Split ratios

01Every doubling has a price

A passive optical splitter does one thing: it takes a single input and divides the light among several outputs. No power, no moving parts, no active management. The elegance is real. The physics, though, extracts a toll every time you divide — and the ratio you choose is ultimately less about optics and less about engineering than it is about how many subscribers you need to break even on a feeder fibre.

The math is straightforward. An ideal 1:2 split loses 3 dB — half the light gone by definition, before you account for insertion loss in the coupler itself. A 1:4 costs 6 dB. A 1:32 costs around 15 dB from splitting alone, plus coupler excess loss of roughly 0.5–1 dB per stage, plus the fibre attenuation across whatever distance separates the exchange from the furthest premise. Stack those numbers and the optical power budget starts to look tight. Most PON standards operate with a link budget somewhere in the range of 28–32 dB depending on the variant; the split ratio, the fibre length and the connector losses all compete for that same margin.

02Architecture: one stage or two

Deployments split their light in one of two ways. A single-stage approach places the full splitter — say 1:32 or 1:64 — at a single point, typically in a cabinet or an underground enclosure close to the exchange. The feeder fibre carries all traffic to that point, and from there individual drop fibres run to each premise. Simple to manage, but it concentrates the division in one place and offers no flexibility once built.

A two-stage or cascaded approach does a 1:4 or 1:8 split at the first enclosure, then completes the ratio at a second splitter closer to the homes it serves. A 1:4 followed by a 1:8 reaches 1:32 overall; a 1:4 into a 1:16 reaches 1:64. The cascaded model suits longer trunk routes or areas where premises clusters are geographically spread — fewer fibres need to be run long distances. The penalty is cumulative insertion loss from two coupler stages instead of one, and an extra enclosure to seal, maintain and eventually fault-find.

What ratio is actually chosen? In GPON deployments the most common target has been 1:32 or 1:64. XGS-PON architectures tend to favour 1:32 on first build, preserving optical budget for higher downstream speeds. The actual split ratio installed in the field often differs from the design ratio because operators routinely leave splitter ports unpopulated, connecting fibres only as customers take service. That means a 1:32 splitter with eight subscribers on it is sharing neither the capacity it will eventually carry nor the capacity the network was costed around — which matters for both performance today and upgrade planning tomorrow.

03Where the commercial decision hides

The ratio is where the business model meets the glass. Push it higher — 1:64 or beyond — and the per-home infrastructure cost falls: fewer feeder fibres, fewer exchanges, less active equipment at the headend. Push it too high and the optical budget runs dry before you reach the furthest subscriber; you either shorten your reach or buy more expensive optics at both ends. Each step up in split ratio also subdivides the downstream bandwidth across more users. At 2.5 Gbps downstream on GPON shared across 64 subscribers, each home's theoretical share is under 40 Mbps before any oversubscription model is applied — acceptable in 2010, increasingly uncomfortable today.

There is also a regulatory dimension in markets with access-sharing obligations. A high split ratio at a single point can simplify the unbundling arrangement; alternatively it can concentrate control in a single physical location that a competitor must access. The choice of where the split happens is never purely optical.

Operators upgrading a GPON build to XGS-PON often find they can reuse the existing passive splitters without touching them, since splitting is wavelength-agnostic and the two standards use different, non-overlapping wavelengths that can coexist on the same fibre. That reuse is one of the key arguments for passive architecture — the expensive, buried glass stays in place while the active electronics at either end absorb the upgrade.

Getting the split ratio right at design stage matters precisely because changing it later means touching the outside plant again — trenching, jointing, repopulating enclosures. Choosing it well once avoids choosing it again.