U.S. patent application number 09/748973 was filed with the patent office on 2001-06-14 for control method and control device for a disk storage device.
This patent application is currently assigned to FUJITSU LIMTED. Invention is credited to Takaishi, Kazuhiko.
Application Number | 20010003497 09/748973 |
Document ID | / |
Family ID | 26333083 |
Filed Date | 2001-06-14 |
United States Patent
Application |
20010003497 |
Kind Code |
A1 |
Takaishi, Kazuhiko |
June 14, 2001 |
Control method and control device for a disk storage device
Abstract
This invention comprises a head 3 for reading information from a
disk storage medium 2, an actuator 5 for moving the head 3, and a
control circuit 8 for calculating a control signal from the
position signal, that was read by the head, using an eccentricity
observer 21. The control circuit 8 selects a first eccentricity
estimation gain during seek control and selects a second
eccentricity estimation gain during following control. This makes
it possible to perform seek control that is not affected by the
eccentricity error.
Inventors: |
Takaishi, Kazuhiko;
(Kawasaki, JP) |
Correspondence
Address: |
Patric G. Burns
GREER, BURNS & CRAIN, LTD
300 S. Wacker
25th Floor
Chicago
IL
60606
US
|
Assignee: |
FUJITSU LIMTED
|
Family ID: |
26333083 |
Appl. No.: |
09/748973 |
Filed: |
December 27, 2000 |
Current U.S.
Class: |
360/77.04 ;
G9B/5.221 |
Current CPC
Class: |
G11B 5/59611 20130101;
G11B 5/59627 20130101 |
Class at
Publication: |
360/77.04 |
International
Class: |
G11B 005/596 |
Foreign Application Data
Date |
Code |
Application Number |
Jun 30, 1998 |
JP |
10-185046 |
Jan 19, 1999 |
JP |
PCT/JP99/00164 |
Claims
What is claimed is:
1. A control method for a disk storage device, having a disk
storage medium, ahead for reading information on said disk storage
medium, an actuator for moving said head, and a control circuit
using eccentricity estimation observer control including a model of
said actuator and a model of the eccentricity for calculating a
control signal for driving said actuator based on a position signal
that is read from said disk storage device by said head;
comprising: a step of selecting a first eccentricity estimation
gain during seek control, and of selecting a second eccentricity
estimation gain during following control; and a step of calculating
state signals, that include estimated position, estimated velocity,
estimated bias and estimated eccentricity signals, based on an
error between said position signal and estimated position signal,
the actuator estimation gain for estimating said actuator
operation, and said selected eccentricity estimation gain; and of
calculating said state signals and said control signal.
2. The control method for a disk storage device of claim 1 wherein
said selection step comprises: a step of setting said first
eccentricity estimation gain to `0` during said seek control.
3. The control method for a disk storage device of claim 1 further
comprising: a step of detecting the position error between the
target position and current position; and a step of comparing said
position error with a specified value and determining that control
is following control when said position error is within the
specified value, and determining that control is seek control when
said position error exceeds the specified value.
4. The control method for a disk storage device of claim 1 wherein
said calculation step comprises: a first calculation step of
calculating the estimated position signal, estimated velocity
signal and estimated eccentricity signal for the next sample from
said error and the estimated position signal, estimated velocity
signal and estimated eccentricity signal for the current sample as
well as from said actuator estimation gain and said eccentricity
estimation gain; and a second calculation step of calculating said
control current from said estimated position signal, estimated
velocity signal and estimated eccentricity signal for the current
sample.
5. The control method for a disk storage device of claim 4 wherein
said first calculation step comprises: a third calculation step of
calculating the estimated position signal and estimated velocity
signal for the next sample from said error and the estimated
position signal and estimated velocity signal for the current
sample, as well as from said actuator estimation gain; and a fourth
calculation step of calculating the estimated eccentricity signal
for the next sample from said error and the estimated eccentricity
signal for the current sample, as well as from said eccentricity
estimation gain.
6. The control method for a disk storage device of claim 1 wherein
said calculation step comprises: a fifth calculation step of
calculating a corrected state signal based on said error,
previously calculated state signal, said actuator estimation gain
and said eccentricity estimation gain; a sixth calculation step of
calculating said control signal from said corrected state signal;
and a seventh calculation step of calculating the next state signal
from said control signal and said corrected state signal.
7. The control method for a disk storage device of claim 6 wherein
said fifth calculation step comprises: an eighth calculation step
of calculating a corrected estimated position signal and estimated
velocity signal from said error, the estimated position signal and
estimated velocity signal for the previous sample, as well as from
said actuator estimation gain; and a ninth calculation step of
calculating the estimated eccentricity signal for the next sample
from said error, the estimated eccentricity signal for the previous
sample and from said eccentricity estimation gain.
8. The control method for a disk storage device of claim 1 wherein
said calculation step comprises: a tenth calculation step of
calculating the corrected estimated position signal and estimated
velocity signal from said error, the previous estimated position
signal and previous estimated velocity signal, and said actuator
estimation gain; an eleventh calculation step of calculating said
control signal based on said corrected estimated position signal
and previous estimated eccentricity signal; and a twelfth
calculation step of calculating the next estimated position signal,
next estimated velocity signal and next estimated eccentricity
signal based on said error, said corrected estimated position
signal and said corrected estimated velocity signal and previous
estimated eccentricity signal.
9. The control method for a disk storage device of claim 1
comprising: a save step of saving in memory said estimated
eccentricity signal for the head before changing, according to a
head change instruction; and an execution step of reading said
estimated eccentricity signal for said head before change, and
executing said calculation steps with said read estimated
eccentricity signal as the initial value.
10. The control method for a disk storage device of claim 9 wherein
said save step comprises: a step of saving said converted value
after said estimated eccentricity signal has been converted to the
estimated eccentricity signal for the reference sector; and wherein
said execution step comprises: a step of correcting said stored
estimated eccentricity signal to the estimated eccentricity signal
for the current sector.
11. A control device for a disk storage device having a disk
storage medium; a head for reading information on said disk storage
medium; an actuator for moving said head; and a control circuit
using eccentricity estimation observer control including a model of
said actuator and a model of the eccentricity for calculating a
control signal for driving said actuator based on a position signal
that is read from said disk storage device by said head; wherein
said control circuit selects a first eccentricity estimation gain
during seek control, and selects a second eccentricity estimation
gain during following control; and calculates state signals, that
include estimated position, estimated velocity, estimated bias and
estimated eccentricity signals, based on an error between said
position signal and estimated position signal, the actuator
estimation gain for estimating said actuator operation, and said
selected eccentricity estimation gain; and calculates said control
signal from said state signals.
12. The control device for a disk storage device of claim 11
wherein said control circuit sets said first eccentricity
estimation gain to `0` during said seek control.
13. The control device for a disk storage device of claim 11
wherein said control circuit detects the position error between the
target position and current position, compares said position error
with a specified value and determines that control is following
control when said position error is within the specified value, and
determines that control is seek control when said position error
exceeds the specified value.
14. The control device for a disk storage device of claim 11
wherein said control circuit calculates the estimated position
signal, estimated velocity signal and estimated eccentricity signal
for the next sample from said error and the estimated position
signal, estimated velocity signal and estimated eccentricity signal
for the current sample as well as from said actuator estimation
gain and said eccentricity estimation gain, and calculates said
control current from said estimated position signal, estimated
velocity signal and estimated eccentricity signal for the current
sample.
