US 20010003497 A1 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. Claims(20) 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 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 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 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 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 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 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 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 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 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 13. The control device for a disk storage device of claim 11 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 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 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 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 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 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 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 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 [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. [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.
[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.
[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 ω0 is taken to be the eccentric angular velocity, then the state equations are given by equations (5) and (6) below.
[0012] Here, when x4=cos(ω0·t) and x5=sin(ω0·t), then sx4 =−ω0·sin(w0·t) and sx [0013] In equation (5) the eccentricity is estimated by the sinusoidal transfer function (1/(S [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.
[0015] Here, T is the sample period. As shown in equation (11), the coefficients in equations (9) and (10) are A, B and C.
[0016] Here, the observer is expressed by the equations (12), (13) and (14) below.
[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 [0020] The position signal (servo signal) y[k] that is read by the magnetic head is output from the plant [0021] The second gain multiplier [0022] This state signal px[k+1] for the next sample is delayed one sample by an delay circuit [0023] Furthermore, the state signal px[k] is multiplied by the feedback coefficient F by a first multiplier [0024] In this way, the observer [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. [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.
[0035] In this equation (15), by calculating the state variables X4 and X5 for eccentricity, the equation (16) below is obtained.
[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), ω0T 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. [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. [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 [0067] A voice coil motor (VCM) [0068] The control circuit (called the processor below) [0069] A read/write circuit [0070] A hard disk controller [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 [0072] The position signal (servo signal) y[k] that is read by the magnetic head [0073] A second gain multiplier [0074] This state signal px[k+1] for the next sample is delayed one sample by the delay circuit [0075] Furthermore, the state signal px[k] is multiplied by a feedback coefficient F by a first multiplier [0076] An error calculator [0077] A gain multiplier u[k]=−Co (Vo−PX [0078] A switch [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 [0081] (S2) The processor [0082] (S3) The processor [0083] (S4) When determined to be following, the processor [0084] (S5) The processor [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 [0086] (S6) When determined to be seeking, the processor [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 [0088] (S8) The processor [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 [0090] In this way, as shown in FIG. 4, when seeing, the estimated eccentricity gain L4. L5 of the observer [0091] It is preferable for the eccentricity estimation gains L4 and L5 of the observer [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.
[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).
[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.
[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 L [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) [0109] The position signal (servo signal) y[k] that is read by the magnetic head [0110] An adder [0111] This state signal qx[k+1] for the next sample, is delayed one sample by a delay circuit [0112] Furthermore, the state signal px[k] of the corrected current sample is multiplied by a feedback coefficient F by a first multiplier [0113] An error calculator [0114] A gain multiplier [0115] A switch [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 [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.
[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.
[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.
[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.
[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 ω0 and also a cycle double that 2ω 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.
[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 [0144] This process is explained in detail using FIG. 9. [0145] (S40) The processor [0146] (S41) The processor [0147] In other words, as shown in FIG. 11, there are a plurality of sectors ST [0148] (S42) The processor [0149] (S43) The processor [0150] (S44) The state variables that are read from memory are values for the reference sector (sector No. 0) so the processor [0151] (S45) The processor [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 [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. [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. Referenced by
Classifications
Legal Events
Rotate |