OTDR Cannot See Past the Splitter
You are commissioning a residential FTTx PON built as 2.0 km of feeder from the OLT to a 1:8 splitter in a street cabinet, then 380 m of distribution to the last ONT. The previous technician tested from the OLT with a 1310 nm OTDR and a 100 ns pulse and recorded a 10.9 dB event at 2.03 km with no further events, reporting no drop cable visible and a suspected break at the splitter. He also saw a reflective spike at 2.03 km that he labelled a ghost. You must confirm or reject the fault before a crew is dispatched. The far-end ONT is powered and its LOS LED is red. You have a 1550 nm OTDR with 5 ns to 10 microsecond pulses, a live-fibre detector, an optical power meter, a 1310 nm source, and a 1625 nm port with a live filter. A 1490 nm video overlay is in service downstream of the splitter and 1310 nm upstream burst traffic is active.
1. Explain why no drop cable appears beyond 2.03 km and what the 10.9 dB event and the reflective spike most likely are.
Read the worked answer
A 1:8 splitter has a theoretical splitting loss of 10 log10(8) = 9.03 dB, plus excess loss of roughly 1.0 to 1.8 dB, so a 10.5 to 10.9 dB event at the cabinet position is exactly what a healthy splitter looks like. It is not a break.
Downstream fibres are invisible for two reasons. First, the splitter divides the OTDR pulse eight ways and divides the returning backscatter eight ways again, so the drop backscatter is attenuated by roughly twice the splitter loss, about 21 dB, and falls below the trace noise floor. Second, the splitter event is a large reflective feature, and with a 100 ns pulse the event and attenuation dead zones (tens of metres) swallow anything immediately after it.
The reflective spike is likely the splitter input connector or a connector at the cabinet, plus a ghost. A ghost is a second reflection that re-enters the launch port and appears at a false distance; the test is to change the pulse width. Ghosts move or disappear with pulse width, real events stay at the same distance.
2. Set out a test plan, using the instruments available, that will actually locate the fault without disrupting the 1490 nm and 1310 nm live traffic.
Read the worked answer
- Record the suspected splitter event distance precisely and use short pulses (10 to 30 ns) at 1550 nm from the OLT side to sharpen the cabinet area, but accept that the drops will remain invisible.
- Test each drop from the ONT end, since a drop is a single fibre with no splitter in the way. Use the live-fibre detector first, then a filtered 1625 nm OTDR port so that live 1490 and 1310 traffic is not disturbed. This shows the drop fibre, the splitter output port event and the feeder.
- Use the power meter at the ONT with the known 1310 nm source at the OLT, or the ONT's own reported receive power, and compare against the calculated budget of feeder, splitter and drop. A shortfall of roughly 10 to 20 dB points to a disconnected or dirty splitter port, a macrobend in the drop, or a failed splitter leg.
- Inspect the cabinet: splitter input and output connectors, end-face condition, port numbering against the as-built, and any pigtail that has been unplugged. A red LOS LED with a healthy feeder and a healthy splitter usually means the path is broken or disconnected after the splitter, or the OLT port itself is down, which should be checked at the OLT.
- Only after all of that should a cable fault be booked for excavation.
3. The crew proposes to dig at 2.03 km. Evaluate that proposal and state where the repair effort should go first.
Read the worked answer
The proposal is wrong. The 2.03 km position is where the splitter sits, and 10.9 dB matches the splitter's insertion loss. Excavating there would disturb a healthy feeder and would not restore the customer, because the missing signal is on the downstream side of a device that the OTDR cannot look through.
Priority order for the repair team:
- Cabinet inspection, because a loose or contaminated splitter port, a wrongly patched port or an unplugged drop pigtail is the cheapest and most common cause, and it is consistent with a suddenly red LOS at one ONT.
- Drop-side OTDR from the ONT at 1625 nm with the live filter, and a power-meter reading with the reference source, to confirm whether loss is in the drop or at the splitter.
- OLT port and upstream check if the whole PON is down rather than one subscriber.
Excavation should be authorised only if a real event at a real cable distance is evidenced by two-way testing, and the as-built route records support that distance.