Module 9

Test & Measurement: OTDR & Friends

You can't see light inside a fiber, and a buried cable can run for kilometers underground. So how do you know the link works — and if it doesn't, where the trouble is and what kind it is? This module is about the small set of instruments that turn invisible light into numbers you can trust: one tool that says pass or fail, and another that says here, at 3.2 km, is your bad splice.

OTDR: radar for fiber

An OTDR (Optical Time-Domain Reflectometer) is, in one phrase, radar for a fiber. It fires a short, high-power light pulse into one end and times the light that comes back. Because it works from a single end, it's a single-ended instrument — no partner needed at the far end.

It turns each round-trip time into a distance using the speed of light in glass:

d = c · t / (2 · n)

  d = distance to the event        n = refractive index of the fiber
  c = speed of light in a vacuum   t = round-trip time (down AND back)

The 2 is the whole trick: the pulse goes down and the echo comes back, so the measured time covers twice the distance. Halve it and you have distance-to-fault. This is why the OTDR is special — the only common tool that gives both the distance to a fault AND its type.

Two ways light comes back

Light returns through two different mechanisms; telling them apart is most of the skill of reading a trace.

So: the sloping line is backscatter (attenuation), and the spikes are reflections (interfaces). Keep those two pictures in mind and the rest of the trace explains itself.

Reading a trace

A trace is a simple graph: X = distance down the fiber, Y = relative returned power in dB. Light fades as it travels, so the line slopes down to the right. Here is every feature you will ever meet, on one drawing:

The one question to ask of every feature: is there a spike? A spike means a reflective event (connector, mechanical splice, break, open end). A step with no spike means a non-reflective event (fusion splice, macrobend). The last spike-into-noise is the far end.

A step with no spike could be either a fusion splice or a macrobend — and there's a simple field test that tells them apart: measure the event at two wavelengths. Bend loss gets worse at longer wavelengths, so a macrobend shows noticeably more loss at 1550/1625 nm than at 1310 nm, while a fusion splice loses about the same at every wavelength. If the step grows when you switch to the longer wavelength, you're looking at a bend, not a splice.

L_true = (L_A→B  +  L_B→A) / 2

OLTS: does the link pass?

For the plain question "will this link carry service?" the right tool is an OLTS (Optical Loss Test Set) — a matched pair that launches a known light level at one end and measures what arrives at the other: the end-to-end insertion loss. (Insertion loss = the loss added by inserting a component or link into the light path — here, the whole link from end to end.)

An OLTS is a calibrated source + power meter. Loss = what you put in minus what comes out.

Insertion Loss (dB) = Reference launch power (dBm) − Measured received power (dBm)

This single number is the most representative real-world loss — it mirrors what a transceiver actually sees. One rule you can't skip: the OLTS must be referenced first (1-, 2-, or 3-jumper method) so you measure the link, not your own patch cords.

Tier 1 vs Tier 2

Testing splits into two tiers that answer different questions. Tier 1 is the verdict; Tier 2 is the diagnosis. On important links, do both.

Tier 1 answers pass or fail; Tier 2 answers where. They are partners, not rivals.

Reflectance, return loss, ORL

So far we've tracked light that's lost. Now the light that bounces back — too much of it destabilizes the transmitter's laser. Three terms describe it; they differ only in scope and sign. The first two are the same event seen two ways:

• Reflectance — the reflected vs. incident power at one interface, written as a negative dB number (e.g. −50 dB for a good connector). The trap is that "more negative" is easy to misread, so be explicit about the endpoints: −14 dB is the LEAST negative = worst (most light bounced back, a glass-to-air break); −60 dB is the MOST negative = best (almost nothing bounced back). A good UPC connector sits around −50 dB.
• Return loss (RL) — the same single-event quantity, written as a positive dB number. Reflectance −50 dB ≡ RL 50 dB (RL is just reflectance flipped in sign). Higher RL = better.
• ORL (Optical Return Loss) — the total reflected power from the entire link: every reflection plus all that distributed Rayleigh backscatter, added up. This is what the laser actually "sees." A distant reflection contributes less to ORL than a near one, because its echo crosses the fiber's attenuation twice.

The pattern is intuitive once the sign clicks: a clean APC connector reflects almost nothing (RL 60+), while a snapped fiber reflects a lot (RL ~14). Bigger RL magnitude = less light bounced back = better.

VFL & inspection

Two humble tools catch a surprising share of real-world faults — and neither is fancy.

VFL — the fiber that glows

A VFL (Visual Fault Locator) is a handheld visible red laser (~650 nm) you couple into the fiber. Where the fiber is broken or sharply bent, the escaping red light makes the fault glow red — often visibly through the cable jacket. It's perfect for problems sitting inside the OTDR's near-end dead zone, and for simply identifying which fiber is which.

Inspect before you connect

The unglamorous truth: dirty end-faces are the single most common cause of link problems. A speck of dust on a connector's polished glass face scatters light, adds loss, and can pit the glass when two faces press together.

The defense is a tool plus a discipline. A fiber inspection scope (a probe microscope) images the end-face — core, cladding, ferrule — to reveal dirt and scratches first. Modern scopes auto-grade against IEC 61300-3-35, which splits the face into concentric zones (strict Zone A = the ~9 µm single-mode core).