Title: Head positioning control method for a storage device and head positioning control device
Abstract: A head positioning method and device positions a head which reads disk-type storage medium at a specified location, and accurately estimates estimates a bias value during settling control. A disk device includes a disk medium, a head, an actuator , and a control circuit, settling control is performed based on a detected position after coarse control without integral compensation or bias compensation having been performed. The position of the head for a next sample is estimated, and an initial bias value is estimated from the difference between the detected position and the estimated position. This initial bias value is then used to perform settling control together with integral compensation of bias compensation. Since the accurate initial bias value at the start of settling is estimated, it is possible to reduce the time for correcting the shift in the bias during settling control, and greatly reduce the time required for settling control.
Patent Number: 6,995,944 Issued on 02/07/2006 to Takaishi, et al.
| Inventors:
|
Takaishi; Kazuhiko (Kawasaki, JP);
Saito; Shunji (Kawasaki, JP)
|
| Assignee:
|
Fujitsu Limited (Kawasaki, JP)
|
| Appl. No.:
|
713578 |
| Filed:
|
November 16, 2000 |
Foreign Application Priority Data
| Nov 17, 1999[JP] | 11-327169 |
| Oct 20, 2000[JP] | 2000-321037 |
| Current U.S. Class: |
360/78.06 |
| Current Intern'l Class: |
G11B 5/59.6 (20060101) |
| Field of Search: |
360/7806,780.9,780.5,789
318/636,434
|
References Cited [Referenced By]
U.S. Patent Documents
| 4894599 | Jan., 1990 | Ottesen et al.
| |
| 5111124 | May., 1992 | Kurosawa.
| |
| 5126897 | Jun., 1992 | Ogawa et al.
| |
| 5150266 | Sep., 1992 | Albert.
| |
| 5381282 | Jan., 1995 | Arai et al.
| |
| 5680271 | Oct., 1997 | Yatsu.
| |
| 5859742 | Jan., 1999 | Takaishi.
| |
| 6166876 | Dec., 2000 | Liu.
| |
| 6429996 | Aug., 2002 | Iwashiro.
| |
| Foreign Patent Documents |
| 2-265079 | Oct., 1990 | JP.
| |
| 3-288913 | Dec., 1991 | JP.
| |
| 5-165528 | Jul., 1993 | JP.
| |
| 5-274040 | Oct., 1993 | JP.
| |
| 06139729 | May., 1994 | JP.
| |
| 7-57414 | Mar., 1995 | JP.
| |
| 07192412 | Jul., 1995 | JP.
| |
| 08255023 | Oct., 1996 | JP.
| |
| 08329630 | Dec., 1996 | JP.
| |
| 09139032 | May., 1997 | JP.
| |
| 10125020 | May., 1998 | JP.
| |
| WO99/44194 | Sep., 1999 | WO.
| |
Primary Examiner: Hudspeth; David
Assistant Examiner: Wong; K.
Attorney, Agent or Firm: Greer, Burns & Crain, Ltd.
Claims
What is claimed is:
1. A head positioning control method for a storage device for positioning a head
at a specified location on a storage medium, comprising:
a step of performing coarse control by observer control based on a present position
and an estimated position of said head without performing bias compensation;
a step of estimating a position of said head for a next sample, and estimating
an initial bias value from a difference between a detected position and said estimated
position at the start of settling; and
a step of performing settling control with said bias compensation by using said
initial bias value,
wherein said step of performing said settling control performs settling control
by observer control.
2. The head positioning control method of claim 1, wherein said step of performing
settling control comprises:
a step of supplying at least one of a target trajectory and feed forward current,
whose size is proportional to an initial position or initial velocity at a start
of said settling control, to a control system for performing said settling control.
3. The head positioning control method of claim 1, wherein said step of performing
said coarse control is velocity control of said head.
4. A head positioning control device for a storage device for driving an actuator
to position a head at a specified location on a disk, comprising:
a detection means for detecting a present position of said head; and
a control means that performs coarse control without bias compensation and then
performs settling control of said actuator based on said detected position and
an estimated position of said head,
wherein said control means performs settling control with bias compensation by
estimating the position of said head for a next sample, and estimating an initial
bias value from a difference between said detected position and said estimated
position; at the start of settling.
5. The head positioning control device of claim 4, wherein said control means
supplies at least a target trajectory or feed-forward current, that is proportional
to the initial position or initial velocity at the start of said settling, to a
control system that performs said settling control.
6. The head positioning control device of claim 4, wherein said coarse control
is velocity control of said head.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates a head positioning control method and ahead positioning
control device for positioning a head at a target position of a storage medium,
and particularly to a head positioning control method and a head positioning control
device for reducing the seek time.
2. Description of the Related Art
Disk devices that have a head for reading disk-type storage medium are widely
used. For example the magnetic disk drives used as a storage devices for computers
comprise a magnetic disk, a spindle motor for rotating the magnetic disk, a head
for reading from and writing to the magnetic disk, and a VCM actuator for positioning
the head on a track of the magnetic disk. The storage density of these kinds of
disk drives is rapidly increasing, as well as is the track density of the magnetic
disks. Therefore, it is necessary to perform high-precision positioning at high velocity.
When a read or write command is received from the computer, the disk drive moves
the magnetic head from its current position to the target position. This is called
the seek operation. This seek operation is transition operation which moves from
coarse control to following control by way of settling control.
