This paper describes Bellcore's generic requirements and technical analysis programs as they apply to passive fiber optic components, including couplers, wavelength-division multiplexers (WDMs), switches, isolators, terminators and attenuators. Connectors, splices and multimode components are not covered.

Passive optical branching devices are being used in many commercial applications, such as optical fiber communications, optical fiber amplifiers and lasers, and interferometric fiber sensors. In order to demonstrate the capability of these devices to perform reliably over their application lifetime, they should pass a comprehensive reliability test program which attempts to duplicate the stresses that the device will experience over its operating life (but in a reasonably short test period). At Bellcore, we have undertaken such a program in which sample devices from several manufacturers are being subjected to test sequences intended to simulate at least twenty years of service. The results of this test program have generated investigations which are leading to techniques for reliability improvements; these techniques are the subject of this paper.

Refinement and new product development of splitters made by the fused biconical taper process are discussed relative to the emerging technologies of optical fiber amplifiers and practical fiber to the curb telecommunication systems.

Conventional test methods use optical switches for testing large port count devices such as couplers and splitters. This method is both capital and labor intensive and the cost of such a system runs between $250 to $300 per test channel. This paper describes a very reliable and low-cost approach for testing a large number of fibers using a mechanical indexer. The method not only reduces the setup time considerably but also offers significant savings in capital equipment. Using this concept, a test system capable of testing 600 channel has been built at a cost of less than $50 per test channel.

This paper details the performance of a new high specification compact broadband 1 X 6 fused fiber coupler. Also reported is confidence testing carried out on the 1 X 6 couplers against the Bellcore document `Generic requirements for fiber-optic branching components' and testing of 1 X 2/2 X 2 couplers against the Bellcore long term reliability document, especially the temperature-humidity cycling test, considered to be the most testing environment for couplers.

Optical sensor systems based on interferometric or resonant effects have been proven to be both feasible and accurate in optical measurement technique. New developments use all-fiber set-ups, as in the all-fiber Brillouin ring-laser gyroscope. The passive components discussed here are resonator couplers and 3 X 3 symmetrical couplers required for a ring resonator sensor system. The miniaturization of passive components is of special interest for sensor applications.

Experimental results concerning the light coupling between a curved single-mode fiber and a planar waveguide with different thicknesses are presented. Different theoretical approaches (an equivalent planar waveguide geometry and a real waveguide geometry couple mode theory, as well as the beam propagation method) are applied to describe the light propagation.

Experimental and theoretical results concerning the characteristics of polished single-mode fiber-optic couplers with an intermediate waveguiding optical layer between the two fiber half- blocks are presented. Different theoretical approaches are compared and a suitable equivalent planar geometry is chosen and treated in the limits of coupled-mode theory. Mechanism of light transfer from first to the second coupler channel is investigated. Both the experimental and theoretical results are presented concerning the properties of the glued optical fiber coupler for three different thicknesses of the glue layer. These couplers are compared in respect to their spectral characteristics.

A new method for the design of binary phase grating is presented. The series intensity formulae of the binary phase grating are obtained from the general complex amplitude transmittance formulae through Fourier integral. This approach simplifies the operation of the design of phase grating and provides a better understanding of the general formulae. Some examples of the design and preparation of the phase grating are discussed.

The LIGA process is a new technology for the fabrication of precisely shaped polymer or metal structures. Using x-ray lithography, electroforming and micromolding as basic technologies, an accuracy in the submicron range can be achieved. For applications in fiber to fiber or fiber to chip coupling technology, mechanical structures with an accuracy of down to 0.3 micrometers have been realized and investigated using different techniques and materials. Additionally, passive waveguide components for optical fiber networks have been fabricated by molding techniques in polymer materials. These components show a low insertion loss, a well defined power splitting ratio and a compact device geometry.

