# OTDR Trace: How to Read Events, Dead Zones and Apparent Gain

**Author:** Omar Molina

**Published:** October 20, 2026

**Last updated:** October 20, 2026

**Category:** FTTH Design

**URL:** https://draftech.com/blog/otdr-trace

**Title tag:** OTDR Trace Reading: Events, Dead Zones and Gainers 2026

**Meta description:** OTDR trace reading explained: launch fibers, backscatter slope, reflective events, ghosts, dead zones, pulse width, averaging and bidirectional splice analysis.

**Visible word count:** 2518

---

A clean event table can still leave an engineer with the wrong explanation for a fiber problem. A splice can appear to add light, and a reflection can appear beyond the physical cable, so we read the plotted signal before accepting labels whose accuracy depends on what the acquisition settings let the instrument distinguish.

This guide follows the trace from the launch connection to the far end, then explains which settings change the evidence. We focus on interpreting a point-to-point fiber measurement and identifying when another acquisition is needed. For contractual pass/fail decisions, use our separate guide to [OTDR testing acceptance criteria and fiber splice loss](https://draftech.com/blog/otdr-testing-acceptance-criteria-fiber-splice-loss).

## What an OTDR trace tells you

An OTDR trace plots returned optical signal against distance along a fiber. Its 2 axes let you locate changes and compare backscatter levels around an event. Read the launch first, follow the descending baseline, inspect each interruption, and confirm where usable backscatter ends before assigning a physical cause.

### Distance comes from travel time

The instrument sends a pulse and measures the light returning toward its test port. Round-trip travel time becomes distance using the configured group index of refraction, a setting we check against the fiber information before using an event location to direct field work. A wrong index shifts the distance scale even when the event shapes look credible.

**Optical distance and route distance differ.** Slack stored in a closure and the fiber's path inside the cable affect that comparison. We match the event to the cable record before assigning it to a structure on a map. A cursor reading cannot identify the enclosure. That limitation matters especially where several closures sit near one another.

### Read the backscatter slope between events

Rayleigh backscatter supplies the distributed signal. Between discrete features on a uniform fiber section, the baseline trends downward as attenuation accumulates. EXFO's article *What is an optical time domain reflectometer (OTDR)?* explains this relationship between backscatter level and attenuation. A local step deserves separate analysis from the gradual slope across the surrounding span.

We place measurement regions on settled sections of that baseline. A region that crosses a reflection's recovery tail or reaches into noise can distort the fitted slope and the loss assigned to a nearby event. It does not improve the underlying measurement resolution, so a larger view of an unresolved feature remains unresolved.

## Establish the launch and link boundaries

The trace begins at the instrument connection, which may produce a strong reflection and receiver recovery region. A launch fiber provides measurable backscatter before the first network connector. We identify the end of that lead explicitly so its length and attenuation are kept separate from the installed link being reviewed.

### Give the first connector a measurable baseline

Choose a suitable launch lead. It must suit the selected acquisition and connection being measured, because simply attaching a lead does not prove the first connector is measurable. Check that the receiver has recovered before that connector and that enough settled backscatter exists to support the measurement.

The far end has the same boundary problem: a receive fiber supplies backscatter beyond the final network connector so that its loss can be evaluated from that baseline. EXFO's *A beginner's guide to OTDR Testing: Acquisition, trace analysis and automation*, revision 20210115V5, marked October 2022, illustrates both lead arrangements. With a receive lead attached, the last network connector and the final fiber end are different locations.

**Keep the original acquisition.** Save the native trace before changing its analysis markers. We want the next engineer to see the original signal and understand why an event boundary was moved.

Once the boundaries are established, use the shape of each interruption to decide what evidence to gather next. The table distinguishes observations from diagnoses. We still compare each feature with the planned splice arrangement; an automated connector symbol does not establish that a connector was installed there.

| Trace feature | Initial interpretation | Useful next check |
| --- | --- | --- |
| Launch reflection | Instrument connection and receiver recovery | Confirm a settled launch baseline |
| Descending baseline | Distributed attenuation along the fiber | Keep fitting regions clear of events |
| Peak with a level change | Reflective connection or discontinuity | Compare location with the connection record |
| Step without a peak | Non-reflective loss or backscatter change | Compare wavelength and reverse direction |
| Repeated isolated peak | Possible ghost reflection | Find the earlier reflective source |
| Backscatter disappears | Possible fiber end or insufficient signal | Verify expected length and acquisition reach |

## Recognize OTDR trace events without overcalling them

### Reflective and non-reflective changes

A reflective event sends a concentrated return toward the instrument and appears as a peak. Connector interfaces can create that feature, but a crack or an open fiber end can also reflect. Assess the surrounding backscatter separately. Its levels before and after the peak describe event loss, while the peak height concerns reflectance; those measurements describe different behavior.

