GPON Far-End Subscriber: Optical Margin and a 1:64 Proposal
A newly built GPON network in a housing colony near Tezpur serves 22 subscribers from one OLT PON port through a 1:32 splitter. The link to the farthest subscriber is 4.2 km of G.652D fibre with six fusion splices at 0.10 dB each, four mating connectors at 0.30 dB each, and a 1:32 splitter with 17.0 dB insertion loss. The OLT port launches +3.0 dBm at 1490 nm and the ONT receiver sensitivity is −27.0 dBm. The operator's design rule is that at least 3 dB of margin must remain after installation. During commissioning the far-end customer reports that the ONT LED goes off whenever it rains heavily, and an OTDR trace taken from the OLT shows a reflective event of 0.6 dB at 1.9 km followed by a clean end. The colony association now wants 20 more connections without adding another PON port and asks whether the splitter can be changed to 1:64.
1. Compute the end-to-end optical loss and the margin at the far-end ONT, and state whether the link satisfies the 3 dB design rule.
Read the worked answer
Build the loss budget item by item:
| Element | Count | Loss each | Total |
|---|---|---|---|
| Fibre 4.2 km at 0.35 dB/km | - | - | 1.47 dB |
| Fusion splices | 6 | 0.10 dB | 0.60 dB |
| Mating connectors | 4 | 0.30 dB | 1.20 dB |
| 1:32 splitter | 1 | 17.0 dB | 17.00 dB |
| Total | 20.27 dB |
Receive level = +3.0 − 20.27 = −17.27 dBm.
Margin = receive level − sensitivity = (−17.27) − (−27.0) = 9.73 dB.
Against the 3 dB rule the link passes comfortably with 9.73 dB spare. Two points of judgement belong in the answer even though the sum is clean:
- The 0.35 dB/km figure is a planning value; a real cable at 1490 nm is usually better, but connectors at the subscriber end are frequently worse than 0.3 dB once they are outside and dirty.
- The 9.73 dB figure is the theoretical budget, not a measurement. The OTDR and a live optical power meter reading at the ONT are what actually close the job; the power meter reading should be within about 1 dB of −17.3 dBm.
2. The ONT drops out in heavy rain and the OTDR shows a 0.6 dB reflective event at 1.9 km. Give the most likely cause, explain why rain affects a fibre plant at all, and state your corrective action.
Read the worked answer
Most likely cause: a reflective (not transmissive) event of 0.6 dB is the signature of a connector pair or a mechanical joint rather than a fusion splice - a good fusion splice shows almost no reflection. At 1.9 km it is plausibly a connectorised joint, an access closure, or a dirty/wet adapter in the drop section.
Why rain matters to fibre: rain does not attenuate light in the glass, so this is not radio fade. What rain does is raise humidity and drive water into a poorly sealed closure, adapter or drop-cable entry. Water in the gap changes the refractive-index match and scatters light, and the loss swings with temperature and moisture. A drop cable also shrinks and moves in wind and cold, so a tight bend or a strained connector can lift loss at 1490 nm at the same time. That is why the symptom is weather-correlated.
Corrective action, in order:
- Inspect and clean both connector faces with a dry one-click cleaner and an inspection scope; a single dirty ferrule can easily cost 0.5–1 dB.
- If the event is a mechanical connector in a closure, cut it out and fusion splice; replace the adapter if the ferrule is damaged.
- Check the drop cable for bend radius violations and reseal the cable entry and closure.
- Re-measure with the power meter at the ONT at both 1490 nm and 1310 nm - a loss that differs sharply between wavelengths points to a bend rather than contamination - and repeat the OTDR to confirm the event is gone.
3. Evaluate the association's 1:64 proposal: recompute the margin, give a decision, and state two non-optical reasons that should influence it.
Read the worked answer
Optical recomputation: a 1:64 splitter has about 20.5 dB insertion loss, i.e. 3.5 dB more than the 1:32.
New total = 20.27 + 3.5 = 23.77 dB, receive level = +3.0 − 23.77 = −20.77 dBm, margin = 6.23 dB.
Decision on paper: 6.23 dB still clears the 3 dB rule, so 1:64 is not technically impossible. In practice it is thin - the extra 3.5 dB consumes most of the comfortable headroom, and one dirty outdoor connector plus a future re-splice can eat 1–1.5 dB. It also leaves almost nothing for ageing or for a second splitter stage.
Two non-optical reasons:
- Shared bandwidth. The PON carries 2.488 Gbit/s downstream; 64 subscribers give a theoretical peak of about 38.9 Mbit/s each, and the 1.244 Gbit/s upstream gives about 19.4 Mbit/s each. If any of the 20 new customers is sold a 100 Mbit/s plan, the contention promise cannot be met at peak.
- Growth and fault domains. A second PON port costs one card or one port and keeps the split ratio at 1:32, preserving margin and halving the blast radius of any future fibre fault or splitter failure.
Recommendation: if the 20 new subscribers are low-usage, 1:64 with a re-measured and documented power budget is acceptable. If any of them is a business or high-tier plan, add a second PON port instead.