JP2639170B2 - Learning control method for fixed magnetic disk drive - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial applications]

The present invention relates to a so-called learning control method in which a fixed magnetic disk drive (also referred to as a hard disk drive, also abbreviated as HDD) uses a characteristic parameter (called a learning parameter) taken from a positioning mechanism in the fixed magnetic disk drive to optimize a control parameter. The present invention relates to a learning control method for a fixed magnetic disk drive that enables acquisition of learning parameters necessary for performing stable control without using a simple control device and without always increasing startup time. .

[Prior art]

FIG. 1 is a block diagram showing a main configuration of an HDD control circuit according to one embodiment of the present invention. Hereinafter, the prior art will be described with reference to FIG. Conventionally, 21 in FIG.
No. to 23 means are not provided. In FIG. 1, reference numeral 1 denotes a read / write head for reading / writing a disk (not shown) (simply referred to as a head, and read / write is also referred to as R / W). 2 denotes a head for increasing the read / write signal of the head 1. The amplifier 3 is a pulse peak detector that binarizes the read signal 2R of the disk output from the head amplifier 2. Reference numeral 4 denotes a data separator. The separator 4 separates clock and data from data read from the disk from the pulse peak detector 3, sends the data to a host (not shown) via the host interface 5, and connects the host interface 5 to the host. For example, code conversion of the disk write data received from the host via the host is performed and the converted data is given to the head amplifier 2. Reference numeral 6 denotes a servo signal amplifier for amplifying the read signal 2R in order to extract a servo signal from the disk read signal 2R from the head amplifier 2. Reference numeral 10 denotes a timing control unit that extracts and outputs a timing signal from the amplified output signal of the servo signal amplifier 6, and 7 inputs the timing signal from the timing control unit 10 and the amplified output signal of the servo signal amplifier 6. A servo peak detector 8 detects the peak value of the servo signal, and 8 converts the peak value detected by the servo peak detector 7 into a digital value servo signal 8a in accordance with the timing signal from the timing controller 10. A D / A converter 9 for performing arithmetic control for positioning control of the R / W head 1, a D / A converter 11 for converting the control output data DiM of the arithmetic control unit 9 into an analog value, Reference numeral 12 denotes a VCM drive amplifier which amplifies the output signal of the D / A converter 11 and supplies the control output current (also referred to as VCM current) iM to a boil coil motor (abbreviated as VCM) 13 for driving the R / W head 1 In . In the control circuit shown in FIG. 1, when an R / W command is input from the host via the host interface 5 via the host interface 5, the arithmetic control unit 9 receives the servo signal 8a on the disk via the means 1, 2, 6, 7, 8, and 10.
Is read, and the VCM 13 is driven through the aforementioned means 11 and 12 to read the R / W
It performs seek control for causing the head 1 to reach the target track and tracking control for holding the head 1 on the target track. When the target track is reached, the read signal from the disk is read by the host via the means 1 to 5, and
The write data from the host is written to the disk via the means 5, 4, 2, 1. By the way, the HDD conventionally controls the VCM 13 and the like with fixed control parameters, and the variation (learning parameter) of each mechanism has been adjusted by a variable resistor or the like on a printed board of the control device. In this control, only a time of about 3 to 4 seconds until the rotation of the disk drive motor (spindle motor) stabilizes is required as a waiting time for the power-on time. However, recent control devices have a configuration that eliminates the adjustment unit based on software servos (there is no printed circuit board with variable resistance, etc.) in order to improve control and reduce costs. The control output is adjusted by software using the result. There are various learning parameters of such a positioning mechanism that affect seek / tracking control, but the typical one is that a flexible printed board for the wiring and a bearing for the support are added to the VCM13. Variation in the applied force (also referred to as external force or torque offset), and variation in the torque offset depending on the position of the VCM, and therefore whether the track position of the head 1 is on the inner or outer circumference side, and the servo signal from the head amplifier 2 Variations in the read output 8a (variations in the performance of the head 1 and in the disk). FIG. 4 is a characteristic diagram showing a relationship example between a cylinder number (horizontal axis) as a track position of the head 1 and the torque offset. Note that this torque offset can be grasped as a value of the control output current iM in a steady state where the head 1 is tracking on the track, and thus a value of the control output data DiM. 5 is a characteristic diagram showing an example of the relationship between the cylinder number (horizontal axis) and the voltage value (vertical axis) of the servo signal 8a read from the head 1. The data shown in FIGS. 4 and 5 are acquired for all cylinders and all heads on the disk, and the goal is to obtain the performance of performing stable seek and tracking on the seek target track in a short time. It is expected that it takes time on the order of minutes to capture data over all tracks as shown in FIG. Therefore, as an example of the required time
To start up the HDD within 12 seconds, it is necessary to greatly reduce the time. In reality, 3 to 4 seconds required to start up the spindle motor are subtracted from 12 seconds, and only 8 to 9 seconds, which is more than 6 to 7 seconds with a margin, are used to obtain the characteristic parameters. Is being implemented.

