Vertical track zoning for disk drives

[0001] Priority is claimed from U.S. patent application Ser. No. 10/053,220 filed Jan. 17, 2002, which is incorporated by reference herein in its entirety.

[0002] The present invention relates generally to the storage of information on storage media, and more particularly to storage of information on rotating magnetic media such as disks in a disk drive.

[0003] Data storage devices such as disk drives are used in many data processing systems for data storage. Typically a disk drive includes a magnetic data disk having recording surfaces with concentric data tracks, and a transducer head paired with each recording surface, for writing data to, and reading data from, the data tracks. Each paired magnetic head and media surface couples to provide a unique data recording capability which depends on the fly height of the head from the recording surface, the quality/distribution of magnetic media on the recording surface, and the magnetic properties of the magnetic head.

[0004] Conventional methods of recording data using the paired head and recording surface are inefficient because they do not take into consideration the differences in data recording capabilities between one pair of head and recording surface, and another head and surface pair. Though the heads are designed to perform identically in read/write operations, in practice different heads in a disk drive can have different read/write performance capabilities. Lower performing heads cannot read/write data as that of other heads in the disk drive. Typically, a single error rate level and a single storage capacity level are used to record data for all the pair heads and surfaces. This results in inefficient data storage for those pairs of heads and surfaces that can store more data. It also lowers the qualification yields of the disk drives because one or more pairs of heads and surfaces do not record data at the qualifying error rate and capacity levels.

[0005] Further, in high data rate design of disk drives, as the recording density (i.e. bits-per-inch and/or tracks-per-inch) is increased, maintaining transducer head tolerances has become a challenge. Variance in the relative head performance distribution increases with increasing data density. In conventional disk drives, the drive yield and capacity suffers as a result of head performance variations in disk drives.

[0006] One method of increasing the data storage capacity of a disk drive includes increasing the areal density of the data stored on the media surfaces (bits/sq. in.BPSI). Areal density is the track density which is the number of tracks per radial inch (TPI) that can be packed onto the media/recording surface, multiplied by the linear density (BPI) which is the number of bits of data that can be stored per linear inch.

[0007] Conventional processes for qualifying disk drives scrap a disk drive when the measured disk capacity of the disk drive is less than a target disk capacity. Conventionally, each recording surface is formatted to store the same amount of data as every other recording surface. Thus, a recording surface that has a low error rate is formatted to the same TPI and BPI levels, as a recording surface having a high error rate, even though it can store more data. However, by adopting a single TPI and BPI level for every recording surface, conventional processes fail to account for the differences in sensitivity and accuracy of the paired head and recording surface, which results in less data storage and more waste of space on each recording surface. This also results in lower overall yields of disk drives because if even a few of the recording surfaces do not meet their targeted capacity, the sum of the surface capacities of all the media surfaces will be less than the target capacity, causing the entire disk drive to fail.

[0008] Some conventional disk drives utilize Variable Bits Per Inch to optimize utilization of the linear density capabilities of the heads. However, with increasing TPI, it is difficult to control tolerance of the head width relative to the shrinking track pitch. As a result, either head yield and/or drive yield suffer.

[0009] There is, therefore, a need for a method of storing data in a disk drive which improves disk drive yield while meeting the desired target drive capacity or increasing the drive capacity while meeting a desired drive yield by taking advantage of the head performance variation.

[0010] The present invention utilizes Vertical Zoning to improve the yield/performance of storage devices such as disk drives by optimizing the TPI and optionally BPI of each head/media pair in the storage device. In one embodiment, the present invention provides a method of implementing Vertical Zoning which applies to disk drives with multiple heads. For single head disk drives, the same method of Vertical Zoning can be used to trade off TPI against BPI to improve drive yield and performance.

[0011] In one version, a method of defining storage format in a data storage device having a plurality of storage media and a plurality of corresponding data transducer heads is provided, wherein each transducer is head for recording on and playback of information from a corresponding storage medium. A storage format is defined in at least one region on each storage medium, wherein each region includes a plurality of concentric tracks for recording on and playback of information. The method includes the steps of: moving each storage medium with respect to the corresponding transducer head and reading data from each storage medium with the corresponding transducer head; measuring a record/playback performance capability of each transducer head; and selecting a group of track densities, one track density for each region on a storage medium, based on the measured record/playback performance capability of the corresponding transducer head.

[0012] In another version, the TPI density is optimized across portions of a single media surface. A TPI is selected and data is recorded on a portion of the media surface at the selected TPI. The level of track density (TPI) can be one of fixed number of preselected levels or can be derived from an algorithm that is based on the location of a portion of the media surface. Thereafter, the recorded data is read and an error rate of the recorded data is measured. The measured error rate is compared to an acceptable error rate, and if the measured error rate is greater than the maximum acceptable error rate, the previous steps are repeated for another track density value, for example, the originally selected value less a decrement. This process continues until the measured error rate is less than or equal to the acceptable error rate, to provide a maximum recordable track density of data for a particular portion of the media surface.

