Open app

PoE device won't power, or reboots under load

very common

“Camera powers up, runs a minute, then reboots. The one next to it won't come up at all. Same panel.”

Likely causes

  • PSE port budget, class negotiation or port configuration

    ~25%

    The switch reports the port at a lower class than the device needs, or denies power outright, and the chassis is at its budget. Read the PSE first: it is free and it is the answer about a quarter of the time.

  • Open or high-resistance conductor on one powered pair

    ~20%

    Works at idle, browns out when the load comes on. DC loop resistance reads high for the measured length, and it is more often a marginal IDC contact than a broken conductor.

  • Two-pair PSE feeding a four-pair load

    ~15%

    The device is a high-class Type 3 or a Type 4 load but the PSE is two-pair only. Type 4 is four-pair by definition and Type 3 may be either, so read both part numbers rather than assuming.

  • DC resistance unbalance, pair to pair or within a pair

    ~15%

    The certifier flags unbalance while every transmission parameter passes. Four-pair PoE shares current across pairs, so unbalance drives one pair hot and trips protection under load.

  • Undersized conductors, copper-clad aluminium, or reduced-diameter cords

    ~15%

    Voltage measured at the device sags well below the PSE output, loop resistance is high for the length, and a scraped conductor shows silver under the copper. 28 AWG cords carry their own PoE and length derating from the maker. Get it, don't guess it.

  • Bundle heat rise or a hot pathway eating the budget

    ~10%

    Works in the morning, drops in the afternoon. Large bundle, fully loaded, ceiling well above 20 C, and it clears overnight.

What to bring

  • Certifier with DC resistance and unbalance tests
  • Multimeter
  • PoE load tester if you carry one
  • IR thermometer
  • Switch CLI or NMS access
  • Spare certified cords of known gauge
  • Punch tool with correct blade
  • Cable and cord data sheets

Safety

Never re-terminate a powered pair. Unplug the port first. Breaking a loaded pair arcs and pits the contacts, and a Type 4 PSE delivers real energy at just under 60 V. Type 4 is four-pair by definition and Type 3 may be two-pair or four-pair, so treat all four as live until the port is out. Hot, heavily loaded bundles in a summer ceiling are also a heat-stress hazard: water, a spotter, short stints.

Steps

  1. Step 1: Read the PSE before you touch the cabling

    Port power state, negotiated class, measured port power, chassis budget, and any inline power policing. Then move the device to a known-good port on the same switch.

    If that doesn’t do it

    PSE is healthy and the device fails on every port: go to the link.

  2. Step 2: Swap the device and the cord

    Proves whether the fault follows the device, the cord or the link. An unknown or 28 AWG cord on a high-power load is a real cause, not a long shot.

    If that doesn’t do it

    Fault stays with the link: get the certifier on it.

  3. Step 3: Measure DC loop resistance and resistance unbalance

    Run the DC tests, not just the transmission sweep. Compare against the limit your tester applies for the selected class rather than a remembered figure. High loop resistance for the measured length points at undersized or clad conductor; unbalance points at one bad contact or pair.

    If that doesn’t do it

    DC results clean: measure what actually arrives at the device.

  4. Step 4: Measure voltage at the device under load

    PSE output sits just under 60 V; what matters is what is left at the far end with the load running. A sag that only appears under load is voltage drop, and voltage drop is resistance, length or gauge.

    If that doesn’t do it

    Voltage holds under load: look at the environment and the budget.

  5. Step 5: Re-seat and re-terminate the suspect end

    Unplug the port first. PoE is unforgiving of a marginal IDC contact: pitted, cold-flowed or half-punched contacts pass a static wiremap and fail under current. Punch to fresh contacts rather than re-seating an old one.

    If that doesn’t do it

    Both ends freshly terminated and it still fails under load: verify the cable itself.

  6. Step 6: Check what the conductor actually is

    Jacket print, spool label, listing and third-party verification marks. TIA requires solid copper conductors; clad conductor has higher DC resistance for the same gauge, which under PoE is voltage drop and heat, not just a performance hit. Nick and scrape and look for silver.

    If that doesn’t do it

    Cable is genuine copper and correctly marked: measure the heat.

  7. Step 7: Measure the bundle and escalate it as a design item

    IR thermometer on the bundle, count the cables and the load per cable. Heat rise reduces both the power you can deliver and the length you can deliver it over. The derating is in TSB-184-A and the cable maker's data, so look it up. Then put the temperature, the bundle count and the measured voltage in writing and ask for smaller bundles, a different pathway, local power or a bigger PSE.

    If that doesn’t do it

    Design change refused: record the measurements and the derating basis and get the exception signed off.

References

  • IEEE 802.3 cl. 145 (PoE, Types 1 to 4)
  • IEEE 802.3 cl. 33 (PoE and PoE+, Types 1 and 2)
  • TIA TSB-184-A (power delivery and bundle heat rise)
  • ANSI/TIA-568.2-D cl. 6 (field test requirements, DC resistance and unbalance)
  • NFPA 70 art. 725 (power over communications cabling, bundle ampacity)

Check it with a calculator

More Copper & Certification guides

All 54 guides