Cutoff wavelength measurement
WILLIAM B. GARDNER, AT&T
Singlemode fibers can perform as multimode fibers at wavelengths below the "cutoff wavelength." In theory, higher-order modes could produce modal noise, which would degrade system performance. In practice, most fiber-optic cable lengths in networks are long enough to attenuate higher-order modes, so that modal noise is negligible. Nevertheless, the cutoff wavelength is an important test parameter for singlemode fibers. But the measurement of cutoff wavelength has always been a vexing measurement problem.
One complication is that the measured value depends heavily on the length and configuration of the test fiber. This dependence has led to the evolution of several standard test configurations, such as cabled and uncabled, and straight and looped. These tests are described in the Telecommunications Industry Association (TIA) Test Methods FOTP-80 for fiber and FOTP-170 for cable.
Another complication is that the spectral transmission curve of the fiber is used to define the cutoff wavelength, and this curve can be distorted by spurious "humps." Steven Jacobs and David Peckham at AT&T Bell Laboratories have demonstrated that interference from coating modes can generate these humps. The good news is that the coating modes can be attenuated in practical systems. The bad news is that the coating modes are difficult to eliminate in measurements that involve only a few meters of fiber.
Early efforts to eliminate these modes were abandoned as impractical. The humps produced an erroneously high value of cutoff wavelength. Because these humps moved around in the spectrum, they also reduce measurement precision.
At the September 1992 National Institute of Standards and Technology Symposium on Optical Fiber Measurements, a joint paper by Corning Inc. and AT&T (see page 217 of the Symposium Technical Digest) offered a fitting algorithm for minimizing the effects of the humps on the measured cutoff wavelength.
Tom Hanson of Corning has been working to introduce this simplex algorithm into the TIA Test Methods, International Electromechanical Commission SC 86A and International Telecommunication Union (ITU) Study Group 15. The algorithm is now part of TIA`s FOTP-80. In the ITU, British Telecom has expressed a preference for a least-squares algorithm. At a scheduled November 1995 meeting, ITU Study Group 15 is expected to modify Recommendation G.650 to allow either algorithm to be used. In the absence of humps in the cutoff wavelength measurement, no algorithm is required.
A new technique for measuring cutoff wavelength without using a spectral transmission curve is beginning to gain favor. This technique involves joining the test fiber to a fiber of known cutoff wavelength. The apparent joint loss is then measured with an optical time-domain reflectometer. If the measurement is performed in both directions and at two different wavelengths, the cutoff wavelength of the test fiber can be calculated. This test method also yields the mode field diameter of the test fiber.
Hanson has documented this technique in Informative Test Method ITM-6, which has been published by TIA`s Subcommittee FO-6.6. This method will be presented at a future meeting of the International Electromechanical Commission`s Working Group SC 86A.
Because the new technique incorporates a comparison method, it will not totally replace the existing FOTP-80 cutoff wavelength test method. The comparison process relies on known data for the type of fiber being measured, such as depressed cladding and dispersion-shifted details. q