Coarse control controls the velocity to the target position. Coarse control
comprises velocity control, PD control, or observer control that does not include
steady-state bias estimation. There is no integral element (integral compensation
or bias compensation) included in the control system. Coarse control switches the
control mode among acceleration, constant velocity and deceleration. In the acceleration
mode, current flows to increase the velocity. In the constant-velocity mode, the
current is '0' or a previously measured bias compensation current flows, to keep
the velocity constant. In the deceleration mode, the current flows in the opposite
direction of the current during acceleration, to bring the velocity to almost zero
near the target position. When the distance is very short, the constant-velocity
mode is eliminated.
Following control controls the magnetic head so that it follows the target
position. Following control comprises PID control, PI×Lead Lag, or observer
control that includes steady-state bias estimation. This control is characterized
by there being an integral element (integral compensation or bias compensation)
included in the control system. Settling control is the control mode that connects
coarse control with following control. In settling control, there may be an integral
element in the control system.
The average time required for this seek operation for a 2.5-inch HDD is on the
order of 10 of milliseconds. As the time required for coarse control is determined
according to the distance of movement, and the maximum current value of the power
amp, it is difficult to reduce this time. In order to reduce the seek time, it
is necessary to reduce the time required for settling control and short distance seek.
The following prior method has been proposed for shortening this settling time.
First, a steady-state bias (external force) applied to the actuator influences
the settling time. In other words, the actuator is applied constantly a steady-state
bias (external force) from an external. To output the head signal from the actuator,
a flexible cable (FPC) is provided to the actuator. This FPC shows spring-like
characteristics. Moreover, as the disk rotates, the wind that is, generated is
applied to the actuator. These external forces also affect the friction and grease
condition of the actuator bearings. The bias of the actuator due to these causes
changes depending on the position of the head. When this bias value is not corrected,
the settling time becomes-long.
Therefore, in the prior art, an average bias value is measured beforehand
for each position and stored in a table, then the value from the bias tables, Bias
Table (y[k]), that corresponds to the detection position y(k) is read and the instruction
current value u(k) is corrected. (Disclosed for example in Japanese Unexamined
published patent No. H2-232875 and H11-25623)
In addition, since the bias value fluctuates, a method of estimating the shift
in this bias value during settling time, and correcting the instruction current
value is also known. This shift is corrected by an integrator or a bias estimator
in the control loop. When using an observer control system, a method of setting
the current position and velocity as the initial observer value at the start of
settling, and setting the estimated bias value to of '0' is performed. The reason
for initializing the estimated bias value to '0' is because the shift in the bias
value is not known at the beginning of settling, so it is set to '0'. In this method,
the current output is the sum of the status feedback current and the pre-measured
bias value from the table, so the estimated bias value gradually converges according
to the status feedback response, and so the position also converges to the target position.
Second, it has been attempted to shorten the settling time by changing the
response characteristics of the feedback during settling control. For example,
there is the method of setting the gain of the integrator in the loop to '0' or
decreasing it during settling control (as disclosed in Japanese Unexamined published
patent No. H2-278582), or the method of increasing the gain in the loop during
settling control (as disclosed in Japanese Unexamined published patent No. H7-153211
and H1-133272).
Moreover, it has been proposed to use feed-forward control in settling
control and short-distance seeking, in which a feed-forward current is made to
flow for correcting the position offset on the target trajectory, and thus making
it possible to shorten the time required for settling control and short-distance
seek. For example, this is proposed in Japanese Unexamined published patent No.
H9-139032 or in the technical report 'Fast Seek Control Taming Actuator Vibrations
for Magnetic Disk Drives' (IEEE Intermag 99).
However, in the prior art, there are the following problems. FIG. 56 shows
the characteristics of the steady-state bias.
(1) As shown in FIG. 56, the steady-state bias value not only changes depending
on the position of the head, but that bias value also differs, even for the same
track, depending on the where the head was located in the past. In other words,
as shown in FIG. 56, it has hysteresis characteristics. Therefore, in the prior
method of using a bias table of average bias values, the shift in bias values is
large. For example, for a 2.5-inch HDD, there is 1 to 3 mA error. Since this error
cannot be estimated in advance, correction is not possible. Therefore, in a feedback
control system in settling control, there must be time to correct this shift in
bias value, and so there is the problem of not being able to shorten the settling time.
(2) Moreover, in the prior method of using a bias estimator, this shift in the
bias value is estimated and corrected. However, since the bias estimate at the
beginning of settling is not known, the initial bias estimate value is set to '0'.
Therefore, it takes time for the estimated bias value to converge to the correct
value. Also, during this time to convergence, there is a shift in bias value, so
as a result of estimating and correcting for shift, overshooting or undershooting
occurs. Therefore, it takes time to correctly estimate the bias value, and so there
was the problem of not being able to shorten the settling time.
(3) Furthermore, in the prior method of changing the response characteristics
during settling control, the dynamic characteristics of the feedback control system
are improved, however the position error increased and position offset occurred,
making it impossible to shorten the settling time. Also, since it is necessary
to correct the shift in the bias value, there is the problem of not being able
to shorten the settling time.
(4) In the feed-forward control, the velocity at the start of feed-forward is
not considered and is assumed, for example, to be zero. In addition, the prior
idea is thought that the velocity error could be suppressed by feedback control.
However, the track width becomes more narrow and it is not possible to ignore the
effect on the velocity even when there is only small external vibration. When the
velocity is large, there is the problem that it takes time for convergence, and
it is difficult for the velocity error to be restored within the target time. The
effect of this velocity error is especially large during short-distance seek, and
in normal feedback control the convergence time is approximately 2 to 3 ms. For
short-distance seek, it takes 1 ms or less to seek one track, for example, so there
is the problem that the effect of velocity error is large, and that it is not possible
to shorten the convergence time.
