Fiber-Optic Communication Systems Exercises

These exercises practise fiber-optic communication calculations as engineering release evidence. The focus is not only whether an optical link comes up. The focus is whether received power, dispersion, event loss, overload margin, FEC evidence, WDM launch power, route latency and acceptance records are strong enough for service operation.

Assume simplified screening models unless an exercise states otherwise. Real fiber links also require connector inspection, reference-method control, calibrated optical power meters, launch-condition control, OTDR trace review, transceiver vendor limits, wavelength confirmation, temperature testing, repair history and service monitoring.

Release Evidence Notes

Use these worked problems as service-acceptance screens until the field record proves the same optical boundary. A credible fiber release package should connect design budget, measured loss, trace evidence, transceiver class, receiver margin, dispersion, reflection, route latency, repair history and operating alarm thresholds.

For fiber-service decisions, the strongest evidence is agreement between calculation, traces and measured power:

• transmitter class, wavelength, launch condition, receiver sensitivity, overload limit and reserve match the installed optics;
• OLTS insertion loss, OTDR events, bidirectional splice evidence, connector inspection and cleaning records explain the same loss budget;
• chromatic dispersion, PMD, multimode bandwidth, compensation loss and OSNR are checked against the real bit rate and optic tolerance;
• pre-FEC BER, post-FEC errors, FEC threshold, received-power margin and receiver alarm limits describe the same optic state;
• bend-sensitive events are retested at the service wavelength and compared with a second wavelength when diagnosis depends on wavelength response;
• DDM trends, calibrated power measurements, trace files, dates, reference methods, technician notes and acceptance thresholds remain attached to the service record.

When those evidence paths disagree, do not release the link from one favorable trace or one live-light test. Identify the wrong reference method, dirty connector, bad launch condition, gainer event, undocumented patch, wrong wavelength, optic substitution, tight bend or drifting transceiver before accepting the service.

How to Use These Exercises

For each problem, define:

1. the optical boundary: transmitter output, patch-panel handoff, receiver input, route segment or service path;
2. whether values are dBm, dB loss, linear power, ps/nm, seconds or service metrics;
3. the relevant acceptance rule: sensitivity, overload, dispersion, event loss, latency, margin or trace agreement;
4. the validation evidence that would support release;
5. the operational consequence if the calculation fails.

The common mistake is to treat fiber as a passive pipe with one attenuation number. A practical fiber service can fail because of overload, connector contamination, dispersion, reflection, route diversity, receiver bandwidth or documentation gaps even when total loss looks acceptable.

Engineering Boundary Notes

Fiber-optic evidence must state the optical boundary being checked: transmitter output, connector interface, patch panel, route segment, receiver input, transceiver diagnostic value or end-to-end service path. A dB loss, dBm received power, OTDR event, OLTS insertion loss and DDM trend are related, but they do not prove the same condition unless their reference method is controlled.

Loss budget, dispersion budget and error budget are separate release boundaries. A link can pass total loss and still fail receiver overload, chromatic dispersion, PMD, OSNR, return loss, pre-FEC BER, connector cleanliness, bend sensitivity or route latency. Conversely, a live link with no post-FEC errors may still have too little guarded receiver margin.

Trace evidence should explain the physical plant. Connector inspection, cleaning record, bidirectional OTDR traces, OLTS reference method, splice averaging, gainer events, wavelength-dependent bend response, patch history and transceiver class should support the same route and service record. If they disagree, the release decision should stop at diagnosis.

Common Release Mistakes

• accepting one favorable OTDR trace while OLTS insertion loss or bidirectional evidence disagrees;
• using total attenuation only while receiver overload, dispersion, OSNR or FEC threshold governs;
• ignoring dirty connectors, reference-method errors, launch conditions and unrecorded patch changes;
• treating DDM power as calibrated acceptance evidence without meter and reference control;
• assuming a bend event is harmless without comparing wavelength response and retest after correction;
• using post-FEC zero errors without pre-FEC BER, confidence duration and receiver-margin evidence;
• releasing route diversity without proving physical diversity, protection switching and latency impact.

