My classes often hear the story of my encounter with a desperate instrumentation manager at a local oil refinery. For several months he had been fighting a fiber optics problem throughout his plant. Contractors were replaced weekly. All of them terminated connectors; re-pulled cable sections, and tested his digital transmission system for his programmable logic controllers.

The manager was concerned by poor signal quality over the link (increased bit error rate), which he felt was due entirely to the cable plant and its installation. Ironically, what he did not realize was that the "installation" was too good! In fact, he had exaggerated the problem by striving to improve the installation!

This situation occurred because all optical receivers have a preferred dynamic range. When the received optical signal power is low, the detected signal lies closer to the internal noise (primarily either thermal noise or shot noise) of the receiver electronics. This lowers the signal to noise ratio, and thus, the decoded bit error rate. However, when too much signal power is presented, the receiver is "blinded" with photons creating numerous free electrons, which saturate the receiver. All electrons from a pulse may not have been removed from the photodetector before the next data bit cycle begins, and this causes the bit error rate to worsen. In other words, if the received signal is strong enough, the saturation electrons could take several clock cycles to be extracted. As a result, digital zero would be interpreted as digital one instead.

Attenuators are so popular because, when adjusted properly, they eliminate this problem. However, many times in the field, attenuators are not readily available. One quick way to determine any operational improvement by attenuation is to back out the connector from the receiver (or transmitter) slightly. The longitudinal gap with any radial offset in the axis will quickly add attenuation. If the receiver or link begins to behave, the problem is usually due to overdriven receivers. In multimode systems, a slight yet careful bend of the patchcord cable close to the receiver will typically introduce enough microbend-induced attenuation to produce the same result. Incidentally, I have seen technicians "wiggle" and bend the cable at the receiver,  trying to determine if a connector is bad. They note that when the cable is bent way over to the side, the connector allows a bad termination to pass light through a cracked fiber and the receiver works. They are actually demonstrating that the connector is good, and the receiver is being overdriven until the fiber is bent and so attenuated to the point of receiver operation. Later a fixed attenuator can be inserted into the system or the transmitter’s drive current decreased, thus reducing the transmitter power.

Kim Jovanovich

President,  Omni Technologies, Inc.