A non-reflective event usually appears as a step without a distinct peak, with a fusion splice as a familiar candidate, although a bend can also produce localized loss. Fluke Networks' knowledge-base article *Event types and its possible causes - DTX Compact OTDR Module*, dated March 3, 2014, describes these possible causes. The shape narrows the investigation, but it does not identify the hardware by itself.

For a suspected bend, compare acquisitions at the wavelengths required by the test plan and inspect the corresponding location. Greater loss at a longer wavelength can support a bend diagnosis. Keep the event positions aligned before comparing them. We also consult the [splice-point placement in the low-level design](https://draftech.com/blog/ftth-lld-splice-point-placement-guide) to distinguish a planned joint from an unexpected loss location.

### Ghosts repeat a reflection

A ghost is an apparent event created by repeated reflections, rather than a separate physical connection at the plotted distance. Look for related distances. Little or no corresponding loss offers another clue. FOA's *OTDR FAQs* describes ghosts that appear in the noise beyond a cable's end. We look upstream for the strong reflection that could explain the extra travel time.

Keep suspicious peaks in the table. A complicated table is no reason to delete one. Inspect the suspected source connection and repeat the acquisition after any corrective action. If the feature persists and its origin remains uncertain, keep it marked as unresolved rather than turning a plausible interpretation into a confirmed splice location.

### The end of fiber needs corroboration

An open end can produce a final reflection followed by noise, whereas a low-reflection termination or an angled break may end without producing a prominent peak. Backscatter disappearing early can also mean the instrument has run out of usable signal. We compare the observed endpoint with the expected optical length and the receive-lead arrangement before calling it a break.

Check the selected range. A window ending too soon can hide the actual endpoint. Extend the range when the link exceeds the displayed window, then assess whether the signal remains distinguishable from noise. Where the physical route and the trace disagree, retain both records because the discrepancy remains useful evidence until the lead lengths and fiber identity have been reconciled.

## Separate pulse width, averaging and dead zones

Acquisition settings change the evidence available for interpretation. We choose them around the feature that needs resolving, while preserving enough reach to inspect the far end. FOA's *OTDRs* reference recommends starting with the shortest pulse that can reach the cable's end with reasonable averaging. A single overview acquisition may leave closely spaced features ambiguous.

### Pulse width changes reach and separation

A longer pulse puts more energy into the fiber and helps the instrument measure farther into a lossy link, but it also worsens separation of nearby events and lengthens dead zones. A shorter pulse improves detail but may leave the far end buried in noise, so we retain complementary acquisitions when a short pulse resolves the near end and a longer pulse is needed farther away.

Nearby connectors and splices expose this tradeoff. With a broad pulse, their responses can overlap and appear as a combined feature. Increasing the display magnification cannot separate information the acquisition never resolved. Our preference for a readable overview has a weakness here: it can hide the very event that explains a local problem.

### Averaging changes noise

Longer acquisition time lets the instrument combine more returns and reduce random noise. VIAVI's knowledge-base note *Improving a poor OTDR trace*, dated October 16, 2025, recommends checking pulse width and increasing acquisition time when the trace is unclear. Averaging can stabilize the baseline. Receiver saturation remains a limitation. Events merged by an overly long pulse remain inseparable.

**Change one setting at a time.** If a feature changes after both pulse width and averaging time change, the comparison cannot isolate which adjustment helped. Keep each acquisition's settings attached to its file.

### Event and attenuation dead zones answer different questions

The event dead zone describes the separation needed to distinguish a subsequent reflective event after a reflection. The attenuation dead zone describes the recovery needed to measure a following loss accurately. EXFO's OTDR fundamentals article distinguishes these specifications and their test conditions. Detecting another peak therefore does not guarantee a trustworthy loss measurement at that location.

Read the conditions alongside dead-zone specifications. Their pulse width and reflectance conditions matter; a brochure's shortest figure cannot be assumed for every field acquisition. When the receiver is still recovering, label the measurement limitation and obtain another suitable trace. For an FTTH path containing splitters, verify the test topology and instrument capability before interpreting overlapping branch returns as one continuous point-to-point route.

## Resolve gainers with bidirectional event averaging

An upward step at a passive splice is an apparent gain caused by the returned signal, not optical amplification. Fibers with different backscatter characteristics can make a joint look better from one direction and worse from the other. Fluke Networks explains this in *Cable Testing 101: There's No Gain with ?Gainers?*. We keep the signed values until the paired measurement is complete.