SUMMARY OF THE INVENTION
The objective of this invention is to provide a head positioning control method
and a head positioning control device for shortening the seek time.
Moreover, another objective of this invention is to provide a head positioning
control method and a head positioning control device that make it possible to more
quickly and accurately estimate the steady-state bias and to reduce the settling time.
Another objective of this invention is to provide a head positioning control
method and a head positioning control device that make it possible to more quickly
and accurately estimate the steady-state bias and reduce the fluctuations in settling time.
A further objective of this invention is to provide a head positioning control
method and a head positioning control device for controlling feed forward during
settling control, and for shortening the settling time.
Yet a further objective of this invention is to provide a head positioning control
method and a head positioning control device for supplying a target trajectory
that corresponds to the initial velocity and initial position, and performing a
high-speed seek operation.
In order to accomplish these objectives, one form of the head positioning control
method for a disk device of this invention comprises: a step of performing coarse
control without performing integral compensation or bias compensation based on
the current position of the head; a step of estimating the position of the head
for the next sample, and estimating the initial bias value from the difference
between the detected position and the estimated position; and a step of performing
settling control together with integral compensation or bias compensation by using
the estimated initial bias value.
One form of the head positioning control device for a disk device of this invention
comprises: a detection means for detecting a current position of the head; and
a control means for performing settling control of an actuator based on the detected
position, after performing coarse control without integral compensation or bias
compensation. The control means estimate the position of the head for the next
sample, and estimates the initial bias value from the difference between the detected
position and the estimated position, and use the initial bias value to perform
settling control together with integral compensation or bias compensation.
In the prior settling control method, it is not possible to avoid the effect
on
the shift in the steady-state bias. In other words, since the bias values has differing
values for the same target position, it is not possible to avoid the effect on
the shift of the steady-state bias. To solve this problem, it is necessary to accurately
estimate the bias value at the start of settling.
In control where integral compensation or bias compensation is not performed,
the difference between the detected position and estimated position is proportional
to the bias value, so this invention estimates the initial bias value at the start
of settling from the difference between the detected position and estimated position
in control where integral compensation or bias compensation is not performed. Therefore,
even when there are differing bias values for the same target position, the initial
bias value for settling control, in which integral compensation or bias compensation
are performed, is accurately set. In this way, since it is possible to reduce the
time for correcting the bias shift during settling control, it is also possible
to greatly reduce the time required for settling control.
It is also possible to prevent variations in seek time since the bias value for
settling control, in which integral compensation or bias compensation are performed,
is accurately set.
In another form of the head positioning control method of this invention, the
step of performing settling control comprises a step of supplying at least one
of the target trace or feed forward current, that are proportional to the initial
position or initial velocity at the start of settling, to the control system that
performs settling control.
In another form of the head positioning control device of this invention, the
control means supply the target trace or feed forward current, that are proportional
to the initial position or initial velocity at the start of settling, to the control
system-that performs settling control.
In this form of the invention, feed forward control is used during settling control.
In the prior art, since only feedback control is used, it is not possible to speed
up the response. In this form of the invention, the response time is positively
reduced by feed forward control. However, it is not possible to reduce the time
by feed forward control alone. In this form of the invention, the settling time
is reduced by generating at least a target trace or feed forward current, that
is proportional to the initial position or initial velocity during settling.
In another form of the head positioning control method of this invention, the
step of performing settling control performs settling control by observer control.
Observer control can accurately predict the control object, or in other words,
the movement of the actuator, and can estimate the unobservable velocity. In addition,
status feedback control is possible, and it is possible to design the pole arrangement
of closed-loop characteristics. Also, by using observer control for the settling
control, high-precision feedback control becomes possible, and it is effective
in reducing the settling time.
In another form of the invention, the head position control method of a disk
device
for positioning the head at a specified position on the disk comprises: a step
of generating a position trajectory and feed-forward current based on a current
position and current velocity of the head; and a step of supplying the position
trajectory and feed-forward current to the feedback control system that calculates
the amount of control from the position error between the current position of the
head and the target position.
The head positioning control device of a disk device, that drives an actuator
of another form of the invention for positioning the head at a specified position
on the disk, comprises a detection means for detecting the current position of
the head; and a control means for performing seek control of the actuator based
on the detected position; in which the control means generates a position trajectory
and feed-forward current based on the current position and velocity of the head,
and supplies the position trajectory and feed-forward current to the feedback control
system that calculates the amount of control from the position error between the
current position of the head and the target position.
In this form of the invention, feed-forward control is used during short-distance
seeking and settling control so high-speed response becomes possible. In addition,
since the initial velocity is also used as a parameter for trajectory generation
together with the initial position, feed-forward control, that is capable of shortening
the convergence time, is possible even though the initial velocity is different.
In the position control method of this invention, it is desirable for the supply
step to comprise a correction step of correcting the position error by the position
trajectory; and a step of adding the feed-forward current to the amount of control
that is calculated by the feedback control system from the corrected position error,
and by doing so an effective high-speed response is possible.
Furthermore, in the position control method of this invention, it is
desirable for the generation step to be a step that is executed during relatively
short-distance seeking or during settling control in relatively long-distance seeking.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a top view of the disk drive of an embodiment of the invention.
FIG. 2 is a cross-sectional view of the disk drive in FIG. 1.
FIG. 3 is a block diagram of the disk drive in FIG. 1.
FIG. 4 is drawing showing the transition of the seek operation of an embodiment
of the invention.
FIG. 5 is a drawing of a model of the observer control of an embodiment of the invention.
