Research On Polarization Controller And Its Control Algorithm

Polarization Controller (PC) is an indispensable component for the state of polarization (SOP) control in optical communication and optical sensor systems. It requires fast response speed, high extinction ratio, and small fluctuation with endless reset or reset-free SOP control. In this thesis, we focus on the development of a PC system base on the transparent ferroelectric ceramics.Using Ti'2000 series product TMS320F2801, a novel embedded system is designed for the PC using transparent ferroelectric ceramics. According to the practical requirements, operation mode of analog to digital (AD) module is optimized. Digital to analog (DA) module is designed using General Port of Input and Output (GPIO) provided by the TMS320F2801. Logarithm amplifier is used to detect the small light signal. Several effective methods are employed to solve the EMC problem and the problem of signal integrity in the design of high frequency digital circuit.For the PC using transparent electro-optic ceramics, the stimulated anneal algorithm is investigated in detail by computer simulations. Limitation of stimulated anneal algorithm on convergence is discussed. In order solve the problem induced by poor convergence of the simulated annealing algorithm, a more general model of the PC system is proposed for the practical situation that the x-axis of the applied fiber polarization beam splitter is not identical with the reference axis of the Poincare sphere decided by the fast axes of the wave-plates. In addition, a novel fast location algorithm is developed to overcome the limitation in the system using the simulated annealing algorithm.Experiments are implemented. The efficiency of the proposed algorithm is verified by experimental results. A large amount of data are use to analyze the process of the given algorithm. A comprehensive comparison between the proposed fast location algorithm and the previous stimulated anneal algorithm is implemented to illustrate the improvement owing to the use of the fast location algorithm.The presented experiment results show that within the same amount of sample periods (20～30), the PC with the fast location algorithm performs much better than that without the fast location algorithm. Therefore it is evident that the presented fast location algorithm can successfully overcome the shortcomings (e.g., poor convergence and low efficiency) of the stimulated anneal algorithm.The results presented in this dissertation are helpful to the development of new type of PC.