Scenario Map

Exercise	Main Calculation	Release Question
1	chromatic dispersion and optic substitution	does the spare optic preserve eye margin?
2	OLTS and OTDR reconciliation	do end-to-end loss and event evidence agree?
3	attenuator sizing	can the receiver avoid overload and weak-signal failure?
4	connector event margin	does a new event consume too much operating reserve?
5	route latency after protection switching	does the restoration path meet added-delay limits?
6	full optical link budget	is sensitivity margin still positive after reserve?
7	PMD comparison	is older fiber still acceptable at the same length?
8	multimode rise-time budget	does bandwidth-distance support the bit period?
9	optical return loss	does reflection evidence pass after cleaning?
10	bidirectional OTDR splice loss	is a gainer event acceptable after averaging?
11	PON splitter and distribution loss	does an ONT still have margin after engineering reserve?
12	optical amplifier OSNR	does ASE noise leave enough optical signal-to-noise margin?
13	dispersion compensation and added loss	does fixing dispersion consume too much receiver margin?
14	wavelength-dependent bend loss	does a bend event pass after tray correction and retest?
15	transceiver diagnostics power drift	will received-power drift reach warning or action margin soon?
16	WDM launch power and nonlinear loading	can OSNR margin be improved without exceeding per-channel or total span power limits?
17	OTDR dead-zone resolution	can close events be measured separately with enough trace dynamic range?
18	pre-FEC BER and receiver power margin	does zero post-FEC error evidence still have enough optical reserve?

Validation Package Checklist

Before treating a fiber calculation as release evidence, collect:

1. optical boundary, wavelength, service rate, transceiver class and route identifier;
2. source power, receiver sensitivity, overload limit, reserve and alarm thresholds;
3. OLTS insertion loss, OTDR traces, bidirectional events and reference method;
4. connector inspection, cleaning record, splice loss, patch history and repair notes;
5. chromatic dispersion, PMD, OSNR, return loss, bend loss and latency evidence;
6. pre-FEC BER, post-FEC counters, DDM trends and monitoring window;
7. calibration record, uncertainty allowance, acceptance threshold and trace storage;
8. release decision, cleaning or repair action, route restriction or retest requirement.

Exercise 1: Chromatic Dispersion and Optic Substitution

A 10\ \text{Gbit/s} service uses a single-mode fiber route of:

at 1550\ \text{nm}. The chromatic dispersion coefficient is:

The approved optic has a dispersion tolerance of:

A proposed spare optic has an effective source spectral width:

The engineering screening rule limits dispersion-induced pulse spreading to:

Check route dispersion tolerance and the proposed spare optic.

Solution

Route accumulated dispersion per nanometer is:

Compare with the approved transceiver tolerance:

The route passes the approved transceiver tolerance with:

of margin.

Now estimate pulse spreading for the spare optic:

Compare with the screening limit:

The spare optic fails this dispersion-spread screen.

Engineering Comment

The route can pass a transceiver dispersion tolerance while a broad-spectrum substituted optic still produces too much pulse spreading for the receiver decision margin. A release decision should therefore record the exact optic type, wavelength, source spectral width or vendor dispersion class, not only the fiber length.

Plausibility Check

At 10\ \text{Gbit/s}, one bit period is about:

A calculated spread of about 103\ \text{ps} is too large to ignore. It is comparable to a full bit period, so the fail result is physically plausible.

Exercise 2: Reconciling OLTS and OTDR Evidence

An installed 1550\ \text{nm} route is tested with an optical loss test set, or OLTS. The measured end-to-end insertion loss is:

The OTDR event table gives:

Item	Loss
fiber attenuation	7.6\ \text{dB}
connector events	1.1\ \text{dB}
splice events	0.7\ \text{dB}
bend event near closure C17	0.95\ \text{dB}

The acceptance rule is that OTDR event-sum loss should agree with OLTS insertion loss within:

Check agreement and identify the review action.

Solution

Sum the OTDR event losses:

Difference between OTDR and OLTS:

This is within the acceptance rule:

The loss evidence agrees.

The bend event is still significant:

It should be reviewed against the per-event acceptance limit, not ignored because total agreement is good.

Engineering Comment

OLTS is usually the stronger acceptance measurement for end-to-end insertion loss, while OTDR is stronger for locating events. Agreement between them supports the trace, but it does not automatically accept every event. A localized 0.95\ \text{dB} bend can become worse after temperature change, tray movement or repair work.

Plausibility Check

An OTDR/OLTS mismatch of 0.15\ \text{dB} is small compared with field uncertainty and reference-method variation. The event location, not total loss disagreement, is the engineering concern.

Exercise 3: Attenuator Sizing for Receiver Overload

A short fiber service uses a transmitter with:

and:

The passive path loss before any fixed attenuator is:

The receiver overload threshold is:

The release rule requires at least 3.0\ \text{dB} margin below overload under maximum transmitter power. Receiver sensitivity is:

and the weak-signal design reserve is:

Size a fixed attenuator and check weak-signal margin.

