Choosing wavelengths for making repeatable chromatic dispersion measurements.

Monte Carlo simulations suggest that the number, range and symmetry of wavelengths used to measure chromatic dispersion of optical fibers have a large effect on measurement repeatability of zero dispersion wavelength (lambda sub 0) and dispersion slope (S sub 0). Although increasing the number of measured wavelengths improves repeatability, it also consumes measurement time. On the other hand, simply doubling the wavelength range can almost halve measurement uncertainties without affecting measurement time. Specifically, YAG measurements in the shop are now made at 1270, 1310 and 1350 nm--an 80 nm range. Substituting these with 1185, 1310 and 1435 nm--a 250 nm range--could reduce the standard deviation for lambda sub 0 from 1.0 to 0.4 nm and for S sub 0 from .007 to .001 psec cdot nm sup (-2) cdot km sup (-1) without increasing measurement time. Such a strategy would also reduce sensitivity to asymmetries in the measurement wavelengths about the fiber's lambda sub 0.