[0013] Yet in another version, the present invention provides a data storage device having a plurality of storage media and a plurality of corresponding data transducer heads, each transducer head for recording on and playback of information from a corresponding storage media. A storage format is defined in one or more regions on each storage media, wherein each region includes a plurality of concentric tracks for recording on and playback of information, by steps including: measuring a record/playback performance capability of each transducer head; and selecting a group of track densities, one track density for each region on each storage media, based on the measured record/playback performance capability of the corresponding transducer head; wherein said multiple regions on each storage media are arranged as concentric regions, each region having an inner and an outer boundary at different radial locations on the storage media, such that each storage media includes the same number of concentric regions as other storage media in that data storage device, wherein the boundaries of radially similarly situated regions on all the storage media in that data storage device are essentially at the same radial locations.

[0014] These and other features, aspects and advantages of the present invention will become understood with reference to the following description, appended claims and accompanying figures where:

[0015] FIG. 1 shows an example partial schematic diagram of a disk drive with an example data storage format according to the present invention;

[0016] FIG. 2 shows another example schematic of the disk drive of FIG. 1 illustrating disk drive electronics;

[0017] FIG. 3 shows an example surface format for data storage according to the present invention;

[0018] FIG. 4 shows an example flow chart of an embodiment of steps of defining a data storage surface format according to the present invention;

[0019] FIG. 5 shows an example flow char of an embodiment of determining storage capacity according to the present invention;

[0020] FIG. 6 shows a conventional data storage format;

[0021] FIG. 7 shows an example layout of data storage format for a disk drive with multiple heads according to the present invention

[0022] FIG. 8 shows an example storage format including capacity zones;

[0023] FIG. 9 shows an example of variable servo track with variable data track data storage format layout;

[0024] FIG. 10 shows an expanded view for an example data storage layout including fixed servo track pitch with variable data track pitch for a zone on a disk surface;

[0025] FIG. 11 shows an example data storage layout including fixed servo track vs. variable data track for 2-head disk drives;

[0026] FIG. 12 shows a conventional logical cylinder format LBA access model for 4-head disk drives;

[0027] FIG. 13 shows an example virtual cylinder data storage surface format LBA access model for 4-head disk drives according to the present invention;

[0028] FIG. 14 shows another example block diagram depicting logical block address accessing scheme according to an example surface format according to the present invention for a two head disk drive;

[0029] FIG. 15 shows another example block diagram depicting logical block address accessing scheme according to an example surface format according to the present invention for a four head disk drive;

[0030] FIG. 16 shows an example capacity zone for storage surface in disk drives according to the present invention;

[0031] FIG. 17 shows an example flowchart of embodiment of steps of optimizing recording density per zone; and

[0032] FIG. 18 shows an example flowchart of embodiment of steps of determining head performance.

[0033] Data storage devices used to store data for computer systems include, for example, hard disk drives, floppy disk drives, tape drives, optical and magneto-optical drives, and compact disk drives. Although the present invention is illustrated by way of an exemplary magnetic hard disk drive, the present invention can be used in other storage media and drives, including non-magnetic storage media, as apparent to one of ordinary skill in the art and without deviating from the scope of the present invention.

[0034] Referring to FIGS. 1-2, an exemplary hard disk drive 100 is diagrammatically depicted for storing user data and/or operating instructions for a computer system 54. The hard disk drive 100 comprises an electromechanical head-disk assembly 10 as including one or more rotating data storage disks 12 mounted in a stacked, spaced-apart relationship upon a rotating spindle 13. The spindle 13 is rotated by a spindle motor 14 at a predetermined angular velocity.

[0035] Each disk 12 defines at least one media surface 23, and usually two media surfaces 23 on opposing side of each disk 12. Each media surface 23 is coated with magnetic or other media for recording data. The spindle drive motor 14 turns the spindle 13 in order to move/rotate the disks 12 past magnetic transducer heads 16 suspended by suspension arms 17 over each media surface 23. Generally, each magnetic head 16 is attached to the suspension arm 17 by a head gimbal assembly (not shown) that enables magnetic head 16 to swivel to conform to the media surfaces on the disks 12. The suspension arms 17 extend radially from a rotary voice coil actuator (not shown). An actuator motor 20 rotates the actuator and head arms and thereby positions the magnetic heads 16 over the appropriate areas of the media surfaces 23 in order to locate and read or write data from or to the storage surfaces 23. Because the disks 12 rotate at relatively high speed, the magnetic heads 16 ride over the media surface 23 on a cushion of air (air bearing). Each magnetic head 16 comprises a read element (not shown) for reading magnetic data on magnetic storage media surfaces 23 and a write element (not shown) for writing data on the media surfaces 23. Most preferably, although not necessarily, the write element is inductive and has an electrical writing width which is wider than an electrical reading width of the read element, which is preferably of magnetoresistive or giant magnetoresistive material.