[Problems to be solved by the invention]

However, as described above, since only the limited learning parameter data is fetched due to the limitation of the drive start-up time, there are insufficient aspects in the data amount, data fetch timing, and the like, and the parameters are not always optimal. It is also possible. For example, see Fig. 4 and 5
This is because, in order to compensate for the shortage of measurement points for the characteristics as shown in the figure, there is a large deviation from the straight line in the portion where the characteristics are approximated and controlled by a straight line, and a problem such as an increase in seek operation time has occurred. . In addition, the system in mind of the present invention is a case where the control of the drive control device is performed by one CPU. In particular, in this method, when the power is turned on, the CPU devotes itself to the drive startup and parameter acquisition processing, so the above processing must be completed within 12 seconds to respond to the command from the host that will come after 12 seconds. Must be completed and a system in place to respond to requests from the host must be taken. Most current HDDs have two CPUs, one for drive control and the other for host control. In these drives, even when the drive is starting up, the host communication dedicated CPU sends a signal to the host as if the start-up was completed. The output can be used to time the communication with the host. That is, the interface LSI immediately after startup
Since a request (command) that does not need to move the drive is sent out, this process is processed by the host-compatible CPU. However, this is not possible with one CPU. From such a background, in order to read parameters within a limited time, the absolute number of readable parameters and the measurement accuracy are insufficient. This is considered to be the main cause of variations in seek time and increase in seek time.
There is a demand for improvements in the quality and quantity of the parameters that are taken. Accordingly, the present invention provides a learning control method for a fixed magnetic disk drive capable of solving the above-mentioned problems. Specifically, (1) it takes time on the order of minutes to capture the optimal parameters for HDD control. Describes a method of realizing this in a state in which normal operation can be performed beforehand and realizing a normal power-on in a few seconds. (2) The second invention checks the performance of an HDD which should be controlled by optimal parameters. It is an object of the present invention to provide a method and a method of realizing backup when the performance is poor.

[Means for Solving the Problems]

In order to solve the above-mentioned problem, the method of the first invention is as follows.
"In a learning control method in which a control device of a fixed magnetic disk device (such as an arithmetic control unit 9) uses learning parameter data taken from a positioning mechanism (such as a VCM13 or a R / W head 1) to optimize a control parameter, An input switch for instructing the reading of the learning parameter is provided, and only when the input switch is turned on in advance and the power is turned on, the learning parameter of the positioning mechanism is read. At the predetermined position. On the other hand, when there is no instruction, the learning parameter data written on the disk is read out, and the control parameter is set to an optimum value using the learning parameter data. "
The method according to the second aspect of the invention is based on a learning method of reading various learning parameters affecting the seek tracking control of a fixed magnetic disk drive from a positioning mechanism into a control device and optimizing the control parameters. In the control method, data indicating a relationship between a seek amount in a normal read / write operation and an allowable seek time for the seek amount is set in advance (by a table TB1 or the like). The required seek time is measured for each seek operation and compared with the permissible seek time. If the former time exceeds the latter time,
(Via the learning parameter read / write means 22).

[Operation]

1) Regarding the first invention: Only when the learning command input switch 21 is turned on in advance and the power is turned on, the learning parameters required for stable control are taken over a minute-order time and taken in. Data is written to a predetermined position on the disk, and during normal start-up, the switch 21 is kept off so that the host can be called in a short time by simply reading out the write data. Things. 2) Regarding the second invention: It is always monitored whether or not the seek time is within the allowable time, and when the seek time exceeds the allowable time, learning parameters are taken.

