The Physics of PoE Power Delivery
Engineering analysis of DC Loop Resistance, Thermal Dissipation, and IEEE 802.3bt Implementation Strategy.
PoE Power Grid Matrix
Verify DC power delivery for Type 1-4 PoE devices. Model real-world voltage drop across various cable gauges, distances, and thermal environments.
PoE Voltage Drop Calculator
Calculate voltage loss over Ethernet cable runsVoltage Drop Profile
Voltage drop increases linearly with cable length. The chart shows how voltage loss varies at 0.6A load current across typical installation distances. AWG 23 (thicker) maintains lower voltage drop than AWG 24.
Engineering Principles
Power over Ethernet voltage drop depends on cable resistance, which is affected by wire gauge (AWG), length, and temperature. Lower AWG numbers indicate thicker conductors with lower resistance and better voltage preservation over distance.
Loop resistance includes both conductors in the twisted pair.
Thermal Advisory
High ambient temperatures increase conductor resistance, potentially worsening voltage drop beyond calculated values.
Technical Standards & References
"You are our partner in accuracy. If you spot a discrepancy in calculations, a technical typo, or have a field insight to share, don't hesitate to reach out. Your expertise helps us maintain the highest standards of reliability."
Contributors are acknowledged in our technical updates.
The Ethernet Power Paradigm: Beyond Data
Power over Ethernet (PoE) has evolved from a convenience for low-power VoIP phones into a critical industrial utility powering 90W digital displays, high-performance Wi-Fi 7 access points, and remote IoT sensors. This evolution, codified in the IEEE 802.3bt (PoE++) standard, pushes the physics of copper cabling to its thermal and electrical limits.
Unlike traditional power wiring, Ethernet uses 24 or 23 AWG twisted pairs designed for high-frequency signal integrity, not heavy current loads. Understanding the voltage drop dynamics is no longer just an electrical formality—it is a mandatory step in ensuring network uptime and preventing catastrophic thermal failure in high-density cable trays.
The Ohm's Law Matrix
The fundamental principle governing PoE power loss is Ohm's Law. As current flows through the copper conductor, it encounters resistance. The resulting voltage drop is directly proportional to both. In a PoE circuit, we must consider the Loop Resistance, which accounts for the current traveling out to the device and back through the return path.
The Joulean Loss Equation
Energy isn't just lost—it's converted to thermal radiation. Every Watt lost to resistance increases the internal temperature of the cable bundle.
In the equation above, $\alpha$ is the Temperature Coefficient of Resistance for copper ($0.00393$ per $^\circ C$). This reveals a dangerous feedback loop: as the cable heats up due to current flow, its resistance increases, leading to even higher voltage drop and more heat generation. This "Thermal Runaway" is why TIA-184-A strictly limits temperature rise in cable bundles to 15$^\circ$C.
Standards Evolution: 802.3af to 802.3bt
The IEEE 802.3 standards body has progressively increased power delivery capabilities by optimizing how current is distributed across the 8 conductors of an Ethernet cable.
Uses 2 pairs (4 wires). Standard for VoIP and basic IP cameras. Max loop resistance: 20 $ \Omega $.
Uses 4 pairs (8 wires). Supports 802.11ac APs and small DPU nodes. Significantly improved thermal efficiency.
The Industrial Limit. Requires AWG 23 and precise thermal management in bundles to avoid fire risks.
Critical Engineering Note: While 802.3bt provides 90W at the PSE source, the standard only guarantees 71.3W at the PD input after 100 meters of worst-case resistance. If your simulation shows arrival power below this, your cabling infrastructure is likely sub-standard or overheating.
The CCA Safety Crisis
The rise of Copper Clad Aluminum (CCA) cable in low-cost consumer markets represents the single greatest threat to PoE infrastructure. CCA cables look identical to pure copper but consist of an aluminum core with a thin copper wash.
Troubleshooting DC Power Dynamics
When a PoE device fails to power up or cycles power (Boot-Looping), engineers should follow a structured diagnostic protocol:
Measure Static Resistance
Use a DC ohm-meter to verify pair resistance. For Cat6a at 100m, you should see roughly $7 \Omega$ per conductor ($14 \Omega$ loop). If it's significantly higher ($25 \Omega+$), you have a bad punch-down or CCA cable.
Monitor Inrush Current
Devices draw a massive spike of current when capacitors charge at boot. If your voltage is marginal, this inrush will trigger a voltage dip that reboots the PSE port before the device even initializes.