FIG. 6 is a drawing of a model of the settling control of an embodiment of the invention.
FIG. 7 is a drawing showing the response waveform when bias is applied (1/2).
FIG. 8 is a drawing showing the response waveform when bias is applied (2/2).
FIG. 9 is a drawing showing the relationship between the bias and shift in position.
FIG. 10 is a drawing showing the relationship between the bias and shift in velocity.
FIG. 11 is a block diagram of the track and feed forward current of an embodiment
of the invention.
FIG. 12 is a drawing showing the frequency characteristics of the FIR filter
in FIG. 11.
FIGS. 13A and 13B are drawings explaining the basic wave in FIG. 11.
FIG. 14 is a drawing explaining an example of the design track of an embodiment
of the invention (1/4).
FIG. 15 is a drawing explaining an example of the design track of an embodiment
of the invention (2/4).
FIG. 16 is a drawing explaining an example of the design track of an embodiment
of the invention (3/4).
FIG. 17 is a drawing explaining an example of the design track of an embodiment
of the invention (4/4).
FIG. 18 is a flowchart of the task control of an embodiment of the invention.
FIG. 19 is a flowchart of the coarse control in FIG. 18.
FIG. 20 is a flowchart of the settling control in FIG. 18.
FIG. 21 is a flowchart of the following control in FIG. 18.
FIG. 22 is a drawing showing the response results of the prior art (1/2).
FIG. 23 is a drawing showing the response results of the prior art (2/2).
FIG. 24 is a drawing showing the response results of the invention (1/2).
FIG. 25 is a drawing showing the response results of the invention (2/2).
FIG. 26 a histogram of the settling time of the prior art.
FIG. 27 a histogram of the settling time of the invention.
FIG. 28 is a drawing of a model of the control of another embodiment of the invention.
FIG. 29 is a drawing of a model of the control of yet another embodiment of
the invention.
FIG. 30 is a drawing of a model of the control of a further embodiment of the invention.
FIG. 31 is a drawing of a model of the positioning control system of another
embodiment of the invention.
FIG. 32 is a schematic diagram of the long-distance seek control unit in FIG. 31.
FIG. 33 is a schematic diagram of the following control unit in FIG. 31.
FIG. 34 is a block diagram of the short-distance seek/settling control unit
in FIG. 31.
FIG. 35 is a drawing explaining the trajectory tables in FIG. 34.
FIG. 36 is a drawing explaining the trajectory design simulation model of another
embodiment of the invention.
FIG. 37 is a drawing explaining an example of a model of the current amp in
FIG. 36.
FIG. 38 is a drawing explaining a model of the resonance characteristics in
FIG. 36.
FIG. 39 is a drawing showing the frequency characteristics of the FIR filter
in FIG. 36.
FIG. 40 is a drawing explaining the rectangular waveform of the position trajectory
generation in FIG. 36.
FIG. 41 is a drawing explaining the position trajectory generation in FIG. 36.
FIG. 42 is a drawing explaining the rectangular waveform of the velocity trajectory
generation in FIG. 36.
FIG. 43 is a drawing explaining the velocity trajectory generation in FIG. 36 (1/3).
FIG. 44 is a drawing explaining the velocity trajectory generation in FIG. 36 (2/3).
FIG. 45 is a drawing explaining the velocity trajectory generation in FIG. 36 (3/3).
FIG. 46 is a drawing explaining an example of the design trajectory for the
0 to 4 track seek in FIG. 36 (1/2).
FIG. 47 is a drawing explaining an example of the design trajectory for the
0 to 4 track seek in FIG. 36 (2/2).
FIG. 48 is a drawing explaining an example of the design trajectory for the
5 to 8 track seek in FIG. 36 (1/2).
FIG. 49 is a drawing explaining an example of the design trajectory for the
5 to 8 track seek in FIG. 36 (2/2).
FIG. 50 is a drawing explaining an example of the design trajectory for the
9 to 12 track seek in FIG. 36 (1/2).
FIG. 51 is a drawing explaining an example of the design trajectory for the
9 to 12 track seek in FIG. 36 (2/2).
FIG. 52 is a drawing explaining an example of the design trajectory for the
13 to 16 track seek in FIG. 36 (1/2).
FIG. 53 is a drawing explaining an example of the design trajectory for the
13 to 16 track seek in FIG. 36 (2/2).
FIG. 54 is a drawing explaining an experiment example of another embodiment
of the invention (1/2).
FIG. 55 is a drawing explaining an experiment example of another embodiment
of the invention (2/2).
FIG. 56 is a drawing of the steady-state bias characteristics.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The embodiments of this invention will be explained by dividing them into the
disk device, positioning control system, positioning control process, example of
bias estimation, positioning control for another embodiment, positioning control
for a further embodiment, trajectory generation method, example of trajectory control,
and another embodiment.
Disk Device
FIG. 1 is a top view of the disk device of an embodiment of the invention, and
FIG. 2 is a cross-sectional view of that disk device. In this example, a hard disk
drive is used as the storage device.
As shown in FIG.
1 and FIG. 2, magnetic disks
6 are such that they
form storage layers on a substrate (disk plate). The size of the magnetic disks
6 is 2.5 inches, and there are three disks inside the drive. A spindle motor
5 supports and rotates the magnetic disks
6. A magnetic head
4
is located on the actuator. The actuator comprises a rotary-type VCM (voice coil
motor)
3, arm
8 and flexure (suspension)
9. The magnetic head
4 is attached to the tip of the flexure
9.
