[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-poe-power-budget-and-loss-calculation":3},{"id":4,"slug":5,"title":6,"category":7,"summary":8,"tags":9,"publishTime":13,"views":14,"seoTitle":15,"seoDescription":16,"seoKeywords":17,"content":18},4,"poe-power-budget-and-loss-calculation","PoE Power Budget and Loss Calculation","engineering","Understand PoE power budgeting and cable loss: PSE budget, I-squared-R losses, distance derating, and a worked example you can reuse.",[10,11,12],"power budget","cable loss","PoE engineering","2026-03-22",10,"PoE Power Budget and Cable Loss Calculation Guide","Calculate PoE power budgets and cable loss with I-squared-R, distance derating and a worked PoE+ example for accurate network design.","PoE power budget, PoE cable loss, I squared R, PoE distance derating, PoE calculation, PSE budget","\u003Cp>A PoE deployment only works if the power that reaches each device, after cable losses, still exceeds what that device needs. This article explains how to budget PoE power at the switch and how to estimate how much is lost in the cable, with a worked example engineers can adapt.\u003C\u002Fp>\u003Cimg src=\"\u002Fbrand\u002Fnet\u002Fethernet-switch.jpg\" alt=\"ethernet switch\" loading=\"lazy\" \u002F>\n\u003Ch2>Two Budgets to Track\u003C\u002Fh2>\n\u003Cp>There are two distinct budgets in any PoE design. The first is the \u003Cstrong>switch (PSE) power budget\u003C\u002Fstrong> - the total wattage the switch can source across all ports simultaneously. A switch may have 24 ports but only enough power supply to run, say, 370 W total, so you cannot assume every port runs at maximum. The second is the \u003Cstrong>per-link budget\u003C\u002Fstrong> - how much of a port's sourced power actually arrives at the device after the cable takes its share.\u003C\u002Fp>\n\u003Ch2>Where the Power Goes\u003C\u002Fh2>\n\u003Cp>The IEEE standards already account for worst-case cable loss in their headline figures. That is why an 802.3af PSE sources 15.4 W but only guarantees 12.95 W at the PD - the difference is reserved for the cable. The same logic gives 30 W to 25.5 W for Type 2, 60 W to 51 W for Type 3, and 90 W to 71.3 W for Type 4. In percentage terms, roughly 15-20% of sourced power can be dissipated as heat in a full 100 m run at the higher tiers.\u003C\u002Fp>\n\u003Cimg src=\"\u002Fbrand\u002Fnet\u002Fcoiled-cables.jpg\" alt=\"coiled cables\" loading=\"lazy\" \u002F>\u003Ch2>The Loss Equation\u003C\u002Fh2>\n\u003Cp>Cable loss is resistive heating, governed by P = I²R, where I is the current per conductor and R is the round-trip (loop) DC resistance of the powered pairs. Two facts follow directly:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Loss rises with the \u003Cstrong>square\u003C\u002Fstrong> of current, so doubling current quadruples the heat.\u003C\u002Fli>\n\u003Cli>Loss rises linearly with cable length and with conductor resistance, so thinner wire (higher AWG number) and longer runs both increase loss.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>A typical Cat5e (24 AWG) conductor is about 8.3 Ω per 100 m, giving a powered-pair loop resistance near 16.6 Ω for two-pair PoE (or about half that when four pairs share the load in 802.3bt).\u003C\u002Fp>\n\u003Ch2>Reference Loss Figures\u003C\u002Fh2>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\u003Cth>Class \u002F Type\u003C\u002Fth>\u003Cth>PSE sources\u003C\u002Fth>\u003Cth>Delivered to PD\u003C\u002Fth>\u003Cth>Approx. cable loss\u003C\u002Fth>\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\u003Ctd>Type 1 (af)\u003C\u002Ftd>\u003Ctd>15.4 W\u003C\u002Ftd>\u003Ctd>12.95 W\u003C\u002Ftd>\u003Ctd>~2.45 W\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Type 2 (at)\u003C\u002Ftd>\u003Ctd>30 W\u003C\u002Ftd>\u003Ctd>25.5 W\u003C\u002Ftd>\u003Ctd>~4.5 W\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Type 3 (Class 6)\u003C\u002Ftd>\u003Ctd>60 W\u003C\u002Ftd>\u003Ctd>51 W\u003C\u002Ftd>\u003Ctd>~9 W (~15%)\u003C\u002Ftd>\u003C\u002Ftr>\n\u003Ctr>\u003Ctd>Type 4 (Class 8)\u003C\u002Ftd>\u003Ctd>90 W\u003C\u002Ftd>\u003Ctd>71.3 W\u003C\u002Ftd>\u003Ctd>~18.7 W (~20%)\u003C\u002Ftd>\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\u003Cimg src=\"\u002Fbrand\u002Fnet\u002Fswitch-cabling.jpg\" alt=\"switch cabling\" loading=\"lazy\" \u002F>\u003Ch2>Worked Example\u003C\u002Fh2>\n\u003Cp>Suppose a Type 2 (PoE+) port sources power over 100 m of Cat5e to a 22 W access point. Per the standard, the PD is guaranteed 25.5 W at 100 m, so 22 W is within budget with about 3.5 W margin - acceptable. Now estimate the loss directly: at 25.5 W usable and roughly 50-55 V at the PD, the current is approximately 0.45-0.5 A per conductor. With a two-pair loop resistance near 12.5 Ω for the powered pairs, P = I²R gives on the order of 4-4.5 W dissipated in the cable - consistent with the 30 W to 25.5 W reservation. Shorten the run to 50 m and the loop resistance halves, so cable loss roughly halves and more power reaches the device.\u003C\u002Fp>\n\u003Ch2>Distance Derating and Design Margin\u003C\u002Fh2>\n\u003Cp>Because loss scales with length, the available power at the PD falls as runs approach 100 m. Practical guidance:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Keep critical high-power runs short; reserve the full 100 m for lower-power endpoints.\u003C\u002Fli>\n\u003Cli>Use lower-gauge cable (23 AWG Cat6 instead of 24 AWG Cat5e) to cut resistance by roughly 17% per conductor.\u003C\u002Fli>\n\u003Cli>Prefer 802.3bt four-pair delivery for high-wattage loads - spreading current across four pairs lowers I²R loss versus two pairs.\u003C\u002Fli>\n\u003Cli>Always leave 10-20% switch-budget headroom for surge, classification rounding, and future endpoints.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Putting It Together\u003C\u002Fh2>\n\u003Cp>Sound PoE budgeting means confirming the switch can source the aggregate load, then verifying each link delivers enough after distance and gauge are accounted for. Our custom PoE power modules are specified with these losses in mind, sizing the converter and conductors so the rated voltage and current are dependable at the far end of a real-world cable run.\u003C\u002Fp>"]