A photonic switching fabric is described which includes an interconnection network made-up of electro-optic switches. The main characteristics both of the whole interconnection network and of the components are summarized. An electronic processor is designed in order to exploit the self-routing behavior of the network. Finally, the timing problem is considered and the delay performance is evaluated.

The problem of monomode fiber polarization produced by the noncircularity of core and fiber strain has been discussed by M. Moneric and S. R. Norman. The experimental results have shown that the frequency shift effect is caused by a fiber core which has a few nonlinear optical activity substances in monomode fiber and it decreases after being shone by monowavelength photo. The frequency shift effect is not Doppler frequency shift or dispersion frequency shift. In this paper, the experimental results have been analyzed using the molecular's obstructing internal rotation theory.

Mechanical failure of silica lightguides is almost always initiated at the fiber surface. Because of this it is important to understand those processes which can lead to surface `damage' and thus to reductions in strength. A consideration of such processes, as well as their time dependence, is undertaken here.

Current models for optical fiber reliability are mostly based upon power law growth kinetics of sharp, stress-free cracks. The physical basis of these models is critically examined and is found to have limitations in describing the behavior of both high strength `pristine' fiber and weak fiber. In particular, the models do not account for the abrupt strength loss sometimes observed in harsh environments for both types of fiber. Recent advances in understanding the behavior of such fibers are discussed. In particular the addition of colloidal silica particles to the coating material is shown to dramatically improve reliability.

Optical glass fibers can exhibit an accelerated static fatigue behavior at long times under moderate stresses. Similarly, optical fibers can exhibit a pronounced strength degradation due to zero-stress aging. The onset of significant strength loss due to zero-stress aging in water occurs at about the same time as the static fatigue transition. The two-point bending static fatigue and zero-stress aging behaviors of two commercial telecommunications fibers were measured at 85 degree(s)C in water immersion and in 60% and 85% r.h. In tests extending out to 680 days, neither fiber has shown evidence of a transition occurring in relative humidity. Similarly, both fibers show modest changes in their two-point bending strength after aging for 1.5 years in the same humidity environments. A model was developed which accounts for the simultaneous effects of zero-stress aging and stress corrosion on crack growth. This model can be used to predict the occurrence of the static fatigue transition, and was applied to both the 85 degree(s)C water immersion and the 85 degree(s)C, 85% r.h. data.

Twelve existing lifetime models for optical fibers, based on a power law to describe the stress- induced crack growth, are studied and compared in COST-218, WG1. The number of models is reduced to only 1 basic model, taking into account the effects of proof test. An alternative model, to be used for proof testing on-line, is given as a worst-case limit. A choice of 3 testing methods to obtain information about the weak-flaw distribution is given: dynamic- fatigue and variable screen-testing of long lengths or using the failure number during proof test. The presented model can also be worked up for different (more harsh) environments during service, without the need to know the corrosion parameter B, provided that stress free aging does not occur. It is explained which additional measurements in environments as during service and (proof) tests have to be performed. The presented model, based upon weak flaw statistics, is also shown to be useful in many short length applications, e.g., splice enclosures and optical amplifiers.

In this work, we studied aged fibers using several different techniques: dynamic tensile testing, two-point bending, an AFM, and a scanning electron microscope (SEM). In particular, we sought to identify the mechanical causes of the low fiber strengths measured after nine months of aging in 80 degree(s)C water. By using a SEM to look at the broken ends and at the glass surface close to the breaks with the lowest failure stress, we gathered new information about the condition of the glass surface of the aged fibers.

A radically different mechanical test, the `distributed strain test,' has been studied theoretically, showing promising results. Owing to the range of loads applied to the fiber, and the convenient data acquisition, the procedure of measurement and evaluation of the fatigue of fiber may be extremely simplified compared to regular tests.

During the past years several actions have been taken by Swedish Telecom in order to ascertain the reliability of optical fiber networks. The status of some current and future works are briefly described in the paper, focusing on the characterization of optical fibers.

