Key equations and principles
L_disp = t_fib_max·1000 / (|D|·Δλ)
Frequently asked questions (FAQ)
What is a fibre link budget?
A link budget is the most basic way to judge whether an optical link will work: take the difference between transmitted optical power and receiver sensitivity (the available budget) as the total allowance, then add up all the losses along the way — fibre attenuation, splice points, connectors and splitters — item by item. If the total loss is below the budget, the link is viable and the difference is the margin. It does not involve bandwidth, dispersion or nonlinearity and is the first step in engineering selection.
Why can’t the margin be zero?
Because real links degrade: ageing and temperature drift of optical modules, contamination of connector end faces, drift of splice points with temperature and stress, additional attenuation from too-small fibre bend radii, and temporary losses during installation and later maintenance. In practice a 3 dB design margin is usually allowed, with 6 dB or more for long-haul backbone or important lines. The ‘design margin’ in the tool is the second threshold used for this judgement.
How do I tell whether a link is attenuation-limited or dispersion-limited?
Compare the L_loss and L_disp given by the tool: the smaller one is the bottleneck. In 1550 nm long-haul high-speed systems, because the G.652D dispersion coefficient is about 16.7 ps/(nm·km), dispersion often becomes limiting first, showing up as ‘distance still has margin but bandwidth is already insufficient’. In that case use a narrow-linewidth source (such as DFB/EML), dispersion-compensating fibre (DCF) or dispersion-managed fibre.
What are reasonable values for splice and connector losses?
Mechanical splices are about 0.1 dB each and fusion-protected splices about 0.02~0.05 dB, with the tool defaulting to 0.05 dB. Connectors (mated pairs) are about 0.3~0.5 dB each, defaulting to 0.5 dB; MPO/MTP multi-fibre connectors should be taken from the manufacturer’s specification. Conservative values are recommended at the design stage to avoid failing measured acceptance.
How do I count the number of connectors?
Count by ‘mated pair’, i.e. two mated ferrules count as one connection point. So one patch cord at the transmit end plus one at the receive end makes two connectors, and an intermediate patch adds two more. This tool distributes them equally between the two ends of the link; if they are all concentrated in the equipment room (e.g. at the ODF), the curve shape will differ, but the total loss is unchanged and the budget result is unaffected.
Why does the splitter loss grow so quickly?
A PLC splitter is a passive power divider whose ideal loss is 10·log10(N): splitting into N ways leaves only 1/N of the power in each. A 1:64 splitter ideally loses 18 dB and typically about 19.5 dB in practice. Therefore in PON systems the splitter, rather than the fibre itself, is often the main factor determining the ODN reach.
Why is the dispersion coefficient of multimode fibre negative?
The dispersion of multimode fibre arises mainly from waveguide and material dispersion varying in opposite directions with wavelength, so the net effect is negative around 850 nm. This does not mean ‘dispersion can be cancelled’; it only means the dispersion power trends oppositely with wavelength. The formula uses the absolute value |D| because the limiting factor is the magnitude of the broadening.
What dispersion does this tool calculate?
Only chromatic dispersion, i.e. pulse broadening caused by different propagation velocities of different wavelength components. Multimode fibre also has modal dispersion, which often dominates in short-reach high-bandwidth applications, but calculating it requires parameters such as the V number, core diameter and refractive index profile, which this tool does not cover.
Why is the rise time taken as 0.7/B rather than 1/B?
This is an engineering rule of thumb for NRZ coding: 0.7 corresponds to about 70% of the symbol interval as the effective rising edge; for RZ coding, whose duty cycle is halved, the factor is 0.35. It is not an exact derivation but a widely used industry design baseline; real systems must also allow margins for equalisation, clock recovery and so on.
Is the rise-time budget method too conservative at high bit rates?
Yes, and systematically so — this is the point this tool most needs to flag. 0.7/B is a first-order conservative screen: the true dispersion tolerance of directly detected NRZ systems roughly follows D·L ≈ 10⁵/B² (ps/nm, with B in Gb/s). The real tolerance of 10G NRZ is about 800~1600 ps/nm, corresponding to roughly 50~90 km on G.652; 40G NRZ drops to about 60 ps/nm, only 3~4 km. Therefore the L_disp given by this tool is usually smaller than the distance actually achievable in engineering above 10G, being a conservative first-order estimate. In coherent systems the tolerance is set by DSP electronic compensation (reaching tens of thousands of ps/nm) and no longer follows 1/B², so this tool’s model does not apply to coherent links.
Why do 100G links often show ‘no solution’?
Because t_allow for 100G NRZ is only 0.007 ns (7 ps), whereas the transmit and receive rise times of real 100G optical modules are typically in the 30~60 ps range, so t_allow² ≤ t_tx² + t_rx²: the transceiver itself is already over budget and no fibre length satisfies the rise-time criterion. This is not a formula error but an inherent limitation of this criterion at high bit rates. In practice 100G links should instead be assessed using the dispersion tolerance (ps/nm) given in the optical module datasheet, or use a coherent approach; the results from this tool for 100G are for reference only.
Why is modal dispersion in multimode fibre not calculated?
This tool calculates only chromatic dispersion (broadening caused by differences in propagation velocity between wavelengths). Multimode fibre also has modal dispersion, which often dominates in short-reach, high-bandwidth scenarios; calculating it requires parameters such as the V number, core diameter and refractive index profile. The multimode standards OM3/OM4/OM5 specify reach in terms of effective modal bandwidth (MHz·km) at 850 nm rather than a dispersion coefficient, so the dispersion figures in these templates are for reference only and are not used to judge multimode links.
Are segmented links with Raman/EDFA repeaters supported?
The current version performs single-span link budgets only (no segmented calculation with optical amplifiers). If repeater segmentation, or gain and noise accounting for erbium-doped fibre amplifiers (EDFAs) and Raman amplifiers, is required, separate modelling is needed. For ultra-long distances, first assess the ‘maximum single-span distance’ and then design segment by segment according to the number of spans.
What are the unit conventions?
Length is uniformly in km (for short data-centre reaches, enter 0.15 for 150 m); attenuation coefficient in dB/km; splice and connector losses in dB per point; optical power in dBm; margin and loss in dB; dispersion coefficient in ps/(nm·km); spectral width in nm; bit rate entered as Gb/s in the interface and converted internally to bit/s.
Why is the calculated margin negative?
It means the total loss has exceeded the available budget and the received optical power is below the sensitivity, so the link is down. Common causes are excessive distance, too many fixed connection losses, or too large a split ratio. In that case shorten the distance, reduce intermediate connection points, switch to a lower-loss splicing method, or simply replace the modules with ones of higher transmit power / lower sensitivity threshold.
Can the results be used to order fibre directly?
No. The attenuation coefficients and insertion losses in this tool are taken from published typical values and are intended only for quick estimation and scheme comparison. Real engineering must be based on the measured values in the selected manufacturer’s datasheet, on-site measured splice losses and the actual parameters of the modules used, confirmed by measured acceptance results.


