Saturation and Frequency Weighting in Adaptive Control of Laser Beam Jitter

摘要

In recent experimental research on adaptive control of jitter in laser beams, sufficiently high levels of high frequency sensor noise have produced high frequency spikes in the output errors. The phenomenon has been observed for both adaptive and high bandwidth linear-time-invariant (LTI) control loops. Recently, the source of the problem has been discovered to be saturation associated with the MEMS fast steering mirror used as the control actuator. Results in this paper demonstrate that the spikes in the output error are eliminated by a recently developed frequency-weighting method for the tuning signal used to determine adaptive control gains. The method places more weight on jitter in frequency ranges where large sensor noise otherwise produces the unwanted response. The frequency-weighted adaptive control loop is based on a recursive least squares lattice filter that implicitly identifies the disturbance statistics from real-time sensor data. The adaptive controller achieves both fast adaptation and true minimum variance steady state performance. Results from an experiment with a MEMS fast steering mirror used in current free space optical communications systems illustrate suppression of jitter with simultaneous multiple bandwidths produced by multiple jitter sources.