[0036] Referring to FIG. 3, each media surface 23 is divided into a plurality of concentric circular tracks 30 that each have individually addressable portions 35, such as sectors, in which data is stored in the form of magnetic bits. The data sectors 35 are separated by embedded narrow servo sectors or spokes 25 which include a series of phase-coherent digital fields followed by a series of constant frequency servo bursts. The servo bursts are radially offset and circumferrentially sequential, and are provided in sufficient numbers such that fractional amplitude signals picked up by the read element from portions of at least two bursts passing under the read element enable the controller 57 (FIG. 2) to determine and maintain proper head position relative to a data track 30. One example of a servo burst pattern for use with an inductive write element/magneto-resistive read element head 16 is provided by commonly assigned U.S. Pat. No. 5,587,850, entitled: Data Track Pattern Including Embedded Servo Sectors for Magneto-Resistive Read/Inductive Write Head Structure for a Disk Drive, incorporated herein by reference.

[0037] The drive controller 57 controls operation of the pairs of magnetic heads 16 and media surfaces 23 to read and write data onto each media surface 23. The drive controller 57 preferably comprises an application specific integrated circuits chip which is connected by a printed circuit board 50 with other chips, such as a read/write channel chip 51, a motors drive chip 53, and a cache buffer chip 55, into an electronic circuit as shown in FIG. 2. The controller 57 preferably includes an interface 59 which connects to the host computer 54 via a known bus structure 52, such as ATA or SCSI.

[0038] The controller 57 executes embedded or system software comprising programming code that monitors and operates the controller system and driver 100. During a read or data retrieval operation, the computer system 54 determines the address where the data is located on the disk drive 100, i.e., magnetic head number, the track 30, and the relevant portion(s) 35 of the track 30. This data is transferred to the drive controller 57 which maps the address to the physical location in the drive, and in response to reading the servo information in the servo sectors 25, operates the actuator motor 54 and suspension arm 17 to position a magnetic head 16 over the corresponding track 30. As the media surface 23 rotates, the magnetic head 16 reads the servo information embedded in each spoke 25 and also reads an address of each portion 35 in the track 30. When the identified portion 35 appears under the magnetic head 16, the entire contents of the portion 35 containing the desired data are read. In reading data from the media surface 23, the read element (not shown) senses a variation in an electrical current flowing through a magnetoresistive sensor of the read element (not shown) when it passes over an area of flux reversal on the surface 23 of the media. The flux reversals are transformed into recovered data by the read/write channel chip 51 in accordance with a channel algorithm such as partial response, maximum likelihood (PRML). The recovered data s then read into the cache memory chip 55 of the disk drive 100 from whence it is transferred to the computer system 54. The read/write channel 51 most preferably includes a quality monitor function which enables measurement of the quality of recovered data and thereby provides an indication of data error rate. One channel implementation which employs channel error metrics is described in commonly assigned U.S. Pat. No. 5,521,945 to Knudson, entitled: Reduced Complexity EPR4 Post-Processor for Sampled Data Detection, incorporated herein by reference. The indication of recovered data error is used in order to select linear data density, track density and/or error correction code levels, in accordance with principles of the present invention, as more fully explained hereinbelow.

[0039] Writing or storing data on the media surface 23 is the reverse of the process for reading data. During a write operation, the host computer system 54 remembers the addresses for each file on the media surface 23 and which portions 35 are available for new data. The drive controller 57 operates the actuator motor 54 in response to the servo information read back from the embedded servo sector 25 in order to position a head 16, settles the head 16 into a writing position, and waits for the appropriate portions 35 to rotate under the head 16 to perform the actual writing of data. To write data on the media surface 23, an electrical current is passed through a write coil in the inductive write element (not shown) of the head 16 to create a magnetic field across a magnetic gap in a pair of write poles that magnetizes the magnetic storage media coating the media surface 23 under the head 16. When the track 30 is full, the drive controller 57 moves the magnetic head 16 to the next available track 30 with sufficient contiguous space for writing of data. If still more track capacity is required, another head 16 is used to write data to a portion 35 of another track 30 on another media surface 23.

[0040] In one aspect, the present invention increases the data storage capacity and yield of data storage devices having a plurality of media surfaces 23, such as hard disk drive 100 including disks 12 covered with magnetic media. In one method, shown by example in FIG. 4, TPI density for each portion 35 of a media surface 23 is individually selected by measurement to optimize the data storage capacity of that particular portion 35. Initially, values of TPI density are predefined and stored in a table of values that is input to a testing and formatting program. Generally, these values are incremental or decremental values of one another; for example, a maximum value or maxima of TPI density of data can be the highest number in a series of five TPI density values. The values of TPI density can be a fixed number of preselected levels or can be derived from an algorithm that is based on a particular pair of magnetic head 16 and media surface. The TPI can be continuously variable, depending on track radius or radial data tack zone. In addition, an acceptable error rate value, which represents the greatest error rate than can be tolerated, is also input into the testing and formatting program.

