Wavelength-Dependent Loss After a Riser Re-Route
A 6.4 km GPON feeder from an OLT cabinet to a residential society was partly re-routed last week. The contractor moved the final 180 m from an overhead span to a wall route inside the society pump room and up a riser, securing it with nylon cable ties at roughly 30 cm spacing. Commissioning with a power meter and a stable 1490 nm source gave an end-to-end loss of 6.3 dB against a design value of 2.5 dB. Suspecting a bad splice, the splicer opened the joint closure: bidirectional OTDR showed every splice between 0.04 and 0.09 dB. A two-wavelength measurement was then taken on the same link: 1310 nm = 2.3 dB, 1490 nm = 3.8 dB, 1550 nm = 6.4 dB. An OTDR run at 1550 nm shows a smooth rising slope over the final 170 m rather than a step, while at 1310 nm the same section looks normal. The society electrician describes the route as mostly straight with a few tight corners behind the panel.
1. The splicer suspects a bad splice. Using the measurement pattern, identify the actual loss mechanism and explain why a splice, a dirty connector and intrinsic fibre attenuation are each inconsistent with the data.
Read the worked answer
The signature is excess loss that grows with wavelength, appears as a distributed slope rather than a discrete step, and is confined to the newly dressed 170 m. That is macrobending.
- Why bending fits: at longer wavelengths the mode field is wider (about 9.2 micrometre MFD at 1310 nm against about 10.4 micrometre at 1550 nm), so light is less tightly confined and couples out of the core more readily at a tight bend. Loss therefore rises 1310 < 1490 < 1550, exactly as measured.
- Why a splice does not fit: a splice is a point event producing a step, not a slope, and its loss is close to wavelength-flat. The OTDR already shows all splices at 0.04-0.09 dB, and a splice cannot produce loss only over 170 m of continuous fibre.
- Why a dirty connector does not fit: contamination also gives a discrete, mostly reflective step and would appear at the same distance at both wavelengths.
- Why intrinsic attenuation does not fit: Rayleigh scattering falls with wavelength as roughly 1/lambda^4, so a 170 m length would show slightly lower loss at 1550 nm, not 4 dB more.
The 1550 nm OTDR slope localises the problem to the re-routed section, so the inspection starts behind the panel and at the riser entry, not in the closure.
2. You are writing the re-dressing instruction for the crew. State the bend radius, tie and clearance requirements with numbers, and say which measurements you will repeat to close the job.
Read the worked answer
Work to the cable manufacturer's data sheet where available; where it is not, use the standard rule of not less than 20 times the cable outside diameter during installation and not less than 10 times the diameter in service. For a 3.0 mm indoor/riser cable that means at least 60 mm while pulling and at least 30 mm permanently. Practical instruction to the crew:
- Re-route the section so no part of the cable is pressed against a corner, conduit edge, pipe or panel frame; fit bend-radius saddles or a radius guide at every direction change.
- Replace nylon ties with hook-and-loop or use hand tension only. The sheath must not deform or neck; spacing about 300 mm on the riser and up to 1 m on horizontal runs.
- Remove any staple, nail clip or cable cleat that indents the sheath, and keep at least 50 mm separation from hot water pipes and from electrical conduit.
- Leave a service loop at the riser top and bottom so the cable is never under tension.
Close-out measurements: power meter at 1310, 1490 and 1550 nm against the 2.5 dB design value; OTDR at 1550 nm to confirm the slope has gone; and if the instrument supports it, a 1625 nm check, because bend loss peaks there and it is the most sensitive early warning of a remaining tight radius.
3. The operator plans to add a 1550 nm video overlay on this link. Using the measured values, advise whether the link can carry it, and state the margin position before and after the fix.
Read the worked answer
Use a Class B+ style budget of 28 dB and add up the planned end-to-end loss:
- feeder and distribution fibre: about 2.5 dB after the fix
- 1:8 splitter: about 10.5 dB
- three connector pairs: about 1.5 dB
- design margin to keep: at least 3 dB
That totals roughly 17.5 dB, leaving about 10.5 dB of headroom at 1490 nm. Before the re-dress, the 1550 nm figure was 6.4 dB, which consumes most of that headroom by itself and makes the video overlay marginal to unusable; the 1490 nm figure of 3.8 dB already exceeded the 2.5 dB design value.
Advice: do not commission the overlay until the bend is corrected, and do not sign off on the 1490 nm reading alone. Bend loss is smallest at 1310 nm, so a link that passes at 1310 nm can still fail at 1550 nm. Re-measure at 1550 nm after the re-dress; expect about 2.7-3.0 dB, marginally above the 1490 nm value because a small amount of bend sensitivity usually remains. Accept only when at least 3 dB of margin survives at the highest wavelength the operator intends to use.