Gig device links at 100 Mbps

Likely causes

• A pair on 4/5 or 7/8 is open or shorted

  35%

  1000BASE-T needs all four pairs; 100BASE-TX only needs two. So the link comes up at 100 and looks perfectly healthy. Wiremap shows the fault and the tester's length on the bad pair reads short to the break. Know what does NOT cause this: 1000BASE-T automatically corrects pair transpositions and pair polarity reversals, so if that is all the tester flags, keep looking: the PHY has already fixed them.

• Split pair from wiring in conductor order instead of by the colour code

  15%

  Passes continuity and passes wiremap on many testers, and fails NEXT badly, so a cheap continuity tester will tell you the cord or the run is fine when it is not. This is a transmission fault, not a mapping fault, which is why it takes gig down when a transposition does not.

• Patch cord is 2-pair or damaged

  15%

  Swapping the cord fixes it on the spot. Worth knowing: a T568A to T568B mix is not this; that swaps pairs 2 and 3 wholesale and makes a crossover, which gigabit links fine through auto-MDIX and which only bites on legacy gear that has none.

• One end hardcoded to 100/full

  15%

  Switch shows the port not auto-negotiating, or the device's own config page shows a fixed speed. Set both ends to auto and gig comes back. Usually a leftover from a previous tech chasing a different fault.

• Wrong plug type for the conductor: a stranded-rated plug crimped on solid conductor

  5%

  Contact reliability fault: the plug's tines pierce the wrong way for solid conductor, giving intermittent or open contact. Reproducible by wiggling the plug, fault distance right at the plug. NEXT can look perfectly fine, so do not clear this one on a good NEXT result.

• Excess untwist at the termination

  5%

  Workmanship fault, a different animal from the wrong plug: wiremap is clean, NEXT and return loss margin are gone, and the tester puts the fault distance at practically zero.

• The port or the device is not capable of the speed you expect

  10%

  Datasheet says 100 Mbps, or that port is a 10/100 port on a mixed chassis, and adjacent ports negotiate gig with the same cord. Same class of answer when an AP reports a 1G uplink and wants 2.5G: the port may not do 802.3bz or may need firmware for it. Check the datasheet and the running version before you touch cabling.

What to bring

• Two known-good 4-pair patch cords
• Certifier with wiremap, NEXT and fault distance
• Switch CLI for speed, duplex and autoneg state
• Punchdown tool, spare jacks and plugs rated for the conductor type
• Tone and probe

Steps

1. 1

   Step 1: Swap the patch cords at both ends

   Known-good, 4-pair, correct category. Gig returns: it was the cord, and you are done in two minutes. This is always the first move.

   If that doesn’t do it

   Try the device on a known gig port

2. 2

   Step 2: Move the device to a port you have seen negotiate gig, using the same cords

   Gig there means the port or its config. Still 100 means the horizontal run or the device itself. Two minutes and you have halved the problem.

   If that doesn’t do it

   Check autoneg state at both ends

3. 3

   Step 3: Check speed and duplex settings at both ends

   Both should be auto. One end hardcoded to 100 caps the link and nothing on the cabling side will change it. Gigabit depends on auto-negotiation; hardcoding a gig link is not normal practice on most platforms.

   If that doesn’t do it

   De-energise any PoE, then wiremap and certify the permanent link

4. 4

   Step 4: If the port is powering a PD, disable PoE and confirm the channel is dead, then wiremap the permanent link watching 4/5 and 7/8

   Look for opens, shorts and splits. Transposed pairs and polarity reversals are auto-corrected by 1000BASE-T and are not why the link is at 100, so do not re-terminate on the strength of them. Run the full certification, not just continuity: a split pair passes wiremap on some testers and you will hand back a fault.

   If that doesn’t do it

   Read the tester's fault distance

5. 5

   Step 5: Use the fault distance to work out which end is bad

   A fault at 0 to 2 m is a termination. Mid-run is damage: a staple, a screw, a crushed tie-wrap, or a coupler someone pulled apart. Allow for slack and service loops when you convert distance to a location.

   If that doesn’t do it

   Re-terminate the suspect end

6. 6

   Step 6: Re-terminate the end the tester points at, then re-test

   Hold the pair twist right up to the termination and use plugs and jacks rated for the conductor type and category; a stranded-rated plug on solid conductor is its own separate fault. Re-test the full channel before you leave: a re-termination that fixes wiremap can still fail NEXT.

   If that doesn’t do it

   Replace the run

More PoE & Network guides

• PD will not power up on the port
• That port was never a PoE port
• Device reboots when it draws peak power
• Port stays dead after you fixed it
• Link and power good, device never gets on

All 54 guides