The magnetic head
4 reads and writes data from and on the magnetic disks
6. The actuator
3 positions the magnetic head
4 at a desired
track on the magnetic disks
6. The actuator
3 and spindle motor
5
are located on a drive base
2. A cover
1 covers the drive base
2
and separates the internal working of the drive from the outside. A printed circuit
board
7 is located under the drive base
2, and it contains circuits
for controlling the drive. A connector
10 is also located under the drive
base
2 and connects the control circuits with the outside.
This drive is compact with dimensions of about 90 mm (Horizontal)×63 mm
(Vertical)×10 mm (Width). It is used as the internal disk drive of a personal computer.
FIG. 3 is a block diagram of the control circuits on the printed circuit board
7 and in the drive. A HDC (hard disk controller)
18 receives commands
from the host CPU and generates internal magnetic disk drive control signals for
controlling the interface with the host CPU such as receiving data, and for controlling
the read/write format of the magnetic disk medium
6. A buffer
17
is used for temporarily storing write data from the host CPU and for temporarily
storing the data read from the magnetic disk
6.
A MCU (micro controller)
19 comprises a microprocessor (MPU), DA converter
and AD converter. The MCU (called the MPU below)
19 performs servo controls
for positioning the magnetic head
4. The MPU
19 executes the program
stored in memory, detects the position signal from the servo demodulation circuit
16, and controls the VCM control current of the actuator
3 for positioning.
Furthermore, the MPU
19 controls the drive current for the SPM drive circuit
14.
The VCM drive circuit
13 comprises a power amp for flowing drive current
to the VCM (voice coil motor)
3. The SPM drive circuit
14 comprises
a power amp for flowing drive current to the spindle motor (SPM)
5 that
rotates the magnetic disk
6.
A read channel
15 is the circuit for reading and writing. The read channel
15 comprises a modulation circuit for writing the write data from the host
CPU to the magnetic disk medium
6, a parallel-to-serial conversion circuit,
a demodulation circuit for reading data from the magnetic disk medium
6
and a serial-to-parallel conversion circuit. The servo demodulation circuit
16
is a circuit for demodulating the servo pattern written to the magnetic disk medium
6, and comprises a peak hold circuit and an integrating circuit.
It is not shown in the figures, however in the drive HDA there is a head IC that
comprises a writing amp that supplies current to the magnetic head
4 for
writing, and a preamp for amplifying the read voltage from the magnetic head
4.
Here an example of a magnetic disk drive is explained as the disk device, however,
it is also possible to use an optical disk drive such as a DVD or MO, a magnetic
card device or an optical card device. Here a device capable of reading and writing
is shown, however a read-only device (reproduction device) could also be used.
[Positioning Control System]
Next, the positioning control system executed by the MPU
19 is explained.
FIG. 4 is drawing showing the transition of the seek operation, and FIG. 5 is
a drawing of a model of the observer.
As shown in FIG. 4, the seek operation is the transition between coarse control,
settling control, and following control. Coarse control controls the velocity to
the target position. Coarse control comprises velocity control, PD control or observer
control that does not include steady-state bias estimation. There is no integrating
element (integral compensation or bias compensation) included in the control system.
The unit for velocity in the SI unit system is meters/sec. However, in the case
of a disk drive, the unit for reading or writing data is tracks, so the velocity
is expressed as the number of tracks per sample, tracks/sample.
As shown in FIG. 4, coarse control switches the control mode among acceleration,
constant velocity, and deceleration. In the acceleration mode, current flows to
increase the velocity. In the constant-velocity mode, the current is '0' or a previously
measured bias compensation current flows, to keep the velocity constant. In the
deceleration mode, the current flows in the opposite direction of the current during
acceleration, to bring the velocity to almost zero near the target position. When
the distance is very short, the constant-velocity mode is eliminated.
Following control controls the magnetic head
4 so that it follows
the target position. Following control At comprises PID control, PI×Lead Lag,
or observer control that includes steady-state bias estimation. This control is
characterized by there being an integral element (integral compensation or bias
compensation) included in the control system. The unit of the target position in
the SI unit system is meters. However, in the case of a disk drive, the unit for
reading or writing data is tracks, so it is expressed in tracks or cylinders.
Settling control is the control mode that connects coarse control with following
control. In settling control, there is an integral element included in the control system.
Next, observer control as suitable position control of this invention will
be explained. Observer control is a method for simulating a control model for realizing
a status feedback control. The actuator of the hard disk drive (HDD) is expressed
by the transfer function of equation
##EQU1##
In the equation above, 'y' is the position, 'Bl' is the power constant, 'm' is
the mass, 's' is the Laplace operator, and 'u' is the value of the drive current.
In this expression, a rotation response is converted to a linear response.
When this expression of the transfer function is expressed in status format,
the following equations (2) result.
##EQU2##
Here, 'x' and 'y' are position and 'v' is the velocity.
These equations are in SI units, where position is in meters, velocity is in
m/sec, and current is in ampere. However, in an actual device, the target position
is in track units, and when the sample period is taken to be T(s), then it is better
for the velocity to be expressed in units of track/sample. Furthermore, since the
current output is set to a digital-analog converter (DAC), it is best for the maximum
current to be normalized to '1'. In this way, by converting units, equations (2)
are converted to equations (3). The status variables are converted to new units.
##EQU3##
When this analog model is converted to a digitized digital model, the following
equations (4) are obtained.
##EQU4##
Here, Ka=Bl/m.
It is possible to construct an observer control system based on these equations.
In order to actually apply observer control, the current observer construction
of the following equations (5) are used.
##EQU5##
The model of these equations is the model shown in FIG.
