Long Distance Fiber: Loss Budgets and Real Reach Limits

A long distance fiber link is an engineering exercise, not a cable purchase. The glass itself has remarkably low loss, about 0.2 dB per kilometre at 1550 nm, so a 100 km span consumes only 20 dB before connectors and splices. What limits the link is the combination of that loss, the optical signal to noise ratio, chromatic dispersion and the capability of the transceiver at each end. Two links of the same length can need completely different hardware, because one has six splices and the other has none.

What Counts as Long Distance

Distance classes are informal. Short reach stays inside a building, typically under 2 km. Metro spans run 2 to 40 km between buildings or central offices. Long haul runs 40 to several hundred kilometres between cities, and ultra long haul goes beyond that with amplifiers and regeneration. The class matters because it changes the technology: short and metro links use simple intensity modulated optics, while long haul uses coherent detection, forward error correction and optical amplification. A 40 km link may sit in either class depending on the loss and the speed.

Metro, Long Haul and Ultra Long Haul

Metro links usually fit inside a single loss budget with ER optics. Long haul needs amplification every 80 to 100 km and careful dispersion management. Ultra long haul adds Raman amplification or coherent systems with advanced coding to preserve OSNR over thousands of kilometres.

Why the Route Matters More Than the Map

A 60 km straight line may pass through six splice closures and two patch panels, adding 3 dB and two failure points. A 90 km route with a single splice can be easier. Design from the splice schedule and the OTDR trace, not from the distance between the two buildings.

Loss, OSNR and the Power Budget

Every link starts with a power budget: launch power minus receiver sensitivity. Subtract fibre attenuation at the operating wavelength, connector loss for each mated pair, splice loss for each joint, and a design margin. At 1550 nm, OS2 loses about 0.2 dB per kilometre, so 100 km costs 20 dB. Connectors cost 0.3 to 0.5 dB each and fusion splices about 0.1 dB. The remaining budget determines whether the receiver can distinguish signal from noise, which is the OSNR question.

Working the Numbers

A 10GBASE-ZR link over 80 km at 1550 nm consumes about 16 dB of fibre, 2 dB of connectors, 0.5 dB of splices and 3 dB of margin, for 21.5 dB against a budget near 23 dB. The link closes, but barely. Replacing two splices with a single patch panel can be the difference between working and marginal.

OSNR and Amplifiers

An optical amplifier adds gain but also adds amplified spontaneous emission, which lowers OSNR. Each amplifier stage degrades the ratio, so a chain of eight spans needs more launch power or a better receiver. OSNR, not loss, is usually the real limit on amplified long haul.

Dispersion and Coherent Detection

Chromatic dispersion spreads pulses as they travel, and at 10 Gbps it becomes significant beyond about 80 km of standard single mode fibre. Dispersion compensating modules were the traditional fix; coherent detection now handles it electronically. A coherent receiver mixes the signal with a local oscillator and digitises the result, which allows electronic dispersion compensation, polarisation demultiplexing and much higher spectral efficiency. That is why 100G and 400G long haul is coherent rather than a faster version of the old 10G link.

Dispersion Limits at 10G

Standard single mode fibre has about 17 ps per nm per km of dispersion at 1550 nm. A 10G signal tolerates roughly 800 to 1000 ps per nm before compensation is needed, which corresponds to about 50 to 60 km. ER and ZR optics include some tolerance, but the route still needs checking.

Why Coherent Changed the Design

Coherent optics compensate dispersion in the digital domain, so the link no longer needs dispersion compensating fibre or careful wavelength planning. The trade-off is higher cost and power per port, which is why coherent sits at the core and simple optics sit at the edge.

Route Design, Splicing and Testing

Long haul routes are built from cable sections joined at closures, with patch panels at each end and sometimes at intermediate sites. Each closure is a future fault, so the splice schedule should be as short as the route allows. Aerial cable is cheaper to install and easier to repair but exposed to wind and ice; duct cable costs more and survives better. Test every span with an OTDR in both directions and store the traces, because a comparison after a fault is the fastest way to locate damage.

Splice Loss and Closure Practice

A good fusion splice loses about 0.05 to 0.1 dB. Poor cleaving, dust or a misaligned arc can push it to 0.5 dB, which multiplies across a long route. Use a fusion splicer with an arc calibration and inspect each splice before closing the tray.

OTDR Baselines

Record bidirectional OTDR traces at acceptance and keep them with the route documentation. A later trace that shows a new 0.4 dB event at 37 km locates the fault without digging. Without a baseline, the same event looks like normal variation.

Long distance fiber reach by technology

Real-World Application Scenarios

A regional provider links two towns 72 km apart with 10GBASE-ZR over a duct route that has four splices and two patch panels, and keeps 3.5 dB of margin. A utility runs 100G coherent between substations 95 km apart, with an amplifier at a midpoint hut. A campus connects a remote research building 14 km away using 10GBASE-ER on a single span, because the route has no accessible joint and the budget closes with room for a future repair splice.

Related Guides on This Site

For cable and optic selection, start with the single mode fiber overview and the OS2 fiber guide. Work the numbers with the fiber loss budget calculator guide, and compare short reach alternatives in single mode vs multimode.

Frequently Asked Questions (FAQ)