[0041] In one version of the present invention, the TPI density is optimized across portions 35 of a single media surface. As shown in FIG. 4, a TPI is selected (step 85) and data is recorded on a portion of the media surface 23 at the selected TPI (step 90). The level of track density (TPI) can be one of fixed number of preselected levels or can be derived from an algorithm that is based on the location of a portion 35 of the media surface 23. Thereafter, the recorded data is read (step 101) and an error rate of the recorded data is measured (step 105). The measured error rate is compared to an acceptable error rate (step 110), and if the measured error rate is greater than the maximum acceptable error rate, the previous steps are repeated for another track density value, for example, the originally selected value less a decrement (step 115). This process continues until the measured error rate is less than or equal to the acceptable error rate, to provide a maximum recordable track density of data for a particular portion 35 of the media surface 23.

[0042] Preferably, in the first iteration, the selected track density is a maximum value for the pair of magnetic head 16 and media surface 23 (step 125). The maxima is calculated or estimated from statistically compiled data of measured track density for a population of pairs of magnetic heads 16 and media surface 23. It is preferred to start with the maximum track density to provide the highest track density value in each portion 35 of the media surface 23 in the fastest time, assuming that the worst media surface 23 has a track density value closer to the maxima than the minima.

[0043] Because of a skew angle attributable to geometrical relationships between the surface 23 and the rotary actuator, track density values can be increased radially from the innermost tracks 30a (FIG. 3) near the center of a media surface 23 to the outermost tracks 30b near its periphery. The outer tracks 30b may have the same number of portions 35 as the inner tracks 30a, they can be made thinner in the radial direction and more closely spaced, thereby providing higher data storage capacities.

[0044] The track density can also be varied from one media surface 23 to another media surface 23. Track density is increased by decreasing either of the track width or the spacing between adjacent tracks 30. Preferably, the track density is varied by varying the spacing between adjacent tracks 30, because the width of the tracks 30 is determined by, and its variation limited to, the writing width or geometry of the write element of the magnetic head 16. The variation in track densities from one media surface 23 to another can be customized, or selected from the number of preselected levels of track density.

[0045] In a preferred method of determining the maximum recordable track density, the embedded servo sector 25 are initially written on a media surface 23 during a factory servo-writing process at a servo track density that is higher than the data track density, as illustrated in FIG. 3. Servo bursts within each servo sector 35 are provided in such number and placement to enable accurate positioning of the magnetic head 16 in a full range of positions across the media surface 23. Given the particular effective width and characteristics of the read element of a particular head (the read element width typically being narrower than the writer) information in the embedded servo sector 25 is read by the magnetic head 16 and passed to the drive controller 57 which directs the actuator motor 20 to readjust the position the suspension arm 167. This is important because high track densities require highly accurate positioning of the suspension arm 17, and the data track density cannot be greater than the servo track density. Generally, as shown in FIG. 3 example, the servo track density is about 150% of the maximum possible data track density. In FIG. 3 five servo tracks Sa, Sb, Sc, Sd and Se are shown in relation to three data tracks Tk1, Tk2 and Tk3. Servo track density is determined by determining the minimum read or write width of a population of magnetic heads 16. After writing the servo wedges 25 at the servo track pitch, the actual data track 30 can be written at any disk radial position between the servo tracks, not just at null position where equal amplitudes are observed from two different servo bursts reads from a servo wedge. Additional tests, as described above, are performed to determine the optimum data track density of the media surface 23. Each servo track comprises radially similarly situated servo information in servo wedges 25 (e.g., the set of servo information Se at essentially same radial distance from the disk center form a servo track circumferrentially, set of servo information Se at essentially same radial distance from the disk center form another servo track circumferrentially, etc.).

[0046] Most preferably, every disk drive is servo written at the factory at a second track density (servo TPI) which is sufficiently high to provide accurate positioning at any radius for the fill range of acceptable read/wrote widths of the read and write elements of a particular head 16. Data track density (data TPI) is then decoupled from servo TPI by writing data tracks centered at non-null positions of the servo pattern. Micro-jig techniques are employed by the controller 57 in order to carry out the desired positioning over the data track locations. Initial servo TPI is determined by determining an minimum read element width of an acceptable population of heads (as also by determining a maximum write width of the same acceptable population, if untrimmed servo bursts are employed in each servo sector 25). More servo bursts an be provided to ensure adequate linearity of servo position error signal (PES) derived by reading relative burst amplitudes at any particular disk radius for a worst case read element and head.

[0047] While an example servo track density is presently approximately 150% of the data TPI, the present invention provides increasing servo TPI relative to average data TPI to ensure that a read element on the narrow end of the distribution has sufficient width of linear response to provide a useable PES for use by the controller 57.