【Example】

Hereinafter, an embodiment of the present invention will be described with reference to FIGS. In FIG. 1, it is assumed that a learning command input switch 21 for inputting a command to input a learning parameter to the arithmetic control unit 9 is added to the arithmetic control unit 9 and that a part of the function is shared in the arithmetic control unit 9. A learning parameter read / write unit 22 and a seek time measurement / comparison unit 23 are provided as partial function units considered. 1) Regarding the first invention: As a learning function of the HDD, individual data is obtained for a part of the mechanical unit and the electric circuit unit where the variation occurs, and control is performed according to the taken data.
One of the main learning parameter data is torque offset data as shown in FIG. By incorporating such a torque offset in advance and incorporating it in the control output, it is possible to realize optimal control for reducing the seek time. Another main learning parameter data is a servo signal (reception voltage) 8a as shown by a solid curve in FIG. The servo signal 8a is input to an arithmetic and control unit 9 via a servo signal amplifier 6 as a signal amplifier different from the data part to implement head positioning control. It varies depending on the variation of the read head and whether the head is on the outer or inner circumference on the disk surface.
As can be seen from FIG. 5, for example, these two fluctuation factors usually cause a three-fold fluctuation. Automatic gain control control for the servo signal is performed to adjust (equalize) the level of the servo signal 8a.
This is not a closed-loop system that adjusts the gain by looking at the captured signal. Instead, an amplifier that amplifies the servo signal by checking the gain signal amplitude on the inner and outer circumferences of each head and media (disk) in advance. A method of setting gain is adopted. The reason is that if automatic gain control of a normal closed-loop integration method is adopted for adjusting the level of the servo signal, the integration for detecting the reception signal level based on the presence / absence of read / write data on the disk and the data pattern of the servo signal. The voltage value charged to the capacitor for the power supply fluctuates, and an error occurs in the adjustment of the automatic gain control. If this error is to be eliminated, adjustment and switching of the integration time constant are required, and the control becomes complicated. Therefore, in order to perform stable control over the entire area of the disk, the amplitude of the servo signal 8a must be changed from the outside of the disk to the gain data of the automatic gain control amplifier.
It is necessary to take in the inside. Then, the gain of the amplifier is determined according to the amplitude of the fetched servo signal data, and the gain of the amplifier is set for each location where the head is on-track, and the read level of the servo signal is set regardless of the position of the head. Can be kept constant. The dashed curve in FIG. 5 shows the required amplifier gain characteristic corresponding to the servo signal 8a indicated by the solid curve in FIG. By performing such a gain setting, correct head positioning control can be realized. However, if the detailed data shown in Fig. 4 and Fig. 5 (solid curve) is taken in when the power is turned on, it takes time to process the data, and the host must be started within 12 seconds, which is the allowable waiting time at the time of startup. Becomes difficult. For this reason, in the first invention, a learning data (learning parameter) capturing mode is created separately from the normal power-on by the procedure shown in FIG. Data is written in a portion not used for writing, for example, inside the innermost circumference of the read / write track or outside the 0 track (0 cylinder). This may take several minutes. Once the learning data is written in this mode, the learning data can be obtained only by reading this data at the next normal power-on, so that the learning data can be obtained.
Startup can be realized in a short time. This mode switching is the first
This is performed using the learning command input switch 21 shown in FIG. Referring now to FIG. In the following, reference numerals S1 to S8 indicate steps in FIG. That is, when the power is turned on, the arithmetic control unit 9 performs an initial process (S1), and determines whether or not the mode is the learning mode by looking at the learning command input switch 21 (S2). If the mode is the learning mode (branch Y), While allowing the head 1 to perform sufficiently stable tracking from the inner circumference of the disk to the tracks on the outer circumference, for example, 16 cylinders at a time, the torque offset data as shown in FIG. 4 is taken in each time (S5), and then the solid line curve in FIG. The servo signal read voltage 8a as described above is taken in (S6). Then, the fetched torque offset data and servo signal data 8a (accordingly, servo gain data) are written to predetermined positions on the disk as described above (S7, S7).
8), end the read / write mode of this learning parameter.
Steps S5 to S8 correspond to the operation of the learning parameter read / write means 22 in FIG. If the learning mode is not set in step S2 (branch N, that is, when the power is normally turned on), the learning data (learning parameters) written in steps S7 and S8 is read from the disk, and a new The control parameters are determined and set (S3). Next, RAM clear, I / O
Performs drive initial processing such as data set (S
4) This completes drive startup. In step S8, the learning data write operation is not once terminated as described above, but the process proceeds to step S3 as indicated by the broken line, the learning data is read, and then the normal operation is performed. Can also be taken. 2) Regarding the second invention: The learning data (learning parameters) taken in for the control parameters may deviate from the optimum values as the temperature of the mechanism rises or other ambient environment changes. In the second invention, "deviation" from this optimum value
Is detected, and when the “displacement” is equal to or more than a predetermined value, the learning parameter is again fetched. FIG. 6 shows an example of the relationship between the cylinder number (horizontal axis) and the seek time (vertical axis). As shown in the figure, the standard seek time generally becomes like this solid curve. In the control circuit, an allowable seek time as shown by a broken line in the figure, which allows for some variation with respect to the solid line curve, is provided as data in the control circuit, and output for each read / write command as shown in the procedure of FIG. If the former time exceeds the latter time by comparing the required time for executing the given seek command with the allowable seek time in the control circuit, it is determined that all the captured parameters have deviated from the optimum values. Then, when the command being executed is completed, the learning parameters are retaken. Here, FIG. 3 will be specifically described. The following S11 to S19
Reference numerals indicate steps in FIG. That is, when a seek command is given from the host, the arithmetic and control unit 9 sets an allowable timer value according to the seek amount with reference to a preset table such as TB1 in FIG. The timer is started (S12). When the seek is completed (S13), it is determined whether or not the timer has timed out (S14). If the timer has not timed out (branch N), and if it has timed out (branch Y), the learning is performed. After setting a flag to re-acquire parameters (S15), read / write processing is performed (S15).
16). Next, it is determined whether or not the re-acquisition flag is set (S17). If it is not set (branch N), and if it is set (branch Y), the learning parameter is reset. After fetching (S18), and then clearing the re-fetching flag (S19), this process ends. Steps S11 to S14 correspond to the operation of the seek time measurement comparing means 23 in FIG.

