Hardware Reference

In-Depth Information

of one of the modes. Choice of time optimal or near time-optimal controller

for the seek mode is explained in chapter 2. The PTOS ([54], [209]) discussed

there is a solution that ensures smooth transition between a nonlinear seek con-

troller and a linear state feedback controller for track-following. Since all states

of the head positioning actuator are not available as measured variable, the

states are estimated using an observer. Many well defined and eļ¬cient algo-

rithms are available for calculating the parameters of state feedback controller

and the observer. However, state feedback is only one of the many standard

and well studied methods available for designing a linear controller. Tradi-

tionally, the industry finds itself more comfortable with PID(Proportional-

Integral-Derivative) controller or PID-type compensators. One can realize the

track-following servomechanism using one such controller. A simple feedback

controller does not always provide the best solution to meet design objectives

of the HDD servomechanism. Feed-forward control or piggy-back correction

algorithms in addition to the usual feedback control are widely used for better

performance of the HDD servomechanism.

3.1 Review of Design Methods

Design of controller for a plant with known model involves two steps - se-

lecting the structure of the controller and determining the parameters such

that the given robustness and perfomance objectives are fulfilled. While de-

ciding the structure of the controller, one must consider different factors such

as knowledge about the plant dynamics, type of reference signals, nature of

disturbances and desired performances. For example, integral action should

be included in the controller for a first order plant if it is desired to have zero

steady state error in presence of constant input disturbance. There are sev-

eral approaches that can be used to find the appropriate parameters once the

structure is decided.

In a mechatronic system, the plant is a continuous-time system but the

controller is usually implemented using a digital processor. In such case, two

approaches can be taken to design and realize the controller - (1) by discretiz-

ing an analog prototype controller or (2) by direct digital design applied to a

discretized model of the plant. The choice between these two approaches de-

pends on the relative rate of sampling with respect to the desired closed-loop

bandwidth. One particular concern in the design of discrete-time controller

for continuous-time plant is the delay introduced by the discretization process

and implementation of control algorithm. The inter-dependence of sampling

frequency, desired bandwidth and choice of design approach is explained in the

following.