Knowledge base · Measurement
Dead zone: why the OTDR cannot see an event that is there
After every strong reflection the instrument is blinded for a while and tells nothing apart. That is the dead zone — not a fault and not a shortcoming of the model, but a consequence of how an OTDR works. It is also the main reason two connectors a couple of metres apart arrive in the file as a single row.
One word covers two different figures: the distance at which two events are still distinguishable as two, and the distance after an event at which loss can be measured again. A datasheet lists them side by side, and they differ several times over.
There are two dead zones, not one
Both are defined by IEC 61746, both follow the pulse width — and they mean different things.
Event dead zone
The distance at which two neighbouring reflections are still distinguishable as two. For a reflective event it is measured across the peak, 1.5 dB down from its top. Inside it the instrument can still say that an event is there, but not how much each of the two cost.
Attenuation dead zone
The distance after an event at which the trace returns to undisturbed backscatter — usually within 0.1 or 0.5 dB. This is the minimum distance at which a splice can be measured after a connector. It is always longer than the event dead zone, and the stronger the reflection, the longer it gets.
Front-end dead zone
The same attenuation dead zone, but at the instrument itself: it starts at the OTDR's own connector. Because of it the first metres of the trace are not measured at all — and it is exactly what a launch cable moves out of your link.
Where its length comes from
The main control is the pulse width. With a long pulse the dead zone is roughly the pulse itself: while the instrument transmits, it does not listen. With a short pulse the pulse is no longer what limits it — the photodiode needs time to recover from a bright flash, and at short pulses the dead zone comes out five to six times longer than the pulse.
Hence the trade made on every job. A short pulse gives a short dead zone and separates close events, but carries less energy, more noise and less reach. A long pulse reaches the far end and glues together everything that stands close. This is why one span is shot with several pulse widths: a long one to see the whole of it, a short one to resolve the head.
The second quantity behind the attenuation dead zone is the strength of the reflection itself. The more light comes back, the longer the trace takes to recover. A connector with an air gap blinds the instrument for a long while; a splice, which barely reflects at all, barely blinds it. That is why datasheet metres describe the instrument's best case — its shortest pulse and a single reflection of a stated strength.
What this looks like in the file
An event that falls inside another one's dead zone is usually not lost: the instrument reports it in one row together with its neighbour. Some instruments mark such a row, and we show that mark — the event reads «Merged». The loss beside it belongs to the group as a whole and the reflectance to the strongest component, and we say so plainly rather than printing them as an ordinary measurement of one event.
How close that is can be seen in our corpus: across the files that carry such rows, the distance between the merged components sits between four and fifteen metres. Connector pairs and short patch cords, which is to say exactly what stands in a cabinet and in a closure. Nothing can separate them after the fact — the file holds one row, and no processing will produce a second one. What it takes is another acquisition at a shorter pulse.
And one consequence better named up front: within the front-end dead zone, in the first metres of the trace, our own event analysis looks for nothing either. The instrument does not resolve events there, and drawing one where there is no data would pass a guess off as a measurement.
Next
A dead zone is not a stretch that is «not shown»: the curve is there and is drawn in full. What does not work over that stretch is the measurement, not the display.