A novel algorithm for fiber-optic alignment automation

An efficient and robust fiber-optic active alignment algorithm for locating the optimal fiber-optic coupling position is critical to the fiber-optic packaging automation, and thus to the cost effective manufacturing of fiber optic components and modules. A robust and speedy fiber-optic alignment algorithm must be capable of avoiding local power peak trapping and simultaneous multidegree of freedom alignment. The currently most widely employed fiber-optic alignment algorithms are based on the so-called hill-climbing method. Because the hill-climbing method is a "one-dimensional-at-a-time" gradient searching method, it can only deal with the alignment on one direction at one time, so it is usually time-consuming and often the real peak cannot be detected because of local peak trapping. Another disadvantage of the hill-climbing method is that it is especially difficult when dealing with angular alignment and arrayed device alignment. Our novel approach is based on the application of the Nelder-Mead Simplex optimization method. The performance of the algorithm is investigated for the first time for three different fiber-optic alignment conditions, i.e., 1) two-dimensional (X and Y) alignment in the presence of side-modes for a single channel fiber-optic device; 2) simultaneous six-dimensional (i.e., three linear, two angular, and one rotational) alignment for a wedged fiber to a laser diode; 3) three-dimensional (X, Y, and rotation) alignment of eight-channel fiber-arrayed devices. We have found that this method has the advantage of fast convergence and easy-implementation. Simulation results indicates that the algorithm converges five times faster than the conventional hill-climbing method; it works well in the presence of side-modes; and it converges to the maximum coupling efficiency within 15 Simplex iterations for the six-DOF alignment; for the arrayed devices, it also only requires less than 15 iterations to reach a balanced optimal coupling position for each channel.