Hardware Reference

In-Depth Information

Figure 2.34: Schematic diagram of mode switching control.

integrating the error signal. Hence the order of the estimator remains the same

as that of equation 2.51. For HDD, a second order estimator is required for

estimating position and velocity. In that case, the estimator gain L is a 2 × 1

column vector and the feedback gain K is a 1×3 row vector. On the contrary,

the estimator in the second method uses an augmented model increasing the

order of the estimator by 1, but the feedback gain for the additional state of

the estimator is fixed as 1. For the case of HDD, one needs to find a 3 × 1

column vector as the estimator gain L. And the feedback gain K,a1×2row

vector, should be found using the plant model prior to augmentation. Several

well defined and eﬃcient algorithms are available for finding these gains [54],

[152].

2.6.4 Mode Switching Control with Initial Value Com-

pensation

The Mode Switching Control is an approach that uses several controllers with

different structures and a switching function that decides changing from one

controller to another depending on certain conditions in the plant and control

states. This scheme, illustrated in the block diagram of Figure 2.34, can be

used in HDD servomechanism to design the track seek control and the track

following control independently.

The two modes of operation in HDD servomechanism requires different

set of objectives to be met. The seek controller must be designed such that

the actuator is moved from one track to another as fast as possible with due

consideration to the power amplifier saturation. On the contrary, the track

following mode is required to make the position error as small as possible. It

is relatively easy to design and optimize the controllers one at a time with-

out taking the other into consideration. The seek controller can be either a

non-linear controller or a linear, two-degree-of-freedom controller to achieve a

near time-optimal performance. Track-following controller, on the other hand,

should be optimized to achieve a bandwidth as high as possible so that tracking

error is minimized. A linear controller such as PID or H
∞
control can be used

for precise head positioning during track-following. When the two controllers