Duplex link dark: polarity reversed

Likely causes

• Duplex jumper not crossed: transmit landing on transmit

  most common

  Power meter finds the light on the leg that should be receive-side dark, and nothing on the receive leg. VFL into the transmit port at one end comes out the transmit-side port at the far end. Flip the duplex clip at one end only and the link comes up.

• Both ends built A-to-A

  common

  Flipping one end fixes it and flipping both ends breaks it again. Behaves the same on every port in the panel.

• Panel terminated against its own labelling at one end

  common

  Every pair in that panel is reversed, not just yours. The odd/even print doesn't agree with what the VFL says.

• MPO cassette or trunk polarity method mixed

  common

  A jumper straight between the two switch ports comes up fine; through the structured path it won't. Position mapping is straight-through where it should cross, or fully mirrored.

• Optic fault, or optics that don't match each other

  occasional

  Power meter shows nothing leaving one transceiver. Or both receivers see correct power and it still won't link: check the two optics are the same wavelength and reach class, that the reach class suits the fiber, and that the port isn't administratively down.

• Broken fiber, unseated connector, or the wrong port patched

  occasional

  VFL shows no red at the far end on one leg, or red appears two rows away from the port on the label. Reseat every connection firmly before you chase anything else; a connector backed part-way out reads like a dead fiber.

What to bring

• VFL
• optical power meter
• spare duplex jumpers
• mode-conditioning patch cord for legacy multimode
• inspection scope
• port map and as-built

Steps

1. 1

   Step 1: Meter both legs at both ends and note where the light is

   This is the measurement that decides the whole call, so read it properly. Correct power on the leg you expect to be receive, at both ends, means polarity is RIGHT: the fault is the optic, the wavelength or reach pairing, or the port config. Light present on the leg that should be TRANSMIT, and nothing on the receive leg, is polarity reversed and that is your diagnosis. Nothing on either leg means dark fiber, dark optic or wrong patch. Note the level as well as its presence: receive power above the optic's maximum is its own fault, the receiver-overload sequence.

   If that doesn’t do it

   No light anywhere in either direction: prove the fiber with a VFL or an OTDR before you chase polarity any further. If the trace ends short, the break-location sequence takes over.

2. 2

   Step 2: Flip the duplex connector at ONE end

   Free, five seconds, and it is the answer more often than everything else combined. One end only: flipping both puts you back where you started.

   If that doesn’t do it

   Still dark: trace the actual fiber path rather than guessing at it.

3. 3

   Step 3: Map the path with a VFL

   Red into the port you believe is transmit, then see which far-end port glows. That tells you in seconds whether the crossover happens somewhere in the path, happens twice, or never happens at all.

   If that doesn’t do it

   Path maps correctly and power is present on both receive legs: the fault is the optics or the switch port, not the cabling.

4. 4

   Step 4: Check the jumper's own construction

   A-to-B duplex crosses; A-to-A does not. Read the legs, don't trust the boot. Uniboot jumpers can be re-polarised by whoever had it last, so assume nothing about a cord out of the spares bin.

   If that doesn’t do it

   Jumper is correct: move to the panel terminations.

5. 5

   Step 5: Verify panel and cassette polarity against the print

   Confirm the polarity method of every cassette, trunk and jumper in the path, and that all of it is the same method. Mixed methods are the usual cause on any structured MPO path.

   If that doesn’t do it

   Methods are mixed or unknown: the MPO polarity sequence.

6. 6

   Step 6: Confirm the transceivers match the fiber and each other

   Same wavelength, same reach class, right fiber type at both ends. An 850 nm short-reach optic on singlemode shows almost nothing and never links. A 1310 nm long-reach optic on multimode is the dangerous one: it will usually link and may pass traffic, then error under load or over distance because of differential mode delay. On legacy multimode, 1000BASE-LX needs a mode-conditioning patch cord. Without one it comes up and misbehaves, which is why this fault gets logged as intermittent rather than as a mismatch.

   If that doesn’t do it

   Optics match, cabling maps correctly, and it still won't behave: if the fiber and optic classes don't agree, the fiber-type sequence. If they do, hand it to the network side: the port or the config is the fault.

More Fiber Optic guides

• Link fails the loss budget
• Dirty endface that won't clean up
• Bend loss: loss climbs at the longer wavelength
• Link errors because there is too much light
• Hard break or cable cut

All 54 guides