An optical time domain reflectometer works on a fundamentally different principle than an OLTS, and understanding that difference is the key to reading a trace correctly. An OTDR sends a short pulse of laser light down the fiber and continuously samples the light that comes back, over time, at the same end the pulse was launched from. Two physical effects produce that return signal: Rayleigh backscatter, a small, continuous, distance-dependent scattering of light off microscopic density variations in the glass itself, and Fresnel reflection, a much stronger, localized reflection that occurs at any point where the fiber's refractive index changes abruptly, such as a connector interface, a mechanical splice, a crack, or the far end of the fiber. The OTDR converts elapsed time into distance using the fiber's group index of refraction, and it converts returned optical power into a logarithmic trace where continuous backscatter appears as a gently sloping line and each reflective or non-reflective event appears as a distinct feature on top of that slope.
The slope of the backscatter line itself represents fiber attenuation, and its steepness in dB per division corresponds to the fiber's loss per kilometer at the test wavelength. Events appear as either a sharp upward reflective spike followed by a drop, in the case of a connector or an open mechanical splice with an air gap, or as a simple downward step with no spike, in the case of a fusion splice, a bend, or any junction with good index matching and no significant Fresnel reflection. A testing specialist has to read both the slope and the events correctly, because the slope alone tells you about the fiber's intrinsic quality and the events tell you about the workmanship and hardware installed along the route, and conflating the two leads to misdiagnosis.
Reading Events: Reflective, Non-Reflective, and Everything in Between
A reflective event on a trace shows a sharp rise above the backscatter line, often followed immediately by a drop below where the backscatter would have continued had the event not been there, and the OTDR software calculates both an insertion loss (the vertical step down in the trace, from just before the event to just after it, once the reflective spike has settled) and a reflectance value (how strong the spike itself was, expressed as a negative dB figure) for each such event. Connectors nearly always show up as reflective events because even a well-mated PC connector has some residual Fresnel reflection, typically in the negative 30 to negative 55 dB range as covered in the previous module, and a poorly mated or contaminated connector shows a taller spike and can even saturate the OTDR's receiver briefly, which is one reason a badly performing connector is often obvious on a trace well before its exact loss value is calculated.
Non-reflective events show only a step down in the backscatter trace with no spike at all, which is the signature of a good fusion splice, since a well-executed fusion joint leaves no meaningful air gap or index discontinuity to reflect light. A macrobend severe enough to cause measurable loss but not a break also shows as a non-reflective, or sometimes weakly reflective, step, and this is one of the trickier interpretation calls in the field because a bend event and a mediocre fusion splice can look very similar on a trace; distinguishing them sometimes requires testing at two wavelengths, since macrobend loss increases significantly at 1550 nm relative to 1310 nm while splice loss does not change nearly as much between wavelengths. A testing specialist who sees a non-reflective event with unusually different loss at 1310 versus 1550 nm should suspect a bend rather than a splice, even without visiting the physical location.
Dead Zones, Ghosts, and the Limits of What an OTDR Can See
Every OTDR has a dead zone, a short distance immediately following a strong reflective event during which the receiver is still recovering from the intensity of that reflection and cannot reliably detect or measure another nearby event. There are two dead zone specifications that matter: event dead zone, the minimum distance after a reflective event at which another event can be detected as separate at all, and attenuation dead zone, the somewhat longer distance required before the trace returns cleanly to the backscatter line and loss measurements become accurate again. This matters enormously in short premises links, where two connectors or a connector and a nearby splice can sit closer together than the OTDR's dead zone allows, causing the instrument to merge them into what looks like a single event or to miss one entirely. Using a shorter pulse width narrows the dead zone at the cost of reduced dynamic range and distance capability, which is why pulse width selection is itself a judgment call tied to the length and event density of the specific link under test, not a single default setting appropriate for every job.
Ghost events, sometimes called ghost reflections, are a separate and often confusing phenomenon: a strong reflective event, particularly an unterminated connector or an open fiber end, can send enough light back toward the OTDR that a portion of it reflects again off the OTDR's own front-panel connector or another strong reflector closer to the instrument, and that re-reflected light shows up further out on the trace as a fake event that does not correspond to any real physical location on the fiber. Ghosts are recognizable because their apparent loss is often negative, showing the trace rising rather than falling, since no real fiber event returns power that exceeds what physics allows at that distance, and because a ghost's apparent position is mathematically related to the distance of the real strong reflector that caused it, often appearing at roughly double that distance or at a distance matching a reflection path through the instrument itself. A testing specialist who fails to recognize a ghost as a ghost can end up reporting a fault at a location where nothing is actually wrong, sending a splice crew to open a closure that does not need attention.
Launch Cables, Bidirectional Testing, and Near-End Blind Spots
An OTDR connected directly to the near end of a cable plant under test has a serious limitation: the first event on the fiber, typically the near-end connector itself, sits inside the instrument's own dead zone, making it difficult or impossible to accurately measure. This is solved with a launch cable, sometimes packaged with a companion receive cable as a dead zone box: a length of known, good-quality fiber, often 100 meters to a kilometer or more, connected between the OTDR and the link under test, so that the near-end connector of the actual cable plant now sits safely past the OTDR's dead zone and can be measured accurately, with the OTDR reading and then subtracting the known length and loss of the launch cable itself from the final result.
Bidirectional OTDR testing, testing a link from both ends and combining the two traces, addresses a different problem: a single-direction OTDR trace can misjudge the loss of an individual splice or event because backscatter coefficient can differ slightly between two different fibers spliced together, causing the instrument to see an apparent gain or an exaggerated loss depending on which fiber it is looking from. Averaging bidirectional results, which most modern OTDR software can do automatically when both traces are loaded together, produces a splice loss figure much closer to the true physical loss than either single-direction trace alone, and this bidirectional average is the standard, defensible method for certification-grade splice loss reporting, mirroring the same directional-averaging principle covered for insertion loss testing with an OLTS.