Knowledge base · Acquisition

OTDR pulse width: what the instrument picks for you

A short pulse tells close events apart; a long one reaches the far end. You cannot have both, and in automatic mode the instrument picks the pulse itself from a built-in table.

See the pulse width in your file

That table is transcribed below in full: 303 combinations from the Anritsu µOTDR firmware. It is how one particular instrument behaves; there is no advice of ours in it.

By Ilya (engineer, 6 years in fibre optics) · Published September 11, 2026

What each pulse width trades off

Each of the consequences below is visible in the finished file.

A sketch, not a recording. One and the same link shot with two pulses: the short one separates close events but sinks into noise sooner; the long one reaches further but merges close events into a single row.

Dead zone

A longer pulse blinds the receiver for longer, and events behind a reflection merge into one. The instrument writes them as a single row, and that is how we show it; there is nothing to split apart afterwards.

Sampling step

The pulse sets how finely it is worth writing points down. If the recorded step is coarser than the pulse allows, events closer together than that step cannot be told apart, and the app says so in a caveat under the table.

Reach

A short pulse carries less light, and on a long line its far end sinks into noise. Hence the instrument's rule: the longer the range, the longer the pulse.

What the instrument sets on its own

Rows are the acquisition range, columns are the splitter on the line. Each cell is the pulse the instrument will choose without asking. A dash marks a combination the instrument does not offer at all.

Working wavelengths: 1310, 1490 and 1550 nm

The pulse the instrument picks on the working wavelengths: rows are the acquisition range, columns the splitter
Range, km	No splitter	1×2	1×4	1×8	1×16	1×32	1×64	1×128
0.5	10	10	10	10	50	50	50	50
1	10	10	10	20	50	50	50	50
2.5	20	20	20	50	100	100	100	100
5	50	50	50	100	100	100	200	200
10	50	50	50	100	100	200	500	500
25	50	100	100	100	200	500	1000	1000
50	100	100	200	500	1000	2000	2000	2000
75	1000	—	—	—	—	—	—	—
125	20000	—	—	—	—	—	—	—
250	20000	—	—	—	—	—	—	—

Pulse widths are given in nanoseconds throughout, so that neighbouring cells compare at a glance; the instrument's own interface writes the long ones in microseconds. The three working wavelengths are treated identically: on no combination do they differ in pulse or in number of averages, which is why one column serves all three.

Out-of-band wavelengths: 1625 and 1650 nm

The same figures on the out-of-band wavelengths PON is measured on under load
Range, km	No splitter	1×2	1×4	1×8	1×16	1×32	1×64	1×128
0.5	10	10	10	20	50	100	100	100
1	10	10	20	50	50	100	100	500
2.5	20	20	50	50	100	100	100	1000
5	50	50	100	200	200	200	200	1000
10	50	100	100	200	200	500	500	2000
25	50	100	200	200	500	1000	1000	5000
50	100	200	200	1000	2000	2000	2000	5000
75	1000	200	200	1000	2000	2000	5000	5000
125	10000	—	—	—	—	—	—	—
250	20000	—	—	—	—	—	—	—

On the out-of-band wavelengths the instrument takes a longer pulse on 31 of the 59 combinations where there is anything to compare; on 27 it is the same, and on exactly one it is shorter: at 125 km with no splitter. The 75 km row here is longer than in the table above: a splitter at that distance is offered on the out-of-band wavelengths only.

It picks the number of averages the same way, and for three modes at once: on the half-kilometre range that is 10 passes in the fast mode and 20 in the precise one; at 250 km, 90 and 180. The average count is written into the file, and the app prints it next to the pulse.

And what it lets you pick by hand

Automatic mode sets one value, but it does not forbid the rest. The header of the same file lists what the instrument will allow on each range.

Pulse widths available for manual selection on each acquisition range
Range, km	Pulse, ns: what you may set by hand
0.5	5, 10, 20, 50, 100
1	5, 10, 20, 50, 100
2.5	5, 10, 20, 50, 100
5	5, 10, 20, 50, 100, 200
10	5, 10, 20, 50, 100, 200, 500
25	5, 10, 20, 50, 100, 200, 500, 1000
50	5, 10, 20, 50, 100, 200, 500, 1000, 2000
75	5, 10, 20, 50, 100, 200, 500, 1000, 2000, 5000, 10000
125	5, 10, 20, 50, 100, 200, 500, 1000, 2000, 5000, 10000, 20000
250	5, 10, 20, 50, 100, 200, 500, 1000, 2000, 5000, 10000, 20000

The shortest pulse is forbidden nowhere, not even on the longest range: the instrument will let you set it at 250 km. The converse rule, that automatic mode stays inside this list, does not hold: on 12 combinations it sets itself a pulse the manual list for that range does not offer at all, and all 12 are 1×128 on the out-of-band wavelengths. A 128-way splitter eats so much light that a short pulse will not show the line even at a kilometre.

Where you see this in your own file

Both the pulse width and the number of averages are written into the file, and the app prints them in the settings panel as Pulse width and Averages. Read them together with the distance: the same fibre shot with two different pulses gives two different event tables, because these are two different acquisitions.

The dead zone follows from the pulse, and from it the two caveats the app prints under the table: that the first event fell inside the dead zone, and that the sampling step is coarser than the pulse allows. Both mean that no analysis can tell some events in this recording apart.

Whose choice this is, and what it does not mean

The table belongs to one instrument. The one next to it has its own, and they need not agree: these are acquisition settings, not acceptance limits, just like the detection thresholds we do not fill in either.

We do not recommend a pulse width and we write nothing into your file. The acquisition happened before us; our business is to read what the recording says and to name what it makes unmeasurable.