Knowledge base · Events
Why your splice is missing from the OTDR event table: the detection threshold
The step is plain to the eye on the curve, yet the table has no row for it. Most of the time the cause is an acquisition setting: the instrument was told to look for events larger than a given figure, and this one came out smaller.
Look at the thresholds in your own file
One word, "threshold", covers two different things. An acceptance limit answers whether an event passes. A detection threshold answers whether it enters the list at all, and it acts first, back on the instrument.
By Ilya (engineer, 6 years in fibre optics) ·
The thresholds the instrument writes into the file
All three sit in the acquisition block of the .sor file, next to the pulse width and the number of averages. They belong to the instrument's settings: the fibre does not produce them, and we do not compute them.
Event loss threshold
The downward step from which the instrument calls something an event. Anything smaller stays fibre as far as it is concerned, so a good splice of a few hundredths of a decibel will not reach the table. Among files that record the figure at all, 0.1 or 0.2 dB is the commonest setting, noticeably coarser than a good splice.
Reflectance threshold
How strong a reflection has to be for the instrument to count it as an event. More often than not the file carries no such figure: the format has the field, the instrument wrote nothing into it, and then the file cannot tell you what reflections it was looking for.
End-of-fibre threshold
How many decibels the level has to drop for the instrument to declare the end of the link. The instrument compares the drop with a single figure and does not read the shape of the curve: the manuals offer either automatic detection or a figure from 3 to 20 dB.
Who sets the figure
Whoever ran the acquisition, in the instrument's settings, before shooting. We neither choose it nor change it: it arrives with the file and describes a measurement that has already been made. The same instrument carries different figures in different configurations.
An example out of a VIAVI release. The seven demonstration traces the instrument carries with it record a loss threshold of 0.05 dB and an end-of-fibre threshold of 6 dB, and no reflectance threshold at all. The configurations shipped alongside them hold 0.10 dB, twice as coarse as the instrument's own files. None of this is a recommendation, only the settings of particular configurations. What to set is decided by the test procedure, not by a program that reads a finished file.
How wide these numbers go is visible in the entry field itself. Yokogawa takes an event loss threshold from 0.01 to 9.99 dB, a reflectance threshold from 20 to 70 dB, and an end-of-fibre threshold from 3 to 65 dB. Those are only the limits of the field: whatever the operator set is what the file carries.
The same instrument has a fourth threshold that never reaches the file: a bend threshold, from 0.01 to 9.99 dB. It looks for bends by that number, but the acquisition block has no slot for it and three get written. So from someone else's file there is no telling whether the instrument looked for bends at all, or from what step.
A row can also be missing because of the dead zone: an event may be not smaller than the threshold but too close to the previous one for the instrument to tell them apart. Then one row stands for two joints, and the loss in it belongs to the group as a whole.
How this shows up here
All three thresholds sit in the Parameters panel: Loss threshold, Reflectance threshold, End-of-fibre threshold. They are usually not among the top twelve fields: open show all →. They cannot be edited. Rewriting a threshold would claim the instrument had been looking for something other than what it was looking for, and the events in its table would not change for it.
A blank field is not a zero. A threshold the instrument never recorded is not shown at all: we put no figure of our own in its place. The same rule keeps our end-of-fibre estimate and our optical return loss silent until the experimental mode is switched on.
Our own detector adds to the instrument rather than replacing it, and it does not inherit the file's threshold: with the instrument's settings it would stay silent exactly where the instrument did. What we find is visible only in the experimental mode and labelled as ours; in the ordinary view the table holds the instrument's own rows.
The other "threshold" lives on the neighbouring Limits tab: those are acceptance limits, and you set them. This program ships no default for what a splice may lose, a blank field is not checked, and there is no verdict without a limit.
Next
A one-move check on somebody else's file: open the settings and see which threshold the instrument was looking from. If it is coarser than the step you are after, arguing with the table is pointless: the link has to be shot again with a different setting.