### Match the same physical event from both ends

Bidirectional event analysis uses measurements from A to B and B to A at the same wavelength, matching each joint by its physical position within the link while accounting for the opposite distance origin. Because event numbering may reverse and software may detect different numbers of events, we check that the pair represents the same joint before calculating an averaged loss.

**The averaged event loss is (A-to-B loss + B-to-A loss) / 2.** Retain a negative sign on an apparent gainer; do not convert it to a positive loss or replace it with zero. The signed average offsets the directional backscatter bias by averaging corresponding event losses in decibels, a calculation that differs from overlaying entire plots or averaging unrelated peaks.

Both measurements need usable baselines. A paired result remains questionable if one direction places the splice in a dead zone or below the noise floor. We request a better acquisition. Arithmetic cannot cure weak data. Temporal averaging within an acquisition and bidirectional averaging across an event solve different problems.

Trace review becomes easier when the design identifies which fibers meet at each joint. Our [fiber network design and connection planning](https://draftech.com/services/fiber-network-design) connects those records to the intended optical path. We recommend carrying the measured event back to that connection record so a direction-dependent anomaly does not become an unsupported instruction to rebuild a splice.

## Choose the next action and preserve the evidence

**For field technicians:** Reacquire an ambiguous event before leaving the accessible endpoint. Use the setting change that addresses the actual limitation, and preserve the original file beside the replacement. If the task needs a reverse-direction measurement, arrange access to the other end rather than closing the issue from an incomplete event table.

**For design reviewers:** Reconcile the fiber path before recommending rework. Our guide to [what fiber as-built drawings should include](https://draftech.com/blog/fiber-as-built-drawings-what-to-include) explains the records needed to connect measured events with installed assets. Resolve a disagreement between the trace and the splice schedule explicitly; neither file should quietly overwrite the other.

**For network owners:** Require trace files that can be reopened and associated with the tested fiber. Preserve the direction and acquisition settings. Keep them available with the result. Assign unresolved features to someone who can obtain the missing evidence, and apply the project's acceptance requirements only after the measurement supports a defensible interpretation.

Draftech's engineering is 100% in-house, so the connection plan and the engineering review have a common point of accountability. We connect unexplained trace locations with the design records that field teams need to investigate them. Where the scope includes construction, Draftech delivers full turnkey construction through Draftech-managed subcontract crews. Active in 24 states. Available across all 50 U.S. states.

If a new route needs that connection planning before field testing begins, consider our [qualifying-route design offer](https://draftech.com/free-design). Draftech engineers the first 20,000 linear feet at no cost, from feasibility and field survey through permit approval. The owner reviews each request before Draftech commits the package, so the proposed route and its engineering scope are reviewed together. [Email your route scope](mailto:info@draftech.com) or use our [contact page](https://draftech.com/contact) to start that review.

## Frequently Asked Questions

### What does the slope on an OTDR trace mean?

The descending baseline represents attenuation along a fiber section as returned backscatter decreases with distance. Read the 2 axes together and place analysis regions on settled fiber between events. A localized step requires a different interpretation from the gradual span slope, and regions extending into a reflection recovery tail or the noise floor can produce misleading measurements.

### Why does a splice appear to have negative loss?

Different backscatter characteristics on either side of a splice can create an apparent gainer from one direction. Measure the same event in 2 directions at the same wavelength and average the signed event losses. Preserve the negative value in that calculation, and confirm that both acquisitions resolve the event before relying on the result for engineering review.

### Can longer averaging remove an OTDR dead zone?

Longer averaging reduces random noise by collecting more returns, but it does not restore information hidden during receiver recovery. The 2 dead-zone specifications describe event detection and attenuation measurement, which are different capabilities. To examine a nearby feature, review pulse width and the reflection causing recovery, then acquire another trace with a setup suited to that separation.

### How can I distinguish a ghost from a real event?

Compare the suspicious peak with earlier reflections and check whether it has a corresponding loss step. Looking at 2 related peaks can reveal repeated travel distances, but spacing alone is insufficient proof. Inspect the suspected reflective source and repeat the measurement after correction; preserve an unresolved label if the available evidence cannot establish whether the event is physical.

### Why use both launch and receive fibers?

The 2 leads provide backscatter on opposite sides of the installed link: a launch fiber before the first network connector and a receive fiber after the last. Those baselines support connector-loss measurements when the receiver has recovered. Identify the network boundaries explicitly so lead lengths remain separate from installed fiber length and the final connector is distinguished from the receive-fiber end.