JAMES J. REFI
Recent announcements by cable-TV and telephone companies confirm that optical fiber will prove central to delivering huge quantities of information to the home. Single-mode fiber is the fiber of choice for building these broadband networks, and fiberoptic splices and connectors will ensure that the fibers are reliably joined.
Loss mechanismsOptical loss at a joint (splice or connector) arises from intrinsic and extrinsic sources. Intrinsic loss occurs because of different optical properties between fibers being connected, while extrinsic loss arises because of the tools and techniques used to join the fibers. The sensitivity of a joint to these loss mechanisms depends on the mode-field diameters of the fibers being joined (see Table 1). Mode-field diameter relates to the "width" of the optical power distributed across the core and cladding of a single-mode fiber.
Fibers are divided into several categories: matched- and depressed-clad refractive-index-profile designs for dispersion-unshifted fibers (1310-nm zero-dispersion wavelength), ring and pedestal index-profile designs for dispersion-shifted fibers (1550-nm zero-dispersion wavelength), dispersion-upshifted-wavelength fiber (1561-nm zero-dispersion wavelength), and nonzero-dispersion fiber such as AT&T's TrueWave fiber. (A fiber's zero-dispersion wavelength identifies the operating wavelength at which the fiber's chromatic dispersion equals zero.)Intrinsic Loss. Manufacturing tolerances on mode-field diameter range from .5 to .0 µm, depending on fiber design and operating wavelength. Consequently, a matched-clad fiber having a mode-field diameter at the low end of its range (8.8 µm at 1310 nm) may occasionally be joined to one having a mode-field diameter at the high end (9.8 µm). Joining two such fibers produces a theoretical intrinsic loss of 0.05 dB (see Table 2).
Worst-case intrinsic losses of 0.16 dB, while possible, rarely occur in practice. Actual mode-field diameter variations are much tighter than those shown in Table 1, and the statistical likelihood of joining fibers with mode-field diameters more than a few tenths of a micron apart is small.Extrinsic Loss. Extrinsic losses arise because of the way a joint is made (see Fig. 1). For example, lateral offset of the two fiber axes, angular tilt between the two fibers, separation between fiber ends, and the quality of the fiber's end surfaces all contribute to a joint's loss. Technicians, by virtue of their skill and joining hardware, can control the magnitude of a joint's extrinsic loss. Just as the intrinsic loss of a joint depends on the fibers' mode-field diameter, so too does the sensitivity of a joint's extrinsic loss to lateral offset.
Core/clad concentricity measures how well a fiber's core is centered in the cladding. An eccentric core produces a misalignment that depends on the fiber's orientation in the groove. Like core/clad concentricity, the misalignment caused by cladding noncircularity depends on the angular orientation of the oval fiber in the groove.
Finally, fiber curl is the tendency for an unsupported length of fiber to have a "set"—a natural curvature. For example, a fiber hanging from the end of a V-groove (as used for fusion splicing) may curve up, down, or sideways. Although curl has no adverse effect on the loss of connectors, mechanical splices, or fusion splices made using splicers that actively align individual fibers, curl can introduce offset when simultaneously fusion splicing many fibers (mass fusion).
Reflectances arise when two fibers are joined and the glass of the first fiber has a slightly different index of refraction than the glass of the second. If a gap also resides between the two fiber ends, two reflectances can occur, and the total reflectance is the sum of the two plus an interaction term. Reflectances can be undesirable when they are large enough to adversely influence operation of the laser source or when they introduce too much noise at the receiver.
Reflectances are frequently measured with an optical time-domain reflectometer (OTDR). While the discrete, or Fresnel, reflectances produced by some mechanical splices and connectors usually attract the most attention, it should not be forgotten that the fiber itself reflects optical power. This Rayleigh backscatter is a fundamental property of silica fiber. So, there becomes a point of diminishing return where the total reflectance of an installed fiber path is dominated by the fiber itself and not by splices and connectors.
Superior reflectance performance is easier to achieve on connectors installed in the controlled environment of a factory than on connectors installed in the field (see photo at top of this page). The increasing demand for excellent reflectance performance may mobilize greater use of cables and pigtails with factory-installed connectors.
Measuring splice loss
The measure of a successful splice is its optical loss. This loss can be determined in several ways: using an OTDR, a local injector and detector, or the loss estimation programs built into the more sophisticated fusion splicers. Caution is advised when making one-way OTDR measurements and when using fusion splice loss estimators.
It is well known that OTDRs do not measure optical power loss, but instead measure optical power backscatter. Because a fiber's mode-field diameter influences its backscatter, fibers on either side of a joint may produce different backscatter levels—thereby obscuring the joint's true loss. This OTDR measurement artifact can be negated by measuring the joint's loss from both directions and then averaging the two results. However, to be more productive, technicians sometimes measure the joint from only one direction and place too much confidence in this single number.One-way OTDR measurements can be very misleading. In fact, the errors that arise from fibers having mismatched mode-field diameters can be ten times larger than the intrinsic joint loss itself. Table 3 shows the theoretical worst-case one-way OTDR errors that can arise because of tolerances on the mode-field diameters of the fibers being joined. While errors as high as those in Table 3 are possible, the statistical likelihood of joining fibers with mode-field diameters at the extreme limits of their specification is small. Nevertheless, a comparison of Tables 2 and 3 shows that one-way OTDR errors can completely obscure the intrinsic joint losses arising from mode-field mismatches.
Some fusion splicers use a profile-alignment system for imaging fibers and measuring geometrical parameters. By viewing the fibers from two perpendicular directions, the images are computer processed and analyzed to determine cladding offset, core deformation, variation in fiber outer diameters, and other key parameters. These are used in an algorithm to estimate splice loss. Depending on the splicer and its loss estimation algorithm, the estimated joint loss and the true joint loss can differ sizably.
A method for overcoming some of the limitations of OTDRs and splice-loss estimators is to use a local injection and detection system that accounts for the uncertainties of coupling light into and out of fibers. Light is launched into the fiber on both sides of the joint being measured, and the signals that cross over the joint are detected on the opposite sides. By simultaneously detecting the launch signals and the crossover signals on both sides of the joint, the joint's loss can be measured directly without additional assumptions.
As fiber penetrates the last mile to the subscriber, fiber lengths will decrease, and more fiber ends will require termination. This scenario will result in more splices and connectors being used per length of fiber deployed.
Additionally, the number of fibers in cables continues to grow—potentially increasing splicing installation time. To keep installation costs competitive with alternative wireline media, splicing productivity will continue to improve by using more multifiber joining.
Higher digital transmission speeds and the increasing use of analog AM signals in hybrid fiber-coax access networks impose tighter reflectance requirements. This will be an impetus for the increased use of fusion splicing, for the development of new index-matching materials used in mechanical splices, and for more connectors being installed at the factory instead of in the field.
Finally, the larger number of fibers being terminated at central offices and headends will motivate the development of smaller, more space-efficient connectors. Some manufacturers have already shown, or are exploring, such miniature connectors.
James J. Refi is a distinguished member of the technical staff at AT&T Network Cable Systems, Norcross, GA 30071.