How to test fiber with a light source and power meter
A light source and power meter measure link loss the way the network will see it. The whole test depends on one thing done right: setting the reference.
What you need
- A light source at the wavelengths you will test (1310 and 1550 nm for single-mode; 850 and 1300 nm for multimode).
- A power meter set to the same wavelength.
- Reference test cords with connectors that match the link, of the same fiber type, plus mating adapters.
- Cleaning supplies and an inspection scope. Every reading is only as good as the cleanest connector.
Setting the reference (one-cord method)
- Clean and inspect the source port, the meter port and both test cords.
- Connect the source to the meter with one test cord. Let the source warm up for a few minutes.
- Set the meter's reference (often labeled REF or zero). The meter now reads 0.00 dB.
- Disconnect the cord at the meter only. Never disconnect it from the source after referencing.
- Connect a second test cord to the meter. Connect cord 1 to one end of the link and cord 2 to the other.
- Read the loss. It includes the fiber, every splice, and the connectors at both ends of the link.
Worked example
A 3 km single-mode link reads 1.9 dB at 1310 nm. Its loss budget at TIA maximums is 3 km × 0.5 + 2 × 0.75 + 4 splices × 0.3 = 4.2 dB, so it passes. But the typical figure is about 1.05 + 0.6 + 0.2 = 1.85 dB, so the reading is right where a healthy link should be.
Other reference methods
The one-cord method is the one TIA-568 prefers, because it counts both end connections. When the link's connectors do not match the meter's (so one cord cannot be used), testers use a two-cord reference, which leaves out one end connection, or a three-cord reference, which leaves out both. Write down which method you used: the same link reads lower with the two- and three-cord methods, and the acceptance limit has to match.
Multimode launch conditions
Multimode readings depend on how the light fills the fiber. Standards call for controlled launch conditions, done with a mandrel wrap on the source cord or an encircled-flux compliant source or launch cord. Without them, multimode readings can be off by enough to pass a bad link or fail a good one.
Mistakes that give bad readings
- Negative loss (a reading better than zero): the reference was set with a dirty or poorly seated connection, or the source cord was disconnected after referencing. Re-reference.
- Wavelength mismatch between the source and the meter.
- Dirty test cords. Reference cords wear out; inspect them every time and replace them when they will not clean.
- Reading in dBm, not dB. The meter shows absolute power (dBm) until a reference is set. See dBm vs dB.
- Only testing one wavelength. A bend shows up much more at 1550 nm than at 1310 nm.