15. The control device for a disk storage device of claim 14
wherein said control circuit calculates the estimated position
signal and estimated velocity signal for the next sample from said
error and the estimated position signal and estimated velocity
signal for the current sample, as well as from said actuator
estimation gain; and calculates the estimated eccentricity signal
for the next sample from said error and the estimated eccentricity
signal for the current sample, as well as from said eccentricity
estimation gain.
16. The control device for a disk storage device of claim 11
wherein said control circuit calculates a corrected state signal
based on said error, previously calculated state signal, said
actuator estimation gain and said eccentricity estimation gain;
calculates said control signal from said corrected state signal;
and calculates the next state signal from said control signal and
said corrected state signal.
17. The control device for a disk storage device of claim 16
wherein said control circuit calculates a corrected estimated
position signal and estimated velocity signal from said error, the
estimated position signal and estimated velocity signal for the
previous sample, as well as from said actuator estimation gain; and
calculates the estimated eccentricity signal for the next sample
from said error, the estimated eccentricity signal for the previous
sample and from said eccentricity estimation gain.
18. The control device for a disk storage device of claim 11
wherein said control circuit calculates the corrected estimated
position signal and estimated velocity signal from said error, the
previous estimated position signal and previous estimated velocity
signal, and said actuator estimation gain; calculates said control
signal based on said corrected estimated position signal and
previous estimated eccentricity signal; and calculates the next
estimated position signal, next estimated velocity signal and next
estimated eccentricity signal based on said error, said corrected
estimated position signal and said corrected estimated velocity
signal and previous estimated eccentricity signal.
19. The control device for a disk storage device of claim 11
wherein said control circuit saves in memory said estimated
eccentricity signal for the head before changing, according to a
head change instruction, then reads said estimated eccentricity
signal for said head before change, and executes said calculation
steps with said read estimated eccentricity signal as the initial
value.
20. The control device for a disk storage device of claim 19
wherein said control circuit saves said converted value after said
estimated eccentricity signal has been converted to the estimated
eccentricity signal for the reference sector, then reads the said
corrected estimated eccentricity signal from said memory and
corrects said estimated eccentricity signal to the estimated
eccentricity signal for the current sector.
Description
FIELD OF THE INVENTION
[0001] This invention relates to a control method and control
device for controlling an actuator to move a head to a target
position in a disk storage device, that reads or reads/writes
information from a disk storage medium by the head.
BACKGROUND OF THE INVENTION
[0002] Disk storage devices such as magnetic disk drives or optical
disk drives are widely used as storage devices for computers and
the like. In these kinds of disk storage devices, eccentricity of
the disk medium occurs. This eccentricity occurs when the center of
rotation of the disk medium that was recorded the position
information shifts when that of writing the position
information.
[0003] In the sector servo method, the position information (servo
information) for detecting the actuator position is recorded on
each disk surface. This position information is formed on
concentric circles. When the center of rotation of the disk matches
with the center of rotation of the disk when the position
information was written, then ideally no eccentricity will
occur.
[0004] However, in actuality, the centers of rotation do not match
and eccentricity occurs. The reason for this is probably due to
thermal deformation of the disk medium and a spindle shaft, or
shifting of the disk due to external impact. When there is
eccentricity, it can be seen from the actuator's point of view that
sinusoidal disturbance on the order of integral multiples of the
rotation frequency is applied. Therefore, a technique for
correcting this eccentricity is necessary.
[0005] Control using an eccentricity estimation observer
(estimator) has been known as a technique for correcting this
eccentricity. In the control by this eccentricity estimation
observer, steady position control by the estimated values is
required.
[0006] FIG. 12 is a configuration drawing of this prior art, and
FIG. 13 is a drawing for explaining this prior art.
[0007] Position control of a magnetic head by the use of an
eccentricity estimation observer is described in detail in Japanese
Unexamined Published patent No. 7-50075 (U.S. Pat. No. 5,404,235).
Therefore, the eccentricity estimation observer will only be simply
explained here.
[0008] First, an ideal actuator model that does not include
resonance will be considered. Here, when `x1` is taken to be the
position, `x2` the velocity, `y` the observed position (detected
position), `u` the control current and `s` the Laplace operator,
then the state equations are given by equations (1) and (2) below.
1 S ( X 1 X 2 ) = ( 0 1 0 0 ) ( X 1 X 2 ) + Kp ( 0 1 ) u ( 1 ) 2 y
= ( 1 0 ) ( X 1 X 2 ) ( 2 )
[0009] Here, `Kp` is the acceleration constant when rotating type
actuator of the model is considered to be an equivalent linear type
actuator.
[0010] When considering the current feedback, the steady-state
current (bias current) state `x3` is added to the state equations,
and thus the state equations are given by equations (3) and (4)
below. 3 S ( X 1 X 2 X 3 ) = ( 0 1 0 0 0 K p 0 0 0 ) ( X 1 X 2 X 3
) + Kp ( 0 1 0 ) u ( 3 ) y = ( 1 0 0 ) ( X 1 X 2 X 3 ) ( 4 )
[0011] Furthermore, the eccentricity disturbance state is added to
these state equations. When `x4` and `x5` are taken to be the state
variable of the eccentricity, and `.omega.0` is taken to be the
eccentric angular velocity, then the state equations are given by
equations (5) and (6) below. 4 S ( X 1 X 2 X 3 X 4 X 5 ) = ( 0 1 0
0 0 0 0 Kp Kp 0 0 0 0 0 0 0 0 0 0 1 0 0 0 - 0 2 0 ) ( X 1 X 2 X 3 X
4 X 5 ) + Kp ( 0 1 0 0 0 ) u ( 5 ) y = ( 1 0 0 0 0 ) ( X 1 X 2 X 3
X 4 X 5 ) ( 6 )
[0012] Here, when x4=cos(.omega.0.multidot.t) and
x5=sin(.omega.0.multidot- .t), then sx4
=-.omega.0.multidot.sin(w0.multidot.t) and sx5=.omega.0
.multidot.cos(.omega.0.multidot.t), so sx4=-.omega.0.multidot.
x5=sx5=.omega.0.multidot.x4. Therefore, the state equations (5) and
(6) are given by equations (7) and (8) below. 5 S ( X 1 X 2 X 3 X 4
X 5 ) = ( 0 1 0 0 0 0 0 Kp Kp 0 0 0 0 0 0 0 0 0 0 - 0 0 0 0 0 0 ) (
X 1 X 2 X 3 X 4 X 5 ) + Kp ( 0 1 0 0 0 ) u ( 7 ) 6 y = ( 1 0 0 0 0
) ( X 1 X 2 X 3 X 4 X 5 ) ( 8 )
[0013] In equation (5) the eccentricity is estimated by the
sinusoidal transfer function (1/(S.sup.2+.omega.0.sup.2)). As shown
in FIG. 13, equation (7) shows the movement in rectangular
coordinates (x4,x5) of a point on the circle with radius
(x4{circumflex over ( )}2+x5{circumflex over ( )}2) that is
rotating at constant velocity.
