Small-business network triage · September 6, 2026
When a security camera, desk phone, or Wi‑Fi access point keeps rebooting, the network cable may be carrying data perfectly while failing at its second job: delivering enough power. Before replacing endpoints or factory-resetting a switch, check the Power over Ethernet budget, the port’s standard and allocation, the device’s peak demand, and the cable path.
Guidance checked September 6, 2026, at 1:11 p.m. Eastern / 17:11 UTC. Menus and power-allocation behavior vary by switch model and firmware; use the exact datasheets and live telemetry for your equipment.
Fast answer: A PoE switch has both a total wattage budget shared across active PoE ports and a per-port limit.[1][2] A device can lose power even while Ethernet link lights still appear normal if the total budget is depleted, the port cannot supply the required PoE class, a startup surge crosses a limit, or the switch detects an electrical fault.[1][2][3]
The costly problem: intermittent power can look like five different failures
PoE powers devices such as IP phones, wireless access points, and network cameras over the same copper Ethernet cable that carries their data. Standards-based equipment commonly uses IEEE 802.3af (PoE), 802.3at (PoE+), or 802.3bt (PoE++); a switch that supports only an older/lower-power mode cannot automatically satisfy every newer, higher-power device.[2]
The symptoms are expensive because they scatter: a camera records gaps after dark, a phone restarts during a call, an access point disappears when several clients connect, one door controller stays offline after a switch reboot, or the last-added device never powers up. Replacing the endpoint may not fix any of those if the shared power source is the bottleneck.
Cisco’s troubleshooting guidance says to compare connected powered devices with available PoE budget, review port and power-supply status, inspect logs, test a known-good port and cable, and consider startup current when a device powers briefly and then disconnects.[1] That is a better order than randomly rebooting equipment.
Know the four numbers that matter
Phone readers: swipe the table left to see every power check.
| Number | Where to find it | Why it matters | Common mistake |
|---|---|---|---|
| Total PoE budget | Switch datasheet and live management page/CLI | The wattage pool shared by all active PoE ports. NETGEAR defines the budget as the total power the switch can supply across active ports.[2] | Counting PoE-capable ports instead of available watts. |
| Available, used, and remaining power | Live switch telemetry | Shows whether today’s power supply, redundancy mode, and connected load leave enough capacity. Cisco recommends checking available, used, and per-port power rather than relying on the model name alone.[1] | Assuming the brochure maximum equals the current installed power configuration. |
| Per-port standard/class and limit | Switch port details plus endpoint datasheet | PoE, PoE+, and PoE++ expose different power ranges; the powered device and switch port must be compatible.[2] | Moving a high-power device to another port without checking whether that port supports the same mode. |
| Device maximum demand | Endpoint datasheet; observed peak when supported | Axis distinguishes typical consumption from maximum consumption.[4] Features such as IR illumination, heaters, motors, fans, audio, storage, and multiple streams can participate in the maximum scenario.[4] | Budgeting from a daytime “typical” reading and missing the nighttime peak. |
Stop before changing power limits, forcing nonstandard modes, or attaching an injector. Practical safety guidance: do not experiment with an undocumented voltage, mode, or power source; verify compatibility in the exact switch, injector, and powered-device documentation first.[2][3] Photograph the current configuration and identify every exact model. Practical continuity guidance: if the device controls a door, alarm, phone/911 path, point-of-sale function, or required camera view, arrange a safe service window before intentionally removing power.[1]
An eight-step fix, from observation to change
Freeze the evidence before rebooting
Record the exact device, port, time, LED state, management warning, and what else was happening. Save screenshots of PoE available/used/remaining power, each port’s class or allocation, and recent switch logs. A reboot can clear the pattern you need to diagnose.
Separate data-link trouble from power trouble
Check whether a known-good non-PoE device can establish an Ethernet link through the same cable and port. Cisco recommends this distinction, then testing the powered device on a known-good PoE port and—when practical—with a short known-good cable connected directly to the switch.[1] A working data link does not prove the full cable path can deliver adequate power under load.[1][4]
Identify the real power source
Do not stop at “24-port PoE switch.” Record the exact model, installed power supply or external brick, redundancy mode, stack member, and any midspan/injector. Some platforms let the same switch chassis operate with materially different available budgets depending on the power supply and configuration.[3]
Build a port-by-port load sheet
For each powered device, write the port, model, PoE standard/class, datasheet maximum, and current allocation or observed draw. Sum the allocated or maximum planning values that your switch actually uses—not only momentary typical readings. NETGEAR notes that switches account for the class and may disable lower-priority ports when total requirements exceed the budget.[2]
Read the denial or fault reason
Look for messages equivalent to insufficient power, denied power, overdraw, overload, short circuit, undervoltage, overvoltage, or overtemperature. Cisco documents that a switch can turn off a port and log the event when it detects several electrical fault conditions.[3] Preserve the exact vendor wording; do not translate every port shutdown into “bad switch.”
Test one variable at a time
During an approved window, move the suspect endpoint to a known-good compatible PoE port, try a short certified cable directly at the switch, and temporarily disconnect one noncritical PoE load. Change only one item per test and record the result. Cisco’s procedure uses this progression to separate device, cable/patch-panel, port, and budget causes.[1]
Exercise the peak—not just the idle state
For a camera, test the supported IR, heater, PTZ, analytics, recording, and stream conditions that can raise demand. Axis explains that its typical camera value represents normal conditions while its maximum scenario can include IR, heaters, motors, fans, audio, storage, and other active features; it also says cable loss belongs in a realistic total estimate.[4] For phones and access points, use the manufacturer’s maximum-power guidance and a representative busy condition.
