XGS-PON Overlay on a Long GPON Branch
A regional fiber operator runs a GPON network serving 1,200 premises from one OLT, with a 12 km feeder and 1:32 splitters. Average take-up is 60%, and 35% of subscribers now order the 1 Gbps tier. To compete, the operator overlays XGS-PON on the same ODN using coexistence combiners and N1-class (29 dB) OLT optics. After cutover, subscribers on the longest feeder branch report drops and ONT resets; measured downstream receive power at those ONTs is -27.8 dBm, while the rest sit at -22 to -24 dBm. The field team proposes raising OLT launch power by 3 dB and keeping the 1:32 splitters. A separate team notes that the branch also serves GPON customers who must not be disrupted, and that the 1 Gbps SLA requires 200 Mbps at peak. You must decide the remediation without a truck roll for every customer if possible.
1. Perform an optical budget sanity check for the 12 km, 1:32 branch and explain why raising launch power alone is or is not safe.
Read the worked answer
A rough XGS-PON budget at 1577 nm is: 1:32 splitter about 17 dB, 12 km fiber at 0.25 dB/km about 3 dB, connectors/splices about 1.5–2 dB, and coexistence combiner about 1 dB. That totals roughly 22–23 dB. With N1 optics at 29 dB, the theoretical margin is about 6–7 dB, so -27.8 dBm at the ONT is suspiciously low and points to extra loss, not normal distance alone.
Raising launch power by 3 dB may restore some ONTs, but it is not a safe fix by itself. It consumes margin that should cover aging, temperature, and dirty connectors; it does not correct a bad splitter port, macrobend, or contaminated connector. It can also push the GPON coexistence path outside its designed operating range if the combiner or receiver is already stressed. Use the measurement to isolate the extra loss first, then decide whether power shaping, a splitter change, or higher-class optics is justified.
2. Compare splitting the branch to 1:16 versus replacing the OLT with E1-class (33 dB) optics. Which do you choose and why?
Read the worked answer
A 1:16 splitter reduces insertion loss by about 3–3.5 dB and improves margin for every subscriber on that branch. It also lowers the chance that one noisy ONT or connector affects the whole group. The cost is a maintenance-window outage for that branch and possible rebalancing of 32 ports into two groups. Because XGS-PON is 10G shared, moving to 1:16 does not create a capacity problem for 1,200 premises at 35% 1 Gbps take-up.
E1 optics add 4 dB of budget without touching the ODN, but they require new OLT cards and may not be available on every platform. If the measured excess loss is only 3–4 dB, E1 could be sufficient; if the branch has a physical fault, E1 merely hides it. My choice is 1:16 on the long branch plus a connector/OTDR check, with E1 as a fallback only if the post-split margin is still below about 3 dB. This preserves GPON coexistence and gives headroom for the 200 Mbps SLA.
3. How would you isolate whether the -27.8 dBm is a splitter/connector fault or normal length loss without taking GPON down?
Read the worked answer
Use a live-PON OTDR at 1650 nm through an appropriate filter so the 1490/1310 nm GPON traffic is not disturbed. Compare the trace against the known 12 km distance and splitter position: a step at the splitter is expected, but an unexpected event before it, a high-reflectance connector, or a short drop at the end indicates a fault. Also measure optical power at 1490 nm and 1577 nm at the same ONT. If GPON power is normal but XGS-PON power is low, suspect the coexistence combiner, a macrobend that affects 1577 nm more, or a dirty connector. If both wavelengths are low, the splitter or feeder is more likely.
Then measure at the splitter output with a power meter and inspect/clean the connector pair. Do not change launch power until the loss distribution is understood; the comparison between wavelengths and the OTDR event distance is the judgement point.