Autoadjusting the photodetector in an optical head system for optimum focus

In the assembly of an optical head system, it is required to mount the photodetector integrated chip (PDIC) in the correct position so that the astigmatism phenomenon can be achieved. The conventional means of adjustment requires an experienced engineer to shift the PDIC in the three dimension (3D) space until the S-word profile is obtained. This paper proposes an automatic method for replacing an experienced engineer, which can be formulated as an optimization problem with parameters trained by a neural network method. During the training process, a convex cost function of the 3D position is established and the minimum of this function indicates the correct position of the PDIC. In the mounting stage, the steepest descent method is then applied to search for the optimum of the trained function, so that the PDIC is automatically mounted at the optimal position.