Thermal Bundle Audit
If performance degrades during the day, check cable tray temperatures. A target of 45$^\circ$C is standard, but in unconditioned warehouse spaces, resistance can spike 15% due to ambient heat alone.
Industrial Use-Case: Smart Factory Vision Links
In a Tier 1 automotive factory, 48 machine-vision cameras were powered via 802.3bt (60W) over 90-meter runs. The initial installation used standard 24 AWG Cat5e. Within 6 months, the factory reported erratic "Video Ghosting" and frame loss during high-speed shifts.
Voltage at the camera dropped to 39.5V during peak processing. The camera's DPU throttled frequency to stay within power limits, causing the frame drop.
Re-cabled with **22 AWG Cat7a (S/FTP)**. Voltage stabilized at 48V. Power loss in the cable dropped from 12.5W per run to 4W, reducing HVAC costs in the IDF.
Technical Standards & References
External Standards
Engineering Math
Calculations are based on the **International Electrotechnical Commission (IEC) 60287** standard for cable sizing and thermal modeling. Resistance values assume standard annealed copper at 20$^\circ$C reference.
PoE++ Budgeting for PTZ Cameras and Access Points
Power over Ethernet (PoE) budgeting for high-consumption endpoints such as pan-tilt-zoom (PTZ) cameras with integrated heaters, Wi-Fi 7 (802.11be) access points with 16 spatial streams, and digital signage displays requires precise calculation of the power allocation margin beyond the IEEE 802.3bt Type 4 (Class 8) maximum of 90 W at the PSE (Power Sourcing Equipment). The IEEE 802.3bt-2018 standard defines the maximum delivered power at the PD (Powered Device) as 71.3 W for Type 4 (Class 8), with the 18.7 W delta representing the worst-case cable loss over 100 meters of Category 6A cable at 50°C ambient. However, a typical PTZ camera with an integrated IR illuminator, motorized zoom, and windshield heater can draw 75-90 W at full load (all subsystems active simultaneously), placing it above the 71.3 W Type 4 PD limit. The solution is either: (1) a dual-feed PoE design where the PTZ camera has two RJ45 ports each receiving Type 4 power (2× 71.3 W = 142.6 W max, with the camera orchestrating load sharing across both power interfaces per the IEEE 802.3bt dual-signature PD specification), or (2) a 60 W power injector with local power assistance, where the camera supplements PoE power with a local 24 V AC supply for the heater only. Our PoE budget model includes a Dual-Signature Load Balancer that accepts the PD's per-function power requirements (heater: 30 W, IR illuminator: 15 W, motor: 12 W, camera electronics: 18 W, total: 75 W) and determines whether a single Type 4 PSE port suffices (for short cables under 50 m where cable loss is under 3 W) or whether dual-feed or local power assist is required.
The cable power loss estimation under load is the most frequently underestimated variable in real-world PoE deployments. Per IEEE 802.3bt, the worst-case pair resistance for Category 5e/6 cable at 50°C is 12.5 Ω per 100-meter pair (using all four pairs for Type 3/4). The power dissipated in the cable for a 90 W PSE output is: P_loss = I² × R_cable = (90W/55V)² × 12.5Ω = (1.636 A)² × 6.25Ω = 16.7 W (since the current splits across two pairs in the Type 4 configuration at 55 V nominal PSE voltage). The actual PD-received power = 90 - 16.7 = 73.3 W, consistent with the Type 4's 71.3 W minimum PD power specification (the additional 2 W margin accounts for connector losses and manufacturing tolerance). For a 90-meter Category 6A run in a hot attic space at 60°C (common for PTZ cameras mounted on building exteriors with cable runs through unconditioned spaces), the pair resistance increases by 15.6% (copper's temperature coefficient: 0.393%/°C × 40°C above 20°C reference), raising R_cable to 14.45 Ω and P_loss to 19.3 W. At this cable temperature and length, the PD receives only 70.7 W—below the 71.3 W minimum for Class 8 PDs, potentially causing the PD to power-cycle or enter a reduced-functionality mode. Our voltage drop model incorporates the Ambient Temperature Derating Factor (ATDF) from the NEC Table 310.15(B)(16) for the cable type and the conduit fill ratio, adjusting the per-meter cable resistance for the localized ambient temperature along the cable path, and computing the worst-case PD-received power for the given installation environmental conditions.
