Intermittent GPON Drop With a Suspected Macrobend
A customer on a 1x16 GPON splitter reports intermittent video freezes and occasional ONT reboots. The drop is 180 m of 2.0 mm indoor/outdoor cable from a pedestal to an ONT in a garage; the splitter is 1.2 km from the OLT and uses 1490 nm for downstream video. A power meter at the ONT reads -27.8 dBm at 1490 nm; the ONT’s specified receive range is -27 to -8 dBm. The customer recently installed a wall-mounted TV and “tucked the fiber behind it.” An OTDR launched from the ONT through a 100 m launch fiber shows 0.3 dB at 4 m, 2.6 dB non-reflective at 38 m, and no further events; the splitter at 180 m is not visible. A VFL at the ONT end shows a bright red glow at the cable behind the TV. The technician has a fusion splicer, new SC/APC pigtails, a cleaning kit, and an inspection probe.
1. Interpret the OTDR trace and the VFL result. Where is the fault, and what physical condition does it indicate?
Read the worked answer
The 0.3 dB event at 4 m is likely the first connector or a launch reference; the key event is the 2.6 dB non-reflective loss at 38 m. Connectors usually produce a reflective event, while a bend typically appears as a non-reflective loss. The VFL glow behind the TV at the same general location confirms a macrobend: visible red light is escaping the fiber because the bend radius is too tight.
The power reading of -27.8 dBm is only 0.8 dB above the ONT sensitivity floor of -27 dBm, which explains the intermittent video freezes and reboots. The link is not fully down, so a simple continuity test would pass and mislead the technician. The 2.6 dB loss at the bend is consistent with the low power reading; removing that loss should bring the receive level to about -25.2 dBm.
Before cutting, inspect and clean the connectors as well, but the VFL and non-reflective OTDR event point strongly to the bend as the dominant fault. A dirty connector would normally show a reflective event near an endface, not a glow behind a TV.
2. Explain why the OTDR does not show the splitter at 180 m and what that means for testing this drop.
Read the worked answer
The OTDR trace stops before the splitter because a 1x16 splitter introduces high loss, roughly 13.5 dB, and splits the backscattered light among many outputs. Beyond the splitter, the return signal is too weak for the OTDR to resolve, and the trace may simply end or show a large loss. This is expected behavior, not evidence of a break at 180 m.
That means the technician should not use the absence of events beyond 180 m as a fault indicator. The OTDR is useful for locating the 38 m bend and for checking the drop up to the splitter, but the splitter and the feeder side require a different test approach. A power meter at the ONT, a VFL for bend and continuity confirmation, and an inspection probe for endfaces are the practical tools here.
If a full PON test is needed, the OLT-side OTDR or a dedicated PON tester would be used, but that is outside this drop repair. The technician should also record the launch fiber length and the OTDR settings so the distance readings can be interpreted correctly rather than treated as absolute.
3. Decide between re-routing, splicing, and replacing the cable. Outline the repair and the acceptance test.
Read the worked answer
First try to re-route the cable behind the TV without cutting: release the pinch, restore a bend radius of at least about 20 mm for 2.0 mm cable, and secure it so it cannot be pushed back into the bend. If the jacket or fiber is visibly kinked or the loss remains after re-routing, cut out the damaged section and fusion-splice with a protected splice closure or splice-on pigtail. Replace the cable only if the damage is extensive or the remaining length cannot be re-routed.
After repair, clean and inspect every endface, then measure with the power meter at 1490 nm. Expected receive level is about -25.2 dBm, giving roughly 2.2 dB of margin above the -27 dBm floor. Also check 1310 nm if the service uses it, and confirm the ONT remains stable under video load. Run an OTDR from the ONT side to confirm the 38 m event is gone or reduced to a normal bend signature.
Acceptance: stable ONT registration, no reboots during a soak test, receive power within the specified range with adequate margin, clean endfaces, and a labeled cable with a service loop. Document the before-and-after power readings so the repair is auditable.