Reading an OTDR Trace That Does Not Match the Power Meter
A 1.82 km 12-fibre link from an exchange to a telecom tower has a design loss of 6.0 dB at 1550 nm. Testing is done at 1550 nm with a 100 ns pulse and a 1 km launch fibre. The OTDR reports a total loss of 5.90 dB, with a 0.35 dB event at the launch connector, a 0.10 dB splice at 1.20 km, and a 1.60 dB reflective event at 1.45 km, after which the trace continues normally. There is also a clear event at 2.90 km, even though the cable is only 1.82 km long. A power meter with a stabilised source reads 6.4 dB end to end. A second OTDR run at 1310 nm shows the 1.45 km event as 0.35 dB. The joint closure at 1.45 km contains mechanical splices, and the splice tray is loosely filled with the slack cinched tight under a cable tie.
1. Is the event at 2.90 km a real fault? Explain how you would prove your answer on site, and state the mechanism that created it.
Read the worked answer
It is a ghost, not a fault. Three checks prove it on site:
- Distance arithmetic. 2.90 km is exactly twice 1.45 km, which is the classic signature: a strong reflection at a real event travels back to the instrument, reflects again at the launch connector, travels out again and returns, so it is displayed at twice the distance of the reflector.
- Change the launch condition. Add a longer launch fibre or a known attenuator at the instrument port. A real event stays at the same distance; a ghost moves out by twice the added length or disappears.
- Change the pulse width and test from the far end. Reduce the pulse width so the receiver is no longer saturated; the ghost fades. From the far end, no event is seen at 2.90 km.
Mechanism: the reflective event at 1.45 km (mechanical splice with an air gap and a poorly finished cleave) returns a strong pulse that re-reflects at the launch connector and is re-displayed. The cure is to remove the strong reflector, not to dig at 2.90 km. Record the ghost in the trace notes so the next technician does not chase it.
2. The OTDR says 5.90 dB and the power meter says 6.4 dB. Reconcile the difference rather than assuming one instrument is wrong.
Read the worked answer
Both can be right; they measure different things. The 0.5 dB difference is explained by the end points of the measurement:
- The OTDR measures backscatter up to the far fibre end and stops at the last backscatter point, so the far-end connector pair and the receive-side adapter are outside its total. The power meter includes both connectors as mated pairs plus the reference cords.
- The launch connector at 0.0 km sits inside the OTDR dead zone and is only partially represented by the 0.35 dB shown.
- The OTDR converts backscatter level to distance using an assumed capture factor, so a link with a mechanical splice and mismatched fibre can bias the total slightly.
So a power-meter reading 0.5-1.5 dB higher than the OTDR total is normal for a link with two connector pairs, and here the difference is at the low end of that band.
To reconcile formally: calibrate the power meter with one reference cord and use the same cords for the through measurement, or do a bidirectional OTDR and average. Use the power meter for the acceptance decision against the 6.0 dB design value, and the OTDR for locating and characterising events.
3. Interpreting the 1.45 km event, what is the most likely fault and what will you do about it? State whether the link should be accepted as it stands.
Read the worked answer
The event is reflective, so it is a connector or mechanical splice rather than a pure bend. But it is strongly wavelength-dependent, 1.60 dB at 1550 nm against 0.35 dB at 1310 nm, and a clean mechanical splice or clean connector does not behave that way. The combination points to two faults at the same place: a mechanical splice with a poor cleave or degraded index-matching gel giving the reflection, plus the slack cinched under a cable tie in the tray giving a tight bend whose loss dominates at 1550 nm.
Action: open the closure. Inspect the tray and release the cable tie so the fibre sits at a bend radius of at least 30 mm with slack coiled, not cinched. Remove the mechanical splice and replace it with a fusion splice at 0.1 dB or better, with a correctly centred protection sleeve. Clean and inspect any connector end faces in the path. Then re-test at 1310 and 1550 nm, bidirectional, and re-check for ghosts.
Acceptance: do not accept as it stands. Although the power meter total of 6.4 dB is only 0.4 dB over the 6.0 dB design, the 1.45 km defect is unstable; gel migration and further cinching will increase loss and eventually cause a reflection-driven outage. After repair expect about 4.0 dB total, which restores margin and clears the reflector that produced the ghost.