How to read an OTDR trace
An OTDR trace is a picture of the light scattered back from every point along the fiber. Once you know the five shapes it is made of, you can read any trace, including the ones that are lying to you.
What the axes mean
The horizontal axis is distance from the OTDR, worked out from how long the light took to come back and the index of refraction (IOR) you set. The vertical axis is the backscattered power in dB, on a log scale. Because it is a log scale, a straight line sloping down means steady loss per kilometer, and a step down means loss at one point.
The OTDR sends a pulse, then listens. It never measures loss directly: it infers loss from how much less light comes back from beyond a point than from before it. That is why some readings need checking from the other end.
The five shapes on every trace
| Shape | What it usually is |
|---|---|
| Steady downward slope | The fiber itself. The steepness is attenuation in dB/km. A section that slopes more steeply than the rest can be a different fiber type or a stressed cable. |
| Spike, then a step down | A reflective event: a connector pair, a mechanical splice, or a crack with an air gap. The spike is the reflection; the step is the loss. |
| Step down with no spike | A non-reflective event: a fusion splice or a bend. Compare 1310 and 1550 nm: a bend loses much more at 1550. |
| Step up | A gainer: a splice between fibers that scatter different amounts of light. Not a real gain. Average it with the reading from the other end. |
| Big spike, then noise | The end of the fiber: a connector, a clean break, or the far end of the receive cable. After it the trace drops into the noise floor. |
Worked example
A trace shows the launch cable's connector spike at 300 m, a smooth slope, a 0.08 dB step with no spike at 2.1 km, a 0.45 dB step at 3.4 km that is 1.9 dB at 1550 nm, and a large spike at 4.8 km followed by noise.
Reading it: the first connector is the start of the fiber under test; 2.1 km is a good fusion splice; 3.4 km is a bend, because the loss grows that much at the longer wavelength (go and look at that closure or vault); 4.8 km is the far end. If the link should be 6 km long, 4.8 km is a break.
Dead zones, and why launch cables exist
After a strong reflection the OTDR's detector is briefly blinded, like your eyes after a camera flash. Two dead zones matter:
- Event dead zone: the shortest distance after a reflective event at which the OTDR can tell a second event is there.
- Attenuation dead zone: the longer distance after a reflective event before it can measure the loss of the next one.
The OTDR's own front panel connector is the first big reflection, so without help you cannot measure the first connector of the link at all. A launch cable (a reel of fiber between the OTDR and the link) moves the first connector out past the dead zone. A receive cable at the far end does the same for the last connector, which otherwise has nothing after it to compare against. Shorter pulse widths make dead zones smaller.
Ghosts
A ghost is an echo: light from a strong reflection that bounces back and forth and shows up again at a multiple of the real distance. Signs of a ghost: it appears at exactly twice (or three times) the distance of a big reflective event, it shows a spike but no loss after it, or it appears beyond the end of the fiber. Cleaning the connectors and using APC connectors or a different pulse width usually makes it move or vanish. Real events do not move when you change settings.
Settings that change what you see
- Pulse width: short pulses resolve events close together and shrink dead zones but do not reach far. Long pulses reach far but blur nearby events together. Use the shortest pulse that still shows the end of the fiber clearly above the noise.
- Range: set it longer than the fiber, often about twice as long, so the end and any ghosts are on screen.
- Averaging time: longer averaging lowers the noise so small events stand out. Fifteen to thirty seconds is a common starting point.
- IOR (group index): this sets the distance scale. Use the value from the cable's datasheet for the wavelength you are testing. A wrong IOR puts every event in the wrong place.
- Wavelength: test at 1310 and 1550 nm (and 1625 nm on some networks). Comparing them is how you tell a bend from a bad splice.
Turning OTDR distance into a place to dig
The OTDR measures the length of the glass, not the cable. Fiber in loose tubes is slightly longer than the jacket, and every slack loop in a vault or on a pole adds glass without adding route. To find a fault, convert the distance to sheath footage using the cable's length markings and your slack records, and shoot from the nearest access point to narrow it down.