Solution

Received power without attenuator at maximum transmitter power is:

The maximum acceptable received power with overload margin is:

Required attenuator:

Choose a standard:

Now check minimum received power:

Weak-signal margin after design reserve:

The 7.0\ \text{dB} attenuator satisfies both overload and weak-signal checks.

Engineering Comment

Short fiber links can fail by being too strong, not too weak. The attenuator should be selected from both transmitter maximum and transmitter minimum cases. The release record should also state where the attenuator is installed, its wavelength rating, connector type and label, because later removal can create intermittent receiver overload.

Plausibility Check

The attenuator moves the worst-case receive level from +0.8\ \text{dBm} to about -6.2\ \text{dBm}, which is safely below the overload threshold. The minimum case remains far above sensitivity, so the result is credible.

Exercise 4: Connector Event and Remaining Service Margin

A live fiber service has baseline received optical power:

Receiver sensitivity is:

The operations rule reserves:

After a maintenance patch, an OTDR trace shows a new connector event with added insertion loss:

The release rule also requires at least 2.0\ \text{dB} residual operating margin after reserve. Check whether the service can be released without cleaning or repatching.

Solution

New received power is:

Margin above sensitivity after reserve:

The service remains above sensitivity after reserve, but it does not meet the residual operating margin rule:

The service should not be released without correcting the connector event or accepting a documented exception.

Engineering Comment

This is the kind of case where a link LED can be misleading. The service may pass traffic immediately after maintenance but have too little remaining margin for aging, temperature, vibration or another patch event. The correct action is connector inspection, cleaning, retest and updated OLTS/OTDR evidence.

Plausibility Check

A 1.7\ \text{dB} added connector loss is large for a single event. It is consistent with contamination, poor mating, damaged end face or a bad patch cord rather than normal connector variation.

Exercise 5: Fiber Route Latency and Protection Path Change

A protected service normally uses an 82\ \text{km} fiber route. After a fiber cut, traffic switches to a protection path of:

Use refractive index:

and:

Estimate one-way propagation delay for each path and the added delay after protection switching. The service allows an added one-way delay of:

from protection switching.

Solution

Propagation speed in fiber is approximately:

Normal route delay:

Protection route delay:

Added one-way delay:

Compare with the allowance:

The protection path passes the added-delay requirement.

Engineering Comment

Optical propagation delay is usually small compared with packet buffering, routing, encryption and queuing, but route changes still matter for timing-sensitive services. Protection design should record both availability improvement and delay change, especially for synchronization, control, protection signalling and measurement traffic.

Plausibility Check

Fiber delay is roughly 5\ \mu\text{s/km}. The normal route estimate is:

which agrees with the detailed calculation.

Exercise 6: Full Link Budget and Receiver Margins

A 52\ \text{km} single-mode route at 1550\ \text{nm} has fiber attenuation \alpha=0.22\ \text{dB/km}. The conservative loss budget includes 10 connector pairs at 0.30\ \text{dB} each, 24 fusion splices at 0.06\ \text{dB} each, a 1.2\ \text{dB} filter allowance and a 0.8\ \text{dB} repair allowance. The transmitter minimum is +2.0\ \text{dBm}, receiver sensitivity is -20.0\ \text{dBm} and design reserve is 3.0\ \text{dB}.

For overload, use transmitter maximum +5.0\ \text{dBm} and a minimum path loss of 15.46\ \text{dB}. Receiver overload threshold is -3.0\ \text{dBm} and required overload margin is 3.0\ \text{dB}.

Check sensitivity margin and overload margin.

Solution

Fiber loss:

Connector loss:

Splice loss:

Maximum path loss:

Minimum received power:

Sensitivity margin after reserve:

Maximum received power:

Overload margin:

The link passes the weak-signal reserve by 1.12\ \text{dB} and overload margin by more than the required 3.0\ \text{dB}.

Engineering Comment

The link is acceptable but not generous on sensitivity margin. Future patching, dirty connectors, repair splices or a lower-output optic can consume the remaining 1.12\ \text{dB}. The handover should identify which allowances are reserved and which changes require retest.

Plausibility Check

The fiber loss alone is 11.44\ \text{dB}, and the total conservative path loss is 17.88\ \text{dB}. That means connectors, splices and allowances add 6.44\ \text{dB}, a large enough share that connector practice and repair control matter as much as route length.

Exercise 7: PMD Margin for Modern and Older Fiber

A 96\ \text{km} route is being reviewed for a higher-rate service. Modern route records suggest D_{PMD}=0.08\ \text{ps}/\sqrt{\text{km}}, but an older undocumented segment could behave closer to 0.50\ \text{ps}/\sqrt{\text{km}}. The service screen uses DGD_{tol}=5.0\ \text{ps}.

