Why PoE Switch Sizing Goes Wrong at Three Watts
The most common PoE sizing mistake is also the most invisible. An integrator reads the switch spec sheet, sees "720W total PoE budget", divides by camera count, and assumes the deployment is safely under capacity. The math is correct. The conclusion is wrong, and the field symptoms — random camera reboots during cold mornings, sporadic PTZ stalls, a switch that runs hot only when it rains — all trace back to the same gap. PoE budget is a sum, but PoE delivery is per port, and the difference between the two is where deployments fail.
The Budget vs Per-Port Distinction
A 24-port PoE+ switch with a 720W total budget can deliver up to 30W per port on PoE+ ports, or up to 60W on PoE++ ports if the switch supports it. The 720W figure is the ceiling for all ports combined under steady-state load. What the spec sheet rarely makes explicit is the per-port headroom — how much margin exists between what each port is currently delivering and what it can deliver under inrush, surge, or cold-start conditions.
When a PoE-powered device boots, it draws more current for 100 to 300 ms than its steady-state rating. A camera rated at 12.5W can briefly pull 18W during boot. If the switch is at 90% budget utilization across all ports, the simultaneous boot of three cameras after a brief power blip can push instantaneous draw above the switch's PSU rating. The switch's overload protection kicks in, drops PoE on one or more ports, and the cameras reboot in a cycle until the load disperses.
Heater Load: The Hidden Multiplier
Outdoor PTZ cameras with integrated heaters change the sizing calculation in a way most integrators don't anticipate. A PTZ camera with a heater might pull 12W in normal operation and 28W when the heater is active. In a region where overnight temperatures hit -10°F, the heater can be active continuously from 11 PM to 9 AM. During that 10-hour window, the switch is delivering more than 2x the load it sees during the day. If the switch was sized for the daytime load with a comfortable margin, the nighttime load can put it over capacity — and the sizing report from the integration design phase looked fine.
Quick PoE Headroom Field Worksheet
Before deploying a PoE-heavy site, walk this worksheet. Each row maps to a measurement you can take on the bench or pull from manufacturer specs.
| Measurement | What It Tells You |
|---|---|
| Switch total PoE budget | The ceiling. Stays the same across operating conditions. |
| Per-port maximum delivery (PoE+/PoE++) | What each port can deliver alone. 30W is PoE+, 60W is PoE++. |
| Steady-state load per port (measured) | What each device actually draws under nominal conditions. |
| Boot inrush per port (measured) | Peak draw during the first 300 ms of device power-on. |
| Heater load contribution (where applicable) | Additional load when the device's heater is active. |
| Total instantaneous draw under cold-start scenario | Sum of all per-port loads plus all heater contributions plus 25% inrush margin. |
| Headroom (switch budget minus instantaneous draw) | Should be at least 20% of the switch budget. Below 15% is a deployment risk. |
Why "Class 4" Labels Don't Tell You Enough
PoE class labels — Class 0 through Class 8 — describe the power range a device negotiates with the switch, not what it actually draws. A Class 4 PoE+ camera negotiates for up to 25.5W, but its actual steady-state draw might be 9W. The switch reserves the full Class 4 budget against that port whether the device uses it or not. So a 24-port switch with twenty Class 4 cameras has reserved 510W out of its 720W budget even if the actual aggregate draw is closer to 200W. The remaining 210W of reserved-but-unused capacity is what gets squeezed when another Class 4 device joins the network or when one of the existing devices spikes.
Deployment takeaway: Size PoE switches against per-port peak draw plus heater load plus 25% inrush margin, not against the steady-state aggregate. A switch that looks 60% utilized on paper can be 95% utilized at 3 AM in February.
The PoE++ Sizing Trap
PoE++ (Type 4, up to 90W per port) was introduced largely for PTZ cameras with integrated wipers, deicers, and high-wattage IR illuminators. Switch vendors market the per-port maximum prominently — "up to 90W per port" — without making clear that a switch with a 1,440W total budget cannot deliver 90W on more than 16 of its 48 ports simultaneously. An integrator who specs a 48-port PoE++ switch assuming all 48 ports can deliver 90W is off by a factor of three.
The fix is to read the switch's specifications for both the per-port maximum and the total PoE budget, then divide the budget by the number of ports you plan to run at the per-port max. For deployments where only a subset of ports need full PoE++, this is fine — but the design has to be deliberate about which ports go where on the patch panel, because the high-draw devices need to be distributed across the switch backplane, not concentrated on adjacent ports that may share a power rail.
Cable Length and Voltage Drop
The other half of PoE sizing that the budget number doesn't capture is voltage drop over cable length. PoE+ delivers 48V at the switch port, but by the time the signal reaches a camera at the end of a 100-meter run, voltage has dropped — sometimes to 38V or below if the cable is older Cat5e with high resistance. The camera negotiates with the switch using class signaling, but the actual power delivered depends on the loop resistance. Cameras run on lower voltage by drawing more current, and the switch sees higher per-port draw on cameras that are at the far end of long cable runs than on cameras connected via short patch cables in the IDF.
This is why cold mornings affect PoE-heavy deployments disproportionately. Lower temperatures increase copper resistance modestly — about 0.4% per °C below 20°C — but on a 100-meter run that's already near the voltage drop threshold, the additional resistance can push a camera below its minimum operating voltage. The camera browns out, reboots, and the cycle repeats until ambient temperature climbs.
Designing for Verifiable PoE Headroom
The practice that catches PoE sizing problems before they reach the field is the bench-validate-then-size approach. On the bench, with the actual switch and the actual cameras the deployment will use, measure peak draw during boot, steady-state draw during operation, and incremental draw when the heater (if applicable) is active. Use those measured values — not the spec sheet values — as the inputs to the headroom worksheet above.
For deployments with 25+ PoE devices on one switch, also bench-test the cold-start scenario: power the switch off, then back on, with all cameras connected. The first 5 seconds of power-up are where inrush spikes happen, and if the switch can't handle simultaneous boot, you'll see ports stay dark until the switch sequences them up. A well-sized deployment boots all ports in under 15 seconds with no PoE drops. A marginal deployment shows a stair-step pattern where ports come up in waves of 4 or 5 at a time.
Where This Fits in a Deployment Program
PoE sizing is one of those engineering inputs that determines deployment durability years out. A switch sized correctly for the original camera count and load handles the field upgrades — PTZ retrofits, heater additions, IR illuminator upgrades — without needing replacement. A switch sized to the spec-sheet aggregate gets replaced 18 months into deployment when the first PTZ retrofit pushes it over capacity.
The good infrastructure choices in this space — managed PoE++ switches from NETGEAR and others — give you per-port draw telemetry through the management console. Pull that telemetry weekly during the first 90 days of a deployment, look at the peak per-port draw and the time-of-day pattern, and adjust which ports carry which devices based on the actual data, not the design-phase estimate. The PoE switch catalog separates models by per-port maximum and total budget so the sizing math is explicit before purchase.
Monday morning, pull the per-port PoE draw history from your managed switch for the past 30 days. Find the port with the highest peak draw and the highest peak-to-steady-state ratio. That port is your weakest link — and the data tells you whether you need to redistribute load, upgrade the cable, or replace the camera before it browns out on the next cold night.