A light meter gives you a number — say, 2.4 dB of loss. Is that good? On a 500-meter link with two connectors, it is terrible. On a 10-kilometer span with six connectors and four splices, it is excellent. The number alone answers nothing. The loss budget is what turns a measurement into a verdict: the maximum loss the link is allowed to have, worked out from its length and its components before you measure anything. Measure against the budget, and "pass" becomes a fact instead of a feeling.
A link's budget is the sum of three things: the fiber itself, every connector pair, and every splice. Count connectors as pairs — each mated connection, not each connector end — and count every splice in the span, including the ones in trays you did not open.
| Component | Planning allowance | Notes |
|---|---|---|
| Single-mode fiber, 1310 nm | 0.35 dB/km | Multiply by span length in km |
| Single-mode fiber, 1550 nm | 0.25 dB/km | The wavelength the optics use is the one that counts |
| Multimode fiber, 850 nm | ~3 dB/km | Short links only — attenuation dominates fast |
| Connector pair (mated) | 0.5 dB each | Count pairs, not ends; a dirty pair can read 1 dB or more |
| Fusion splice | 0.1 dB each | A splice reading 0.4 dB is a re-do, not a pass |
Those are planning allowances — the figures you use to judge whether a link is in the right neighbourhood. Your site's specification is the limit that actually governs, and it wins every argument. If the site spec says 0.75 dB per connector pair, use 0.75. The method does not change; only the numbers do.
Take a single-mode span between two data halls: 2.0 km long, four connector pairs (two panels at each end), two fusion splices in trays along the route, tested at 1550 nm.
Budget at 1550 nm: Fiber: 2.0 km x 0.25 dB/km = 0.5 dB Connectors: 4 pairs x 0.5 dB = 2.0 dB Splices: 2 x 0.1 dB = 0.2 dB Total budget: 2.7 dB Measured loss (light source + power meter): 2.4 dB Verdict: PASS, with 0.3 dB margin.
Now change one fact: the same span measures 3.6 dB. That is 0.9 dB over budget — a real fault, even if the link is passing traffic right now. A link running over budget has no margin left: the next dirty connector, the next temperature swing, or the next optic swap can push it over the receiver's sensitivity and take the link down at the worst possible time. Over budget is a fault to fix on this visit, not a note to file.
Work the loss down in order of cheapest fix first — never re-terminate blindly:
A ticket that says "loss OK" cannot be re-checked. A ticket with the budget, the measurement, and the margin can. The minimum useful note:
Span: Hall A ODF-2 ↔ Hall B ODF-5 (2.0 km SM). Budget @1550 nm: fiber 0.5 + 4 connector pairs 2.0 + 2 splices 0.2 = 2.7 dB. Measured before: 3.6 dB (0.9 dB over budget). Found: bulkhead connector at Hall B panel, 1.2 dB (allowance 0.5 dB), contamination on scope. Action: cleaned + re-inspected (clean), re-terminated same panel position. Measured after: 2.5 dB @1550 nm — PASS, 0.2 dB margin.
The shape to copy: the budget calculation itself, the before-and-after measurements, the wavelength, the faulty event with its own allowance, and the margin. That note lets the next tech verify the pass, spot the thin margin, and know exactly which panel to check first when this span acts up again.
See RackReady Free sample Fiber no-link checklist
Support: support@hasherventures.com
Free tool: turn your loss-budget findings into a structured ticket note with the RackReady Ticket Builder.