Long-period fiber gratings (LPFGs) have recently been utilized as optical bend sensors by observing changes in their transmission spectra as the fiber is bent. One such spectral change reported is the "splitting" of each attenuation notch into two. To date, explanations given for this apparent notch splitting have proven unsatisfactory. In this communication, we show that the apparent notch splitting is due neither to a splitting of degenerate cladding modes nor to the phase-matching condition being satisfied at multiple wavelengths for a given cladding mode. In contrast, bending causes new notches to be formed at nearby wavelengths as a result of coupling to asymmetric cladding modes that are not coupled to in a straight UV-induced LPFG. With increased bending, these new notches' central wavelengths shift in the opposite direction as the original notches, thus causing the apparent splitting of the latter. We use a numerical analysis to show that the cladding modes of a fiber undergo large spatial changes when the waveguide is bent. These changes allow coupling in a bent fiber between modes with differing azimuthal symmetry even with a uniform UV-induced index perturbation. All of the primary experimental effects published thus far are
well-described with this analysis. This improved understanding of bent LPFGs will be important for the development of devices and is also relevant whenever there is interaction with the cladding modes in a curved optical fiber.
We have developed long-period fiber gratings (LPFGs) utilizing the photoelastic effect and have demonstrated polarization-independent operation. The LPFG is made by pressing a standard, jacketed single-mode fiber between a flat plate and a plate with grooves mechanically machined with a suitable period. The grating's transmission spectrum is easily tuned by adjusting pressure, grating tilt, and length. Furthermore, the grating can be completely erased by removing the pressure from the fiber. Grating attenuation greater than 25 dB has been demonstrated with a notch-location polarization dependence of +-4 nm. In this paper we report drastic reduction in this polarization dependence by two different approaches. Passing through the grating a second time after reflecting off a Faraday rotator mirror was successful; this method may be used with other types of LPFGs. The second approach utilizes our mechanical grating's ability to be double-passed with two fibers side-by-side. Between passes, a fiber-loop half-wave plate aligned at 45 degrees to the plane of the grooved plate swaps power between x- and y-polarization states. The resulting output's measured polarization dependence was smaller than +/- 0.2 nm. Further improvement is expected through careful tuning of the wave plate. We also report a computer model of the filter spectrum and its polarization dependence, which takes into account non-uniform index perturbation, lossy cladding modes, cladding index perturbation, as well as the polarization dependence of the photoelastic effect, characteristics not usually present in UV-induced LPFGs. The model generates transmission spectra that agree quite well with experimental results.
We propose a novel cascaded amplifier system for long-haul PSK or FSK transmission that equalizes WDM channel powers completely passively. An unflattened EDFA is asymmetrically placed within an all-fiber Sagnac interferometer. The Kerr nonlinearity of the Sagnac-loop fiber induces a net phase difference between the counterpropagating signals. Individual channels have independent, nonlinear transfer functions upon exiting the loop. This system provides higher gain for weak channels, while strong channels receive reduced gain. Thus, all channels approach and maintain a steady-state power level through successive amplifications. This system requires no active feedback mechanisms to maintain channel power equality. Its performance is not affected by changes in the gain spectrum of the optical amplifier and, unlike all other power equalization or gain-flattening schemes, the degree of equalization improves with increasing number of amplifications. This presentation will discuss the operating principle of this device, theoretical predictions of its properties, and work in progress towards an experimental proof of principle.