OTDR Acceptance: A Gainer and a Ghost
You are accepting a 2.45 km single-mode span at 1310 nm with an OTDR, using a 1 km launch fiber and a 100 m receive fiber at 100 ns pulse. Trace from end A shows: launch connector 0.4 dB; splice at 0.62 km, 0.08 dB; splice at 1.18 km, 0.06 dB; an event at 1.62 km reading -0.31 dB; the far-end connector at 2.45 km at 0.90 dB and reflective; and a second reflective spike at 4.90 km with no loss step, after which the trace ends. The two-point span loss covering fiber and splices is 0.64 dB. The design allowance at 1310 nm is 0.40 dB/km, 0.10 dB per splice, and 0.75 dB per mated connector pair.
1. What is the event at 4.90 km, and how do you prove it?
Read the worked answer
It is a ghost - not a real event.
Evidence:
- 4.90 km is exactly twice 2.45 km, the location of the reflective far-end connector.
- A ghost appears at a multiple of the distance to the strong reflector: light reflects at the far end, travels back, reflects again off a connector near the OTDR, and re-enters the fiber - so the OTDR records it as an event at twice the distance.
- There is no loss step across the 4.90 km spike. A real splice or connector in a fiber produces a step; a ghost does not, and it usually shows as a pure reflection sitting on an otherwise clean trace.
The 0.90 dB far-end connector is what feeds it - a reflective UPC connector, likely dirty or damaged.
Proof and fix:
- Change the pulse width (e.g., to 30 ns and to 300 ns). A real event stays; a ghost changes amplitude or disappears.
- Nudge the refractive index setting slightly - the ghost moves to a different multiple while real events stay put.
- Add a longer receive fiber or bend the far-end jumper to change the reflection and watch the 4.90 km spike die.
- Clean and re-inspect the 2.45 km endface (and the launch endface), then re-test. Solve the 0.90 dB connector - it is above the 0.75 dB allowance for a mated pair and it is generating the ghost.
Do not report the ghost in the acceptance document; do report the far-end connector loss.
2. The 1.62 km event reads -0.31 dB. How do you report the splice loss at 1.62 km?
Read the worked answer
A negative loss (a 'gainer') is physically impossible - report a bidirectional average, not the single-direction reading.
Why it happens:
- The OTDR computes splice loss from the difference in backscatter level on each side of the event. If the two fibers have a different backscatter coefficient (different manufacturer, mode field diameter, or a fiber whose bend/geometry scatters differently), the step in backscatter is mistaken for gain or loss. The splice itself can be fine.
- The error is symmetric: from the other end, the same splice reads positive by roughly the same magnitude.
Procedure:
- Test the splice from end B. Suppose B-to-A reads +0.37 dB.
- Average the two directional values: (-0.31 + 0.37) / 2 = 0.03 dB, which is the true splice loss.
- Report 0.03 dB, note that a gainer was observed, and state that the value is a bidirectional average - or, better, get the averaged value from the OTDR's automatic bidirectional function.
Why it matters practically: the -0.31 dB is also masking real loss elsewhere on the span. If the splice were actually 0.3 dB and the backscatter mismatch added -0.6, a single-direction trace would call a bad splice good. Bidirectional averaging is the only way to separate splice loss from backscatter mismatch on mixed-fiber spans. Accept the splice at 0.03 dB and record the fiber mismatch in the as-built; it will affect every future OTDR reading on this route.
3. Using the design allowance, make the acceptance decision for the span and list anything you would still fix or document.
Read the worked answer
Predicted loss (fiber and splices only, as the OTDR measures):
- Fiber: 2.45 km x 0.40 = 0.98 dB
- Splices: 2 x 0.10 = 0.20 dB
- Allowed = 1.18 dB
- Measured two-point span loss = 0.64 dB (the -0.31 gainer makes it look better than it is, but even normalizing the 1.62 km splice to its true ~0.03 dB, the span is around 0.98 dB and still inside allowance).
Verdict: span passes on fiber and splice loss.
Still open:
- Far-end connector at 0.90 dB exceeds the 0.75 dB per-pair allowance and is producing a ghost. Clean, inspect, and if the loss persists, re-terminate that end. Re-test.
- OLTS comparison. The OTDR excludes the launch/receive connectors by reference; an OLTS measures the whole channel including both mated pairs. Run a light source/power meter test, correctly referenced, and accept on the number the end user's optics will actually experience - not on the OTDR span loss alone.
- Documentation. Record the gainer and the bidirectional average, the ghost and its cause, the test wavelengths and pulse widths, launch/receive fiber lengths, OTDR serial number and calibration date, and the test date/technician. Label the span per the customer's naming scheme and attach the trace files (native format, not screenshots) to the closeout package.
- Flag the fiber mismatch at 1.62 km in the as-built so the next tester is not surprised by the backscatter step.