[0014] The observer is designed to transfer the state equations (7)
and (8) to a discrete form. The equations are transferred into a
discrete form by estimating the zero-dimension hold. In other
words, it performs Z conversion. By considering the time lag from
when the position is detected until current is output to the
actuator, state equations become 6 dimensional. Even when not
considered, the state equations are given by equations (9) and (10)
below. 7 ( X 1 [ K + 1 ] X 2 [ K + 1 ] X 3 [ K + 1 ] X 4 [ K + 1 ]
X 5 [ K + 1 ] ) = ( 1 T KpT 2 2 Kp 0 2 ( 1 - cos ( 0 T ) ) Kp 0 2 (
0 T - sin ( 0 T ) ) 0 1 KpT Kp - sin ( 0 T ) ) 0 Kp 0 ( 1 - cos ( 0
T ) ) 0 0 1 0 0 0 0 0 cos ( 0 T ) - sin ( 0 T ) 0 0 0 sin ( 0 T )
cos ( 0 T ) ) ( X 1 [ K ] X 2 [ K ] X 3 [ K ] X 4 [ K ] X 5 [ K ] )
+ Kp ( T 2 / 2 T 0 0 0 ) u [ k ] ( 9 ) y [ k ] = ( 1 0 0 0 0 ) ( X
1 [ K ] X 2 [ K ] X 3 [ K ] X 4 [ K ] X 5 [ K ] ) ( 10 )
[0015] Here, T is the sample period. As shown in equation (11), the
coefficients in equations (9) and (10) are A, B and C. 8 A = ( 1 T
KpT 2 2 Kp 0 2 ( 1 - cos ( 0 T ) ) Kp 0 2 ( 0 T - sin ( 0 T ) ) 0 1
KpT Kp - sin ( 0 T ) ) 0 Kp 0 ( 1 - cos ( 0 T ) ) 0 0 1 0 0 0 0 0
cos ( 0 T ) - sin ( 0 T ) 0 0 0 sin ( 0 T ) cos ( 0 T ) ) , B = Kp
( T 2 / 2 T 0 0 0 ) C = ( 1 0 0 0 0 ) ( 11 )
[0016] Here, the observer is expressed by the equations (12), (13)
and (14) below. 9 ( PX 1 [ K + 1 ] PX 2 [ K + 1 ] PX 3 [ K + 1 ] PX
4 [ K + 1 ] PX 5 [ K + 1 ] ) = A ( PX 1 [ K ] PX 2 [ K ] PX 3 [ K ]
PX 4 [ K ] PX 5 [ K ] ) + B u [ k ] + ( L 1 L 2 L 3 L 4 L 5 ) ( y [
k ] - PX 1 [ k ] ) ( 12 ) py [ k ] = C ( PX 1 [ K ] PX 2 [ K ] PX 3
[ K ] PX 4 [ K ] PX 5 [ K ] ) ( 13 ) 10 u [ k ] = - ( F 1 F 2 1 1 0
) ( PX 1 [ K ] PX 2 [ K ] PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) ( 14
)
[0017] Here, px1 is the state variable for position (estimated
position), px2 is the state variable for velocity (estimated
velocity), px3 is the state variable for the bias current
(estimated bias current), px4 and px5 are state variable for the
eccentricity (estimated eccentricity), `u` is the control current,
`y` is the observed position (detected position) and `py` is the
estimated position.
[0018] Moreover, L1 to L5 are the estimation gains of the observer,
where L1 is the position estimation gain, L2 is the velocity
estimation gain, L3 is the bias estimation gain, and L4 and L5 are
the eccentricity estimation gains. Furthermore, F1 to F5 are the
state feedback matrix.
[0019] When this is shown in a block diagram, a diagram as shown in
FIG. 12 is obtained. In other words, a plant 90 shows the portion
that performs real positioning of the head of a magnetic disk
drive, and includes an actuator, AMP and magnetic heads. The
observer (estimator) 91 estimates the position, velocity, bias and
eccentricity from the current state, and outputs control
current.
[0020] The position signal (servo signal) y[k] that is read by the
magnetic head is output from the plant 90. The observer 91
calculates the error (y[k]-px1[k]) between the position signal y[k]
and the estimated position py[k] (=px1[k]). The error is input to
the fourth gain multiplier 96. The gain multiplier 96 multiplies
the error by the estimation gain L (L2 to L5) (see equation (12)).
L1 to L3 is the estimated operating gain of the actuator, and L4
and L5 are the eccentricity estimation gain.
[0021] The second gain multiplier 94 multiplies the control current
u[k] by the coefficient B (see equation (12)). The third gain
multiplier 95 multiplies the state signal px[k] at the time of this
sample by coefficient A (see equation (12)). An adder 98 adds the
outputs from the three adders 94 to 96. From this the state signal
px[k+1] for the next sample of equation (12) is output.
[0022] This state signal px[k+1] for the next sample is delayed one
sample by an delay circuit 99 to obtain the state signal px[k] for
this sample. This state signal px[k] is then multiplied by a
feedback coefficient `C` by a fifth multiplier 97. From this, the
estimated position py[k] for this sample, as given by Equation
(13), is obtained.
[0023] Furthermore, the state signal px[k] is multiplied by the
feedback coefficient `F` by a first multiplier 93. From this, the
control current u[k] for this time, as given by Equation (14), is
obtained. This control current u[k] is supplied to the plant
90.
[0024] In this way, the observer 91 is comprised by an eccentricity
estimation observer including the actuator model and eccentric
model, and it predicts the next state from the error between the
detected position and estimated position, the control current and
the state variables, and creates a control current from the state.
In this way, the eccentricity is corrected in real time making it
possible to quickly correct. In this prior art, a control signal
was created by a similar configuration during seek control time as
well.
[0025] However, this prior art had the following problems.
[0026] First, during seeking, the actuator moves at high speed so
it may move more than 50 tracks per sample. When it moves at high
speed like this, it is not possible to accurately detect the
position for each sample. Therefore, the eccentricity estimation
observer estimates the amount of eccentricity from the position
error and thus error occurs in the estimated amount of
eccentricity. It is a long time that the eccentricity state
converges if there is the error, because convergence is slow at
about 90 Hertz. This becomes a problem in that convergence at the
end of seeking becomes slow.
[0027] Second, when seeking over a long distance, the output
current of the current amp becomes saturated. The maximum current
during saturation differs depending on the power-supply voltage or
the actuator resistance, and it is greatly affected by variations
in differences in the environment and equipment. Therefore, since
the eccentricity estimation observer does not predict the
saturation of the output current of the current amp, errors occur
in the amount of eccentricity that is estimated from the position
error. Therefore, there was the problem the convergence becomes
slow after seeking.
DISCLOSURE OF THE INVENTION
[0028] The objective of this invention is to provide a control
method and control device for a disk storage device that prevent a
slow convergence when seek end.
[0029] Another objective of this invention is to provide a control
method and control device for a disk storage device that prevent a
slow convergence when seek end even when eccentricity correction is
performed.
[0030] A further objective of this invention is to provide a
control method and control device for a disk storage device that
prevent a position error from affecting the estimated value for the
eccentricity.
[0031] The disk storage device of this invention comprises: a disk
storage medium, a head for reading information on the disk storage
medium, an actuator for moving the head, and a control circuit,
that uses eccentricity estimation observer control that includes an
actuator model and eccentric model, for calculating a control
signal for driving the actuator based on a position signal that is
read from the disk storage medium by the head.
[0032] In addition, the control method and device of this invention
comprises: a step of selecting a first eccentricity estimation gain
during seek control, and for selecting a second eccentricity
estimation gain during following control; and a step of calculating
a state signal, which includes the estimated position, estimated
velocity, estimated bias signal and estimated eccentricity signal,
and which is based on the estimated actuator gain, that estimates
the actuator operation, and the selected eccentricity estimation
gain, and of calculating the control signal from the state
signal.
