Wavelength-Dependent Event at 1.1 km
You are characterizing a 2.4 km OS2 link between a hospital building and a utility vault. The link has a splice closure at 1.1 km with two fusion splices and a patch panel at each end. The OTDR trace at 1310 nm shows a non-reflective event of 0.25 dB at 1.1 km. At 1550 nm the same event measures 2.8 dB, and the trace after it is noisy but continuous to the end. At 1625 nm the event is 4.1 dB. There is no reflective spike at the event. The link passes a 1310 nm insertion-loss test with 1.9 dB total, but the 1550 nm insertion-loss test reads 4.4 dB. The cable is loose-tube outdoor, 12-fiber, with a 1.5 m service loop at the closure. The customer needs 10G at 1550 nm. The contractor suggests increasing the OTDR launch power to 'burn through' the event.
1. Identify the most likely cause and the trace evidence that supports it.
Read the worked answer
The event is almost certainly a macrobend or a tight bend at the closure, not a simple bad splice. Evidence: the loss is non-reflective, it is located at the closure, and it rises strongly with wavelength: 0.25 dB at 1310 nm, 2.8 dB at 1550 nm, 4.1 dB at 1625 nm. Bend loss increases approximately with wavelength and can be tens of dB at 1625 nm, while a normal fusion splice is only weakly wavelength-dependent and would not typically jump by >10×. The continuous trace after the event rules out a complete break. The 1310 nm insertion-loss test hiding the problem is also consistent: the link passes at 1310 nm but fails at the operating wavelength. Inspect the service loop, splice tray radius, cable ties, and bend radius at the closure.
2. Why is the 1310 nm insertion-loss result insufficient, and what test set would you specify for acceptance?
Read the worked answer
The system operates at 1550 nm, so acceptance must be at the operating wavelength; 1310 nm can under-report bend loss and connector effects. Specify insertion loss at 1550 nm (and 1625 nm if available) using a calibrated source and power meter, with reference cords and one-reference method. Also run bidirectional OTDR at 1550 nm to locate and measure the event; single-direction OTDR splice loss can be biased by backscatter differences. For the bend, an OTDR at 1625 nm is a sensitive detector. The acceptance criterion should be the design loss budget at 1550 nm, not the 1310 nm result. If the link is for 10G, verify the 1550 nm loss and optical return loss.
3. Evaluate the contractor's proposal to increase launch power, and give the corrective action.
Read the worked answer
Increasing OTDR launch power does not remove a real bend loss; it only improves dynamic range and can make the trace look cleaner after the event by pushing more light through. It can also overdrive the receiver, distort the trace, or create ghosts. The link loss at 1550 nm is already 4.4 dB, which will remain. The corrective action is physical: open the closure, inspect the 1.5 m service loop and splice tray, and restore the minimum bend radius (typically 30 mm for 250 µm fiber, or per cable manufacturer). Re-route and re-clamp without zip-tie tension. Then re-test at 1550 and 1625 nm. If the bend is in the cable route, locate it with the OTDR and fix the cable support. Only after the event loss drops to a normal splice value should the link be certified.