Compare RMS differential group delay and PMD margin for both assumptions.

Solution

Length factor:

Modern-fiber DGD:

Modern-fiber margin:

Older-fiber DGD:

Older-fiber margin:

Engineering Comment

The same route length is comfortable if the modern PMD coefficient is valid, but almost out of margin if the older-fiber assumption governs. That is a documentation and validation issue: route records, fiber type, PMD test evidence and service bit rate must be tied together before release.

Plausibility Check

PMD scales with \sqrt{L}, not linearly with length. The older coefficient is 6.25 times larger than the modern coefficient, so the DGD estimate is also 6.25 times larger at the same route length.

Exercise 8: Multimode Rise-Time Budget

A 10\ \text{Gbit/s} multimode link is 0.30\ \text{km} long. The fiber bandwidth-distance product is 4700\ \text{MHz km}. Transmitter rise time is 35\ \text{ps}, receiver rise time is 40\ \text{ps} and filter rise time is 15\ \text{ps}. Use the screen:

Check the rise-time budget.

Solution

Available fiber bandwidth:

Fiber rise time:

Total rise time:

Bit period:

Allowed rise time:

Margin:

Engineering Comment

The link passes the simplified rise-time screen, but it is not proof of a clean eye diagram. Launch condition, modal noise, connector mode conditioning, transceiver standard, equalization and measured BER or eye margin still need validation.

Plausibility Check

The fiber contribution is smaller than the transmitter and receiver contributions, which is plausible for a short 300\ \text{m} link using high-bandwidth multimode fiber.

Exercise 9: Optical Return Loss Before and After Cleaning

At a connector panel, incident optical power is 0.0\ \text{dBm}. Before cleaning, reflected power is measured as -34.0\ \text{dBm}. After inspection and cleaning, reflected power is -41.0\ \text{dBm}. The return-loss requirement is at least 35\ \text{dB}.

Calculate return loss before and after cleaning and decide whether the connector event can be accepted.

Solution

For powers in dBm, return loss is the incident power minus reflected power:

Before cleaning:

Margin before cleaning:

After cleaning:

Margin after cleaning:

Engineering Comment

The connector fails before cleaning and passes after cleaning. This is exactly why inspection and cleaning records matter. A link can have acceptable insertion loss while reflection remains poor enough to disturb sensitive optics or create intermittent behavior.

Plausibility Check

Lower reflected power means higher return loss. Moving from -34\ \text{dBm} reflected to -41\ \text{dBm} reflected improves return loss by 7\ \text{dB}, matching the calculated change from 34 to 41\ \text{dB}.

Exercise 10: Bidirectional OTDR Splice Interpretation

An OTDR splice event is measured as +0.18\ \text{dB} from end A and -0.04\ \text{dB} from end B. The bidirectional-average event limit is 0.15\ \text{dB}, and the allowed directional discrepancy is 0.30\ \text{dB}.

Calculate bidirectional average loss, directional discrepancy and release decision.

Solution

Bidirectional average:

Directional discrepancy:

Average-loss margin:

Discrepancy margin:

The event passes both the bidirectional-average loss limit and the directional-discrepancy limit.

Engineering Comment

The one-direction +0.18\ \text{dB} value alone would look like a failure against a 0.15\ \text{dB} limit. Bidirectional averaging is important because OTDR event readings can be biased by backscatter differences across spliced fibers. The release record should still keep both traces.

Plausibility Check

A negative event in one direction can occur when the fiber after the splice backscatters more strongly than the fiber before it. That does not create optical gain; it is an OTDR interpretation artifact that bidirectional testing is meant to handle.

Exercise 11: PON Splitter Budget and ONT Margin

A passive optical network uses an OLT launch power of:

The distribution path to the ONT includes:

• 18\ \text{km} of fiber at 0.35\ \text{dB/km};
• one 1:32 splitter with loss 16.8\ \text{dB};
• six connector events at 0.25\ \text{dB} each;
• twelve splice events at 0.05\ \text{dB} each.

The ONT receiver sensitivity is:

and the engineering reserve is:

Compute received power and margin after reserve. Then check whether replacing the splitter with a 1:64 splitter of loss 20.5\ \text{dB} would still pass.

Solution

Fiber loss:

Connector loss:

Splice loss:

Total loss with the 1:32 splitter:

ONT received power:

Margin after engineering reserve:

The 1:32 design passes with 2.80\ \text{dB} of residual margin.