The PSE power budget allocation across multiple high-power PoE ports on a switch is constrained by the switch's internal power supply capacity, not just the per-port Type 4 limit. A typical 48-port PoE++ switch with a 1,200 W internal power supply can deliver at most 25 W per port on average when all 48 ports are active (since 48 × 25 = 1,200 W). This means that while any single port can deliver 90 W (Type 4), only 13 ports can operate at 90 W simultaneously without exceeding the power supply's 1,200 W capacity. The Power Allocation Priority per port (critical, high, low per IEEE 802.3bt power classification) determines which ports are powered first when the total PSE power demand exceeds the supply capacity. Our budgeting model implements the Power Shedding Algorithm: when the aggregate PD power demand exceeds the PSE supply (after accounting for the PSE's internal power conversion overhead of 10-15%), ports with "low" priority are power-cycled (PoE disabled for 100 ms, then re-enabled in a staggered sequence) until the total demand falls below the supply threshold. For a floor switch powering 10 PTZ cameras (90 W each), 20 Wi-Fi 7 APs (60 W each), and 18 standard VoIP phones (15 W each), the total demand is 10×90 + 20×60 + 18×15 = 900 + 1,200 + 270 = 2,370 W, exceeding a 1,200 W PSE by 1,170 W (97.5% oversubscription). The power-shedding sequence disables the 10 PTZ cameras (lowest priority) first, reducing demand by 900 W to 1,470 W, still exceeding the PSE capacity, then disables 5 of the 20 Wi-Fi 7 APs (300 W), bringing the total to 1,170 W—within the PSE's 1,200 W limit.
The midspan power injector alternative for high-power PDs bypasses the switch PSE limitations entirely. A single-port 90 W midspan injector (IEEE 802.3bt Type 4 compliant) costs $30-60 and connects inline between the switch port and the PD, drawing power from a local AC outlet rather than from the switch's internal PSE budget. For a dense deployment of 10 PTZ cameras requiring 90 W each, dedicated midspan injectors add $300-600 in material cost per switch location but eliminate the PSE oversubscription risk entirely. However, each injector represents an additional point of failure (the injector's power supply has an MTBF of 50,000-100,000 hours, compared to 300,000+ hours for the switch PSE) and adds 3-5 W of standby power loss per injector (the injector's own AC-to-DC conversion efficiency is 85-90%). Our tool compares the Total Cost of Ownership for three power delivery strategies: (1) oversubscribed PSE with power-shedding risk (lowest CAPEX, highest operational risk), (2) higher-capacity PSE (e.g., dual 1,200 W PSUs in a switch that supports redundant power supplies, increasing CAPEX by $500-1,000 but eliminating oversubscription), and (3) midspan injectors (medium CAPEX, moderate operational risk from injector failure). The TCO model includes the power cost of each approach over a 5-year horizon at $0.12/kWh, enabling the network architect to select the power delivery topology that minimizes both the capital expenditure and the operational risk of PD power starvation under full-load conditions.
Related Engineering Resources
"You are our partner in accuracy. If you spot a discrepancy in calculations, a technical typo, or have a field insight to share, don't hesitate to reach out. Your expertise helps us maintain the highest standards of reliability."
Contributors are acknowledged in our technical updates.
Explore Physical Layer Tools
PoE Class Negotiation and Power Allocation Under Dynamic Load
Power over Ethernet (PoE) class negotiation, defined in IEEE 802.3bt-2018 (PoE++), determines the maximum power that the PSE (Power Sourcing Equipment — typically a PoE switch) allocates to each PD (Powered Device). The negotiation occurs during the hardware classification phase, which takes place before the PSE applies full operating voltage to the pairset. The PSE briefly applies a 15-20V classification voltage and measures the current drawn by the PD, which indicates the PD's class. IEEE 802.3bt defines 9 power classes: class 0 (15.4W default), class 1 (4.0W), class 2 (7.0W), class 3 (15.4W), class 4 (30.0W, Type 2), class 5 (45.0W, Type 3), class 6 (60.0W, Type 3), class 7 (75.0W, Type 4), and class 8 (90.0+W, Type 4). The class negotiation determines the PSE's per-port power budget: a PSE with a total power budget of 720W (24-port switch, 30W per port for Type 2) allocates 30W per port for class 4 PDs, even if the PD only draws 10W under typical load. The power allocation is not dynamically adjusted — the PD is classified once during initial handshake, and the PSE's power budget is statically partitioned according to the detected class. This static allocation leads to a significant over-provisioning problem: if all 24 ports detect class 4 (30W), the PSE reserves 24 × 30W = 720W of its 720W budget, leaving zero headroom for a 25th PD. If the actual average power draw of the 24 PDs is only 10W (a typical Wi-Fi 6 AP's maximum load under full traffic), the PSE reserves 3× the actual power, and the 25th port cannot be powered even though the PSE has 720 − 24 × 10 = 480W of physical headroom. The PoE voltage drop tool's power allocation model accepts the PSE's total power budget and per-class port count, and it computes the "budget utilization" U = Σ(N_class × P_class) / P_total, where P_class is the per-class allocated power (not the actual PD draw). When U exceeds 90%, the tool recommends reclassifying low-draw PDs to a lower class via the LLDP (Link Layer Discovery Protocol) power negotiation (IEEE 802.3bc-2016) — the protocol extension that allows the PSE and PD to negotiate a power allocation below the hardware class maximum.