[0033] In this invention, the amount of eccentricity is not
estimated from the position error during seeking since large errors
occur on the amount of eccentricity when estimating it from the
position error during seeking. However, in order for a steady seek
operation, it is necessary to correct the eccentricity. Therefore,
this invention minimizes the eccentricity estimation gains L4, L5
of the eccentricity estimation observer during seek more than
during following, to prevent the position error from affecting the
estimated amount of eccentricity.
[0034] For example, during seeking, the eccentricity estimation
gains L4, L5 are taken to be `0`. By doing this, the state equation
(12) becomes equation (15) below. 11 ( PX 1 [ K + 1 ] PX 2 [ K + 1
] PX 3 [ K + 1 ] PX 4 [ K + 1 ] PX 5 [ K + 1 ] ) = A ( PX 1 [ K ]
PX 2 [ K ] PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) + B u [ k ] + ( L 1 L
2 L 3 0 0 ) ( y [ k ] - PX 1 [ k ] ) ( 15 )
[0035] In this equation (15), by calculating the state variables X4
and X5 for eccentricity, the equation (16) below is obtained. 12 (
PX 4 [ K + 1 ] PX 5 [ K + 1 ] ) = ( cos ( 0 T ) - sin ( 0 T ) sin (
0 T ) cos ( 0 T ) ) ( PX 4 [ K ] PX 5 [ K ] ) ( 16 )
[0036] In equation (16), the state variable X4 and X5 for
eccentricity become unrelated to the observed position y and
estimated position x1. In other words, state variables X4 and X5
that are not affected by the position error (y[k] - x1[k]) are
obtained. In equation (16), .omega.0T is the phase for one sample.
In addition, equation (16) shows that the current state variables
have a phase shift of only one sample. In other words, equation
(16) is a sinusoidal recursive equation.
[0037] Therefore, it is possible to prevent the position error from
affecting the estimated eccentricity value when seeking. Moreover,
it is possible to quicken the convergence operation after seeking.
Also, as shown in equation (16), since the eccentricity is
corrected during seeking, there is no loss of stability during
seeking. The eccentricity estimation gain during seek, does not
need to be `0`, but can be a value near `0`.
[0038] Moreover, another form of the invention further comprises a
step of detecting the position error between the target position
and the current position, and a step of determining that control is
the aforementioned following control when the position error is
within a specified range, and of determining that control is the
aforementioned seek control when the position error exceeds the
specified value by comparing the position error and the specified
values.
[0039] Furthermore, in another form of the invention, the
calculation step comprises: a first calculation step of calculating
the estimated position signal, estimated velocity signal and
estimated eccentricity signal for the next sample from the
estimated position signal, estimated velocity signal, estimated
eccentricity signal, estimated actuator gain and eccentricity
estimation gain of the current sample; and a second calculation
step of calculating the control current from estimated position
signal, estimated velocity signal and estimated eccentricity signal
of the current sample. In other words, the eccentricity estimation
observer is a prediction observer.
[0040] In yet another form of the invention, the aforementioned
first calculation step comprises: a third calculation step of
calculating the estimated position signal and estimated velocity
signal for the next sample from the estimated position signal,
estimated velocity signal and estimated actuator gain of the
current sample; and a fourth calculation step of calculating the
estimated eccentricity signal for the next sample from the
estimated eccentricity signal and from the eccentricity estimation
gain of the current sample.
[0041] In this form of the invention, the calculation of estimating
the actuator operation is separate from the calculation of
estimating the amount of eccentricity, so the amount of calculation
is reduced. Therefore it is possible to quickly calculate the
estimated amount.
[0042] In even yet another form of the invention, the calculation
step comprises: a fifth calculation step of calculating the
corrected state signal based on the error, the previously
calculated state signal, estimated actuator gain and eccentricity
estimation gain; a sixth calculation step of calculating the
control signal from the corrected state signal; and a seventh
calculation step of calculating the next state signal from the
control signal and corrected state signal.
[0043] In this form of the invention, the eccentricity estimation
observer is a current observer. Therefore it is suitable for
processor processes.
[0044] Furthermore, in yet another form of the invention, the
aforementioned fifth calculation step comprises: an eighth
calculation step of calculating the corrected estimated position
signal and estimated velocity schedule from the error, and
estimated position signal and estimated velocity signal from the
previous sample, and the estimated actuator gain; and a ninth
calculation step of calculating the estimated eccentricity signal
for the next sample from the aforementioned error, and the
estimated eccentricity signal from the previous sample and the
estimated eccentricity gain.
[0045] In the current observer of this form of the invention, the
calculation for estimating the actuator operation and the
calculation for estimating the amount of eccentricity are separate,
so the amount of calculations is reduced. Therefore, it is possible
to calculate the estimations quickly.
[0046] In another form of the invention, the calculation step
comprises: a tenth calculation step of calculating the corrected
estimated position signal and estimated velocity signal from the
error, the previous estimated position signal and estimated
velocity signal and the estimated actuator gain; an eleventh
calculation step of calculating the control signal based on the
corrected estimated position signal and previous estimated
eccentricity signal; and a twelfth calculation step of calculating
the estimated position signal, estimated velocity signal and
estimated eccentricity signal for the next time, based on the
corrected estimated position signal, corrected estimated velocity
signal and previous estimated eccentricity signal.
[0047] In this form of the invention, since the calculation of the
estimated eccentricity signal for the next time is performed after
the calculation of the control signal, the time from when the
position signal is sample until the control signal is output is
shortened. Therefore, it is possible to quickly output the control
signal.
[0048] Furthermore, another form of the invention comprises: a
saving step of saving the estimated eccentricity signal for the
head before changing when there is an instruction to change the
head; and a step of reading the estimated eccentricity signal for
the changed head and executing the aforementioned calculation steps
with the read estimated eccentricity signal as the initial
value.
[0049] In this form of the invention, since the amount of
eccentricity varies for each head, or in other words for each
surface of the disk medium, it is necessary to change the state
variable for the eccentricity for each head. When doing this, the
state variable for eccentricity is saved for each head, and by
changing the state variable when changing heads, it is possible to
initialize the eccentricity estimation observer with a state
variable that matches that head.
[0050] In another form of the invention, the saving step comprises
a saving step of saving the conversion value after the estimated
eccentricity signal has been converted to an estimated eccentricity
signal for the reference sector, and the execution step comprises a
step of correcting the saved estimated eccentricity signal to an
estimated eccentricity signal of the current sector.
[0051] In this form of the invention, when saving the initial value
of the eccentricity estimation observer for each head, only two
state variables need to be saved. Therefore, it is possible to
reduce the amount of memory required for saving the state variable
for eccentricity.
BRIEF DESCRIPTION OF THE DRAWINGS
[0052] FIG. 1 is a configuration diagram of an embodiment of this
invention.
[0053] FIG. 2 is a block diagram of the servo control in FIG.
1.
[0054] FIG. 3 is a flowchart of the servo control in FIG. 1.
[0055] FIG. 4 is a diagram that explains FIG. 3.
[0056] FIG. 5 is a flowchart of the calculation process in FIG.
3.
[0057] FIG. 6 is a block diagram of the servo controller of another
embodiment of this invention.
[0058] FIG. 7 is a flowchart of the calculation process in FIG.
6.
[0059] FIG. 8 is a flowchart of another calculation process of the
invention.
