A 42 km Link That Passes by 0.2 dB
You are asked to accept a 42 km point-to-point single-mode link at 1550 nm before it carries traffic. Design records show 20 fusion splices, four mated connector pairs (two patch panels at each end), and cable attenuation of 0.22 dB/km; the design used these typical values plus a 0.3 dB aging allowance, a 0.2 dB repair allowance and 0.5 dB for measurement uncertainty. The transceivers are specified with launch power -1.0 to +2.0 dBm, receiver sensitivity -21.0 dBm and receiver overload -3.0 dBm; the link is far too lossy to overload the receiver. Your one-jumper insertion loss measurement at 1550 nm reads 15.8 dB. The installer's OTDR report claims 15.1 dB and the installer argues that the link therefore meets the design. The customer's acceptance criterion is that the link must retain at least 3.0 dB of margin against the transceiver budget after all allowances are applied.
1. Compute the design loss and the transceiver budget, then determine whether the link passes the acceptance criterion.
Read the worked answer
Transceiver budget (minimum): launch power worst case -1.0 dBm minus receiver sensitivity -21.0 dBm = 20.0 dB. (Overload is irrelevant here: -1.0 dBm launch into a 15 dB link arrives at about -16 dBm, well below the -3.0 dBm overload point.)
Design loss with the values actually used:
- Fiber: 42 km x 0.22 dB/km = 9.24 dB
- Fusion splices: 20 x 0.10 dB = 2.00 dB
- Mated connector pairs: 4 x 0.50 dB = 2.00 dB
- Subtotal: 13.24 dB
- Allowances: 0.30 aging + 0.20 repair + 0.50 uncertainty = 1.00 dB
- Design figure: 14.24 dB
Acceptance with the measured 15.8 dB: 20.0 - 15.8 = 4.2 dB gross margin; subtract the 1.0 dB of allowances = 3.2 dB net margin.
3.2 dB is greater than the 3.0 dB criterion, so the link passes - but by 0.2 dB, which is smaller than the 0.5 dB measurement uncertainty. The pass is technically correct and practically fragile; it should be recorded and flagged rather than presented as comfortable.
2. The installer wants to use the OTDR figure. Which number is the acceptance value, and what does the 0.7 dB difference tell you?
Read the worked answer
Use the 15.8 dB source/power meter result. Insertion loss acceptance is defined by the light source and power meter method; the OTDR infers loss from backscatter, does not fully capture the end connectors, has limited linearity, and is affected by backscatter-coefficient differences between cable types. It is a fault-location tool, not the acceptance instrument for link loss.
A 0.7 dB gap is larger than the combined uncertainty of the two methods and worth explaining, because the two could be measuring different paths. Check that:
- The power meter measurement was one-jumper referenced and includes the two end connections the design assumed. If a two-jumper reference was used, the measurement excludes a mated pair and both numbers must be adjusted to the same basis.
- The OTDR trace covered the whole span and did not stop at a splitter, a high-reflectance event or a dead zone.
- The same fibers were tested - 42 km of cable may have spare fibers, and a different fiber could have been measured.
Write both numbers into the report with the method for each, and make the 15.8 dB the acceptance figure.
3. The customer asks why the design predicted 14.24 dB and the as-built is 15.8 dB. Give the explanation and the consequence for future design work on this route.
Read the worked answer
The design used typical rather than worst-case component values, and the as-built link is simply worse than typical.
Recompute with the specification maxima: fiber 42 x 0.25 = 10.50 dB, splices 20 x 0.15 = 3.00 dB, connector pairs 4 x 0.75 = 3.00 dB, subtotal 16.50 dB, plus 1.00 dB of allowances = 17.50 dB. Against the 20.0 dB budget that leaves only 2.5 dB of margin - the route would not have met the 3.0 dB criterion even if every component had hit its worst-case limit. The 14.24 dB figure was optimistic and was never a valid acceptance target.
Consequence: a design margin must be calculated from maximum specified values, with the allowances on top, or the project has no defensible pass/fail line. For this route, future additions - another splice after a repair, a re-terminated connector - consume a margin that is already only 3.2 dB against a 3.0 dB criterion. Track the margin as a managed quantity, and do not add splices without recalculating.
Also note that a 1.6 dB excess over the typical estimate is a plausible symptom of a tight bend or a poor splice. The link passes today, but the OTDR should be reviewed at 1550 nm (the band most sensitive to macrobending) to confirm the excess is distributed rather than concentrated in one place.