Result
Fill in the fields to see your result.
Every element in an optical path removes a little signal power, measured in decibels because losses along a path simply add up in dB — unlike in linear power units, where they would multiply. The fiber itself attenuates gradually with distance; splices (fused, permanent joints) and connectors (mechanical, reusable joints) each add a small fixed loss; and any coupler or splitter used to divide the signal (as in a PON serving many homes from one fiber) adds a loss that grows sharply with the number of ways it splits.
The link margin is what remains once every loss is subtracted from the launch power and compared against what the receiver needs to work reliably. A design with zero margin is one bad splice or one degraded connector away from failing, which is why real installations target a positive buffer, not just a break-even budget.
Loss = (length × attenuation per km) + (splices × loss per splice) + (connectors × loss per connector) + (couplers × loss per coupler)
Margin = Ptx − Loss − receiver sensitivity
At 0.35 dB/km with 4 splices, 2 connectors and one 17.5 dB splitter, the fiber and hardware alone cost roughly 3.5 + 0.4 + 1 + 17.5 ≈ 22.4 dB, so launch power and receiver sensitivity must be chosen with that in mind.
With minimal splices and no splitter, nearly the full launch power reaches the far end, leaving a comfortable margin.
A splice just joins two fiber ends and passes essentially all the light through; a splitter actively divides the light between multiple output fibers, so a 1:32 splitter necessarily sends roughly 1/32 of the power down each output, an inherent loss no manufacturing quality can avoid — on top of a couple of dB of excess loss from the splitting device itself.
Enough to cover measurement uncertainty, future splices from repairs, connector wear from repeated mating, and some fiber ageing — 3 dB is often cited as a bare minimum for PON deployments, with many operators designing for more.
Updated