[0048] Following the factory servowriting process, additional time during drive self-scan is needed to determine the optimum data TPI for each data surface 23. One preferred method, described further below, is to perform 747 measurements that can be used to determine the optimum track pitch (the expression 747 comes from a similarity in appearance between a resultant data plot and an elevational outline of the Boeing model 747 airplane). The head 16 is moved off track until the error rate exceeds as chosen threshold. The distance to failure is called off track capacity. This process is repeated with adjacent tracks written at smaller spacing until the off track capability drops to zero. The resulting data for off track capability versus track pitch can then be analyzed to determine the optimum track pitch, typically chosen as the track pitch with maximum off track capability. This process is described in more detail in an article by R. A. Jensen, J. Mortelmans, and R. Hauswitzer, entitled: Demonstration of 500 Megabits per Square Inch with Digital Magnetic Recording, IEEE Trans. on Magnetics, Vol. 26, No. 5, September 1990, p. 2169 et seq. However, a simple in-drive erase width measurement may also be used to determine suitable data TPI.

[0049] The optimized track density determined can also be used to optimize the yield or qualifying pass rate of the data storage devices. The example flowchart in FIG. 5, shows steps of an implementation of this process for increasing the yield and data storage capacity of the disk drive 100 including the plurality of media surfaces 23. In this method, in a determining step 150 maximum track density of data (optionally maximum recordable linear density of data) is determined for each media surface 23 using the methods described above. Optionally, the media surface 23 is formatted using the predetermined maxima of track density in a formatting step 152. Then, in a calculation step 155 the surface capacity of each media surface 23 is calculated from the measured, maximum recordable density. The surface capacity is described by the equation: TPIBPI(1ECC)/FE, wherein TPI is track density, BPI is the linear density, ECC is the fractional level of error correcting code used which is typically about 0.1, and FE is the format efficiency which is typically about 0.57.

[0050] After each media surface 23 has been formatted, the calculated surface capacities of all formatted surfaces 23 are summed in a summing step 160 to determine the device capacity, which is the storage capacity of the entire data storage device 100. If the device capacity equals or exceeds a target or desired storage capacity, the data storage device 100 is passed, and it is not necessary to determine optimal TPI, BPI and ECC levels for any more media surfaces 23. However, if the sum of the capabilities of all measured surfaces does not equal or exceed the target capacity, it is determined if all surface 23 have been measured. If all the media surfaces 23 have not been measured, the surface capacity of the next media surface 23 is determined, and if the device capacity is still less than the target capacity, the disk drive 100 is failed. After the disk drive 100 is qualified, testing ends, and the drive controller 57 is programmed for the appropriate track density and linear density for formatting each media surface 23. The drive controller 57 is also programmed to apply a measured or calculated level of error code to each media surface 23 during formatting. The above methods are utilized to manufacture storage devices such as disk drives 100, with storage media surface formats according to the methods described herein.

[0051] In every storage device such as the disk drive 100, there is a distribution associated with head/media pair performance in that disk drive. In another aspect, the present invention takes advantage of that distribution to determine different/variable TPI assignment for heads, and optionally variable BPI.

[0052] As described, in conventional disk drives, the TPI is the same for each head and corresponding disk surface, regardless of the capabilities of different heads in the disk drive. Example FIG. 6 shows conventional layout in disk drives, wherein the TPI is the same for each head and corresponding disk surface, regardless of the capabilities of different heads in the disk drive. In the example of FIG. 6, the disk drive includes N heads, with fixed servo track pitch and fixed data track pitch for each zone for heads 0, . . . , N1. For all heads, there are 45 servo tracks, wherein:

Head 0:	15 Data Tracks	15 Data Tracks per 45 Servo Tracks
Head 1:	15 Data Tracks	15 Data Tracks per 45 Servo Tracks
. . .		. . .
Head N-1:	15 Data Tracks	15 Data Tracks per 45 Servo Tracks

[0053] However, according to the present invention, for a desired disk drive capacity, based on the number of heads/surfaces, a suitable TPI (and optionally BPI) per head-surface pair is selected to satisfy the desired disk drive capacity. Based on the capability of head and corresponding capacity of each disk surface, using variable TPI, a data storage (surface) format per disk surface in the disk drive is then determined.

[0054] As such, for example, once a disk drive 100 with multiple heads 16 is assembled, then each head's recording capability/performance is determined. Then if a head 16 is better performing, then the TPI for that head is increased. And, if a head 16 is has lower performance, then the TPI for that head 16 is decreased. By making TPI per surface portion adjustable to the capability of the corresponding head 16, a higher performing head compensates for a lower performing head, whereby the disk drive capacity remains at the desired capacity. In another aspect of the present invention, variable TPI is utilized to optimize disk drive capacity by providing an optimum TPI for each head 16 in the disk drive 100 according to the capability of the head 16.

[0055] In one example, a higher performing head 16 can record at narrower track pitch than a lower performing head 16. This allows for variable TPI for different disk surfaces, by increasing the number of tracks per inch for the higher performing head, and decreasing the number of tracks per inch for the lower performing head. Overall, the disk drive capacity remains at the desired value or is increased over conventional disk drives.

[0056] For variable TPI, each head's performance is determined during testing (e.g., determining TPI tolerance for each head). For a desired disk drive capacity, an optimization process selects suitable TPI (and optionally BPI) to each head based on that head's measured performance, to achieve (or surpass) the desired disk drive capacity. The optimization process is performed per head 16 per disk drive 100, and can be performed during a self-scan of each disk drive 100.

