Search

This paper describes a fiber optic local area network which has been designed for use as an Ethernet and is currently operating at the Xerox Palo Alto Research Center. The fiber optic network, named Fibernet II, is an active-star configured network which is plug compatible with Ethernet at the transceiver cable interface. Collision detection is performed in the active star and collision presence is transmitted to the Fibernet transceivers. The performance and capabilities of Fibernet II are presented along with the motivation for the particular fiber optic implementation that was realized.

High-speed data links using laser-like LED's, single fiber graded index multimode waveguides, and avalanche photodiodes are described. We describe a new stripe-geometry laser/LED source in which lateral current confinement is achieved by selective diffusion of the n-type substrate prior to the epitaxial growth of the GaAs/GaAlAs layers. Using an H.-P. 3760A/3761A Bit Error Rate system, we observed no errors in a psuedo-random sequence of 1011 bits over a half kilometer fiber at 150 Mbits/sec., the bit rate limit of the H.-P. system. Satisfactory eye diagrams were observed at somewhat higher bit rates. A measurement of the broadening of single laser pulses set an upper limit to the pulse dispersion, for this particular h km fiber, of h nsec/km, which compares to 5 to 25 nsec/km for step-index fibers of similar diameters and index differences. A status report will be presented on an experiment to install a 360 meter cable with six independent waveguide channels thru buried conduit between two laboratory buildings.© (1975) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Under certain conditions a changing speckle pattern exists at the output plane of a multimode fiber, resulting in modal noise which can degrade the error performance of a fiber data link. Fiber motion is the usual cause of such speckle change, and Daino et al. have studied the first-order statistics of such modal noise, assuming a single frequency source. But source frequency variation can also cause modal noise; and source frequency diversity has been shown effective in its reduction. In this paper we use a speckle theory approach to study the frequency dependence of modal noise. We have measured and analyzed the correlation of two speckle patterns as a function of source frequency difference, and the speckle spatial frequency distribution as a function of fiber parameters. We have also measured the speckle contrast as a function of fiber length for several sources and fiber types. Such information permits the prediction of the modal noise statistics, from which corresponding changes in error rates can be derived.

Coupling has been measured for dry-spliced fibers as a function of break angle. The angles were measured interferometrically to about 0.1 degrees accuracy. An analysis is performed, and the measurements are found to be within 1.0% RMS of the calculation.