Link budget is a two-sided engineering check
An optical link budget tests whether received power is expected to remain inside a model-specific operating window after the complete channel is included. The weak-signal check asks whether the minimum transmitter output can survive channel loss and still meet the receiver's minimum requirement. The strong-signal check asks whether the maximum transmitter output, after the minimum path loss, could exceed the receiver's maximum input or overload limit. Both sides matter.
Keep units separate. Transmitter and receiver power are expressed in dBm, an absolute logarithmic power unit. Channel attenuation and margin are expressed in dB, a relative loss. A planning worksheet may express the weak-signal relationship as: minimum expected receive power = minimum transmitter power − maximum channel loss. Compare that result with the exact receiver sensitivity or minimum receive specification under the required FEC and test conditions. Do not add dBm values together or substitute a live DOM reading for a designed maximum loss.
Build the passive-path worksheet
| Input | Evidence to retain | Common mistake |
|---|---|---|
| Fiber length and type | Route survey, fiber designation and wavelength-specific attenuation basis | Using map distance instead of installed length |
| Connections | Count of mated pairs, panels, cassettes and connector grades | Counting individual plugs instead of complete mated events |
| Splices and components | Splice plan plus splitters, filters or other passive-device data | Leaving undocumented events at handholes or panels |
| Engineering margin | Project-approved allowance and its purpose | Hiding unknown route data inside an arbitrary margin |
| Transceiver limits | Exact part/revision data sheet, PMD, temperature and FEC conditions | Using a generic “400G” or “800G” value |
| Field result | OLTS insertion loss, direction, wavelength, reference method and date | Treating DOM or OTDR alone as channel certification |
The planned maximum channel loss normally includes wavelength-appropriate fiber attenuation over the installed length, the permitted loss of every mated connector pair and splice, documented passive components, and an explicit engineering allowance. Values must come from the project's governing cabling standard, component specification and contract. There is no safe universal connector-loss or reserve value that applies to every 400G/800G channel.
Collect the transceiver limits correctly
- Identify the exact optical PMD at both endpoints, not only the form factor or aggregate speed.
- Record minimum and maximum transmitter output, minimum receiver requirement and maximum receiver input/overload where the controlling specification provides them.
- Record the measurement conditions: wavelength, temperature range, lane, modulation, FEC and any stressed-receiver definition.
- Confirm whether the published figures apply per lane or to an aggregate value and whether transmitter penalties are already included.
- If endpoints differ, calculate each direction with the transmitter limits from one end and receiver limits from the other.
Do not mix “typical” and guaranteed limits in the same worst-case calculation. Typical values help describe expected behavior but cannot replace specified minimum/maximum boundaries. If a vendor document omits an input needed for the calculation, mark the budget incomplete and request model-specific evidence rather than filling the gap with a value from another product.
Separate design evidence from field evidence
A budget is a design artifact; an insertion-loss test is evidence about an installed channel. Use an optical loss test set (OLTS) with a recorded reference method, launch/test cords, direction and wavelength to measure end-to-end loss. An OTDR can help locate and characterize events, but its event/dead-zone behavior and test setup require competent interpretation. Connector inspection addresses contamination or damage that power measurements cannot identify by themselves.
Digital optical monitoring (DOM/DDM/CMIS) is useful operational telemetry. It can reveal lane imbalance, trends, alarms and large discrepancies while the link is installed. It is not a substitute for calibrated channel certification: host displays may be rounded, offset, unavailable or interpreted differently, and a single receive-power snapshot does not reveal the loss allocation of the path. Retain raw DOM output as supporting evidence alongside, not instead of, the route record and field test.
Review both low-power and overload risk
For the weak-signal case, use minimum transmitter output and maximum planned channel loss, then compare the calculated receive level with the relevant receiver minimum. The difference is the modeled margin under those assumptions. For the strong-signal case, use maximum transmitter output and the minimum credible path loss, then compare with maximum permitted receive power. A short clean link can create overload risk on some PMDs even when the long-path calculation passes.
Check every optical lane where the specification is lane-based. A total or average value can conceal an outlier. Also examine whether repair splices, extra patch panels, wavelength changes, bend-sensitive routing, aging/temperature allowances or future reconfiguration alter the documented model. ITU-T fiber recommendations define fiber characteristics, but they do not certify an individual installed route or guarantee a particular transceiver pairing.
Approval and evidence boundary
- Sign the route inventory and loss assumptions before ordering production quantities.
- Review the calculation in both directions and record the controlling PMD/module documents and revisions.
- Measure the installed path under the project's accepted procedure and reconcile unexpected margin against the event map.
- Validate the named endpoint configuration, FEC, telemetry and traffic/error behavior.
- Archive the worksheet, raw test files, inspection results and any approved exception as one revision-controlled evidence pack.
A passing calculation is not a compatibility guarantee, and a link-up event is not a channel certificate. This guide describes a review method, not contractual acceptance limits. The customer, system vendor, cabling standard and exact module documentation determine the final criteria.
Primary and authoritative references
- ITU-T G Supplement 39 — optical-system design and engineering considerations.
- ITU-T G.650.3 — test methods for installed single-mode optical-fibre cable links.
- ITU-T G.652 — characteristics of a single-mode optical fibre and cable.
- Fluke Networks loss-budget guidance — practical channel components, margin and testing context.