Can a fibre link fail because there is too much power?

Yes, and it is the failure people lose the most time to. Every receiver has an overload threshold as well as a sensitivity. A transceiver meant for eighty kilometres, patched across a rack, can arrive well above that maximum and produce errors — while the sensitivity margin reads 30 dB, so power is the first thing ruled out. The fix is an in-line attenuator: the link needs less signal, not more.

What engineering margin should I use?

Three decibels is a common planning figure, and it is not padding. It covers repairs — a cable cut adds two permanent splices — connector ageing, later mid-span joints, and the fact that the cable's stated attenuation is a specification rather than a measurement of your drum. A link commissioned with no margin will fail during its service life at a time nobody chooses.

Why is 1550 nm better than 1310 nm?

Two mechanisms cross there. Rayleigh scattering falls as the fourth power of wavelength, so it dominates at short wavelengths; infrared absorption by the glass rises with wavelength and dominates beyond about 1600 nm. The minimum sits where they cross, near 1550 nm — which is also why optical amplifiers were developed for that band.

What attenuation figure should I use?

The one on the cable drum, or from an OTDR trace of the actual route. The defaults here are planning figures, and real fibre varies by manufacturer, production run and age. A number carried into a budget as though it were a constant will be wrong on somebody's cable.

How much loss does a splice add?

A fusion splice is typically around 0.1 dB as a planning figure, but a poor field splice is several times that. Mechanical splices are higher and less stable, and they age. Where a route is already built, the OTDR trace beats any table.

Why does the tool say reducing connectors helps more than better cable?

Because on short routes the fibre is a minority of the loss. Over two kilometres at 0.35 dB/km the glass costs 0.7 dB, while six connector pairs cost over two. On a route like that, taking out patch panels buys far more than a lower-attenuation cable would.

Is anything I enter sent anywhere?

No. Everything is calculated in your browser and nothing is uploaded.

Fibre Optic Loss Budget Calculator

What arrives at the far end of a fibre, and whether the receiver can actually use it — which is two questions, not one. A link can fail because too little power arrives, and it can fail because too much does.

Every default here is a starting point, not a specification Fibre attenuation varies by manufacturer, by production run and with age. Splice loss depends on the machine, the operator and the weather. Use the figures on the cable drum and the measurements from the OTDR where you have them — these defaults exist so the page is usable before you do.

The span

Wavelength

Fibre length (km)

Attenuation (dB/km)

Joints

Fusion splices

dB each

Connector pairs

dB each

Engineering margin (dB)

Splitters, muxes and other passives

The optics

Transmit power (dBm)

Receiver sensitivity (dBm)

Receiver overload (dBm)

All three are on the transceiver datasheet. The overload figure is the one people leave out, and it is the one that catches short links.

The budget

The green band is the receiver's usable window. The marker is where this link lands.

How far could it go?

The failure everybody checks for

Too much loss. The signal arrives below the receiver's sensitivity and the link does not come up, or comes up and errors constantly. It is the obvious failure, it is what a loss budget is normally built to avoid, and it is generally caught before anything is installed.

The failure almost nobody checks for

Too little loss.

Every optical receiver has an overload threshold as well as a sensitivity. Above it the photodiode saturates and the eye closes, and the link produces errors. Take a transceiver specified for eighty kilometres, patch it across a rack into an identical unit two metres away, and it can fail immediately — not despite the enormous power margin but because of it.

What makes this expensive is where people look. The sensitivity margin reads 25 or 30 dB, which is wonderful, so power is the first thing ruled out. Hours go into transceivers, patch leads, line cards and configuration before anyone measures the received level and finds it well above the maximum.

The fix is an attenuator, which is counterintuitive enough that it is worth saying plainly: the link needs less signal, not a cleaner one. A 10 dB in-line attenuator in the receive path, costing a few pounds, and the link comes straight up.

What the margin is for

The engineering margin — typically 3 dB — is not padding. It covers things that are certain to happen and cannot be predicted individually:

Repairs. A cable cut adds two splices. Over twenty years a route through a city will be cut more than once, and each repair is permanent added loss.
Ageing. Connectors degrade with mating cycles; splices drift slightly; hydrogen ingress raises attenuation in some conditions.
Measurement uncertainty. The cable's stated attenuation is a specification, not a measurement of your particular drum.
Later work. A mid-span joint added to serve a new building is loss nobody budgeted for.

A link commissioned with no margin is a link that will fail during its service life, at a time nobody chooses.

Why attenuation depends on wavelength

Two mechanisms, pulling in opposite directions.

Rayleigh scattering falls steeply with wavelength — as the fourth power — because it depends on microscopic density variations frozen into the glass. This dominates at shorter wavelengths and is why 850 nm multimode is an order of magnitude worse than 1550 nm single-mode.

Infrared absorption rises with wavelength as the glass itself begins to absorb. This dominates beyond about 1600 nm.

The minimum sits where the two curves cross, at around 1550 nm, which is why long-haul systems live there and why optical amplifiers were developed for that band rather than another.