Correct the design, then retest
The supported fix may be moving the device to a higher-capability port, replacing a damaged cable/termination, using the manufacturer-approved power supply or injector, redistributing noncritical loads, adding a correctly sized PoE switch, or correcting a documented allocation setting. Keep operating margin instead of designing to the last advertised watt, then repeat the peak test and preserve the before/after telemetry.[1][2][4] That margin is practical guidance, not a universal vendor percentage.[1][2][4]
Good troubleshooting versus an expensive wrong turn
Good
- Exact switch, power-supply, injector, device, and port models recorded.
- Available, used, remaining, and per-port power captured before reboot.
- Device maximum demand compared with port capability and total budget.
- Short known-good cable and compatible port used as controlled tests.
- Night/peak features exercised after the correction.
- Critical phones, cameras, doors, and access points changed only in a planned window.
Bad
- Replacing cameras because video drops while ignoring their shared switch.
- Adding port wattages without checking the switch’s total power pool.
- Budgeting only from a low daytime reading.
- Forcing a port mode or random injector because the plug fits.
- Factory-resetting the switch before saving logs and configuration.
- Power-cycling every device at once and losing the causal evidence.
Affected and not affected
Likely affected: homes and small businesses with IP cameras, VoIP phones, Wi‑Fi access points, door controllers, intercoms, or small downstream switches powered through Ethernet—especially after adding a device, changing a power supply, extending cable runs, enabling high-power features, or seeing failures cluster by time or port priority.
Not necessarily a PoE-budget problem: a device with stable local power, an endpoint that remains powered while only application traffic fails, a recorder with full disks, a phone whose cloud service is down, or a network suffering a broadcast loop. The power audit is a diagnostic branch, not proof of cause.[1]
Two realistic examples
The nighttime camera reboot
Four cameras look healthy all afternoon. After sunset, one repeatedly restarts when infrared illumination activates. The switch log shows a port power event, and the camera’s datasheet maximum is much higher than its daytime reading. A controlled short-cable test and temporary removal of one noncritical load isolate a budget/cable-margin problem before anyone replaces the recorder.
The last-added access point
A shop adds another access point to a nearly full PoE switch. It powers on when tested alone but stays off after all phones and cameras reconnect. The live budget shows no remaining allocation, and port priority explains which device loses power. The fix is to resize or redistribute the power design—not to keep factory-resetting the access point.
Related guides from The IT Guys
- Prove that your cameras record usable evidence with our 20-minute security-camera recording test.
- Keep calling and emergency-location plans separate from power troubleshooting with our VoIP outage and 911 continuity guide.
- If many devices fail together but PoE readings are healthy, follow the Ethernet loop and broadcast-storm checklist.
FAQ
Can a PoE device have network link but still lack enough power?
Yes. Data-link and power delivery are related but separate diagnostic layers.[1] Cisco explicitly recommends testing data link, powered-device behavior, cable path, and switch power budget as distinct checks.[1]
Is the number of PoE ports the same as the PoE budget?
No. Port count describes connection capacity; the PoE budget is the total wattage available across active powered ports.[2] A switch can have free PoE-capable ports but insufficient remaining watts for another device.
Should I use typical watts or maximum watts?
Use the switch’s documented allocation method and the device’s vendor maximum/class for capacity planning. A typical reading is useful telemetry, but Axis shows why cameras can draw more when features such as IR, heaters, motors, fans, storage, audio, or multiple streams are active.[4]
Will a shorter cable fix it?
It can help diagnose cable loss or termination trouble, but it is not a substitute for a correctly sized power design. Cisco recommends a short known-good cable/direct-switch test, and NETGEAR notes that the standard power ranges account for cable length and suggests considering a shorter cable when power is insufficient.[1][2]
Can I just raise the port’s watt limit?
Only when the exact switch and device documentation supports it and the total power supply can safely cover it. Cisco warns that inaccurate manual allocation can create overdraw conditions; a switch may shut down a device that consumes more than the configured maximum.[1][3]
Do I need to replace the switch?
Not until you separate a depleted total budget from a bad cable, one incompatible port, a faulty endpoint, an undersized/incorrect power supply, a configuration limit, or an electrical/thermal fault. Practical replacement guidance: replace or expand hardware only after the measured requirement exceeds the supported design or the switch fails a controlled vendor-supported test.[1][2]
Your next 20 minutes
- Photograph the switch label, power-supply label, and every currently powered port.
- Save the management view showing total, used, remaining, and per-port PoE power.
- Write down each powered device’s exact model, port, PoE standard/class, and datasheet maximum.
- Mark which devices are safety-, security-, phone-, or revenue-critical before testing.
- Schedule one controlled short-cable/known-good-port test and one representative peak-load test.
Treasure Coast help: The IT Guys helps homes and small businesses in Port Saint Lucie, Jensen Beach, Fort Pierce, and Vero Beach turn intermittent PoE failures into a measured port-and-power plan. Privacy guidance: bring model numbers, cable paths, switch screenshots, and timestamps—never passwords or private camera footage unless a specific, authorized support process requires it.[5]
Sources
- Cisco — Troubleshoot Power over Ethernet on Catalyst 9000 Switches.
- NETGEAR — Power over Ethernet (PoE) standards and compatibility.
- Cisco — Power over Ethernet Configuration Guide.
- Axis Communications — Typical and maximum power consumption in Axis cameras.
- Federal Trade Commission — Protecting Personal Information: A Guide for Business.
Evidence boundary: These vendor sources explain standards, allocation, telemetry, fault behavior, and camera power scenarios.[1][2][3][4] They do not identify the cause in any particular installation, guarantee compatibility, or replace the exact manuals and safe-change procedures for your equipment.