | 1.967 / 0.0230 / 0.050 | 100 MHz | 100 m |
| Cat6, Class E | 1.808 / 0.0170 / 0.200 | 250 MHz | 100 m; shorter reach may apply to 10GBASE-T |
| Cat6A, Class EA | 1.820 / 0.0091 / 0.250 | 500 MHz | 100 m |
| Cat7, Class F | 1.800 / 0.0100 / 0.200 | 600 MHz | 100 m |
| Cat8, Class I or II | 1.720 / 0.0100 / 0.250 | 2000 MHz | 30 m |
A positive margin means the modelled insertion loss is below the chosen budget. A margin greater than about 20% of the limit is comfortable planning territory. A smaller positive margin can still pass, but temperature, workmanship, cable bundling and future re-termination deserve attention. A negative margin means the modelled channel exceeds the budget.
Worked example: a 78 m Cat6 channel at 250 MHz
Consider 78 m of solid Cat6 cable, 10 m of stranded cords and four connections tested at 250 MHz. The Cat6 curve produces approximately 32.85 dB per 100 m. The horizontal cable therefore contributes about 25.62 dB. The cords contribute roughly 3.94 dB after the 20% stranded allowance, and four connections add about 2.53 dB. Total insertion loss is approximately 32.09 dB.
Against a 35.9 dB Class E planning limit, the example retains about 3.8 dB. Reducing the cord total from 10 m to 4 m recovers roughly 2.4 dB. Removing two unnecessary connections recovers another 1.26 dB. This demonstrates why tidy topology and sensible cord lengths can be as important as the category printed on the cable jacket.
Assumptions and limitations of this attenuation estimate
The curves represent planning limits rather than measurements of a particular product. High-quality cable may perform better, while damaged, counterfeit or poorly terminated cable may perform worse. Temperature also matters. Copper resistance rises in hot spaces, so a channel designed with almost no margin at 20 °C may become marginal in a warm ceiling void.
Insertion loss alone cannot certify Ethernet. A field tester also evaluates return loss, near-end crosstalk, attenuation-to-crosstalk ratio, propagation delay, delay skew and other parameters. Alien crosstalk is especially important for 10GBASE-T. The calculator also cannot model local electromagnetic interference, crushed cable, excessive untwisting at a jack or an unsupported application.
Use the result to compare layouts and identify where the dB budget is being spent. For an installed link, confirm the design with the correct test limit, approved adapters and a recently calibrated certification instrument.
Ethernet attenuation questions
Why does insertion loss rise with frequency?
Skin effect confines more current to the outside of each conductor as frequency increases, raising effective resistance. Dielectric losses also increase. Together, those effects make the high-frequency end of an Ethernet test sweep the most demanding part of the insertion-loss curve.
Does a higher category automatically allow a longer channel?
No. Better cable can retain more electrical margin, but ordinary twisted-pair Ethernet remains subject to the physical reach specified for its application and channel class. Use fibre, an intermediate switch or a purpose-built extender when a route must exceed that reach.
How should service loops be counted?
Count every installed metre, including vertical drops, tray detours and service loops. The cable length is the routed path rather than the straight-line distance between rooms.
Sources: coefficient forms and channel concepts are based on the ANSI/TIA-568 series, ISO/IEC 11801 and twisted-pair channel requirements in IEEE 802.3.