A GPON splitter at 1:128 is officially supported and runs fine within 20 km on a clean link, but only when three things hold at once: enough optical power budget, a manageable fiber reach, and a per-user bandwidth ceiling you can live with. The standard writes 1:128 as a capability ceiling, not a promise.
A 1:128 split is the standard's stated ceiling, not a guarantee — one OLT port can drag 128 ONUs only if optics, distance, and bandwidth all cooperate.
With a +9 dBm C++++ OLT module and a B+ ONU you get a 37 dB budget; after the ~21 dB splitter loss the link still covers 20 km comfortably.
The real trade-off at 1:128 is per-user bandwidth: 128 homes share 2.5G down, so the saturation per-user peak drops to ~19.4 Mbps.
A passive optical splitter sits in the ODN and divides one OLT downstream fiber into N paths with no power and no amplification. 1:128 simply means one-to-128. GPON follows ITU-T G.984: a single PON port runs 2.488 Gbps downstream and 1.244 Gbps upstream. The higher the split ratio, the thinner the optical power and bandwidth each path receives — that is the physical cost of passive splitting, not a vendor cutting corners.
In a passive optical network (PON) the fan-out happens without electricity, so the split ratio is paid for entirely in light and bandwidth, never in active gear.
Note: Splitter insertion loss grows with N as 10·log₁₀(N): about 9 dB at 1:8, 12 dB at 1:16, 15 dB at 1:32, 18 dB at 1:64, and ~21 dB at 1:128. Each step up roughly doubles the loss penalty.
The OLT transmits, the ONU receives, and every loss in between must stay under the optical module budget. Today's OLTs ship with a C++++ (high-power) PON module transmitting +9 dBm; the ONU end saves cost with a Class B+ receiver at −28 dBm sensitivity. The total usable optical power budget is +9 − (−28) = 37 dB.
A 1:128 splitter eats ~21 dB, connectors and fusion splices take another 2–3 dB, leaving 13–14 dB for the fiber. At 0.35 dB/km that is a 37 km-class reach. GPON's logical coverage is normally capped at 20 km, so within that band a 1:128 split has thick margin and links up without drama. In the C++++ era, distance rarely limits the split ratio anymore — per-user bandwidth does.
Tip: Always size the budget before trusting the standard's "1:128" label. A +9 dBm OLT module raises the first gate high, but old cable, extra flanges, or over-distance runs eat the margin first.
| Dimension | 1:64 | 1:128 |
|---|---|---|
| Splitter insertion loss | ≈18 dB | ≈21 dB |
| Fiber budget left (C++++ +9 dBm OLT, B+ ONU) | ~16 dB (no pressure within 20 km) | ~13 dB (no pressure within 20 km) |
| Actual reach at ≤20 km | Full coverage (~6 dB margin) | Full coverage (~6 dB margin) |
| GPON downstream per-user peak | 2.488G ÷ 64 ≈ 38.9 Mbps | 2.488G ÷ 128 ≈ 19.4 Mbps |
| PON ports needed for 128 users | 2 | 1 |
| Typical use | Suburb / long haul / enterprise | Dense urban FTTH |
The 2.488 Gbps GPON downstream is shared across the whole PON port. If all 128 users saturate at once, the theoretical per-user peak is 2.488G ÷ 128 ≈ 19.4 Mbps; at 1:64 it is 2.488G ÷ 64 ≈ 38.9 Mbps. Double the split, halve the per-user shared bandwidth.
In practice 128 homes rarely all max out together; DBA (dynamic bandwidth allocation) schedules by demand, so residential FTTH with low concurrency barely notices. But for live streaming, video backhaul, or enterprise high-bandwidth services, the 1:128 ceiling shows up fast. XGSPON (G.987) lifts the downstream to 10G, yet the split ratio is still capped by optical power budget — more speed cannot recover lost light.
Warning: Do not spec 1:128 on the assumption that a faster PON class fixes it. XGSPON raises bandwidth but the ~21 dB splitter loss and reach constraints are unchanged; the optical budget is the hard wall.
A 1:128 split is the right call for new urban renewal FTTH: building-to-equipment-room within 20 km, clean fiber, mostly residential broadband with low concurrency. A +9 dBm OLT module with B+ ONUs leaves comfortable margin, and one PON port serves 128 households — halving OLT ports and uplink fiber cores, which clearly lowers TCO.
Step down to 1:64 or 1:32 when: distance exceeds 25 km (rural, mountain, cross-town); the cable is old or has high bend loss; video surveillance or enterprise leased lines make up a large share of users; or you want to reserve ~20% optical margin for device aging (splitter loss grows ~0.1–0.3 dB per year) and temperature drift. Beyond 25 km, the OLT is already at full C++++ power (+9 dBm) — you then upgrade the ONU to C+ (−32 dBm, 41 dB budget) or C++ (−35 dBm, 44 dB budget), or move to a long-reach scheme.
Real example: a 128-home community, 9 km from equipment room to the farthest unit, +9 dBm OLT module, B+ ONU (37 dB budget). Split loss 21 dB + connectors 2 dB + fiber 9 × 0.35 ≈ 3.2 dB = 26.2 dB, leaving 10.8 dB for aging and margin — plenty, stable even in summer heat. Push the same topology to 18 km (fiber loss 6.3 dB, total 29.3 dB) and 7.7 dB remains; B+ still holds. Past 25 km you must upgrade the ONU class.
Rayin sizes GPON split plans by calculating the budget first, then fixing the split ratio — not by copying the standard's upper limit.
What does a 1:128 GPON splitter ratio mean?
It means one OLT PON port drives 128 ONUs through a passive splitter. The GPON standard (ITU-T G.984) sets 1:128 as the maximum allowed split.
What happens if the OLT optical power is insufficient?
The end ONU receives light below its receiver sensitivity (about −28 dBm for B+), so it drops and fails to register repeatedly. A 1:128 split already costs about 21 dB; with a +9 dBm OLT module the budget is not tight, but old cable or over-distance runs burn the margin first.
How do I choose between 1:128 and 1:64?
Use 1:128 for short, dense, low-concurrency urban FTTH to save ports. Drop to 1:64, or even 1:32, for long distance, old cable, or high-bandwidth scenarios where stability matters more.
Does 1:128 crush each household's bandwidth?
At full saturation the per-user peak is about 19.4 Mbps, half of 1:64's 38.9 Mbps. Homes with low concurrency barely notice; high-bandwidth business services feel the ceiling.
A 1:128 split is a capability the standard writes down, but whether it actually runs full depends on three things lining up: optical power budget, coverage distance, and per-user bandwidth. With a +9 dBm C++++ OLT module and B+ ONUs, 1:128 is stable within the 20 km standard reach; only beyond ~25 km, on old cable, or under high bandwidth do you step back to 1:64 for peace of mind. Calculate the budget before you chase the highest split ratio. For GPON OLT and splitter planning, see Rayin's GPON OLT lineup.
Customer Manager at Shenzhen Rayin Technology Co., Ltd.
