GPON Splitter Calculator

Check a PON split design against the class budget: single or cascaded splitters, distance and connector count, with the per-subscriber bandwidth that split implies.

Calculator Fiber & Optical Runs in your browser
Try:

Result

Total split ratio
1:32
Splitter loss
17.8 dB
Total ODN loss
24.4 dB
Class budget
28 dB
GPON Class B+
Margin
3.6 dB
Power at ONU
-21.4 dBm
sensitivity -27 dBm

Optical distribution network loss

ElementLoss (dB)
First stage 1:47.3
Second stage 1:810.5
Fiber 8 km @ 0.35 dB/km2.8
Connectors (6)3
Splices (8)0.8
Total ODN loss24.4

Bandwidth implications

Downstream capacity
2.488 Gbit/s
GPON line rate
Average per subscriber
77.8 Mbit/s
at 1:32 with all ONUs active
Upstream capacity
1.244 Gbit/s
shared TDMA
Design closes
Within budget with usable margin. Verify at the physically longest drop, not the average - the worst-case ONU is what defines the design.

About GPON Splitter Calculator

A PON design lives or dies on the optical distribution network budget. Every split halves the light (and a bit more, from excess loss), and the worst-case subscriber - furthest drop, most connectors - has to stay above ONU sensitivity. This adds it all up against the class budget.

Why splitter loss is not exactly 3 dB per split

Splitting light in two costs 3 dB in theory, but real planar splitters add excess loss and uniformity error. Practical figures are 3.6 dB for 1:2, 7.3 dB for 1:4, 10.5 dB for 1:8, 13.7 dB for 1:16, 17.0 dB for 1:32 and 20.5 dB for 1:64. Cascading a 1:4 and a 1:8 costs slightly more than a single 1:32 - the trade is flexibility in where you place capacity.

Split ratio versus bandwidth

GPON provides 2.488 Gbit/s downstream shared across the PON. At 1:32 that is roughly 78 Mbit/s per subscriber if every ONU pulls at once - fine for residential duty cycles, marginal once you sell gigabit tiers. XGS-PON at 10 Gbit/s symmetric moves the problem, but the optical budget still limits the split.

Common use cases

  • Validating a split plan before a fiber build is committed.
  • Deciding between centralised 1:32 and cascaded 1:4 + 1:8 architectures.
  • Diagnosing an ONU that trains at a low receive level.

Edge cases and gotchas

  • Budget against the worst-case drop, not the average - the furthest ONU defines the design.
  • Upstream at 1310 nm attenuates more than downstream at 1490 nm; using the 1310 coefficient is the safe choice.
  • Splitters are bidirectional and symmetric: the same loss applies to upstream traffic.

Frequently asked questions

Can I run 1:64 on Class B+?
Only on a very short, very clean ODN - 20.5 dB of splitter loss leaves about 7 dB for everything else. In practice 1:64 is a Class C+ or XGS-PON design.
Does an unused splitter port waste light?
No. The split loss is fixed by the component regardless of how many ports are connected, so an unfilled 1:32 costs exactly as much as a full one.