For the 1:64 splitter:

The 1:64 replacement fails the reserve requirement.

Engineering Comment

Splitter loss dominates many access-fiber budgets. A higher split ratio can look attractive for port utilization but consume all operating reserve. The release decision should include actual OLT class, ONT sensitivity class, wavelength, connector inspection, worst-case distribution length and whether future repair splices are already included.

Plausibility Check

Changing from 1:32 to 1:64 adds:

of splitter loss. The original residual margin was only 2.8\ \text{dB}, so the upgraded split should fail by about:

which matches the detailed calculation.

Exercise 12: Optical Amplifier OSNR Screen

A long fiber service enters an optical amplifier with signal power:

The amplifier gain is:

The amplifier noise figure is:

Use a simplified single-amplifier OSNR screen at 1550\ \text{nm} with reference bandwidth:

and photon energy:

Approximate spontaneous-emission factor as:

and amplifier ASE noise power as:

Check whether the output OSNR meets a 28\ \text{dB} requirement.

Solution

Output signal power:

Noise figure in linear units:

Spontaneous-emission factor:

Amplifier gain in linear units:

ASE power:

Convert ASE power to dBm:

OSNR:

OSNR margin:

The simplified OSNR screen passes.

Engineering Comment

This is a first-order amplifier screen, not a complete optical design. Cascaded amplifiers, filters, nonlinear penalties, channel spacing, launch power, receiver implementation and measurement bandwidth conversion can change the final margin. The useful decision is that one amplifier with these assumptions is not the limiting noise source by itself.

Plausibility Check

The amplifier raises the signal from -18\ \text{dBm} to +2\ \text{dBm}. The calculated ASE in the reference bandwidth is about -33\ \text{dBm}, so a mid-30-dB OSNR is plausible for a single moderate-gain amplifier. A margin of 7\ \text{dB} is meaningful, but not enough to ignore downstream penalties.

Exercise 13: Dispersion Compensation Module Loss and Receiver Margin

A 10\ \text{Gbit/s} single-mode service at 1550\ \text{nm} uses a route length of:

The chromatic dispersion coefficient is:

The transceiver dispersion tolerance is:

Before adding any dispersion-compensation module, the receiver power after the installed route is:

Receiver sensitivity is:

The engineering reserve is:

and the release rule requires at least:

of residual margin after reserve. A proposed dispersion-compensation module provides:

with insertion loss:

An alternate low-loss module provides:

with insertion loss:

Check uncompensated dispersion, residual dispersion after each module, receiver margin after each module and the release decision.

Solution

Uncompensated accumulated dispersion:

Uncompensated dispersion margin:

The route fails the uncompensated dispersion screen.

With the proposed DCM:

Dispersion margin after compensation:

Power margin before adding the DCM:

Receiver power after the proposed DCM:

Residual power margin after reserve:

The proposed DCM fixes dispersion but fails the residual power-margin rule:

Additional margin needed:

With the alternate low-loss module:

Receiver power after the alternate module:

Residual power margin after reserve:

The alternate module passes both dispersion and residual power-margin checks.

Engineering Comment

Dispersion compensation is not a free fix. A module can repair chromatic dispersion while adding enough insertion loss to break the optical power budget. The release decision should therefore treat dispersion margin and receiver margin as a coupled tradeoff, then validate the final path with measured power, BER or eye evidence and updated records.

Plausibility Check

The uncompensated route is only 324\ \text{ps/nm} beyond tolerance, so a DCM around -1100\ \text{ps/nm} leaves a large residual-dispersion margin. The power result is also plausible: a 4.6\ \text{dB} module consumes most of the original 5.3\ \text{dB} post-reserve power margin, while a 3.1\ \text{dB} module leaves slightly more than the required 2.0\ \text{dB}.

Exercise 14: Wavelength-Dependent Bend Loss Diagnosis

A service team finds a new non-reflective OTDR event in a fiber closure after maintenance. The service wavelength is 1550\ \text{nm}. The event is measured at two wavelengths before correcting the tray routing:

Measurement before correction	Event loss
1310\ \text{nm} OTDR	0.12\ \text{dB}
1550\ \text{nm} OTDR	0.86\ \text{dB}

After the tray is opened, a tight bend is corrected and the event is retested:

Measurement after correction	Event loss
1310\ \text{nm} OTDR	0.08\ \text{dB}
1550\ \text{nm} OTDR	0.21\ \text{dB}

The release rule requires:

and wavelength-dependent excess no greater than:

where:

The service had only:

of receiver margin after reserve before the bend correction. The release rule requires at least:

after reserve. Check the event before correction, after correction, the recovered service margin and the release decision.