[0057] The aforementioned methods according to the present invention are described in further detail below.

[0058] Vertical Zoning

[0059] Referring to FIG. 7, an example track layout in a disk drive with n heads is shown. Each disk surface 23 is divided into several concentric zones 27 for writing data to and reading data from using a corresponding head 16, wherein each zone 27 includes multiple data tracks 30. Example FIG. 16, described further below, shows another example of several capacity zones 27 for a disk drive surface 23.

[0060] Referring to FIG. 8, according to an embodiment of the present invention, each zone 27 includes a number of concentric virtual cylinders (sub-zones or regions) 29, wherein each virtual cylinder (VC) 29 includes a number of data tracks 30 between radially spaced boundaries for each VC 29. The disk drive includes concentric VCs from ID to OD on all disk surfaces. There are multiple zones 27 per disk surface 23, and there are multiple VCs 29 (e.g., VC0 . . . VCn) per zone 27. Within a VC 29 there are multiple servo and data tracks. Further, as shown in FIG. 1, each VC 29 (e.g., VC1 . . . VCM, . . . , etc.) extends vertically between a first surface 23 of a first disk 12 (e.g., Disk1) and a second surface 23 of the last disk 12 (DiskN) in the disk stack in the disk drive 100.

[0061] Conventionally there is a fixed number of data and servo tracks on disk surfaces, and there is a fixed ratio of data tracks relative to servo tracks in a zone from one surface to the next. However, according to the present invention, in each VC 29, the density of data tracks 30 (TPI) can change from surface 23 to surface 23, the number of servo tracks (e.g., servo tracks per inch) can change from surface 23 to surface 23, and the ratio of the number of data tracks to the number of servo tracks can change from surface 23 to surface 23. Further, on each disk surface 23, the number of data tracks (TPI) can change from VC 29 to VC 29, the number of servo tracks (e.g., servo tracks per inch) can change from VC 29 to VC 29, and the ratio of the number of data tracks to the number of servo tracks can change from VC 29 to VC 29.

[0062] For example, the ratio of data tracks relative to servo tracks in a VC 29 can change from one surface 23 to the next. In another example, each VC 29 may include the same number of servo tracks from one surface to the next, but may have different number of data tracks from one surface to the next in the same VC 29. On the same disk surface, there can be the same number of servo tracks from VC 29 to another, but there may be different number of data tracks from one VC 29 to another.

[0063] In one embodiment the present invention provides a method (Vertical Zoning) to provide different area track densities/formats on different disk surfaces 23 in relation to corresponding heads 16, to match those area densities optimally with the capabilities of each head 16. In Vertical Zoning, the area density is obtained by varying the track density TPI (and optionally BPI) in relation to the heads. As such, a weak head 16 which does not meet the requirement for a selected TPI (and optionally BPI), is assigned to a lower TPI (and optionally BPI), and is compensated by strong head(s) which are capable of more than the selected TPI, by adapting TPI (and optionally BPI) per head such that the same disk drive capacity is maintained.

[0064] Variable TPI

[0065] In one version of the present invention, variable TPI is used to implement Vertical Zoning. In order to provide variable TPI, a surface format (i.e., virtual cylinder format) is utilized instead of conventional cylinder format (FIG. 6) for logical block addressing. With that surface format, variable TPI is supported across different data zones 27 and across different disk surfaces 23.

[0066] Examples of variable TPI implementations according to the present invention, are now described.

[0067] In one example, variable TPI is implemented by varying the servo track pitch profile for each head 16 during servo-writing process. Example FIG. 9 shows ratio of data tracks 30 to servo tracks 31, providing variable number of servo tracks 31 with fixed number of data tracks 30, per head 16. As such, Head 0 and Head 1 have different ratio of number of data tracks 30 per number of servo tracks 31.

[0068] In another example, the number of servo tracks 31 per head-surface pair changes, while the ratio of data tracks 30 to servo tracks 31 remains the same for all surfaces 23 (e.g. 3 servo tracks for each 2 data tracks). Using a fixed ratio between servo tracks 31 and data tracks 30, by increasing/decreasing the number of servo tracks 31, the number of data tracks 30 automatically increase/decrease, and so does the surface capacity. Based on each head's performance, the corresponding disk surface 23 may have a different number of servo tracks 31 (and data tracks 30) than other disk surfaces 23.

[0069] In another example, variable TPI is implemented by maintaining the same servo track pitch profile for all heads/surfaces, and varying the data track pitch relative to the servo track pitch without servo-writing each surface at a different servo track pitch profile.