The deployment of fiber in the subscriber loop will require that an optical fiber network maintain the highest possible level of reliability over time, despite being subjected to extremes of temperature, humidity, and other environmental and mechanical stresses imposed on the outside plant. At the same time, both the initial cost and the ongoing maintenance expenses for loop equipment must be kept low. Fiber in the Loop (FITL) applications will entail increased fiber handling. Cable lengths will be shorter, and fiber counts higher, than has been the case so far in long-distance applications. There will also be more cable sheath openings per unit length of cable and/or fiber, as well as more splicing and connectorization. It may become a common practice that a customer is connected to a cable installed many years earlier. In subscriber loops, cables and fibers will be installed in harsher and more varying environments. Fibers will be exposed to higher humidity and temperature, particularly in splice boxes mounted on building walls, in pedestal cabinets, and in other similar enclosures. Corrosive gases and/or liquids may also be present at some locations and will adversely affect the fibers. The combination of increased handling, greater exposure, and more stressful environments may give rise to a need for new, more stringent requirements for fiber mechanical reliability. These can include increaSed fiber strength, increased aging resistance, and increased fatigue resistance.

The durability of optical fibers is becoming increasingly important as FITL programs grow. UV cured coatings used to protect optical fibers must be resistant to property changes in a variety of environments. This paper describes the behavior of optical fiber protective coatings when exposed to thermo-oxidative and hydrolytic conditions, actinic radiation, a variety of gel filling compounds, as well as high and low pH solutions. Material property changes are measured through dynamic mechanical analysis and weight loss.

Fiber optic image guides (tapers) are one of the most important elements used to build many optoelectronic Instruments for technical purposes and medicine. Application of these elements has croated entirely new technical possibilities for image processing. They are applied in X- ray units, apparatuses for endoscopy, LLLTV cameras as Well as in modern types of night vision devices. The images obtained with the use of image guides may be observed in real time on a television monitor. The images feature high resoution and contrast. The other advantage may also be pointed out, such as the possibility to register and edit the image with the use of computer or the possibility to reduce the exposure tme and intensity in X-ray units.

The essential advantages of Light Injection and Detection (LID) Systems, which are a prerequisite to reliable low loss splicing technique and highly accurate loss measurement of single mode fibers, are transferred to multi-fiber/ribbon technology. In this paper we discuss in detail the main advantages of intelligent LID Systems and compare different techniques for the realization of fiber selective multi-fiber/ribbon LID Systems. The structure and technique of different systems are presented. Novel systems based on bend couplers have been developed and tested in the laboratory. Results of measurements are shown to illustrate the benefits of this technique.

Optical ribbons are light waveguides whose most important advantages are: identification of the fibers, mass splicing, and fibers organized in a parallel arrangement. Consequently, they can be used in cables in which connection and installation lead to a decrease of cost, compared to current techniques. This paper covers two fields: the ribbon as a waveguide, and the ribbon cable inspected in various uses (duct, direct buried, distribution and building).

The current approach to predicting the reliability of optical fiber systems is based, in part, on calculations involving the strength of buffered fiber. In actual applications the fiber is often damaged during the process of termination for use in either splices or connectors. This damage may contribute to unanticipated, early-life failures. A brief overview of the mechanisms of delayed failure in glass is given. Specific examples of splice and connector failures that were initiated by flaws generated during termination are shown. Data are presented which show the variability in fiber damage associated with both the personnel and the type of tools used in stripping buffer coatings from optical fiber. Application of this information to the use of accelerated life testing is reviewed.

Factors limiting the maximum strength of carbon coated fibers are considered. The ultimate inert strength of carbon coated fibers was estimated by using the strength reduction data at high temperature.