Solution

Before correction, wavelength-dependent excess is:

The event fails the service-wavelength event limit:

and it also fails the wavelength-sensitivity trigger:

This is consistent with a bend-sensitive event rather than a simple uniform attenuation change.

After correction:

The corrected event passes the service-wavelength event limit:

and passes the wavelength-sensitivity check:

The loss improvement at the service wavelength is:

Updated receiver margin after reserve:

Margin above the release rule:

The corrected event passes the event rules and just meets the receiver-margin release rule.

Engineering Comment

Bend loss is often more severe at longer wavelengths, so comparing 1310\ \text{nm} and 1550\ \text{nm} traces helps separate a bend from a uniform splice or connector loss. The corrected result is acceptable but thin: a 0.05\ \text{dB} margin is not spare capacity for undocumented patching. The release package should keep the before/after traces, closure photo, tray-routing note, service-wavelength power check and a change-control trigger for any future work in the same closure.

Plausibility Check

Before correction, the 1550\ \text{nm} loss is more than seven times the 1310\ \text{nm} loss, which is a credible bend signature. After correction, the service-wavelength event falls to 0.21\ \text{dB} and the wavelength delta falls to 0.13\ \text{dB}. Recovering 0.65\ \text{dB} raises the service margin from 1.40 to 2.05\ \text{dB}, so the release is mathematically valid but operationally tight.

Exercise 15: Transceiver Diagnostics Received-Power Drift

A service was accepted with transceiver digital diagnostics recorded at commissioning. The received optical power was:

After 150 days, the same port reports:

The receiver sensitivity is:

The engineering reserve is:

The operations rule sets a warning when residual margin after reserve reaches:

and a required intervention when it reaches:

Use digital-diagnostics uncertainty:

as a guard band against optimistic received-power readings. Estimate drift rate, current guarded margin, time to guarded warning, time to guarded action and the operational decision.

Solution

Received-power drift expressed as a positive loss is:

Average drift rate is:

Current residual margin after engineering reserve is:

Guarded residual margin is:

The guarded warning threshold in measured received power is:

Remaining measured-power drop before the guarded warning is:

Time to guarded warning at the observed drift rate is:

The guarded action threshold is:

Remaining measured-power drop before the guarded action limit is:

Time to guarded action is:

The service is still above the guarded action threshold, but the warning threshold is close. The correct decision is to open a planned investigation now, compare the DDM trend with OLTS or a calibrated power measurement, inspect connectors and check whether the drift is common to one optic, one patch path or the whole route.

Engineering Comment

Transceiver diagnostics are useful for trend monitoring, but they are not a replacement for controlled acceptance measurements. Their value is operational: they can show gradual margin loss before the service drops. A drift that leaves 3.4\ \text{dB} guarded margin today but reaches warning in about 40 days should trigger scheduled evidence gathering, not a wait-for-failure response.

Plausibility Check

A 1.5\ \text{dB} received-power decline corresponds to a received-power ratio of:

Only about 71\% of the commissioning optical power remains at the receiver. That is large enough to be a real maintenance signal, but it is not yet a complete loss-of-service condition because the guarded residual margin remains positive.

Exercise 16: WDM Launch Power and Nonlinear Loading

A WDM fiber route is being tuned after an OSNR review. The operations team proposes per-channel launch power into the span of:

There are:

active wavelength channels. The span engineering rule limits total launched optical power to:

and also limits per-channel launch power to:

At the proposed launch power, the measured OSNR is:

The receiver requires:

and the release rule requires:

of OSNR reserve above the receiver requirement. Assume OSNR changes approximately 1\ \text{dB} for each 1\ \text{dB} launch-power change while amplifier noise is otherwise fixed. Check whether the proposed launch power is acceptable, find the maximum allowed per-channel launch power, and decide whether the corrected setting can be released.

Solution

Total WDM launch power for equal channel powers is:

For the proposed setting:

Total-power margin:

The proposed setting exceeds the total launch-power limit. It also exceeds the per-channel limit:

Now find the per-channel power allowed by the total span limit:

The maximum allowed per-channel setting is the lower of the total-limit value and the per-channel limit:

Power reduction from the proposed setting:

The corrected total launch power is:

Estimated OSNR after reducing launch power:

Guarded OSNR requirement:

Corrected OSNR margin:

The proposed +3.0\ \text{dBm} per-channel launch is not acceptable because it violates nonlinear power-loading limits. The corrected +1.97\ \text{dBm} per-channel setting satisfies both power limits and still clears the guarded OSNR requirement by about 0.27\ \text{dB}, so it is only a thin conditional release.