5. An explanation
is given based on FIG.
5. The estimated position x hat[k] and the estimated
velocity v hat[k] for this time are obtained by adding the estimated position x
bar[k] and estimated velocity v bar[k] for this time, that were calculated during
the previous sample, to the difference between the observed position (detected
position) y(k) and estimated position x bar[k] that are w multiplied by the gain
L. The instruction value u(k) for this time is obtained by multiplying the current
estimated position x hat[k] and estimated velocity v hat[k] by the gain F. The
output current value u out(k) is obtained by adding the bias table value BiasTable
(y[k]) to the instruction current value u(k).
The estimated position x bar[k+1] and estimated velocity v bar[K+1] of the next
sample time, that are estimated according to this instruction current u(k), are
calculated from the estimated position x hat[k] and estimated velocity v bar[k+1]
for this time and the instruction current u(k).
The bias table value BiasTable(y[k]) is the average value of previously measured
bias values and is stored in a table. This control system is used in coarse control.
Since the average bias value is corrected even in coarse control, there are few
velocity errors due to bias.
Next, the observer control system used on settling control and following control
and that included bias value estimation will be explained. A steady-state bias
model, or in other words, the following equation (6) is added to the analog plant
model (equations (3)) in units of acceleration. Furthermore, when these equations
are digitized and a current observer is constructed in the same way as described
above, the observer control system of equations (7) are obtained.
SB=0 (6)
##EQU6##
These equations (7) are the result of adding bias 'B' as acceleration to the
model in FIG. 5, and as a model is the same.
In other words, The estimated position x hat[k], estimated velocity v hat[k]
and
estimated bias b hat[k] for this time are obtained by adding the estimated position
x bar[k], estimated velocity v bar[k] and estimated bias b bar[k] for this time,
that were calculated during the previous sample, to the difference between the
observed position (detected position) y(k) and estimated position x bar[k] that
are multiplied by the gain L. The instruction value u(k) is obtained by multiplying
the current estimated position x hat[k], estimated velocity v hat[k] and estimated
bias b hat[k] by the gain F. The output current value u out(k) is obtained by adding
the bias table value BiasTable (y[k]) to the instruction current value u(k).
The estimated position x bar[k+1]) estimated velocity v bar[K+1] and estimated
bias b bar[k+1] of the next sample time, that are estimated according to this instruction
current u(k), are calculated from the estimated position x hat[k], estimated velocity
v bar[k+1] and estimated bias b hat[k] for this time and the instruction current u(k).
The construction of this observer control system is used for settling control
and following control. The steady-state bias is estimated and corrected so the
DC component of the position error becomes '0'. The observer gain (l
1, L
2)
or (L
1, L
2, L
3), and the status feedback gain (Fx, Fv) or
(Fx, Fv, Fb) are obtained from known optimal control theory or optimal pole location method.
Next, the settling control of this invention is explained. FIG. 6 shows a model
of the settling control of an embodiment of this invention, FIGS. 7 and 8 show
simulation of the response to the bias value of the observer control system with
no estimation of steady-state bias, FIG. 9 shows the relationship between the shift
in actual position and estimated position with respect to the bias, and FIG. 10
shows the relationship between the shift in actual velocity and estimated velocity
with respect to the bias.
As shown in FIG. 6, there is an initial bias estimator
21 for the observer
20. The initial bias estimator
21 is explained.
In the deceleration mode in coarse control just before settling control, the
control
system does not include an integrator (or bias estimator). In settling control
and following control, the control system includes an integrator (or bias estimator).
The initial bias estimator
21 calculates the difference between the observed
position x bar[k] of the observer
20, and the actual position y[k] at the
start of or just before the start of settling, then multiplies that position error
(the difference) by a preset gain Kx to calculate the initial bias, and sets that
initial value in the observer
20.
Furthermore, the estimator
21 calculates the product of the position
error and the gain Kv and adds it to the estimated velocity v bar[k] of the observer
20 to correct the estimated velocity. In FIG. 6, the target position 'Y
0'
is subtracted from the detected position y[k] (absolute position), and that value
is taken to be the actual position (relative position) y[k]. The detected position
y[k] is obtained from the position signal of the magnetic head
4.
This theory is explained. In a control system (coarse control) that does not
include an integrator (or bias estimator), when the bias value is added, shifts
occur in the estimated position and velocity value of the observer. This shift
is proportional to the bias value. Therefore, by detecting the position error,
it is possible to estimate the bias value.
FIG.
7 and FIG. 8 are simulation drawings of the response to the bias
value of an observer control system that does not include steady-state bias estimation
when the pole of the status feedback is taken to be 330 Hz. FIG. 7 shows the time
response for the instruction current U (%), position P (track), velocity V (track/sample),
position error (actual position y-observed position x), velocity error (actual
velocity v-estimated velocity vhat), when the initial position and initial velocity
are '0', and when a steady-state bias of +4 mA is applied.
FIG. 8 shows the time response for the instruction current U (%), position P
(track), velocity V (track/sample), position error (actual position y-observed
position x), velocity error (actual velocity v-estimated velocity vhat), when the
initial position is-32 track and initial velocity is '10', and when a steady-state
bias of +4 mA is applied.
In FIG.
7 and FIG. 8, where there are two lines for the position P (track)
and velocity V (track/sample), the top line is the estimated position or estimated
velocity, and the bottom line is the real position or real velocity.
In both FIG.
7 and FIG. 8, the position error between the real position
and the estimated position, and the velocity error between the real velocity and
estimated error show the same response. Also, after 1.5 ms, they show constant
values. The value for the position error is 0.47 tracks, and the velocity error
is 0.40 tracks/sample when the bias value is 4 mA.
