Multirate sampling digital servo of an optical disk drive

Abstract

Full digital implementation of disk drive electronics is expected to improve the performance of drives as well as to reduce their cost, power consumption, and form factor. The Digital Signal Processor (DSP) has recently been used as a servo controller for optical disk drives. Although DSP implementation has the advantage of flexibility, the servo control algorithm has to be simple so that it can finish the computation within a specified time interval. The sampling rate is determined by the servo bandwidth, and is about ten times higher than that of a hard disk drive. Therefore, conventional microprocessors cannot be used in optical disk drives and an efficient algorithm is always necessary for DSP servo. In this paper we apply modem control theory to the design of a DSP servo controller. The combination of a multi-input multi-output (MIMO) model-following control technique and a mutirate sampling technique realizes both high performance and low computational load on the DSP. Unlike the classical control approach, this method assumes a drive mechanism that has two inputs (fine and coarse actuator drive signals), and two outputs (tracking error signal and objective lens position signal). The model-following control technique allows systematic design of an optimal feedback controller for arbitrary characteristics of the mechanism and achievement of a good tracking performance, because an appropriate feedforward signal can be generated to decrease the tracking error. We mathematically derive the controller and demonstrate the seek performance on a 3.5' disk drive, using a velocity estimator. Figures 1 and 2 show the configuration of the servo controller and the experimental results. A multirate sampling servo technique has been used in a hard disk drive to improve the performance when the sampling frequency of the sensor input is limited[ I]. Since the sensor signal is obtained continuously in conventional optical disk drives, we use this method to reduce the sampling frequency of DSP calculation. Since the open loop bandwidth of each feedback loop in this MIMO controller is different and the sampling frequency can decrease for low bandwidth loops, the multirate sampling scheme can be applied to this system in order to decrease the computational load of DSP. The design approach of modern control theory has the advantage of flexibility in performance improvement. An adaptive control scheme is one possible way of achieving higher performance that increases the robustness with respect to parameter variation and/or external vibration. In an optical disk drive, it is desirable to adjust the control parameter according to the circumstances, such as the gain variations of the sensors and actuators. As a preliminary step, we developed a seek algorithm that not only tracks the desired trajectory, but also identifies the gain variation and automatically optimizes the trajectory that minimizes the seek time. This scheme uses the least mean square method for parameter identification during seek, and modifies the model to make the actuators move with the maximum acceleration and deceleration.