[0060] FIG. 9 is a flowchart of the head-change process in FIG.
1.
[0061] FIG. 10 is a diagram that explains the state variables in
FIG. 9.
[0062] FIG. 11 is a diagram that explains the head-change process
in FIG. 9.
[0063] FIG. 12 is a configuration diagram of the prior art.
[0064] FIG. 13 is a diagram that explains the prior art.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0065] FIG. 1 is a configuration diagram of an embodiment of this
invention, FIG. 2 is a block diagram of the prediction observer in
FIG. 1, FIG. 3 is a flowchart of the servo control process and FIG.
4 is a diagram that explains an embodiment of the invention.
[0066] As shown in FIG. 1, a magnetic disk drive 1 comprises a
magnetic disk 2 and magnetic heads 3. The magnetic disk 2 comprises
data surfaces that are embedded the servo signal on the data
tracks. The magnetic head 3 reads information from or writes
information to the magnetic disk 2. The magnetic disk 2 is rotated
by a spindle motor 4.
[0067] A voice coil motor (VCM) 5 moves the magnetic heat 3 in a
direction that crosses the tracks on the magnetic disk 2. A power
amp 6 drives the VCM 5. A spindle-drive circuit 7 drives the
spindle motor 4. A control circuit 8 comprises a microprocessor, an
analog/digital converter, a digital/analog converter and RAM.
[0068] The control circuit (called the processor below) 8 reads the
position signal from the magnetic head 3 to detect the current
position y[k] of the magnetic head, and then creates a control
value (control current value) u[k] that corresponds to the distance
between that position and the position being sought.
[0069] A read/write circuit 9 control reading and writing of the
magnetic head 3 according to instructions from the processor 8. A
position detection circuit 10 demodulates the servo signal from the
magnetic head 3, and outputs the position signal to the processor
8. A ROM 11 stores the data and program required for processing of
the processor 8.
[0070] A hard disk controller 12 performs interface control with
the host computer. There is RAM 13 in this hard disk controller 12.
The RAM 13 stores data from the host computer and data to be sent
to the host computer.
[0071] FIG. 2 is a block diagram of the servo process that is
executed by the processor in FIG. 1. In FIG. 2, the plant 20 shows
the part that performs physical positioning of the head in the
magnetic disk device, and it comprises an actuator 5, amp 6 and the
magnetic head 3. A prediction observer (estimator) 21 estimates the
next position, velocity, bias and eccentricity states from the
current states, and outputs a control current.
[0072] The position signal (servo signal) y[k] that is read by the
magnetic head 3 is output from the plant 20. In the observer 21, an
error computing unit 22 calculates the error (y[k]-px1[k]) between
the position signal y[k] and the estimated position py[k]
(=px1[k]). This error is then input to a fourth gain multiplier 26.
The gain multiplier 26 multiplier multiplies this error by the
estimation gain L (L1 to L5) (see equation (12)).
[0073] A second gain multiplier 24 multiplies the control current
u[k] by a coefficient B (see equation (12)). A third multiplier 25
multiplies the state signal px[k] of the current sample by a
coefficient A (see equation(12)). And adder 28 adds the outputs of
the three adders 24 to 26. By doing this, the state signal px[k+1]
for the next sample of equation (12) is output.
[0074] This state signal px[k+1] for the next sample is delayed one
sample by the delay circuit 29, to obtain the state signal px[k] of
the current sample. This state signal px[k] is multiplied by a
feedback coefficient C by a fifth multiplier 27. By doing this, the
estimated position py[k] of the current sample given by equation
(13) is obtained.
[0075] Furthermore, the state signal px[k] is multiplied by a
feedback coefficient F by a first multiplier 23. By doing this, the
current control current u[k] given by equation (14) is obtained.
This control current u[k] is supplied to the plant 20.
[0076] An error calculator 30 calculates the position error
.DELTA.x between the current position y[k] and the target position
r. The current position y[k] is given from the plant 20. A target
velocity generator 31 generates the target velocity v0 from the
position error .DELTA.x. A velocity calculator 32 calculates
difference in velocity between the target velocity v0 and the
estimated velocity px2[k] from the observer 21.
[0077] A gain multiplier 33 multiplies the velocity difference by a
velocity gain C0. A compensator 34 adds the velocity difference,
the estimated bias current px3[k] from the prediction observer 21
and the estimated amount of eccentricity px4[k] from the prediction
observer 21, inverts it and outputs it as the control current u[k].
The control output u[k] during seeking is given by equation (17)
below.
u[k]=-Co (Vo-PX.sub.2[k])-PX.sub.3[k]-PX.sub.4[k] . . . (17)
[0078] A switch 35 selects the output of the compensator 34 during
seeking, and selects the output of the prediction observer 21
during following. These blocks 21 to 35 are realized by the
programs of the processor 8.
[0079] The servo interrupt process is explained by FIG. 3 and FIG.
4.
[0080] (S1) When a servo interrupt (servo gate signal) is given to
the processor 8, the processor 8 reads the position signal y[k]
from the position detection circuit 10.
[0081] (S2) The processor 8 calculates the difference between the
target position r and the current position (position signal)
y[k].
[0082] (S3) The processor 8 determines whether the absolute value
abs[y-r] is less than four tracks. Here, the judgment criteria for
seeking and following is four tracks. When the absolute value
abs[y-r] is four tracks or less then it is determined that
following is in progress, and operation advances to step S4. When
the absolute value abs[y-r] is more than 4 tracks, then it is
determined that seeking is in progress, and operation advances to
step S6.
[0083] (S4) When determined to be following, the processor 8
replaces the eccentricity estimation gain L4, L5 of block 26 of the
prediction observer 21 with the design value. This eccentricity
estimation gain L4, L5 is designed such that the error (y[k]
-px1[k])converges to zero during following. Also, the processor 8
connects the switch 35 to the `b` switch.
[0084] (S5) The processor 8 calculates the state by the prediction
observer. In other words using the aforementioned equation (12), it
uses the prediction state px[k] from the previous sample and error
(y[k]-px1[k]) to calculate the prediction state px[k+1] (px1[k+1]
to px5[k+1]) for the next sample.
[0085] Next, using equation (14), it uses the prediction state
px[k] of the previous sample to calculate the next control current
u[k]. In addition, it outputs the control current u[k] to the plant
20 (amp 6). Also, it ends this servo interrupt.
[0086] (S6) When determined to be seeking, the processor 8 replaces
the eccentricity estimation gain L4, L5 of block 26 of the
prediction observer 21 with `0`. Also, the processor 8 connects the
switch to the `a` side. The state equation that is calculated in
step S8 becomes equation (15).
[0087] (S7) The absolute value abs [y-r] mentioned above, indicates
the number of remaining tracks. Also, depending on the size of the
absolute value abs[y-r], determines whether it is an acceleration,
constant velocity or deceleration interval. When it is an
acceleration interval, the target velocity for the acceleration
interval is generated by the velocity generator 31. Moreover, when
it is a constant velocity interval, the target velocity for the
constant velocity interval is generated by the velocity generator
31. Furthermore, when it is a deceleration interval, the target
velocity for the deceleration interval is generated by the velocity
generator 31.
[0088] (S8) The processor 8 performs state calculation with the
prediction observer. In other words, using equation (15), it uses
the prediction state px[k] of the previous sample and the error
(y[k]-px1[k]) to calculate the prediction state px[k+1] (px1[k+1]
to px5[k+1]) for the next sample. When doing this, as shown in
equation (15), the eccentricity estimation gain L4, L5 is zero.
