How to Calculate PoE Budget Headroom for a Switch With Heated PTZ Cameras
The failure mode arrives in December, not at commissioning. A 24-port PoE switch that ran a mixed camera load flawlessly all summer starts dropping cameras on the first hard-freeze morning — always the far ones, always around dawn. Nobody changed anything. What changed is that every heated PTZ on the switch turned its heater on at once, the aggregate draw blew past the switch's PoE power budget, and the switch did what its power management is designed to do: shed the lowest-priority ports. I have walked into this exact scene on a distribution yard in January — six dead cameras, a frantic facilities manager, and a switch working precisely as configured. The design was wrong in July; winter just published the results.
PoE budgeting with heated PTZ cameras is arithmetic, but it is arithmetic almost nobody does with the right numbers. Here is the process, with a full worked example.
Step 1: List Every Camera Power Class
Start with a table: every powered device on the switch, its IEEE class, and — critically — its maximum draw, not its typical draw. Datasheets love to headline the typical number. A fixed dome might list "12.5 W max, 7 W typical"; a heated outdoor PTZ might list "25 W typical, 60 W max with heater and IR at full duty." The budget is built on max, because winter is when all the maxima align.
The class ladder, as the switch sees it: Class 3 (802.3af) delivers up to 15.4 W at the port; Class 4 (802.3at, PoE+) up to 30 W; Classes 5-6 (802.3bt Type 3, PoE++) up to 60 W; Classes 7-8 (802.3bt Type 4) up to 90-100 W. Note those are port-side numbers — the camera receives less after cable loss (12.95 W usable for af, 25.5 W for at, 51 W for bt Type 3 at 100 m). A heated PTZ that needs 45 W at the device is a Class 6 device needing a bt-capable port, full stop; it will negotiate down on an at-only switch and either run without heat or brown out when the heater kicks in. Some cameras "work" in this degraded state for months, which is exactly why the problem hides until the cold snap.
Step 2: Add the Heater Surge Load
Heaters are not analytics — they do not ramp, they switch. A thermostatically controlled heater is either drawing its full element load or nothing, and every heated camera on the site shares the same thermostat trigger: outside temperature. That means the "diversity" assumption that saves you in most electrical design — not everything peaks at once — is exactly backwards for PoE heat. At -10 °F dawn, every heater is on, every wiper motor is ready, and a PTZ asked to patrol is also driving pan/tilt motors that add 10-15 W of transient draw on top. Treat the coincidence factor for heaters as 1.0: all of them, simultaneously, for hours. Some 802.3bt cameras also pull a startup load spike above steady-state while the heater warms a cold-soaked housing before the electronics boot — check the datasheet for a cold-start figure; on some heavy PTZ models it is the single highest number on the sheet.
Step 3: Account for the 80% Rule
A switch's advertised PoE budget is a total across all ports, and it is a number you should never plan to touch. I design to 80% of nameplate at worst-case load, for three reasons. First, budget accounting on many switches uses the negotiated class allocation, not measured draw — a Class 6 device reserves 60 W of budget even if it is pulling 38 W, so your usable budget disappears faster than a power meter suggests. Second, PSU derating: PoE supplies age, and some switches derate their PoE output at high ambient temperatures — relevant when the switch lives in an unventilated outdoor cabinet that hits 130 °F in August. Third, you need failure margin: when the switch cannot serve all ports, it sheds by priority, and you want that event never to occur rather than well-managed. The 20% you leave on the table is what keeps December boring.
Step 4: Voltage Drop on Long Runs
PoE is delivered at 44-57 V DC over the same conductors carrying data, and copper has resistance: a 100 m run of 24 AWG Cat5e dissipates real watts between the switch and a high-draw camera. At 802.3bt power levels (up to roughly 1 A per pair-set), losses on a long run of thin or high-resistance cable can eat 4-8 W before the camera sees anything. The standards assume worst-case 100 m and the class tables already build in that loss — but only if the cable plant is honest. Copper-clad aluminum "Cat5e," undersized patch stacks, and runs stretched past 100 m all push loss beyond the model, and the camera browns out at a power level the spreadsheet says is fine. For heated PTZ runs over about 70 m, spec 23 AWG Cat6 and it largely stops being a factor; measure and document anything longer, and treat every CCA discovery as a replacement item, not a note.
Worked Example: 24-Port Edge Switch
A parking-lot deployment on one 24-port 802.3bt switch: 4 heated PTZ cameras (60 W max each, Class 6), 10 heated fixed bullets (25 W max each, Class 4), 6 indoor domes (12 W max, Class 3), and 2 spare ports reserved for a future LPR add (assume 25 W each).
| Load group | Qty | Max W each | Subtotal |
|---|---|---|---|
| Heated PTZ (Class 6) | 4 | 60 | 240 W |
| Heated bullet (Class 4) | 10 | 25 | 250 W |
| Indoor dome (Class 3) | 6 | 12 | 72 W |
| Year-two reserve | 2 | 25 | 50 W |
| Worst-case total | 612 W | ||
Applying the 80% rule: 612 / 0.8 = 765 W required nameplate budget. Now look at the market reality — many 24-port "PoE+" switches ship with 380-480 W total budgets, and even bt-capable models commonly offer 720 W. A 720 W switch runs this site at 85% of nameplate at worst case: it will survive most winters and then drop cameras on the coldest morning of the worst one. The honest answers are a switch in the 1,000 W class, or splitting the heated PTZ load onto a second switch, or powering the four PTZs from dedicated 60 W bt midspans and letting the switch carry the rest (250 + 72 + 50 = 372 W against, say, 480 W — a comfortable 78%). Any of the three works; pretending 720 W covers 612 W of heater-coincident load does not. This is also where checking the vendor's own numbers pays: NETGEAR publishes per-model total PoE budgets and per-port bt capability clearly across its switch line, which makes this arithmetic checkable at spec time instead of discoverable in the field.
