Dynamic magnetic information storage or retrieval – Automatic control of a recorder mechanism – Controlling the head
Reexamination Certificate
2002-02-27
2004-12-14
Sniezek, Andrew L. (Department: 2651)
Dynamic magnetic information storage or retrieval
Automatic control of a recorder mechanism
Controlling the head
C360S078060, C360S078070, C318S561000
Reexamination Certificate
active
06831809
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the invention
The present invention relates to a rotary recording apparatus such as a hard disk drive (hereinafter referred to as “HDD”) and to a method of controlling the apparatus. More particularly, the present invention relates to a technique effectively used to improve the head seek performance of a model following control system used for head position control in HDDs.
2. Background of the Invention
As a method of controlling the position of a head in a rotary recording apparatus such as an HDD, one-degree-of-freedom control methods using only feedback control and two-degree-of-freedom control methods in which a feedforward signal is added to a feedback signal are known. In two-degree-of-freedom control, a step response characteristic (seek performance) is achieved chiefly by means of a feedforward signal while a disturbance response characteristic (following performance) is achieved chiefly by means of a feedback signal. The two characteristics can therefore be optimized separately from each other to improve the control performance or response performance. For this reason, there are now an increasing number of cases of adopting two-degree-of-freedom control methods for HDD head position control, of which a reduction in head seek time is strongly required. As two-degree-of-freedom control methods, target-trajectory-generation filter method, feedforward compensation method, model following method, (called model reference methods in some cases) and other various control methods are known.
A target-trajectory-generation filter control method is a method in which a target position of a feedback loop is changed by a feedforward signal each time sampling is made for a control signal (in a control cycle). This control method achieves head position control at a higher speed and in an improved manner because a target position is changed by a feed forward signal for each sample. This method, however,has the drawback that it is difficult to generate an optimum feedforward signal because the head response characteristics (seek operation characteristics) depend on the characteristics of the feedback loop as well as on the feedforward signal.
A feedforward compensation control method is a method of feeding forward an acceleration signal to a conventional speed control loop. For example, a target speed is differentiated at the time of deceleration to calculate a target acceleration, which is applied as a feedforward signal to the control loop. In this method, various techniques for optimizing the feedforward signal can be devised, for example, the gain set for the feedforward signal can be determined on the basis of an adaptive control rule or a learning rule. However, mode change from speed control to position control is required, which cannot be smoothly performed, and it is difficult to optimize a response characteristic of settling.
A model following control method is a method using a system which has generally an internal control model obtained by modeling a control object, which gives the object a control output and a target trajectory such that control of the model is optimized. That is, in model following control, the same control target is input to a control model and a control object (feedforward control). If the model is perfect, the control model and the control object output equal control values. Generally, however, other factors, i.e., a disturbance, etc., exist, which hinder convergence to the target trajectory. Therefore feedback to the control object is provided (feedback control), thereby enabling seek with the head and following of the head to be controlled separately by feedforward control and by feedback control, respectively.
The model following control method is free from the above-described drawbacks of the target-trajectory-generation filter control method and the feedforward compensation control method, and has the following advantages, which the other two methods lack. That is, if a control object and a control model are equivalent to each other in terms of control characteristics (that is, there is no modeling error), the control object can be controlled in an optimum fashion theoretically. Even if there is a modeling error, a steady-state error in head position caused by the modeling error can be eliminated by the feedback loop. Further, a step response characteristic (seek operation characteristic) can be determined without being influenced by the feedback loop. Enabling determination of a step response characteristic with no influence from the feedback loop is particularly advantageous in a case where, as in the case of an HDD, there is a need for a high-speed seek operation while suppressing vibration due to mechanical resonance in a high-frequency range associated with the seek operation. That is, the model following control method makes it easier to obtain a response characteristic realizing high-speed seek operation while suppressing mechanical vibration of the head. Theoretically, it enables optimization of HDD head control in this manner.
For example, a control model using a state estimator may be mentioned. A control method has been employed in which a position, a speed and a acceleration are estimated by state estimators and respectively combined with a feedforward signal.
As measures for compensating a non-coincidence (modeling error) between a control model and an actual control object, techniques described below are known. For example, in the specification of U.S. Pat. No. 6,031,684, a technique of determining a change in the gain of an actuator during seek operation and effecting asymptotic conversion of the gain to an optimum value is described. A method of compensating only a gain by calibration means is described in the specification of Published Unexamined Patent Application No. 10-312655. Also, techniques of storing variable factors other than gain as a feedforward signal in a table are described in the specifications of Published Unexamined Patent Application No. 9-139032 and U.S. Pat. No. 5,859,742.
A feedforward control output (feedforward signal) is given as a current value or a target acceleration of a voice coil motor (VCM), for example. To achieve faster step response, a feedforward signal preferably includes certain high-frequency components. Mechanical systems for head seek including an arm, etc., however, ordinarily have high-frequency resonance mode. If a signal which can resonate at the resonance frequency of such a mechanical system, corresponding mechanical vibration is increased, which results in an increase in seek time. Therefore there is a need to produce a control output having no resonance frequency. As a control output production means satisfying this requirement, techniques of limiting high-frequency components by using a sine function are disclosed in the specifications of U.S. Pat. No. 6,031,684 and Published Unexamined Patent Application No. 2000-123502. Also, methods of performing filtering on an input signal according to the natural frequency of mechanical vibration of a control object are disclosed in the specifications of U.S. Pat. Nos. 4,916,635 and 5,638,267. A technique of producing a feedforward signal by using a polynomial formed as a time function is disclosed in U.S. Pat. No. 5,469,414.
The inventors of the present invention recognized problems described below when the above-described model following control is applied to an actual HDD head positioning system.
First, there is a problem of a reduction in seek speed due to a modeling error. That is, while high-speed time response is required of the seek operation of HDDs, the feedback loop response currently achieved is not enough for following high-speed seek operation. The highest operating frequency of a feedback control system is generally determined by the zero cross frequency of the open loop. However, components exceeding this zero cross frequency exist in the frequency components of the seek operation of currently available systems. In general, there is a non-coincidence (modeling error) between a control mo
Kagami Naoyuki
Okada Kenji
Semba Tetsuo
Tokizono Akira
Uchida Hiroshi
Feece Ronald B.
Sniezek Andrew L.
LandOfFree
Rotary recording apparatus and method of controlling the... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rotary recording apparatus and method of controlling the..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rotary recording apparatus and method of controlling the... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3277624