[0070] FIG. 10 shows an expanded view for an example fixed servo track pitch with variable data track pitch for a zone 27 (e.g., Zone1), wherein for all heads 16 there are 45 servo tracks 31, such that:

Head 0:	15 Data Tracks	15 Data Tracks per 45 Servo Tracks
Head 1:	18 Data Tracks	18 Data Tracks per 45 Servo Tracks
. . .		. . .
Head N:	12 Data Tracks	12 Data Tracks per 45 Servo Tracks

[0071] FIG. 11 shows example details of servo track 31 and data track 30 ratio for fixed servo track pitch with variable data track pitch. As such, in VC0 there are fixed servo tracks 31 for a 2 head model with variable data tracks 30 (e.g., for Head0 and Head1, there are the same number of servo tracks 31, but data tracks 30 vary for Head0 and Head1). The same format is provided for VC1, wherein the data track density changes and the servo track density remains fixed. The servo tracks 31 for all heads are written the same way, however based on each head's record/playback performance, in self-scan process for putting down data tracks 30, for one head e.g. 3 servo tracks 31 for every 2 data tracks 30 are used, and for another head the ratio is changed to 3 servo tracks 31 for every 2.2 data tracks 30. Therefore, data tracks 30 can be either closer to each other, or further apart, depending on how the data track to servo track ratio is changed. As such, in this case the data tracks 30 are no longer physically aligned vertically in cylinders. And, for logical block sequencing, instead of a conventional cylinder format (FIG. 12), the surface format (i.e., virtual cylinder) is utilized.

[0072] As shown in FIGS. 6 and 12, in conventional disk drives the data tracks are organized into concentric data zones. With multiple transducer heads in a disk drive (e.g., one head per disk surface), the data zones are aligned vertically. Within each data zone, the same TPI is used for all the heads on different disk surfaces. The data tracks on different disk surfaces are aligned vertically, forming logical cylinders in which logical data blocks are accessed sequentially. When accessing data sequentially within logical a cylinder, a head switch is performed between consecutive data tracks. At the last head, a single track seek is performed to read data from the next logical cylinder. In this description, data track pitch indicates distance between two adjacent data tracks 30, and servo track pitch indicates distance between two adjacent servo tracks 31.

[0073] As shown in FIG. 12, in conventional disk drives wherein data tracks on different disk surfaces lineup vertically in cylinders, to access data logically, every time reading data from a logical cylinder on one disk surface is complete a head switch is performed to another disk surface to continue logical data access. For example, in a 4-head, 2-disk, disk drive, as data is accessed in a logical cylinder going down vertically from e.g. head0surface0 on the first disk to head3surface3 on the second disk, after accessing data on a track on surface0 using head0, a switch to head1 of surface1 is performed and data is read from a track in the same logical cylinder from surface1. This process continues (e.g., head2surface 2, head3surface 3) until all data in that logical cylinder is accessed. Then a seek is performed to the next logical cylinder, and data read that from that next logical cylinder in a similar fashion described above.

[0074] However, referring to example FIG. 13, for disk drives with variable TPI according to the present invention, the conventional cylinder format for logical block addressing is undesirable because of performance degradation caused by a head switch which may also involve a track seek. Instead, the surface format according to the present invention for logical block addressing improves the drive performance, and TPI can vary from virtual cylinder 29 to virtual cylinder 29, zone 27 to zone 27 and from surface 23 to surface 23. With the example variable TPI format according to the present invention, the data tracks 30 on different disk surfaces 23 may no longer align vertically. To reduce the head switch time and to improve the drive's performance during logical operations, it is preferable to utilize the surface data format according to the present invention instead of the conventional logical cylinder format. With the surface format, all disk surfaces are divided into said virtual cylinders 29. The virtual cylinders 29 are defined in relation to servo tracks 31 and are aligned vertically from one disk surface 23 to the next. However, within the same virtual cylinder 29, the corresponding data track density (TPI) can be different on different surfaces.

[0075] As shown in the example FIG. 13, when sequentially accessing logical blocks according to the present invention, a single track seek is used instead of head switch within the same virtual cylinder 29 for speed. At the end of a virtual cylinder 29, a head switch occurs and sequential access continues on the surface 23 of another (e.g., next) disk in the opposite direction. In FIG. 13, when all tracks 30 in a VC 29 on one surface (e.g., Surface0Head0) are read, a head switch to the next surface (Surface1Head1) in the same VC 29 is performed to read the tracks 30 therein, until all tracks on all surfaces for that VC 29 are read (or written). FIG. 13 shows example track access in one VC 29.