Engineering Comment

Increasing launch power can improve OSNR, but it is not a free margin source in a WDM span. High total optical power and high per-channel power can trigger nonlinear penalties such as self-phase modulation, cross-phase modulation, four-wave mixing or vendor-specific coherent-system limits. A setting that improves a simple OSNR number can still degrade the link if the span is overdriven.

A credible release should include channel count, per-channel launch powers, total launched power, amplifier settings, gain tilt, spectrum trace, nonlinear or vendor design limits, OSNR measurement bandwidth, receiver FEC margin, temperature range and rollback threshold. The 0.27\ \text{dB} OSNR surplus in this exercise is not enough to ignore measurement uncertainty or future channel additions.

Plausibility Check

Eight equal channels add 10\log_{10}(8)=9.03\ \text{dB} above one channel, so a +3.0\ \text{dBm} per-channel setting creates about +12.0\ \text{dBm} total launch power. Reducing each channel by about 1.0\ \text{dB} lands exactly on the +11.0\ \text{dBm} total-power rule. Losing about 1.0\ \text{dB} of OSNR while recovering nonlinear headroom is a plausible optical-design tradeoff.

Exercise 17: OTDR Dead-Zone and Close-Event Resolution

An OTDR trace shows two events in a closure:

• a reflective connector at (1220\ \text{m});
• a splice or bend event at (1244\ \text{m}).

The event separation is therefore:

A long-pulse OTDR setting gives:

and dynamic range:

A short-pulse setting gives:

and dynamic range:

The route end-to-end loss is:

The trace review requires at least:

of dynamic-range margin beyond route loss. Decide whether each pulse setting can resolve the two events as separate reflections, whether it can measure their losses separately, and which trace should be used for close-event release evidence.

Solution

Long-pulse event-resolution margin:

The long pulse can show the events as separate reflections.

Long-pulse attenuation-resolution margin:

The long pulse cannot reliably measure the event losses separately because the attenuation dead zone is longer than the event separation.

Long-pulse dynamic-range margin:

Surplus beyond the required dynamic-range margin:

The long pulse has strong dynamic range but insufficient close-event loss resolution.

Short-pulse event-resolution margin:

Short-pulse attenuation-resolution margin:

The short pulse can both separate the reflections and measure the close-event losses under the stated dead-zone screen.

Short-pulse dynamic-range margin:

Surplus beyond the required margin:

The short pulse still has enough dynamic range for this route, but with less reserve than the long-pulse trace.

The release package should use the short-pulse trace for the close-event loss decision and keep the long-pulse trace for whole-route visibility and far-end confidence.

Engineering Comment

OTDR pulse width is a tradeoff. A longer pulse improves dynamic range but makes closely spaced events harder to measure separately. A shorter pulse improves event resolution but can reduce trace reach and signal-to-noise ratio. Close-event release evidence should state event dead zone, attenuation dead zone, pulse width, wavelength, launch fiber, receive fiber, route loss and whether the trace is being used to locate an event or to accept event loss.

Plausibility Check

The events are (24\ \text{m}) apart, so both pulse settings can distinguish reflections if the event dead zone is (8\ \text{m}) or less. The problem is attenuation dead zone: (35\ \text{m}) is longer than the separation, so the long pulse cannot support a separate loss measurement. The short pulse reduces attenuation dead zone to (12\ \text{m}), leaving (12\ \text{m}) of separation margin while still keeping (3\ \text{dB}) of dynamic-range surplus beyond the route-loss requirement.

Exercise 18: Pre-FEC BER and Guarded Receiver-Margin Release

A (100\ \text{Gbit/s}) fiber service is being accepted after a repair. During a (300\ \text{s}) traffic test, the receiver reports:

pre-FEC corrected bit errors and zero post-FEC uncorrected errors. The vendor FEC threshold is:

The release policy requires pre-FEC BER to stay below (80%) of the vendor threshold:

The measured received power is:

The receiver sensitivity associated with the vendor FEC threshold is:

Power-meter and DDM comparison uncertainty is:

The operations rule requires at least:

of guarded optical power margin beyond the sensitivity point. Calculate the pre-FEC BER, guarded BER limit, error-count headroom, guarded receiver power margin and release decision.