【The invention's effect】

According to the first aspect of the present invention, there is provided a learning control method in which the arithmetic and control unit 9 of the fixed magnetic disk device uses the learning parameter data taken from the VCM 13 and the R / W head 1 as a positioning mechanism to make the control parameter an optimum value. A learning input switch 21 for instructing reading of the learning parameter is provided, and only when the input switch 21 is turned on in advance and the power is turned on, the learning parameter of the positioning mechanism is read and read. When the learning parameter is written to a predetermined position on the disk, and the power is turned on while the input switch is off,
Since the learning parameters written to the disk are read out and the control parameters are optimized using the learning parameters, it is possible to shorten the normal drive start-up time when the learning command input switch 21 is not turned on. it can. If the drive becomes defective, it can be expected that the cause of the failure can be easily found by reading the learning data via the host. According to the second aspect of the present invention, the arithmetic control unit 9 presets data indicating a relationship between a seek amount in a normal read / write operation and an allowable seek time for the seek amount in the table TB1, The seek time required for each seek operation is measured via the seek time measurement comparing means 23 and compared with the allowable seek time.If the former time exceeds the latter time, the learning is performed via the learning parameter read / write means 22. Since parameter data is taken in, stable control can be maintained at all times.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an essential configuration of a control circuit as an embodiment of the present invention, FIG. 2 is a flowchart showing an embodiment of a processing procedure of the first invention, and FIG. 3 is a process of the second invention. FIG. 4 is a characteristic diagram showing an example of torque offset data, FIG. 5 is a characteristic diagram showing an example of a relationship between a read amplitude voltage of a servo signal and a necessary amplifier gain, and FIG. FIG. 4 is a characteristic diagram illustrating an example of a relationship between a seek amount and a seek time. 1: R / W head, 2: Head amplifier, 3: Pulse peak detector, 4: Data separator, 5: Host interface, 6: Servo signal amplifier, 7: Servo peak detector, 8: A / D Converter, 8a: Servo signal, 9: Operation control unit, 1
0: timing control unit, 11: D / A converter, 12: VCM drive amplifier, 13: VCM, 21: learning command input switch, 22: learning parameter read / write means, 23: seek time measurement comparison means, TB1:
table.

Claims (2)

(57) [Claims] 1. A learning control method for optimizing a control parameter by reading various learning parameters affecting a seek tracking control of a fixed magnetic disk drive from a positioning mechanism to a control device, and instructing reading of the learning parameter. An input switch is provided, and only when the input switch is turned on in advance and the power is turned on, the learning parameter of the positioning mechanism is read, and the read learning parameter is written to a predetermined position on the disk. When the power is turned on with the input switch turned off, a learning parameter written to the disk is read out, and a control parameter is optimized using the learning parameter. Learning control method. 2. A learning control method for optimizing control parameters by reading various learning parameters affecting a seek tracking control of a fixed magnetic disk drive from a positioning mechanism into a control device. The data indicating the relationship between the amount and the seek time allowable for the seek amount is set in advance, and the seek time required for each seek operation is measured and compared with the allowable seek time. A learning control method for a fixed magnetic disk drive, characterized in that the learning parameters are fetched when the time has passed.