How to Test and Certify a PoE Installation

A PoE link can pass a wiremap test, pass a certification test, and still fail to power a camera. Data certification and power verification measure different things, and a commissioning procedure that only does the first will hand over a system with problems that surface weeks later under load. The sequence below separates the cable plant from the electronics so that when something does fail, you know immediately which side to look at.

Stage 1: Certify the Cable Plant

Before any active equipment is connected, certify every channel with a field certifier to the appropriate class - Cat5e, Cat6 or Cat6a - against ANSI/TIA-568 or ISO/IEC 11801. Insist that the report includes the PoE-relevant parameters, not just the frequency-domain results. Two matter most. DC loop resistance tells you how much voltage the run will drop under load. DC resistance unbalance tells you whether current will share properly between conductors; within-pair unbalance is commonly limited to 3 %, and for 4-pair 802.3bt operation pair-to-pair matching is equally important, because significantly unbalanced pairs will not share current evenly.

Unbalance failures are typically workmanship: a poorly seated conductor in an IDC, a partially damaged strand, or mixed cable types spliced mid-run. They pass a wiremap and pass a basic length test, which is exactly why they need to be caught here.

Stage 2: Verify the PSE Under Load

Testing a port with a multimeter alone tells you almost nothing, because a compliant PSE presents no operating voltage until it detects a valid signature. Reading 0 V on an idle port is correct behaviour. Use a PoE loader or a tester that presents a genuine PD signature at a selectable class, and confirm three things: that the port negotiates the expected class, that it sustains the rated power without dropping, and what the voltage is at the far end under that load.

Measurement	Expected result	Indicates if wrong
Idle port voltage	0 V (no valid PD)	Passive injector or fault
Negotiated class	Matches the design class	Budget exhausted or PSE Type too low
Voltage at PD under full load	Within the PD's input range, comfortably above minimum	Excess loop resistance or run too long
Sustained load for 10 minutes	No dropout	Thermal or budget limit at the switch

Stage 3: Test the Worst Cases Deliberately

Commissioning in ideal conditions is how winter faults are created. Force the coincident worst case: enable every heater and illuminator, bring up every radio, and hold the system at full load for long enough for the switch and the cable bundles to reach thermal equilibrium - twenty to thirty minutes is usually sufficient to expose a marginal design. Then power-cycle the switch and confirm every device returns unaided, since simultaneous inrush at restart is a distinct stress case that a staged installation never encounters.

Stage 4: Record the Baseline

Capture per-port consumption from the managed switch once the system is stable, and store it with the handover documentation. This baseline is the most useful diagnostic asset the operator will have. A port whose draw creeps upward over months is usually a failing heater, water ingress, or a degrading power supply, and comparing against the commissioning figures identifies it long before the device fails.

What Belongs in the Handover Pack

Certification reports for every channel including DC resistance and unbalance; a port schedule mapping each switch port to its device, location, class and cable run length; the loaded test results and delivered voltage for the longest runs; the per-port consumption baseline; the earthing and surge protection arrangement with locations of every SPD; the switch PoE budget with the reserved total and remaining headroom; and firmware versions for switches and powered devices. None of this is onerous to produce during commissioning and all of it is expensive to reconstruct later.

The Faults This Catches

Done properly, this procedure catches the four problems that account for most PoE support calls: workmanship faults that pass a wiremap but fail under current, runs that are electrically too long for their class, switches whose system budget cannot support the populated port count at the design class, and devices that negotiate a lower class than intended and silently run in a degraded mode. All four are cheap to fix before handover and expensive to diagnose afterwards from the other end of a phone line.