How to Size an Industrial PoE Switch Power Supply Without Missing the Losses

A 300 W power supply does not automatically provide 300 W of useful power to PoE devices. Some power operates the switch itself, some is lost in conversion and cabling, and the supply may not deliver its full nameplate rating under the cabinet's actual conditions.

Supply selection becomes much easier when every number has a defined boundary. Start with what the endpoints need, move upstream through the Ethernet channels and switch, then check the supply's available output. After that, test the operating states that a spreadsheet cannot establish by itself.

The method below is intended for cabinet builders and integrators selecting an industrial PoE switch power supply. Its worked figures are illustrative. Substitute current data for the exact equipment being purchased before approving a design.

Begin with the load that must operate together

List each endpoint, its required features and the corresponding maximum demand. Camera illuminators, heaters and accessories can matter; access points may have different powering modes. Do not combine a maximum figure from one operating mode with an assumption that the device will actually be used in another.

Retain both the declared PoE requirement and the consumption information. Allocation can differ from observed draw. A switch may need to reserve a stated amount even when a device is using less at the moment. A supply calculation and a switch allocation check therefore belong in the same review, but they answer different questions.

If two loads cannot occur together because of a documented control arrangement, explain that arrangement before taking credit for diversity. An informal expectation that cameras “usually do not all need extra power at once” is not a design constraint. For essential functions, test the concurrent operating state the site actually requires.

Use a boundary-based worksheet

Boundary Quantity to establish Frequent mistake
Powered-device input Required demand for the intended functions Using typical consumption as the maximum
PSE port output Device demand plus relevant channel losses, or the approved allocation method Treating the PD requirement as identical to the switch output
Switch DC input PoE conversion demand plus switch electronics Forgetting the switch itself consumes power
Supply output Continuous capacity at the specified conditions Using the nameplate without its derating curve
Upstream source Input demand and startup characteristics of the selected supply Inferring input current from DC output watts alone

The Ethernet Alliance technical brief identifies the distinction between the source and powered-device interfaces. When using standardized PD allowances, avoid subtracting the standard's channel allowance twice. When using a component-level estimate, identify where each estimated loss occurs.

Editorial engineering bench illustration with an industrial switch, DIN-rail supply and measurement equipment.
Editorial illustration of a supply-sizing review. The numerical example below is a calculation, not a reported laboratory measurement.

Worked example: from eight endpoints to the DC supply

Suppose eight endpoints each require 18 W at their input in the relevant maximum operating state. Their combined demand is 144 W. For this example, assume the engineering estimate for the installed channels adds 12 W, making the required aggregate PSE output 156 W.

Next assume a 92% effective efficiency between the switch's DC input and that PoE output, and 14 W for the switch electronics under the specified configuration. The estimated DC input demand is:

156 W ÷ 0.92 + 14 W = approximately 183.6 W.

The efficiency, channel loss and electronics figures are deliberately stated assumptions. They are not universal PoE constants. If the manufacturer's supplied full-load input figure already includes the electronics and conversion losses, use that defined figure instead of adding those elements again.

For a 48 V input at this load, estimated steady-state current is approximately 3.83 A. That calculation helps review the relevant power path, but it does not establish conductor sizing, protective-device selection or a transient-current requirement. Those decisions need the actual component ratings and installation design.

Now add a defined expansion requirement: two more endpoints with a combined estimated source-side need of 40 W. Applying the same illustrative conversion efficiency adds about 43.5 W to the input requirement. The expanded estimate becomes approximately 227.1 W, assuming the switch electronics allowance remains appropriate.

This is a useful reserve because it represents a named future load. It is more informative than adding an unexplained percentage to every stage of the calculation. A separate engineering margin can still be appropriate, but record what uncertainty it covers and avoid presenting it as an IEEE rule.

A larger nameplate can still be insufficient after derating

Suppose a candidate 300 W supply is allowed to deliver only 75% of nominal output at the installation's specified operating condition. Its available output would be 225 W. That would exceed the initial 183.6 W estimate but fall below the 227.1 W expanded estimate, even before any additional justified margin.

The 75% factor is an example, not a typical value or a recommendation. Use the selected supply's actual curve, mounting instructions and stated conditions. MEAN WELL's selection guidance explicitly calls attention to temperature derating and product-specific protection behavior.

Measure temperature where the component manufacturer defines its ambient condition. The outdoor weather report, cabinet-door surface and air next to the supply are not interchangeable measurements. Consider how filters, neighboring equipment, orientation and spacing affect the installation. A supply that was comfortably loaded on an open bench can face a different thermal environment inside a crowded enclosure.

Also check whether the switch's PoE budget changes with input voltage, supply arrangement or temperature. Adequate external watts cannot override the switch's own output limits. The completed selection has to satisfy both devices at the same operating condition.

Startup is a separate operating case

Once steady-state capacity is established, review how the system starts. Does the supply tolerate the supported load profile? Does the switch admit devices together or in a sequence? Does an endpoint enable a high-demand function shortly after boot?

A repeated rise-and-fall pattern in DC output can indicate a protection cycle, but it is not enough to identify the cause without the selected unit's documentation and measurements. Current limiting, shutdown and automatic recovery are different behaviors. Do not treat repeated restarts as evidence that a larger supply alone will resolve every case; incorrect input voltage, wiring faults or an incompatible load can also be involved.

For a controlled test, record the exact supply, switch configuration, connected load and startup method. Observe switch uptime and endpoint availability. If a restart fails only with all endpoints attached, separate the aggregate case from individual-port behavior before making a change. Keep the failed configuration in the test record so a successful retest has a meaningful comparison.

Redundancy changes the required capacity of each path

Two 150 W supplies do not automatically provide a redundant 300 W system. If essential operation needs 200 W and either supply must sustain it alone, that arrangement cannot meet the stated requirement unless a documented alternative operating behavior applies.

Distinguish a system designed to share capacity from one designed to survive a source failure. PULS explains the difference between paralleling and redundancy. The approved topology and any required decoupling are component-specific; do not improvise parallel connections from wattage arithmetic.

Define the failure case in functional terms. Must every camera keep recording, or may selected nonessential loads be shed? If shedding is allowed, identify the priorities and verify that the remaining path supports the essential load. If uninterrupted full operation is required, size and test each relevant surviving path accordingly.

Five pieces of evidence to request with the quotation

  1. The complete switch and supply ordering codes, with current datasheets and approved input ranges.
  2. The available aggregate PoE budget under the specified input and environmental conditions.
  3. A clearly bounded full-load input figure, or the information needed to calculate it without double counting.
  4. Startup, overload and recovery behavior relevant to the proposed load and redundancy arrangement.
  5. A port schedule linking each endpoint to its required class, operating features and expansion allocation.

For a TODAHIKA industrial PoE switch inquiry, sending this information with the cabinet conditions makes the response more specific. Ask the supplier to identify any assumption that changes the quoted budget, including an optional supply or a restricted operating mode.

Close the calculation with an acceptance record

Validate the completed cabinet with the intended endpoint functions enabled, the final supply arrangement and relevant cable channels. Include startup, sustained load and the approved source-failure case where required. Record temperature and the supported measurement points alongside the result.

The acceptance document should state what passed and what remains outside its scope. For example, a successful full-load test at one cabinet condition does not establish performance at every temperature in a product-family brochure. Keep the calculation, supplier confirmation and test result together so future expansions can be assessed against the same boundaries.

A well-sized supply is not simply a large number of watts. It is a component with enough available output, at the conditions that matter, to support an explicitly defined network and its required failure behavior.


Post time: Sep-15-2026