Coherent links require a broader diagnostic view than a simple receive-power check. A link can have readable module telemetry and adequate optical power while failing because of wavelength, mode, FEC, line-system, optical signal-to-noise ratio (OSNR) or host configuration. Preserve the original state, compare both directions and change one variable at a time.

Classify the failure before changing anything

Record whether the module is absent, rejected, administratively down, unable to acquire signal, failing to achieve frame lock, showing high pre-FEC bit error rate (BER), reporting post-FEC errors or repeatedly flapping.

Capture the exact host alarm, controller state, module inventory, software release, configuration, optical measurements and error counters. Save timestamps and both endpoints. Clearing counters or reseating the optic too early can remove the evidence needed to distinguish a configuration problem from a marginal line.

Confirm the designed coherent application

Verify the exact module and host support at both ends. Record line rate, client rate, modulation, wavelength or frequency, channel spacing, transmit mode, FEC type, line-system mode and expected reach.

Do not assume that two modules described as coherent are interoperable. The operating mode, host support, wavelength plan and line-system requirements must agree. Pluggable coherent optics may support different modes depending on the device and software.

Compare the saved configuration with the approved design. Look for an incorrect channel, a fixed-grid/flexible-grid mismatch, unsupported modulation, inconsistent FEC or an accidental port-mode change.

Evaluate optical power in both directions

Compare transmit and receive power with the limits for the exact module and operating mode. Investigate missing or implausible telemetry as a separate symptom.

Power inside a broad operating range does not prove adequate signal quality. Amplifiers, filters, adjacent channels and accumulated noise can leave power acceptable while OSNR is inadequate. Conversely, an unexpectedly high level can indicate an overload risk or an amplifier setting problem.

Trace the route through patch panels, mux/demux units, amplifiers, reconfigurable optical add-drop multiplexers and monitor ports. Confirm that the intended channel follows the intended path in both directions.

Use OSNR and pre-FEC BER together

OSNR describes signal power relative to optical noise within a defined measurement method. Pre-FEC BER shows the error rate before FEC correction. Both are useful, but the applicable thresholds depend on the module, mode, vendor implementation and design.

Compare current values with a known-good baseline for the same link and configuration. A worsening pre-FEC BER with stable receive power may point to noise, filtering, dispersion, reflections, line-system settings or a marginal endpoint. Post-FEC errors indicate that the correction capability is being exceeded or that another severe fault exists.

Do not publish a universal OSNR or BER pass value. Use the exact module and platform documentation, and note whether the displayed value is measured, estimated or vendor-specific.

Check FEC and error behavior

Confirm that both endpoints use the intended FEC and that the host reports the coherent controller as expected. Review corrected and uncorrected errors, signal-degrade alarms and lane or carrier information where available.

Allow an appropriate observation period after a controlled change. A link that comes up briefly but accumulates errors is not accepted. Preserve counters before and after each step.

Isolate the physical path carefully

Inspect connector condition and patching under an approved procedure. Verify channel routing, polarity where relevant, fiber type, attenuators and amplifier direction. Look for recent work that could have changed a patch, filter, gain setting or channel plan.

Use loopbacks, pseudo-random binary sequence tests or controlled substitutions only when supported by the platform and change plan. Start with the smallest test boundary that can distinguish host, module and line path. A substitution is evidence only when the replacement is verified and all other conditions are held constant.

Decide the next action from evidence

Escalate with a compact evidence package:

• Endpoint devices, cards, ports and software.
• Exact coherent module identities and operating modes.
• Configuration and wavelength or frequency plan.
• Alarms, controller state, power, OSNR and pre-/post-FEC behavior.
• Route, passive and amplified elements.
• Timeline, recent changes and controlled tests.

This prevents multiple teams from repeating disruptive tests. For adoption and program planning rather than incident isolation, use the service-provider optics upgrade checklist.

Frequently asked questions

Can receive power be normal while a coherent link stays down?

Yes. Power alone does not prove correct wavelength, mode, FEC, OSNR, framing, host support or line-system configuration.

Is pre-FEC BER the same as packet error rate?

No. Pre-FEC BER is measured before forward error correction at the optical layer. Packet errors are observed at a later layer and may not appear while FEC is correcting the raw errors.

Does zero post-FEC error mean the link has enough margin?

Not necessarily. Review pre-FEC trend, alarms and the design margin. A link near a threshold can be error-free during a short observation period.

Should OSNR be compared between different vendors?

Only with caution. Measurement methods and estimates may differ. Use vendor documentation and a baseline from the same system where possible.

When is a loopback useful?

When the platform supports it and the test boundary is understood. It can help separate local host/module behavior from the external line, but it does not validate the entire service.

Should the optic be reseated first?

Preserve alarms, counters, configuration and telemetry first. Reseating can temporarily change the symptom and destroy useful evidence.

Build a coherent-link evidence package

Share the endpoint platforms, software, coherent module models, modes, wavelengths, line-system path, optical readings, alarms and error history. Axonode can help organize sourcing and validation requirements for compatible coherent pluggables without treating coding or a single telemetry value as proof of complete interoperability.

Contact Axonode about coherent optics requirements. For general recognition and link-down isolation, see SFP detected but link down.