Village PON Over Budget and Flapping at Night
You are auditing a 1:32 PON serving a village cluster. The OLT port transmits +3.5 dBm at 1490 nm. A calibrated power meter at the last ONT reads minus 24.7 dBm and that ONT is reported to flap in the evening. The documented route is 2.4 km of feeder, the splitter in a cabinet, then 1.1 km of distribution and a 90 m drop. There are 9 fusion splices and 6 connector pairs on the path (OLT, ODN patch panel, splitter input, splitter output, ONT, plus one field connector). Datasheet values: fibre 0.25 dB/km at 1490 nm, splice 0.10 dB, connector 0.30 dB, 1:32 splitter 17.0 dB typical and 17.8 dB maximum. The PON class is B+ with a 28 dB budget and the ONT sensitivity is minus 27 dBm. Your supervisor asks whether to re-splice the feeder, replace the splitter, or change the splitter ratio. The recorded evening temperature drop is 12 °C.
1. Calculate the expected end-to-end loss, compare it with the measurement, and state whether the link meets class B+ with an acceptable margin.
Read the worked answer
Expected loss:
- Fibre: 2.4 + 1.1 + 0.09 = 3.59 km x 0.25 dB/km = 0.90 dB
- Splices: 9 x 0.10 = 0.90 dB
- Connectors: 6 x 0.30 = 1.80 dB
- Splitter: 17.0 dB typical
- Total: 20.60 dB typical, about 21.4 dB if the splitter is at its maximum
Measured loss: 3.5 minus (minus 24.7) = 28.2 dB.
Gap: about 7.6 dB above the typical calculation. This is far too large to be explained by datasheet tolerance or by measurement error on a calibrated meter, so there is a real defect or a wrong component.
Budget check: the received power of minus 24.7 dBm is only about 2.3 dB above the minus 27 dBm sensitivity, and the class B+ budget of 28 dB is already used up at 28.2 dB. A working PON should normally keep at least 3 dB of margin for ageing, temperature and connector repeatability. So the link fails the engineering rule even when it occasionally works, which matches the flapping report.
2. Rank the likely causes of the 7.6 dB gap and describe how you would confirm each one with the instruments on site.
Read the worked answer
1. Wrong splitter ratio. A 1:64 splitter instead of 1:32 adds about 3.5 to 4.0 dB. Confirm by reading the splitter label and the as-built, and by comparing measured power at the splitter input with power at one output port.
2. Dirty or damaged connector. A contaminated end-face can add 3 to 6 dB or more. Inspect every connector in the cabinet with a scope, and clean and re-measure one at a time so that the change is attributable.
3. Macrobend, especially temperature-sensitive. The evening correlation with a 12 °C drop is a strong clue: a drop cable tied too tightly at the pole, or a tight loop in the cabinet, contracts and tightens, pushing the bend radius below the recommended 30 mm and adding loss. Confirm by OTDR at 1625 or 1550 nm on the drop from the ONT end, and by gently relieving the bend and watching the power meter.
4. Poor or mechanical splice. Confirm with bi-directional OTDR, not a single direction.
Isolation method: measure at the splitter input first. If the input is close to the calculation, the feeder and OLT are fine and the fault is at or after the splitter; then measure each splitter output to separate the splitter from the drops. This avoids re-splicing the feeder on suspicion.
3. Given the findings, recommend a repair plan and justify it against the alternative of simply upgrading the PON class.
Read the worked answer
Repair first, then decide on architecture. The audit shows a defect of about 7.6 dB. Repairing it returns the link to about 20.6 dB, which gives roughly 10.9 dB of margin against the 28 dB B+ budget and comfortably fixes the flapping. This is the lowest-cost, lowest-risk action and it needs no change of optics.
Do not blanket re-splice the feeder. The feeder accounts for only part of the calculated loss and is probably healthy; re-splicing it would add work and risk without addressing the real cause. Measure before cutting.
If the splitter turns out to be 1:64 because the village was over-subscribed, the choice is between returning to 1:32 and splitting the area across a second PON port, or moving to a class C+ budget with new OLT optics. Returning to 1:32 is usually cheaper if a spare OLT port exists, but check the subscriber count and the 1:32 port capacity first. Class C+ optics add cost for every port, not just this one, and they also mask poor installation practice.
Close-out: re-measure power at every ONT, record the before and after values, update the as-built with the actual splitter ratio, connector count and route length, and photograph the repaired cabinet bend and connector condition.