When the relationship between the bias value Bias and position error Position
Offset and velocity error Velocity Offset are simulated, the relationship between
bias value Bias and position error Position Offset shown in FIG. 9, and the relationship
between bias value Bias and velocity error Velocity Offset shown in FIG. 10 are
obtained. It can be seen that the shift in estimated position (position error Position
Offset) and the shift in estimated velocity (velocity error Velocity offset) are
proportional to the bias value Bias. This means that in an observer control system
that does not include steady-state bias estimation it is possible to estimate the
bias value by determining the error between the real position and the position
estimated by the observer.
In addition, it is possible to estimate the bias value based on the error between
the real position and the observer estimated position that is measured at the beginning
of settling. The bias value b bar[k] can be obtained from equation (8) below.
b bar[
k]=Kx(
y[k]-x bar[
k]) (8)
Similarly, it is possible to correct the velocity error. The corrected
estimated velocity v bar[k] can be obtained from equation (9) below.
v bar[
k]=va bar[
k]+Kv(
y[k]-x bar[
k]) (9)
Note that it is necessary to switch the gain Kx, Kv depending on the seek direction.
This estimated bias value can be obtained, as described above, in an observer
control system that does not include steady-state bias estimation. Therefore, it
is calculated as the initial value before settling control that is performed by
observer control that includes steady-state bias estimation, or in other words,
at the start or just before the start of settling. From then on, this initial value
is used in settling control by observer control that includes steady-state bias estimation.
In this way, it is possible to accurately estimate the bias value. Moreover,
it
is possible to suppress fluctuation in seek time that accompany the fluctuations
in the bias value during seeking. Therefore, the average seek time is reduced.
For example, the average seek time for a 2.5-inch HDD can be reduced to 10 ms.
Also, here the observer estimated position is used to estimate the bias value.
However, the observer estimated position is expressed by the previous position,
velocity and current value. In addition, since the velocity is small enough, the
observer estimated value can be expressed by the previous position and current.
With this equation, the observer estimated position does not need to be used. In
other words, the position can be similarly calculated from just an equation that
estimates the next sample position from the previous sample position.
Next, in this invention, in order to further reduce the settling time, feed
forward control is used during settling control. In other words, by just estimating
the bias value, there are limits to how much the settling time can be reduced because
the response depends on the control band of the control system and the poles of
the state feedback. For example, when the state feedback pole is located at 500
Hz, then convergence takes 1/500=2 ms.
In order to speed up the response, feed forward control is used in the settling
control. To do this, a track generator
22 is installed as shown in FIG.
6. The track generator
22 generates a target track and feed forward
current (called FF current below) according to the initial position 'x
0'
and initial velocity 'v
0' at the start of settling, and gives them to the
control system.
It is necessary to generate this target track and feed forward current corresponding
to the initial position 'x
0' and initial velocity 'v
0' The track
and current for doing this are designed in advance. In this control, it is necessary
to design by taking into consideration the effect of the cut-off frequency of the
current amp, and the effect of resonance of the actuator.
FIG. 11 is a block diagram for designing the track and FF current. FIG. 12 is
a drawing showing the characteristics of a FIR filter for suppressing resonance
of the actuator. FIG.
13A and FIG. 13B are drawings explaining the basic
wave. FIG. 14 to FIG. 17 are drawings explaining the design track.
First, the track for the target position is designed as follows. As shown
in FIG. 11, a FIR filter
51 is prepared for removing the resonant frequency
component. The FF current is designed by passing an arbitrary waveform to this
FIR filter
51 from a basic wave generator
50. In this way, the resonant
frequency component is removed from the designed waveform.
This FIR filter
51 is such that it contains the zero point of the resonant
frequency. This dampens the frequency component of the zero point or makes it zero.
Here, when an arbitrary waveform is passed through this FIR filter
51, the
waveform that is output from the FIR filter
51 becomes a waveform for which
the gain near the resonant frequency has been suppressed. As shown in. FIG. 11,
the position track is obtained by applying the current that has been passed through
the FIR filter
51 to a plant model
52.
FIG. 12 is a drawing showing the characteristics of the frequency of the FIR
filter
51 that suppresses 5.8 kHz actuator resonance. Here, the zero point
is placed at two frequencies near 5.7 kHz and 5.9 kHz, to remove a wide frequency
component near 5.8 kHz. It is possible to use just one zero point, or it is possible
to use a plurality of zero points placed at different locations.
Next, the trajectory is divided into a position trajectory and a velocity trajectory.
In other words, provided is the first trajectory (position trajectory) when the
initial position is 1 track and the initial velocity is 0 track/sample, and the
second trajectory (velocity trajectory) when the initial position is 0 track and
the initial velocity is 1 track/sample. Moreover, the trajectory when the initial
position is track X
0 and the initial velocity is V
0 tracks/sample,
is synthesized by multiplying the two trajectories by gains of X
0 and V
0.
In this way, it is possible to obtain the trajectories for an arbitrary initial
position and initial velocity by synthesizing one unit track.
Furthermore, the method of generating the target track for settling
control and the method of generating the trajectory for controlling the resonant
frequency component of the actuator are explained.
The resonant frequency is taken to be Fr (Hz). When there are changes in the
resonant frequency due to different solids, aging or temperature change, the average
value of the resonant frequencies is used. In this case, the value obtained by
the following equations is used. The sample period is taken to be T(s). Here, the
sample period is the time interval for changing the current. It may be different
than the period for detecting the position.