[0089] Next, using equation (14), it uses the prediction state
px[k] for the previous sample to calculate the control current u[k]
as a state variable. When seeking, this control current is saved as
a state variable, and is not used as output. Furthermore, using
equation (17), it uses the velocity error and the prediction state
px[k] of the previous sample to calculated the control current
u[k]. When seeking, the control current u[k] that is calculated
using equation (17) is output to the plant 20.
[0090] In this way, as shown in FIG. 4, when seeing, the estimated
eccentricity gain L4. L5 of the observer 21 is made smaller than
the eccentricity estimation gain during following so it is possible
to calculate an estimated eccentricity signal that is not affected
by the error between the detected position and the estimated
position. Therefore, it is possible to correct the eccentricity
during seeking without being affected by the error.
[0091] It is preferable for the eccentricity estimation gains L4
and L5 of the observer 21 during seeking to set zero, however the
gains can be a value near zero.
[0092] FIG. 5 is a flowchart of the process of an example of the
observer in FIG. 2 that has been transformed. The change in the
calculation process of the prediction observer will be explained.
The equations (12) and (14) for calculating the state px[k+1] of
the next sample and control current u[k] become 5 dimensional.
Therefore, the amount of calculations increases, so the calculation
for estimating the actuator operation and the calculation for
estimating the external disturbance are separated.
[0093] In other words, equation (12) is separated into equations
(18) and (19) below. 13 ( PX 1 [ K + 1 ] PX 2 [ K + 1 ] ) = ( 1 T 0
1 ) ( PX 1 [ K ] PX 2 [ K ] ) + Kp ( T 2 / 2 T ) u ob [ k ] + ( L 1
L 2 ) ( y [ k ] - px 1 [ k ] ) , ( 18 ) ( PX 3 [ K + 1 ] PX 4 [ K +
1 ] PX 5 [ K + 1 ] ) = ( 1 0 0 0 cos ( 0 T ) - sin ( 0 T ) 0 sin (
0 T ) cos ( 0 T ) ) ( PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) + ( L 3 L
4 L 5 ) ( y [ k ] - px 1 [ k ] ) , ( 19 )
[0094] Equation (18) estimates the actuator operation, and equation
(19) estimates the external disturbance (bias, eccentricity).
[0095] Similarly, the control current (state) u(k) is separated
into the control signal uob from the calculation that estimates the
actuator operation, and the control current uw from the calculation
that estimates the external disturbance. Also, the control current
uvcm is obtained by adding the control current uob and control
current uw. In other words, equation (14) is transformed to
equation (20). 14 uob = - ( F 1 F 2 ) ( PX 1 [ K ] PX 2 [ K ] ) ,
uw = - ( 1 1 0 ) ( PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) uvcm = uob +
uw . ( 20 )
[0096] It must be noticed that the control current uob in equation
(18) is the control current uob from the calculation for estimating
the actuator operation in equation (20).
[0097] This will be explained using the flowchart in FIG. 5.
[0098] (S10) From equation (18), the state variables px1[k+1] and
px2[k+1] for the next sample are calculated.
[0099] (S11) Next, from equation (19), the state variables px3[k+1,
px4[k+1] and px5[k+1] for the next sample are calculated.
[0100] (S12) Furthermore, with equation (20), the control current
uob that estimates the actuator operation, and the control current
uw that estimates the disturbance are calculated. In addition, the
control current uvcm is obtained by adding the control current uob
with the control current uw.
[0101] When the calculation for estimating the disturbance is
separate in this way, it is possible to separate the calculation
for estimating the actuator operation and the calculation for
estimating the disturbance, and since there is one a maximum of
three equations, it is possible to reduce the number of adding
operations. This makes it possible to calculate the state at high
speed.
[0102] Next, an example of modifying the observer is explained. In
FIG. 2, a prediction observer is explained, however, using a
current observer is also possible. FIG. 6 is a block diagram of
another servo process executed by the processor in FIG. 1, and FIG.
7 is a flowchart of the calculation process of the current observer
in FIG. 6.
[0103] FIG. 2 shows a prediction observer, however FIG. 6 shows the
configuration of a current observer. The items in FIG. 6 that are
the same as those in FIG. 2 are indicated with the same number.
[0104] As is well known, when a prediction observer is defined by
equations (12) and (14), the state equations of a current observer
are defined by equations (21), (22) and (23) below. 15 ( PX 1 [ K ]
PX 2 [ K ] PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) = ( qX 1 [ K ] qX 2 [
K ] qX 3 [ K ] qX 4 [ K ] qX 5 [ K ] ) + ( L 1 ' L 2 ' L 3 ' L 4 '
L 5 ' ) ( y [ k ] - qX 1 [ k ] ) ( 21 ) uob = - ( F 1 F 2 ) ( PX 1
[ K ] PX 2 [ K ] ) , uw = - ( 1 1 0 ) ( PX 3 [ K ] PX 4 [ K ] PX 5
[ K ] ) , uvcm = uob + uw , ( 22 ) 16 ( qX 1 [ K + 1 ] qX 2 [ K + 1
] ) = ( 1 T 0 1 ) ( PX 1 [ K ] PX 2 [ K ] ) + Kp ( T 2 / 2 T ) uob
[ k ] ( qX 3 [ K + 1 ] qX 4 [ K + 1 ] qX 5 [ K + 1 ] ) = ( 1 0 0 0
cos ( 0 T ) - sin ( 0 T ) 0 sin ( 0 T ) cos ( 0 T ) ) ( PX 3 [ K ]
PX 4 [ K ] PX 5 [ K ] ) ( 23 )
[0105] Here, px[k] (px1[k] to px5[k]) are estimated values for the
corrected current sample, qx[k] (qx1[k] to qx5[k]) are estimated
values for the previous sample, and qx[k+1] (qx1[k+1]to qx5[k+1])
are estimated values for the next sample.
[0106] In other words, as shown in equation (21), the estimated
values px[k] (px1(k] to px5[k]) for the corrected current sample
are found from the error, and the estimated values qx[k] (qx1[k] to
qx5[k]) of the previous sample. The estimation gains `L1`' to `L5`'
in equation (21) differ from the estimation gains `L1` to `L5` of
the prediction observer.
[0107] Moreover, as shown in equation (22), the control current
uvcm is obtained from the estimated values px[k] (px1[k]to px5[k])
for the corrected current sample. Equation (22) separates the
equation for estimating the actuator and the equation for
estimating the disturbance, as shown in equation (20) described
above. Finally, as shown in equation (23), the estimated values
qx[k+1] (qx1[k+1] to qx5[k+1]) for the next sample are obtained
from the estimated values px[k] (px1[k] to px5[k]) for the
corrected current sample and the control current uob[k].
[0108] When this is shown in a block, it is as shown in FIG. 6. In
other words, in FIG. 6, a current observer (estimator) 36 estimates
the current state of the position, velocity, bias and eccentricity
from the previous state, and outputs a control current.
[0109] The position signal (servo signal) y[k] that is read by the
magnetic head 3 is output from the plant 20. In the observer 36,
the error calculator 22 calculates the error (y[k]-qx1[k]) between
the position signal y[k] and the estimated position py[k]
(=qx1[k]). The error is input to a fourth gain multiplier 26. The
gain multiplier 26 multiplies the error by the estimation gain L1
(L1' to L5') (see equation (21)).