We have measured radiation-induced losses of all kinds of optical fibers as well as of different kinds of connectors, couplers, and multiplexers. Fiber irradiations were performed at ⁶⁰Co sources with dose rates ranging from <EQ 1 Gy/d up to about 10⁵ Gy/d and temperatures between -195 degree(s)C and +100 degree(s)C, as well as at a flash x-ray facility. We present typical results for all test objects. ⁶⁰Co irradiations up to 10⁶ Gy within about one week showed that there exist high bandwidth graded index and single mode fibers with induced losses of only about 5 dB in lengths of 50 m that are typical, e.g., for satellites or nuclear power plants. Pure silica core fibers with high OH content can even show less than 0.5 dB under the same conditions.

Methods have not yet been developed for predicting the reliability of fiber optic connectors and splices even though they have been installed in increasing numbers over the last decade. There is a need for failure rate prediction methods as these components move into new areas of existing networks and new applications where they will be subjected to a broad range of service environments. These methods must be based on accelerated tests derived in many cases from life-stress models and on statistical models appropriate for the known failure modes. This paper deals with the application of these basic principles to fiber optic connectors and splices. Guidelines are proposed for developing prediction methods.

With many manufacturers supplying fiber optic components such as splices and connectors, it is necessary for users to define requirements for these components for their intended applications. Requirements should be realizable and should not lead to a cost penalty without substantial benefits. These generic requirements provide an understanding of customers' expectations and allows users to gage these products with a well-defined yard-stick. However, the requirements do change with advances in technology and changes in applications. Bell Communications Research (Bellcore) develops and publishes such generic requirements in documents referred to as Technical References to inform the industry of Bellcore's view of proposed generic requirements. This paper describes various Bellcore requirements for fiber optic interconnecting devices and exemplifies some of the criteria development processes and how the requirements have changed over the last decade.

This paper describes the cost and performance issues of mass splicing as compared to discrete fiber splicing. There are several variables in the telco decision to deploy mass splicing. Performance and cost factors for both mass mechanical and mass fusion are evaluated compared to currently available discrete fiber splicing technology. New mass splicing systems offer an attractive option for splicing cables of lower fiber counts than previously considered.

The performance of an optical splice or connector is defined by its insertion loss and reflectance. Insertion loss is well understood and primarily depends on the precision of the alignment of the two fiber cores. Reflectance performance is more complex; it depends strongly on the fiber separation and the refractive index of the media between the fiber endfaces, as well as the preparation of the endfaces. This paper describes the principle sources of loss and reflectance for both mechanical splices and connectors. The connectors considered are those with cylindrical-ferrule, physical-contact design and the mechanical splices considered are those which use either index-matching gels or adhesives at the fiber-fiber interface. The paper shows data illustrating typical performance parameters. Finally, some specific concerns that exist regarding reliability and intermateability issues are considered. These include the issue of fiber motion in connectors and particulate occlusion (e.g., voids, contaminants, droplets) in splices.

Arc fusion splicing is the result of fiber preparation, bare fiber welding, and protection application. Fiber preparation is performed through a sequence of operations like coating stripping (chemical, mechanical, thermal), cleaning and cleaving of stripped fibers. Welding process involves placing the fibers in the splicing machine, alignment and fusion. The length of bare fiber in the splice falls in the range 8 divided by 30 mm depending on splicer type. At the end a protection is applied to seal the bare splice from the environment and to increase its strength. In this paper the influence of the different steps of splicing process on the reliability of the splices is analyzed. All tests have been performed on single-mode reduced-core fibers made by OVD technology and coated by means of a double layer acrylate material or ribboned together by means of an additional common extruded coating.

Today's feeder applications and future distribution applications call for shorter-length, higher fiber count cables, more splices per kilometer, and increased connectorization. Whapham estimates that five to eight splices per subscriber will be required for a branched distribution and loop network. In addition, splices and connectors in the loop will experience harsher environments than the controlled environment of a telephone central office or typical remote site. In the distribution portion, between the remote site and the optical network unit (ONU), the splices can be subjected to a wide range of temperature and humidity extremes, as can the ONU itself. The increased handling and the harsher environments in the local loop place significant new demands on the performance of optical splices.