In any fiber optic interconnection, some loss occurs. Insertion loss for a connector or splice is the difference in power that you see when you insert the device into the system. For example, take a length of fiber and measure the optical power through the fiber. Note the reading (P1). Now cut the fiber in half, terminate the fibers and connect them, and measure the power again. Note the second reading (P2). The difference between the first reading (P1) and the second (P2) is the insertion loss, or the loss of optical power that occurs when you insert the connector into the line. This is measured as: IL (dB) = 10 Log10 (P2 / P1) You must understand these two important things about insertion loss: - **The specified insertion loss is for identical fibers**. If the core diameter (or the NA) of the side that transmits data is larger than the NA of the fiber that receives data, there is additional loss. Ldia = 10 Log10 (diar/diat)2 LNA = 10 Log10 (NAr/NAt)2 where: - Ldia = Loss diameter - diar = diameter receive - diat = diameter transmit - LNA = Loss on optical fiber Additional loss can occur from Fresnel reflections. These occur when two fibers are separated so that a discontinuity exists in the refractive index. For two glass fibers separated by an air gap, Fresnel reflections are 0.32 dB. - **The loss depends on the launch**. The insertion loss depends on the launch, and receives conditions in the two fibers that are joined. In a short launch, you can overfill the fiber with optical energy carried in both the cladding and core. Over distance, this excess energy is lost until the fiber reaches a condition known as equilibrium mode distribution (EMD). In a long launch, the fiber has already reached EMD, so the excess energy is already stripped away and is not present at the connector. Light that crosses the fiber-to-fiber junction of an interconnection can again overfill the fiber with excess cladding modes. These are quickly lost. This is the short-receive condition. If you measure the power output of a short-receive fiber, you can see extra energy. However, the extra energy is not propagated far. The reading is therefore incorrect. Similarly, if the length of the receive fiber is long enough to reach EMD, the insertion loss reading can be higher, but it reflects actual application conditions. You can easily simulate EMD (long launch and receive). For this, you must wrap the fiber around a mandrel five times. This strips the cladding modes. ## Power Budget You can make a rough estimate of a link power budget. For this, you must allow 0.75 dB for each fiber-to-fiber connection, and assume that fiber loss is proportional with length in the fiber. For a 100-meter run with three patch panels and 62.5/125 fiber that have a loss of 3.5 dB/km, the total loss is 2.6 dB, as shown here: Fiber: 3.5 dB/km = 0.35 dB for 100 meters Patch Panel 1 = 0.75 dB Patch Panel 2 = 0.75 dB Patch Panel 3 = 0.75 dB Total = 2.6 dB The measured loss is normally less. For example, the average insertion loss for an AMP SC connector is 0.3 dB. In this case, the link loss is only 1.4 dB. Regardless of whether you run Ethernet at 10 Mbps or ATM at 155 Mbps, the loss is the same. Optical time-domain reflectometry (OTDR) is a popular certification method for fiber systems. The OTDR injects light into the fiber, and then graphically displays the results of detected back-reflected light. The OTDR measures elapsed transit time of reflected light to calculate the distance to different events. The visual display allows determination of loss per unit length, evaluation of splices and connectors, and fault location. OTDR zooms in to certain locations for a close-up picture of portions of the link. While you can use power meters and signal injectors for many link certifications and evaluations, OTDRs provide a powerful diagnostic tool to get a comprehensive picture of the link. But OTDR requires more training and some skill to interpret the display.