[0110] An adder 36 adds the output of the gain multiplier 26 and
the previous estimated state qx[k], to obtain the estimated value
px[k] for the corrected current sample. A second multiplier 24
multiplies the control current u[k] by a coefficient B (see
equation (23)). A third gain multiplier 25 multiplies the state
signal px[k] for the current sample by a coefficient A (see
equation (23)). An adder 28 adds the output of the two adders 24
and 25. By doing this the state signal qx[k+1] of the next sample
of equation (23) is output.
[0111] This state signal qx[k+1] for the next sample, is delayed
one sample by a delay circuit 29 to obtain the state signal qx[k]
for the current sample. The state signal qx[k] for the previous
sample is multiplied by a feedback coefficient C by a fifth
multiplier 27. By doing this the estimated position qy[k] for the
current sample shown in equation (21) is obtained.
[0112] Furthermore, the state signal px[k] of the corrected current
sample is multiplied by a feedback coefficient F by a first
multiplier 23. In this way, the current control current u[k] shown
in equation (22) is obtained. This control current u[k] is then
supplied to the plant 20.
[0113] An error calculator 30 calculates the position error
.DELTA.x between the target position r and current position y[k].
The current position y[k] is given from the plant 20. A target
velocity generator 31 generates the target velocity v0 from the
position error .DELTA.x. A velocity difference calculator 32
calculates the velocity difference between the target velocity v0
and the estimated velocity px2[k] of the prediction observer
21.
[0114] A gain multiplier 33 multiplies the velocity difference by a
velocity gain CO. A compensator 34 adds the velocity difference,
the estimated bias current px3[k] of the current observer 36 and
the estimated amount of eccentricity px4[k] of the current observer
36, and then inverts it and outputs it as the control current u[k].
The control output u[k] during seeking is given by equation
(17).
[0115] A switch 35 selects the output of the compensator 34 during
seeking, and selects the output of the prediction observer 21
during following. These blocks (22 to 36) are realized by the
programs of the processor 8.
[0116] In this embodiment as well, the same process as shown in the
flowchart in FIG. 3 is performed. The eccentricity estimation gains
L4', L5' are set to `0` during seeking, and the eccentricity
estimation gains L4', L5' are set to a setting other than `0`
during following. Therefore, an explanation of this process will be
omitted.
[0117] The process of the observer control shown in FIG. 3 is as
shown in FIG. 7. This process is explained below.
[0118] (S20) As shown in equation (21) the estimated values px[k]
(px1[k] to px5[k]) of the corrected current sample is found from
the error(y[k]-qx1[k]), estimated values qx[k] (qx1[k]-qx5[k]) of
the previous sample and the estimation gains L1' to L5'.
[0119] (S21) Next, as shown in equation (22), the control current
uvcm is obtained from the estimated values px[k] (px1[k] to px5[k])
of the corrected current sample. Also, the control current uvcm is
output to the plant 20.
[0120] (S22) Finally, as shown in equation (23), the estimated
values qx[k+1] (qx1[k+1] to qx5[k+1]) of the next sample are
obtained from the estimated values px[k] (px1[k] to px5[k]) of the
corrected current sample and the control current uob[k].
[0121] In this example as well, as shown in FIG. 3, the
eccentricity estimation gains L4' to L5' are set to zero during
seeking. Equation (21) is transformed to equation (24) below. 17 (
PX 1 [ K ] PX 2 [ K ] PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) = ( qX 1 [
K ] qX 2 [ K ] qX 3 [ K ] qX 4 [ K ] qX 5 [ K ] ) + ( L 1 ' L 2 ' L
3 ' 0 0 ) ( y [ k ] - qX 1 [ k ] ) ( 24 )
[0122] The control current uvcm and the estimated values qx[k+1]
(qx1[k+1] to qx5[k+1]) of the next sample are calculated from
equation (22) and equation (23). In this way, the same effect as
the prediction observer is also possible with a current observer.
In addition, by constructing a current observer, realization by
processor processing becomes easy.
[0123] Next, an example of a modified current observer is
explained.
[0124] FIG. 8 is a flowchart of the process of a modified example
of the current observer in FIG. 6. The modification in the
calculation processes for the current observer will be explained.
In the example in FIG. 6, the state px[k+1] of the next sample is
calculated by equation (21). However, since the frequency of the
eccentricity is low, estimation of the eccentricity by the current
observer can be delayed one sample more than the other state
variables (position, velocity, bias). In other words, for the
eccentricity state, it is possible to estimate the state one sample
ahead.
[0125] Taking this into consideration, it is possible to estimate
the eccentricity by calculating the error (y[k]-qx1[k]) and the
eccentricity estimation gains L4' to L5' when calculating the
estimated variables for eccentricity values qx4[k+1] and qx5[k+1]
of the next sample.
[0126] In other words, in equation (21), finding the state
variables for eccentricity px4[k] and px5[k] of the current sample
is omitted. Therefore, equation (21) is transformed to equation
(25) below. 18 ( PX 1 [ K ] PX 2 [ K ] PX 3 [ K ] ) = ( qX 1 [ K ]
qX 2 [ K ] qX 3 [ K ] ) + ( L 1 ' L 2 ' L 3 ' ) ( y [ k ] - qX 1 [
k ] ) ( 25 )
[0127] Moreover, the state variables for eccentricity qx4[k] and
qx5[k] for the previous sample are used instead of the state
variables for eccentricity px4[k] and px5[k] of the current sample
to calculate the control current uw. Therefore, equation (22) is
transformed to equation (26) below. 19 uob = - ( F 1 F 2 ) ( PX 1 [
K ] PX 2 [ K ] ) , uw = - ( 1 1 0 ) ( PX 3 [ K ] PX 4 [ K ] PX 5 [
K ] ) , uvcm = uob + uw , ( 26 )
[0128] Furthermore, equation (27) below is applied without making
changes to the estimation equations for the estimated position and
estimated velocity in equation (23). However, in equation (23),
calculation of the error (y[k]-qx1[k]) and eccentricity estimation
gains L4' and L5, are added to the eccentricity estimation
equation, and becomes equation (28) below. 20 ( qX 1 [ K + 1 ] qX 2
[ K + 1 ] ) = ( 1 T 0 1 ) ( PX 1 [ K ] PX 2 [ K ] ) + Kp ( T 2 / 2
T ) uob [ k ] , ( 27 ) ( qX 3 [ K + 1 ] qX 4 [ K + 1 ] qX 5 [ K + 1
] ) = ( 1 0 0 0 cos ( 0 T ) - sin ( 0 T ) 0 sin ( 0 T ) cos ( 0 T )
) ( PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] ) + ( 0 L 4 ' L 5 ' ) ( y [ k
] - qX 1 [ k ] ) ( 28 )
[0129] In this way, since the eccentricity estimation values for
the previous sample are used in the calculation of the control
current, it is not necessary to calculate the eccentricity
estimation value for the current sample. Instead, the error is
reflected on the eccentricity estimation value for the next
sample.
[0130] By doing this, it is not necessary to calculate the state
variable for eccentricity before outputting the current to the
actuator. Also, it is possible to quicken the time of the current
output.
[0131] This process will be explained with the process flowchart in
FIG. 8.
[0132] (S30) The state variables px1[k], px2[k] and px3[k] of the
current sample are calculated from equation (25).
