Study of novel algorithms for fiber-optic alignment and packaging automation

Packaging process remains a bottleneck for the cost-effective manufacturing of fiber-optic devices. Automation is a key for yield enhancement and thus for cost reduction. Among many critical automation issues, fiber-optic alignment is a difficult one. The currently widely used conventional fiber-optic alignment method based on Hillclimbing-type algorithms is usually time-consuming at best and it often fails by being trapped at a local sub-optimal peak. It also has difficulties in performing angular alignment and the alignment of multi-channeled fiber-optic devices. Thus efficient searching and intelligent algorithms are needed. The objective of this work is to present the first comprehensive study of fiber-optic active alignment algorithms. By developing various novel methods, the currently existing problems in fiber-optic alignment are resolved. These methods include fast physics-based package-specific laser-fiber alignment methods, novel fiber-optic alignment algorithms based on generalized Hamiltonian dynamics, Genetic adaptive optimization technique and Simplex optimization technique. Different alignment conditions, including aligning fiber and laser diode with long-distance separation, single channeled devices in the presence of side-modes, simultaneous multi-degree-of-freedom alignment, and multi-channeled devices alignment, will be considered. Both simulation and experimental results will be presented. Comparison among different algorithms will be presented. In addition, a novel scheme of the integrated laser-diode-to-fiber alignment and RIN testing automation for the manufacturing of Fabry-Perot (FP) laser diode module hosted in a low cost isolator-free coaxial-type package will be introduced.