Scratched MPO Plug at Turn-Up: Clean, Replace, or Mate It?
You are turning up a 100 m 40GBASE-SR4 link on OM4 between two racks in the same data hall. The channel uses a 12-fiber MPO trunk plus MPO-to-LC harnesses at each end. Before mating, you inspect the trunk's male MPO plug with a 400x probe against the IEC 61300-3-35 zone template for a 50 um multimode PC endface. The image shows a 55 um translucent particle straddling the Zone B/Zone C boundary, several 3-6 um particles scattered in Zone B, and a 12 um linear scratch that starts in Zone A and runs into Zone B. You dry-clean with an MPO one-click cleaner rated for the plug and re-inspect: the 55 um particle is gone, the small particles remain, and the scratch is unchanged. The harness at the far end of the trunk inspects clean. Production traffic is scheduled in 30 minutes. The only spare MPO trunk on site is 300 m long, and you have no field-polishing kit.
1. Using the zone criteria for a 50 um multimode PC endface, which of the three findings decides pass/fail, and what is your decision in the next 30 minutes?
Read the worked answer
Zone A is the 0-50 um core region and permits no scratches and no defects. A 12 um scratch that begins inside Zone A is a fail no matter how short it is, because it sits over the light-carrying area. Zone B (cladding, out to the ferrule boundary) tolerates small scratches and defects below the criteria limit, so the 3-6 um particles would not fail the connector by themselves. The 55 um particle was in Zone C, outside the optical area, and has already been removed.
The scratch is a surface defect in the glass or ferrule, so it will not clean off - stop cleaning it. Do not mate it "just to get traffic up": a scratched or contaminated plug abrades the clean harness it mates with and can transfer a permanent defect to the mating part, so you would convert one failed endface into two.
With no polishing kit, the options are: delay the turn-up and obtain a same-length trunk, or use the 300 m spare as a temporary path at 10GBASE-SR (400 m reach on OM4). The spare cannot carry 40GBASE-SR4, which is specified to 150 m on OM4 - 300 m is out of reach regardless of the 0.6 dB of extra attenuation.
2. The 55 um particle was outside the core. Why was removing it before mating still mandatory, and what would you have done if dry cleaning had not removed it?
Read the worked answer
A particle on the ferrule contact area props the two ferrules apart, creating an air gap. That gap produces a Fresnel reflection and added insertion loss even when the particle never crosses the core, and under mating pressure the particle can be pressed into the endface or dragged across the core area. Loose contamination also migrates: it transfers to the mating connector, and on high-power links absorbed light heats the particle until the endface pits.
If dry cleaning had failed, the sequence is: dry clean, inspect; if still contaminated, wet-clean with a lint-free wipe and 99% isopropyl alcohol, immediately followed by a dry wipe, then inspect again. For a bulkhead or adapter, use a dry swab first, then a solvent-damp swab, then a dry swab, one swab per pass. Never use canned air containing propellant, never reuse a wipe, and inspect after every step. Two or three cycles is the limit - if a defect does not move, it is damage, not dirt, and the assembly is replaced or re-terminated.
3. You decide to install the 300 m spare as a temporary 10GBASE-SR path. What must you verify about the spare before it can carry traffic, and what would disqualify it?
Read the worked answer
Check in this order, because each item can make a physically perfect trunk unusable:
- Fiber count and type - 12 fibers of OM4. OM3 would still be acceptable at 10GBASE-SR for this length but must be recorded; a mismatched type is a documentation and future-upgrade problem.
- Polarity method - the trunk must match the harnesses. A Type A (straight) trunk with Type A harnesses, or a Type B (reversed) trunk with Type B harnesses, gives working polarity; mixing methods swaps transmit and receive on the duplex pair and the link will not come up even though both fibers test good.
- Connector gender and keying - male (pinned) versus female (unpinned) must mate with the harness plugs, and the key orientation must match what the adapters accept.
- Endface condition - inspect both plugs, not just one; a clean near end proves nothing about the far end.
- Insertion loss and return loss - measure the trunk on all 12 fibers. A fiber reading more than about 1.0 dB above its neighbours usually means a dirty or damaged endface rather than fiber loss.
- Physical fit - 300 m coiled in a rack needs a bend radius of at least 10x cable diameter and somewhere legal to store it; a tight coil is both a loss and a future-fault risk.