ω=2
·Fr
S22
ζωs+ω2=0
Z0=exp(
sT)
'Z
0' is expressed the zero point of the filter that corresponds
to the resonant frequency in a discrete system. There is not one 'Z
0' but
rather two or a complex number. Here, they will be expressed As 'Z
01' and 'Z
02'.
Consider the FIR filter that contains 'Z
01' as a zero point.
F(
z)=(1
-Z01Z-1)(1
Z02Z-1)=1-(
Z01+Z02) Z
-1+Z01Z02Z-2
This characteristic is the same as for a notch filter. In FIG. 12 a total of
four zero points 'Z
0' have been designed and synthesized (connected in series
with the filter) at frequencies 5.7 kHz and 5.9 kHz and with ζ set to 0.
When the base wave is passed through this filter, the filter output is a wave in
which the resonant frequency component has been controlled. Here, passing the rectangular
waveform shown in FIG.
13A through the filter as the basic wave, tracks
(FF current, position track) are generated that correspond to the initial position
for settling control. Also, by passing the rectangular waveform shown in FIG.
13B
through the filter, trajectories (FF current, position trajectory) are generated
that correspond to the initial velocity.
The design of the position trajectory is obtained by designing a trajectory such
that only the initial position moves after movement and where the current and velocity
become zero. Also, in designing a trajectory for the initial velocity, it is okay
to design trajectory where the velocity and position both become zero.
In other words, the track for the position is designed by letting a suitable
current
flow. By dividing the current by the distance moved, it is possible to design the
current for the distance 1 track. As shown in FIG. 14, with the initial velocity
zero, the current, velocity and position trajectories for moving 1 track are obtained
by simulation.
Next, the current that will give a constant velocity is designed. After the
current has flowed, constant velocity is maintained. By dividing the distance by
that constant velocity, the track that will give 1 track/sample is obtained. FIG.
15 shows the current, velocity and position trajectories for obtaining a velocity
of 1 track/sample.
Next, when the initial conditions are 0 track, and 1 track/sample, the arrival
position in the minus direction that gave the aforementioned trajectory is obtained.
FIG. 16 shows the current, velocity and position tracks for making the velocity
'0' from an initial position of '0' and initial velocity of 1 track/sample. By
applying the afore mentioned position trajectory in order to make the arrival position
'0', it is possible to design trajectories that will go from an initial position
of 0 track, and initial velocity of 1 track/sample, to end with a position of '0'
track and velocity of 0 tracks/sample. FIG. 17 shows the trajectories (current,
velocity, position) that will go from an initial position of 0 track and initial
velocity of 1 track/sample, and end at a position of 0 track and velocity of 0 tracks/sample.
In this way, it is possible to design a trajectory with initial conditions of
1 track and initial velocity of 0 tracks/sample, and end conditions of 0 track
and 0 tracks/sample, and a trajectory with initial conditions of 0 track and initial
velocity of 1 track/sample, and end conditions or 0 track-and 0 tracks/sample.
FIG. 6 shows an example, of the trajectories shown in the figures, that give
current and position to the control system. In other words, the trajectory generator
22 contains a current track and target position track. Therefore, there
are current unit track generators
36,
38 and target position unit
trajectory generators
30,
31. Moreover, the target position unit
trajectory generators
30,
31 are driven by a clock source, and then
a target position trajectory, that is proportional to the initial position 'X
0'
and initial velocity 'V
0' at the start of settling, is generated by multipliers
32,
33 and an adder
34. This target position trajectory is
given to the adder
23 in a prior step of the observer
20 as a target
value, and the difference between the target position and the detected position
is found and the difference is input to the observer
20 as the observed position.
Similarly, the current unit trajectory generators
36,
38
are driven by a clock source, and then a target position trajectory, that is proportional
to the initial position 'X
0' and initial velocity 'V
0' at the start
of settling, is generated by multipliers
37,
39 and an adder
40.
This target trajectory is given to an adder in the output stage an of the observer
20 as the feed forward current.
Since feed forward control is performed during settling control in this way,
high-speed response is possible. Also, since current amp characteristics as well
as the resonant effects are considered during trajectory design, high-speed response
is possible even when feed forward control is performed.
[Positioning Control Process]
Next, the positioning control process by a program executed by the MPU
19
will be explained. FIG. 18 is a flowchart of the task controls of the positioning
control process. FIG. 19 is a flow chart of the coarse control process. FIG. 20
is a flowchart of the settling control process. FIG. 21 is a flowchart of the following control.
First, the positioning control process will be explained using FIG.
18.
(S
1) This task process is executed when an on-track start command
is received. First, it is judged whether the target value 'y
0' has changed.
(S
2) When the target value 'y
0' has not changed, following
control (described later using FIG. 21) is executed and the process returns to
step S
1.
(S
3) When the target value 'y
0' has changed, the distance
to the target value 'y
0' is checked whether it is 4 tracks or less.
(S
4) When the distance to the target value 'y
0' is not
4 tracks or less, coarse control (described later using FIG. 19) is executed and
the process returns to step S
3.
(S
5) When the distance to the target value 'y
0' is 4 tracks
or less, then the specified settling conditions are checked whether they are satisfied.
When the specified settling conditions are satisfied, processing advances to following
control of step S
2. The judgment conditions will be explained for the settling
control in FIG.
20.
(S
6) When the specified settling conditions are not satisfied,
settling control (described later using FIG. 20) is executed and the process returns
to step S
5.
In the seek process, when coarse control is performed and the distance is within
4 tracks, then processing moves to settling control, and when the settling conditions
are satisfied, processing moves to following con
