Sandia installation uncovers fiber issues
Arlin S. Powell, Jr.
Last fall Sandia National Laboratories (Albuquerque, NM) reached the halfway point in a recabling project that is upgrading a 1950s cabling infrastructure to handle the high-bandwidth needs of advanced research in such fields as photonics, microelectronics, intelligent systems and robotics. Project manager for the recabling is Roger Adams, a member of Sandia`s technical staff. At last winter`s BICSI meeting, Adams and Jon A. Eberhard, a telecommunications systems specialist with Holmes & Narver Inc. (Albuquerque, NM) and part of the team designing the Sandia installation, reported on the lessons learned on the project to date. Those lessons will come as a shock to some, especially to proponents of optical fiber, who have been proclaiming for some time that this medium is virtually without faults or problems.
According to Adams, recabling at Sandia has been proceeding for more than two years, with nine buildings recabled and four new buildings cabled to date. Five additional buildings will be recabled and four new buildings cabled through 1996. Primary cabling media are 4-pair enhanced unshielded twisted-pair cable and 4-fiber optical-fiber cable terminated with ST-type connectors.
Infrastructure test bed
Pointing out that Sandia can be viewed as a test bed for the National and Global Information Infrastructures, Adams said, "We believe our cabling efforts are only unique with respect to time. Within a few years, high-quality cabling activities involving large quantities of cable and connectors will be commonplace. In that respect, Sandia`s cabling can be thought of as an indicator of things to come.
"Sandia has experienced numerous unexpected and unprecedented performance-related cabling challenges since 1993," Adams added. Among them were issues surrounding singlemode optical fiber, viewed by many in the industry as the unchallenged medium of the future.
Sandia decided upon singlemode fiber because its research scientists were demanding virtually unlimited bandwidth, while the laboratory`s management was calling for cost containment, "a very perplexing but common situation," according to Adams. When laboratory staff began to look into singlemode fiber-optic components, they found that the performance data published by many manufacturers was incomplete, and that a statistical failure rate of 16% for connectors resulted from the data that was available.
With 120,000 fiber-optic connectors installed in the first half of the project alone, this failure rate was deemed unacceptable by Sandia, and ways were sought to lower it. The lab`s staff first focused on convincing fiber, coupler and connector vendors to improve manufacturing tolerances, but it soon became obvious that improvements in the submicron range would be required--and would be difficult to achieve. Sandia then turned to field termination procedures, discovering that improvements in cleaving, air-polishing, lubricants, papers and polishing procedures could yield significant reductions in average attenuation.
As a result, says Adams, "more than 50,000 singlemode ST-type connectors were field-installed at Sandia, with performance levels equivalent to factory-produced super-PC ST connectors. We believe similar field-termination process improvements could be applied to all manufacturers` connectors, yielding similarly improved assembled performance beyond the manufacturers` published values." To achieve this goal, though, manufacturers must first publish more comprehensive statistical performance data on their products, and buyers and installers need to acquire a fundamental understanding of statistics so they can better understand cost/performance tradeoffs.
A second area where the Sandia team discovered problems that have not yet been addressed in the field is with the ferrules of fiber-optic connectors. When terminated, the performance of such connectors can be significantly affected by both undercut fibers and fibers protruding from the ferrule. Undercut fibers can lead to an air gap when mated, which theoretically creates a reflection of -14 decibels due to the glass/air interface. A protruding fiber, on the other hand, is stressed during mating. The long-term effects of this stress are currently being investigated.
The Sandia staff found that such problems as undercut and protruding fibers cannot be adequately seen using the conventional tool in such cases, the fiber-optic inspection microscope; an interferometer is required instead. As a result, buyers should specify inspection using interferometry, installers should incorporate such an inspection into their procedures, and fiber-optic field training should include use of an interferometer.
The Sandia team also investigated automatic polishing machines. Surprisingly, it discovered that such machines did not always perform adequately, and could overpolish or underpolish ferrules. Properly executed hand polishing was found to produce exceptional and consistent results, although only for flat-tipped connectors. The domed surface required in some terminations could only be achieved with polishing machines.
One additional conclusion reached by the Sandia team is that air quality can be affected in closed telecommunications closets where large numbers of riser-rated cables are being installed. Such cables contain phthalate plasticizers in their jackets. Adams recommends that cabling manufacturers work with OSHA to set pertinent standards, and that enclosed spaces be properly ventilated during cable installation.