Continuing the thread of measurement uncertainty in fiber optics, this month I discuss measuring the loss of an installed fiber optic cable plant. Optical loss, tested with a light source, power meter and two reference cables, is the most common measurement in fiber optics.


This test works just like a fiber optic communications system. The test source is used as a transmitter, and a power meter is used as a receiver. If you are testing an installed cable plant—usually between patch panels or to an outlet—you use reference launch and receive cables to replace the patchcords used for connecting electronics.


When you measure loss, you must calibrate your test set to establish a 0-decibel (dB) reference, generally by measuring the output of the launch reference cable attached to your test source and setting that on the meter to 0 dB.


Disconnect your meter from the launch reference cable, leaving it attached to the source, and connect the source and launch cable assembly to one end of the cable you want to test. At the other end of the cable, connect a receive reference cable and a meter to the far end of that cable.


The test source and launch cable assembly has a connector on the end, so when you connect it to the cable under test, there will be a connection loss as the light is coupled into the fiber of the cable under test. As the light travels down the cable, the fiber’s attenuation and connection losses in splices and connectors will cause more loss. There is one more connection loss when you connect to the receive reference cable.


Since you have calibrated the output of the launch cable as 0 dB, the meter will read a loss consisting of the connector losses on each end of the cable under test, where the reference cables are connected, and all losses in the cable itself. That’s how you define the loss of a fiber in a fiber optic cable, and you test every fiber in the cable the same way, one at a time.


Each fiber you test gives a loss reading on the meter as a digital display. Fiber optic power meters will show loss as a negative number (e.g., –3.2 dB) because it is measuring lower power, which is a more negative number on a log scale, such as decibels. Some optical loss test sets—combined sources and meters—confuse the issue by making loss a positive number.


The problem with digital readouts is most people believe that number is absolutely correct, but those digital numbers have errors that the operator needs to understand in order to interpret the validity of the data.


Remember my previous discussions of systematic and random errors? Systematic errors affect every measurement in the same way, and random errors vary with each measurement.


In loss measurements, many factors can cause systematic errors. For example, if you set your 0-dB reference improperly, it will affect every measurement the same amount.


One issue that has gotten a lot of attention in the last few years is the modal distribution of the source and launch cable for multimode fiber testing. Modal distribution is a measurement of how light fills the core of multimode fiber. If the light fills more of the core, losses will be higher and vice versa. Without control of the light in the fiber, one can see loss variations of 10 percent or more. Thus, a standard was created called “encircled flux” that has confused practically everyone in the industry. Basically, it says to use a mandrel wrap on the launch cable to control modal distribution and reduce errors.


Other causes of systematic errors can be just as large as modal distribution. These issues include some parameters you have no control over, such as test source wavelength, fiber type (e.g., bend-insensitive or regular fiber) and core size variations within the same type of fiber (e.g., 50/125 micron fiber).


We do have some control over one of the biggest sources of error, the condition of the launch and receive reference cables and mating adapters. If the reference cable connectors are bad, they will cause systematic errors, with higher loss on every measurement. If connectors are dirty and not cleaned and inspected regularly, they can cause random errors. Over time, they wear out, causing a systematic increase in measured loss.


If the equipment is in good condition and properly calibrated, and you follow procedures carefully, the measurement of optical loss will have an uncertainty of about ±10 percent. 


Next month, I’ll cover interpreting the data.