The LLDP-MED (Media Endpoint Discovery) power negotiation, defined in ANSI/TIA-1057, provides the mechanism for dynamic power renegotiation after the initial hardware classification. An LLDP-MED-capable PSE sends a `MED Power Type = PD` TLV with the `Power Requested` field set to the PD's actual power requirement (in tenths of a watt, 0.1W resolution). The PSE responds with a `MED Power Type = PSE` TLV indicating the `Power Allocated` (the power the PSE is committing to that port). The negotiation can be re-triggered at any time: if a PD's power requirement changes (e.g., a PTZ camera starts moving its motors, increasing its draw from 8W to 25W), the PD can send a new LLDPMED power request, and the PSE can either grant the new allocation (if its budget allows) or deny it (maintaining the previous allocation and expecting the PD to stay within it). The LLDP-MED exchange takes approximately 30-100 ms (the LLDP transmission interval is 5-30 seconds by default, but can be reduced to 1 second with `lldp fast-start` on most PoE switches). The voltage drop tool's LLDP model simulates this dynamic renegotiation under various load scenarios: for N = 16 PDs that simultaneously increase their power demand by 2× (e.g., a coordinated alarm event triggering all cameras to IR mode), the PSE's budget must accommodate Σ(P_original × 2) = 2 × P_budget, causing a budget oversubscription if the original LLDP allocations were at the hardware class limits. The tool computes the probability of budget oversubscription P_oversub = P(Σ(P_requested) > P_total) using a Poisson arrival model for the PD power-increase events (λ = number_of_PDs / mean_time_between_power_changes). For N = 48, P_total = 720W, mean_time_between_changes = 60 seconds, and P_requested_mean = 15W per PD, P_oversub = P(N(48 × 15, 48 × 225) > 720) = P(N(720, 10800) > 720) = 50% — meaning half of all alarm events trigger a budget oversubscription, causing some PDs to be denied their requested power increase.
The cable length's effect on the effective power delivered to the PD — specifically, the I²R loss in the Ethernet cable — interacts with the LLDP-negotiated power allocation in a way that can cause PD brownouts at long cable lengths even when the PSE's per-port power allocation appears adequate. The IEEE 802.3bt standard specifies that the PSE must deliver at least P_class × 0.9 at the PD's input (the 10% allocation for cable loss), but this margin is calibrated for the worst-case 100-meter cable at the maximum current for the class. For class 8 (90W at the PSE), the minimum PD input power is 81W, implying a maximum cable loss of 9W. At 100 meters of CAT6A cable (12.5 Ω loop resistance), the current I = P_PSE / V_PSE = 90W / 55V (the highest 802.3bt voltage) = 1.64A, and the cable loss P_cable = I² × R_cable = 1.64² × 12.5 = 33.6W — well above the 9W margin. The standard's margin is based on the PD voltage (50V at the PSE output), giving I = 90 / 50 = 1.8A and P_cable = 1.8² × 12.5 = 40.5W — still far above 9W. The 9W margin is achieved only if the PSE supplies 55V and the PD draws 90W at that voltage (requiring the PD's DC-DC converter to have at least 55V × I output), which is not a guaranteed operating condition. The voltage drop tool's cable loss model computes the actual P_cable for the user's cable length and type and subtracts it from the PSE's power allocation to determine the power at the PD. If P_PD < 0.9 × P_class, the tool flags the port as "risk of brownout" and recommends either increasing the PSE's power allocation via LLDP (if the PSE budget permits) or reducing the PD class assignment so that the cable loss margin is proportional to the actual PSE output current: the solution is to negotiate a lower power allocation via LLDP (e.g., class 6 instead of class 8) to reduce the I²R loss, accepting 60W maximum at the PD instead of 90W, but delivering a stable 60W rather than an unreliable 90W that browns out when the IR camera heaters activate.