Optical Return Loss vs Reflectance: Reading Reports

Summary: The distinction between optical return loss vs reflectance in test reports is both mathematical and physical. Reflectance usually describes returned power from an individual interface and is reported as a negative dB value. Link ORL usually combines returned energy over a defined span and is reported as a positive dB value. Always read the measurement boundary as well as the sign.

A supplier can provide a good connector reflectance result without proving the installed link meets its total ORL requirement. Conversely, a link-level result does not identify which individual connector is responsible for the returned power. This guide explains how to ask for the correct evidence.

Define the ratio before comparing numbers

Let the returned-power fraction at a stated reference plane be r = returned power divided by incident power. With consistent measurement boundaries, reflectance is 10 log10(r), while return loss is −10 log10(r). For ordinary passive fiber interfaces, r is less than one, making the former negative and the latter positive.

The FOA explanation of reflectance and return loss distinguishes an individual reflecting interface from total link return, including fiber backscatter. That boundary distinction is essential: changing the sign of one connector result does not convert it into the ORL of an entire network.

These are mathematical examples, not guaranteed connector grades. A ten-decibel improvement corresponds to ten times less returned power under the same definition. It does not mean ten decibels less insertion loss.

Why multiple reflections cannot be added in dB

Consider a deliberately simplified example with two independent returned-power contributions, each equivalent to −50 dB at the same input reference plane. Assume negligible intervening attenuation, no distributed backscatter and incoherent power addition. Each contribution equals 0.00001 of incident power.

The combined fraction is 0.00001 + 0.00001 = 0.00002. Therefore the corresponding return loss is −10 log10(0.00002), approximately 46.99 dB. It is not 100 dB, and it is not exactly 50 dB. Combining two equal returns increases the total returned power.

Add a stronger contribution

Now add a third hypothetical contribution of −40 dB at that same reference plane. Its fraction is 0.0001, so the new total is 0.00012. The simplified combined return loss is approximately 39.21 dB. The stronger reflector dominates, but the smaller contributors still change the result.

This educational model is not a complete field-ORL prediction. Real links contain distributed backscatter, propagation loss, wavelength dependence and potentially interference effects. The calculation is useful because it exposes why arithmetic addition or averaging of dB ratings is the wrong operation.

Account for the trip to the reflector and back

A local connector reflectance and its contribution at the transmitter reference plane are different quantities. If a hypothetical connector has −40 dB local reflectance and lies after 2 dB of one-way attenuation, its returned contribution at the input is approximately −44 dB, assuming the same attenuation in the reverse direction and ignoring other effects.

The extra four decibels represent the outward and return paths. Record whether a reported number has already been referenced to the input before applying this correction. Applying it twice would understate the returned power. In branched or wavelength-selective systems, use the actual path and equipment measurement method.

Match the instrument output to the purchase requirement

EXFO’s OTDR fundamentals describes event reflectance and total ORL among the instrument’s analysis outputs. They serve different purposes. One helps locate an offending interface; the other characterizes return over a selected span. The report should identify which output is being used for acceptance.

Ask the supplier to name the test method, wavelength, reference plane, included interfaces and far-end termination. For an assembly test, identify the reference connector and whether the result covers each end separately. A value copied from a generic connector datasheet is not a serialized measurement of the supplied assembly.

Separate resolution from uncertainty

A display with two decimal places does not imply uncertainty of 0.01 dB. Request the instrument’s applicable measurement range and uncertainty conditions, together with calibration status and the test setup. Near a contractual threshold, the acceptance rule should explain how uncertainty and retesting are handled.

Do not substitute an instrument’s best-case specification for every field configuration. A highly reflective termination, limited signal range or an unsuitable analysis region can affect the usefulness of the result. Preserve any warnings shown by the instrument rather than copying only the favorable numerical value.

An acceptance sheet that avoids sign mistakes

Create separate rows for insertion loss, individual-interface reflectance and link ORL. Give each row a unit, a sign convention, a boundary and a pass direction. A reflectance requirement might use an upper limit expressed as a negative number, while a positive ORL requirement normally uses a minimum.

For example, under a hypothetical requirement of reflectance no greater than −45 dB, a −50 dB result satisfies the numerical limit and −40 dB does not. Under a separate hypothetical ORL requirement of at least 35 dB, a 38 dB result satisfies the numerical limit and 32 dB does not. These thresholds illustrate reading direction only; they are not recommended application limits.

Keep supplier and installed-link evidence distinct

The factory assembly may be tested against reference connectors under controlled conditions. The installation introduces field mating partners, patch panels and the complete fiber span. Retain both records rather than replacing the factory report with a system measurement or assuming one automatically validates the other.

If the results disagree, compare boundaries and conditions before treating the difference as product deterioration. Identify the specific interfaces included in each test and whether a remote open end was present. This often prevents a lengthy argument over numbers that were never measuring the same object.

Resolve three common report disagreements

Different signs: ask both parties to express the underlying returned-power fraction at the same boundary. If one report uses positive return loss and the other negative reflectance for that identical boundary, the apparent disagreement may be a convention difference. Do not change the sign until the measured object has also been confirmed.

Different span lengths: identify what each test includes. A short assembly result and a long installed-link result have different potential contributions. Record the included fiber, connections and termination, then compare each result with its own applicable limit. Neither value should be substituted into the other’s acceptance row without a justified measurement model.

Different mating partners: a delivered connector is tested as part of a connection. The installed partner may not have the same condition as the reference interface used at the factory. Preserve the assembly identity, inspect both sides and agree on a repeat method before assigning responsibility. Replacing several parts at once can restore service while leaving the cause of the disagreement unknown.

For each case, retain a short decision record containing the original values, definitions, relevant drawings, test conditions and the agreed interpretation. This is especially helpful when procurement and engineering teams use different shorthand. A clear report should remain understandable after the people who performed the initial tests are no longer involved in the project.

For a purchase inquiry, specify the required evidence on our fiber optic patch cords and assemblies page. Include the connector interfaces, polish, operating wavelengths and project acceptance method. Do not request a return-loss number without identifying what must be tested.

The UPC versus APC connector guide addresses polish selection, while the low-loss patch-cord acceptance guide covers transmitted-power evidence. Keeping these topics separate helps the quotation describe the complete assembly requirement without confusing loss, reflection and mechanical compatibility.

The final review should answer three questions: what returned the light, where the ratio was referenced, and which limit applies. Once those are explicit, the signs and calculations become much easier to interpret consistently.