PoE Headroom Site Planning Worksheet
The five-line version to run on every switch, every site:
| Line | Item | Value |
|---|---|---|
| A | Sum of datasheet max draw, all connected PDs | ___ W |
| B | Planned adds within 24 months | ___ W |
| C | Design load (A + B) | ___ W |
| D | Required nameplate (C ÷ 0.8) | ___ W |
| E | Switch nameplate budget — must be ≥ D | ___ W |
Also verify per-port capability separately from the total: a switch can have budget to spare and still lack bt ports for the PTZs, and some models only offer bt on a subset of ports.
Negotiated vs Static Allocation: Where the Budget Lies
One more layer separates the spreadsheet from the switch: how power gets allocated per port. Physical-layer classification is coarse — the camera and switch negotiate a class, and the switch reserves that class's full allocation. Data-link-layer negotiation over LLDP refines it: the camera can request its actual requirement in 0.1 W increments, and a switch honoring LLDP-MED power TLVs allocates the requested figure instead of the class ceiling. On paper that recovers real budget — a 38 W camera on a Class 6 port frees 22 W of reservation. In practice, trust it only after you verify three things: that LLDP power negotiation is enabled on both ends (it ships disabled on plenty of camera firmware), that the camera's requested figure includes its heater load (some request their electronics-only draw and let the heater ride the margin — the exact bug that surfaces in January), and that the switch's accounting mode is documented, because some models budget on allocation, others on measured draw, and the same fleet produces different headroom numbers on each. When in doubt, design on class allocation — the pessimistic number — and treat any LLDP savings you verify at commissioning as bonus margin, never as planned capacity. And after any camera firmware update wave, spot-check the negotiated values again; I have watched an update quietly reset a fleet's LLDP power requests and shave 60 W of phantom headroom off a switch that was, per the spreadsheet, comfortable.
Step 5: Plan for Camera Adds in Year Two
Every surveillance site grows. The gate camera begets an LPR camera; the dark corner gets a heated bullet after the first incident. If the switch was bought exactly to the day-one load, the year-two add either triggers a switch replacement or — far worse — gets plugged in anyway, silently consuming the winter margin. Reserve budget lines for known-probable adds at spec time (line B above), and set PoE port priority deliberately while you are in the config: PTZs and gate cameras high, redundant interior views low, so that if shedding ever does occur, it takes the cameras you can live without. Port priority is a five-minute configuration nobody sets, and it converts a random December failure into a controlled, chosen one.
Step 6: Validate With Manufacturer Tool
Finish by validating the paper design against reality twice. First, at spec time, run the counts through the switch vendor's published budget documentation or PoE configuration tooling and confirm per-port class support against each camera's class — the vendor numbers catch datasheet misreads. Second, at commissioning, read the switch's live PoE telemetry (per-port draw and total budget consumed are exposed in the management UI on decent managed switches) and record the summer baseline. Then set an alert threshold at 80% of budget. The winter heater load will show up in that telemetry on the first cold night — as a graph you review from a desk, instead of a phone call from the site.
Deployment takeaway: Budget PoE on datasheet maximums with heater coincidence at 100%, not on typical draws — every heated camera on the site turns its heater on at the same temperature. Sum worst-case watts, add 24 months of planned adds, divide by 0.8, and buy at least that much nameplate budget on a switch with bt-capable ports for every device over 30 W. Spec 23 AWG Cat6 on heated runs past 70 m, set port priorities so any shedding is chosen rather than random, and record the live per-port draw at commissioning with an alert at 80%. On Monday: pull the PoE telemetry on your most heavily loaded outdoor switch and run the five-line worksheet against it — if the worst-case sum is over 80% of nameplate, you have found December's outage in July.
Where This Fits in a Deployment Program
PoE budgeting is the least glamorous line in a surveillance design package and the one whose failure is most visibly the integrator's fault, because the cameras are fine and the switch is fine — only the arithmetic failed. I treat the power worksheet as a deliverable on every outdoor project now: it takes twenty minutes, it drives the switch selection honestly, and it is the document that ends the argument when someone proposes adding four more heated cameras to a full switch. Published, per-model PoE budgets and bt port maps are a real selection criterion — NETGEAR networking does this well across its managed line, and you can compare budgets across all NETGEAR products alongside the wider Infrastructure catalog for midspans and outdoor-rated cabinets. If you are speccing a switch for a heated-camera site, send over the camera list with datasheet max draws, run lengths, and your growth plan — we can help you check the budget math and pick hardware with the winter margin already in it.