MPO Trunk Polarity: Six Links Pass, Six Fail
Two data halls in a 12,000 m² colocation facility are linked by an 85 m trunk with two 12-fiber MPO connectors at each end, installed in overhead ladder tray. At each end are two 12-fiber MPO-to-LC cassettes, giving 12 duplex LC ports per end. The design drawing says 'Method B polarity.' The installer used a key-up-to-key-down trunk and 'Type A' cassettes. The first six duplex links pass; links 7–12 fail. The contractor proposes swapping TX/RX patch cords on the failed links and documenting it as a field change. The facility manager wants the links live by Monday and asks whether the patch-cord fix is acceptable, what the actual defect is, and how to prove the repair without disrupting the six working links. A spare 12-fiber Type B cassette is on site.
1. Is the patch-cord swap acceptable? Diagnose the likely polarity defect and explain why six links work while six fail.
Read the worked answer
The patch-cord swap is not acceptable as a permanent repair. It creates undocumented transmit-to-receive reversals, defeats the Method B polarity scheme, and will cause failures when a port is moved, a cassette is replaced, or a test is run against the drawing. The design calls for Method B: a Type B trunk reverses fiber order end to end, and Type B cassettes reverse the duplex pairs so that transmit at one end arrives at receive at the other. The installed key-up-to-key-down trunk with Type A cassettes is Method A, not Method B. The split result points to a mixed build rather than a single global error: one 12-fiber group is behaving as wired and the other is not, or one cassette is mounted with the wrong key orientation. Six passing links may also be passing because those ports' electronics perform automatic MDI/MDIX, which is not a reason to accept the plant. Diagnose by tracing fiber 1 and fiber 12 end to end with a visual fault locator, recording key orientation at every MPO, and comparing the two 12-fiber groups. The correction is to re-pin or replace the incorrect group's cassette or trunk so that the entire link is Method B, then remove the field-built patch-cord swaps.
2. How would you prove the repair without disrupting the six working links, and what records should change?
Read the worked answer
Treat the six working links as live and do not use them as test paths. Work on the failed group during a maintenance window. First, use a visual fault locator on the failed group to confirm fiber continuity and correct end-to-end mapping after the cassette change. Second, run an OLTS insertion-loss test on each duplex pair in the repaired group at 850 and 1300 nm, using the same reference method and test cords used for the passing group. Record both directions or use a method that accounts for source and meter reference. Third, if an OTDR is used, add a launch and receive fiber and test bidirectionally; do not judge a short 85 m link from a unidirectional trace with poor dead zone. Fourth, verify that the repaired group matches the drawing's Method B polarity by tracing a known pair, not by observing link lights alone. Update the as-built drawing to show the actual MPO genders, key orientation, cassette type, and polarity method. Replace the 'field change' note with a proper test record. Label both ends of each 12-fiber group and each duplex port per TIA-606. Finally, retain the failed cassette or document its re-pinning so the root cause is traceable.
3. The customer later wants 40GBASE-SR4 over this same 24-fiber trunk. What changes and limitations should you explain?
Read the worked answer
40GBASE-SR4 uses eight fibers per link: four transmit and four receive in an MPO-12 interface. A 12-fiber group can therefore support one 40G link using eight fibers, leaving four fibers unused; the 24-fiber trunk supports two 40G links, not twelve duplex links. The polarity method and harness type must match the 40G optics and the existing Method B scheme. Do not assume that a cassette built for 12 duplex LC ports can be reused; the 40G link needs an eight-fiber Type B harness or a properly wired MPO-to-MPO connection. Also check reach: 40GBASE-SR4 over OM4 is limited to 150 m, so the 85 m trunk is within reach, but the link budget and connector count still need verification. If the customer wants more than two 40G links, or a future path to 100G, a higher-fiber-count trunk or single-mode OS2 with 100GBASE-LR4 is the better design. Explain that the existing 24-fiber count is a hard limit, and that 'MPO' alone does not define polarity, fiber mapping, or application support.