[0133] (S31) The control current uob that estimates the actuator
operation and the control current uw that estimates the disturbance
are calculated with equation (26). Also, the control current uvcm
is obtained by adding the control current uob and the control
current uw.
[0134] (S32) The state variables qx1[k+1] and qx2[k+1] for the next
sample are calculated with equation (27).
[0135] (S33) The state variables qx3[k+1], qx4[k+1] and qx5[k+1]
for the next sample are calculated with equation (28).
[0136] Here, in equation (26), similar to the embodiment shown in
FIG. 5, the control current (state) u[k] is separated into the
control current uob from the calculation for estimating the
actuator operation, and the control current uw from the calculation
for estimating the disturbance. The control current uvcm is
obtained by adding the control current uob and the control current
uw.
[0137] Moreover, as shown in equation (27) and equation (28), the
calculation for estimating the actuator operation (equation (27))
and the calculation for estimating the disturbance, including the
eccentricity, (equation (28)) are separated. By separating the
calculation for estimating the disturbance in this way, it is
possible to separate the calculation for estimating the actuator
operation and the calculation for estimating the disturbance, and
since the equation is at most 3-dimensional, it is possible to
reduce the number of summation operations. This makes it possible
to calculate the state at high speed.
[0138] The method of this embodiment becomes even more effective as
the frequency that is the object of eccentricity correction becomes
two or three. For example, when eccentricity correction is
performed for a cycle .omega.0 and also a cycle double that
2.omega. 0, the state variables of the eccentricity (x6, x7) are
added to (x4, x5).
[0139] In this case, the state variables of equation (28) can be
increased from (x3, x4, x5) to (x3, x4, x5, x6, x7). Therefore,
equation (28) is transformed to equation (29) below. 21 ( qX 3 [ K
+ 1 ] qX 4 [ K + 1 ] qX 5 [ K + 1 ] qX 6 [ K + 1 ] qX 7 [ K + 1 ] )
= ( 1 0 0 0 0 0 cos ( 0 T ) - sin ( 0 T ) 0 0 0 sin ( 0 T ) cos ( 0
T ) 0 0 0 cos ( 2 0 T ) - sin ( 2 0 T ) 0 0 0 sin ( 2 0 T ) cos ( 2
0 T ) ) ( PX 3 [ K ] PX 4 [ K ] PX 5 [ K ] PX 6 [ K ] PX 7 [ K ] )
+ ( 0 L 4 ' L 5 ' L 6 ' L 7 ' ) ( y [ k ] - qX 1 [ k ] ) ( 29 )
[0140] In this way, even when the estimation states of the
eccentricity are increased, the calculation performed before
calculating the control current is still that of equation (25), and
even when the estimation states of the eccentricity are increased,
it is possible to output the control current quickly.
[0141] Next, the operation for changing the head is explained. FIG.
9 is a flowchart of the process for changing the head in another
embodiment of the invention, FIG. 10 is a diagram explaining the
state variables in FIG. 9, and FIG. 11 is a diagram explaining the
head changing operation.
[0142] The waveform of the eccentricity differs for each surface of
the magnetic disk. Therefore, the wave form of the eccentricity
differs for each head. Also, the state variables of the
eccentricity differ for each head. When using the observer
described above, it is necessary to initialize the state
eccentricity variables x4, x5 of the observer when changing the
head. However, when the initial value is `0`, it takes time for the
estimated eccentricity value to follow the eccentricity.
[0143] Therefore, as shown in FIG. 10, the state variables x4, x5
for each head are stored in the memory (not shown in the figure) of
the processor 8 (see FIG. 1). Also, when there is an instruction to
change the head, the current state variables for the eccentricity
are stored in the memory. The state variables of the eccentricity
that are stored in memory for the head to be changed to are read,
and set in the observer. By doing this it is possible to
immediately initialize the observer.
[0144] This process is explained in detail using FIG. 9.
[0145] (S40) The processor 8 obtains the current state variables
(x4, x5) for the eccentricity of the observers 21, 36 according to
an instruction from the host to change the magnetic head.
[0146] (S41) The processor 8 converts the state variables (x4, x5)
for the eccentricity of this current sector number N1 to state
variables (x40, x50) for the eccentricity of sector number 0.
Conversion is performed by equation (30) below. 22 ( PX 4 ' 0 PX 5
' 0 ) = ( cos ( - N 1 0 T ) - sin ( - N 1 0 T ) sin ( - N 1 0 T )
cos ( - N 1 0 T ) ) ( PX 4 [ K ] PX 5 [ K ] ) ( 30 )
[0147] In other words, as shown in FIG. 11, there are a plurality
of sectors ST0 to STn around the magnetic disk 2. The current state
variables are for the sector (sector no. N1) where the magnetic
head is currently located, so the current state variables are
converted to the state variables for the reference sector (in this
case, sector No. 0).
[0148] (S42) The processor 8 stores the converted state variables
(x40, x50) for the eccentricity in the memory area that corresponds
to the current head number.
[0149] (S43) The processor 8 reads from memory the state variables
(x40, x50) for the eccentricity of the head number of the head to
be changed to.
[0150] (S44) The state variables that are read from memory are
values for the reference sector (sector No. 0) so the processor 8
uses equation (31) below to convert them to state variables (px4,
px5) for the eccentricity that correspond to the current sector
(sector No. 2). 23 ( PX 4 [ K ] PX 5 [ K ] ) = ( cos ( N 2 0 T ) -
sin ( N 2 0 T ) sin ( N 2 0 T ) cos ( N 2 0 T ) ) ( PX 4 0 PX 4 0 )
( 31 )
[0151] (S45) The processor 8 sets these state variables (px4, px5)
for the eccentricity as the state variables for the observers 21,
36.
[0152] By doing this, it is possible to immediately follow the
eccentricity even when changing the head. When the state variables
are not converted to the values of the reference sector position,
the sector number is further stored in memory.
[0153] The amplitude of the eccentricity correction current is
calculated from the state variables (px4, px5) that are stored in
memory using the equation square (px4.sup.2+px5.sup.2). This
corresponds with the amount of eccentricity. These state variables,
the amplitude or the amount of eccentricity is such that they can
be read from a host such as a computer.
[0154] Even though it is possible to correct the eccentricity, it
is not possible to correct infinitely large eccentricity.
Therefore, the host sends a warning to the user when the
eccentricity is large. Particularly, in the case of a disk device
that is installed in a portable computer and that is susceptible to
impact, a warning is issued when the eccentricity data exceeds a
pre-determined allowable limit, and saves the data on a separate
disk.
[0155] In addition to the embodiments of the invention described
above, the invention can be changed as follows:
[0156] (1) A magnetic disk device was explained as the disk storage
device, however the invention can also be applied to other disk
storage devices such as a magneto-optical disk device or optical
disk device.
[0157] (2) The observer was shown by processor processing, however
it can also be constructed by a digital circuit.
[0158] The preferred embodiments of the present invention have been
explained, however the invention is not limited to these
embodiments and can be embodied in various forms within the scope
of the present invention.
Industrial Application
[0159] As described above, this invention has the following
effect.
[0160] (1) The eccentricity estimation gains L4, L5 of the
eccentricity estimation observer during seeking are made extremely
small when compared with those during following to prevent the
position error from affecting the estimated amount of eccentricity.
Therefore, it is possible to speed up the convergence operation at
the end of seeking.
[0161] (2) Moreover, since eccentricity correction is performed
during seeking, there is no loss instability of the seek
operation.
* * * * *