Solution

Total tested bits are:

Measured pre-FEC BER is:

The guarded BER release limit is:

Pre-FEC BER margin is:

The corresponding guarded error-count limit for the same test duration is:

Error-count headroom is:

As a fraction of the guarded limit:

The raw optical power margin above sensitivity is:

Guarded optical power margin is:

Reserve deficit is:

The pre-FEC BER passes the guarded FEC threshold and the post-FEC counter shows no uncorrected errors during the test. However, the guarded receiver power margin is only (1.0\ \text{dB}), below the required (1.5\ \text{dB}). The service should be held for optical-margin correction or accepted only with an explicit engineering exception and monitoring action.

Engineering Comment

Zero post-FEC errors are not enough for fiber release when the pre-FEC error rate is already close to the guarded threshold and optical margin is thin. FEC can hide a degrading optical path until temperature, connector movement, additional patch loss or transmitter drift pushes the receiver past the correction cliff.

Release evidence should include raw and guarded received power, pre-FEC BER, post-FEC errored seconds, corrected-codeword trend, transceiver model, FEC threshold basis, temperature, wavelength, connector inspection, OLTS comparison and alarm thresholds. If power margin fails while BER still passes, the correct action is to restore margin or tighten monitoring, not to claim the link is healthy from zero post-FEC errors alone.

Plausibility Check

A (100\ \text{Gbit/s}) service tests (3.0\times10^{13}) bits in (300\ \text{s}), so tens of billions of corrected pre-FEC bit errors can still correspond to a BER near (10^{-3}). The measured (1.4\times10^{-3}) BER is below the guarded (1.6\times10^{-3}) limit, but only by (12.5%) of the guarded error count. A guarded optical margin deficit of (0.5\ \text{dB}) is therefore enough to block unconditional release.

Review Checklist

Before accepting a fiber-optic calculation, confirm:

1. power levels and losses are referenced to the correct optical boundary;
2. both receiver sensitivity and receiver overload are checked;
3. dispersion is checked against the actual optic, wavelength and any compensation-module loss;
4. residual dispersion and receiver margin are both recalculated after compensation;
5. PMD and multimode rise-time screens are included when route length, bit rate or fiber type makes them relevant;
6. OLTS and OTDR evidence are used for the right purpose;
7. OTDR pulse width, event dead zone, attenuation dead zone and dynamic range support the event decision;
8. connector events are inspected, cleaned and retested when margin is thin;
9. return loss is checked where reflection-sensitive optics or poor connector evidence are credible;
10. bidirectional OTDR evidence is used when event interpretation depends on direction;
11. wavelength-dependent bend loss is checked with service-wavelength evidence and a second wavelength when bend risk is credible;
12. route diversity decisions include latency and operations impact;
13. PON splitter ratios are checked against receiver class and reserve, not only port count;
14. optical-amplifier OSNR screens state bandwidth, gain, noise figure and downstream penalties;
15. transceiver diagnostics are compared with commissioning baselines and guarded for measurement uncertainty before triggering operations decisions;
16. WDM launch-power settings satisfy both per-channel and total span-loading limits before OSNR margin is accepted;
17. pre-FEC BER, post-FEC errors and receiver power margin are checked together before accepting a corrected link;
18. validation records include instrument settings, reference method, wavelength, date, technician, trace files and acceptance thresholds.

Common Mistakes

• Mixing dBm power levels with dB losses or margins, then adding and subtracting quantities with different meanings.
• Checking receiver sensitivity but forgetting receiver overload, attenuator sizing and launch power variation.
• Treating OLTS and OTDR results as interchangeable instead of using OLTS for end-to-end insertion loss and OTDR for event location and interpretation.
• Accepting close OTDR events from a long-pulse trace without checking event dead zone, attenuation dead zone and dynamic range.
• Accepting a one-way OTDR event when bidirectional evidence is needed to separate real splice loss from backscatter mismatch or gainer behavior.
• Cleaning a connector but failing to preserve before/after inspection, loss and trace evidence when service margin is thin.
• Approving a spare optic from reach label alone while wavelength, source width, dispersion tolerance, receiver class and vendor limits differ.
• Fixing chromatic dispersion with a compensation module but consuming the receiver power margin with added insertion loss.
• Raising WDM launch power to recover OSNR without checking nonlinear loading, total span power, channel count and vendor limits.
• Ignoring PMD, multimode bandwidth, route latency or protection-path delay because total attenuation alone looks acceptable.
• Diagnosing a bend from a single trace without checking service wavelength, second-wavelength behavior, tray routing and post-correction margin.
• Treating DDM received-power drift as exact calibrated power, or ignoring a trend that predicts warning and action thresholds before failure.
• Accepting zero post-FEC errors while pre-FEC BER headroom, receiver power margin and alarm thresholds are already thin.

A fiber link is ready for service when the calculations, traces, measured power and operational limits tell the same engineering story.