[0076] In another example, head switch from Head0 to Head1 can be direct, wherein e.g., in FIG. 13 reading data started top of first block (VC1 on Surface0), zig-zag down the first block, then across to the second block (VC1 on Surface1) zig-zag up from the bottom of second block to the top of the second block, then across to the top of the third block (VC1 on Surface2) zig-zag down to the bottom of the third block, etc. FIGS. 14 and 15 show other example block diagrams depicting the logical block address accessing scheme using the above example surface format for two-head disk drives and four-head disk drives, respectively. In example FIG. 14, an example logical block addressing (LBA) scheme for two-head, 2 surface, disk drives is shown. For the same virtual cylinder, the data is accessed by Head0 sequentially from Track0 through Track12 on Surface0, then by Head1 from Track0 to Track6 on Surface1, then by Head0 from Track13 through Track25 on Surface0, then by Head1 from Track7 through Track13 on Surface2, etc. In this example, Surface0 has higher TPI density than Surface1 in the same virtual cylinder. In example FIG. 15, another example logical block addressing (LBA) scheme for four-head, 4 surface, disk drives is shown. Sequentially, Head0 reads tracks in a VC 29 on Surface0, Head1 reads tracks in that VC on Surface1, Head2 reads tracks in that VC on Surface2, Head3 reads tracks in that VC on Surface3, Head0 reads tracks in that VC on Surface0, Head1 reads tracks in that VC on Surface1, Head2 reads tracks in that VC on Surface2, Head3 reads tracks in that VC on Surface3, etc. The surfaces can have different TPIs for the same VC 29. For illustration purposes, in the example of FIG. 11, for the two-head disk drive, Head0 for one disk surface 23 supports 6 data tracks 30 for a virtual cylinder VC0, whereas Head1 for another disk surface 23 supports 5 data tracks 30 for that virtual cylinder VC0.

[0077] Variable Data Track Pitch for Variable TPI Implementation

[0078] As aforementioned, one example variable TPI surface format is implemented by varying the servo track pitch, wherein each disk surface can be servo written with a different servo track pitch profile. Example FIG. 9 shows 2 head format having different servo track pitch with different data track pitch. To reduce the complexity of servo writing and to write the servo pattern in a single pass, in an alternative example method the servo track pitch profile remains constant in all surfaces, and the variable TPI is implemented by varying the data track pitch relative to the servo track pitch. By disassociating data tracks from a fixed ratio to the servo tracks, TPI can be determined after surfaces have been servo written. Example FIG. 11 shows a 2 head format having fixed servo tracks and variable data tracks.

[0079] In disk drives with MR-type heads, the servo system can read from any track location depending on the offset between the writer and the reader elements of the heads. However, during writing, the servo typically writes at track center which is a spot with good TMR. To implement variable TPI with varying data track pitch, the servo system must be capable of writing at any desired track location, away from the track center, in locations with less than optimum TMR.

[0080] In one example, the number of data tracks 30 per virtual cylinder 29 also varies from data zone 27 to data zone 27 across the stroke on a disk surface 23. Each data zone 27 can include a fixed number of virtual cylinders 29 for all heads 16. The number of virtual cylinders 29 is the same across different disk surfaces 23 in the disk drive. In this fashion, the surface format with virtual cylinder structure according to an embodiment of the present invention supports variable TPI across the zones and across the disk surfaces. An optimization technique to determine the TPI (and optionally BPI) for each head according to the present invention is provided further below.

[0081] Other example formats according to the present invention include Variable Zone Layout (Vertical Data Zoning) and Vertical Track Zoning. In Variable Zone Layout, areal density variation is implemented by variable recording frequency (BPI) for each head 16 per disk surface 23. In Vertical Track Zoning (i.e., Vertical Zoning with variable TPI), the areal density variation is implemented by variable TPI for each head 16 per disk surface 23.

[0082] In another example according to the present invention, variable BPI and variable TPI are combined to allow each head to be adapted such that the areal density capability of each head is better utilized by allowing the selection of both linear and track densities. With both TPI and BPI as variables, a single head disk drive can also be optimized by trading off TPI against BPI. As such, areal density variation is implemented by both variable TPI and variable BPI for each head per disk surface. In addition, the TPI and BPI can be adaptive across the actuator stroke. In that case, by dividing the disk surface 23 into capacity zones 27 (e.g., FIG. 16) and by calculating capacity in real-time during the self-scan test, the drive capacity can be optimized across the capacity zones depending on the head/media performance. During the self-scan test process, capacity optimization is performed before variable TPI/BPI optimization.

[0083] Optimization Process

[0084] The present invention also provides variable TPI (and optionally variable BPI) optimization process, wherein in one embodiment, an example optimization process based on a 747 geometric model measurement is utilized. An example method 747 measurement is described in a publication titled Measure a Disk-Drive's Read Channel Signals, August 1999, Test & Measurement World, Published by Cahners Business Information, Newton, Mass.

[0085] This optimization process allows optimization of TPI (and optionally BPI at the same time) during the self-scan test of the disk drive to meet Off-track Capacity (OTC) performance and drive capacity requirements. Further, the disk drive capacity can be maximized for a given OTC performance. The optimization process can be applied to disk drives with multiple heads and single head drives, wherein a drive with a single head can be optimized by trading off TPI against BPI.

[0086] In the following description, these terminologies are utilized. Capacity zone is the drive capacity of a zone (each disk surface is divided into many zones). Linear density is the number of bits recorded per inch (BPI). Track Mis-Registration (TMR) indicates allowable position error. Track density is the number of tracks per unit length such as inch which is measured in a direction perpendicular to the direction in which the tracks are read (TPI). UOTC is Unsqueezed Offtrack Capacity. SOTC is Squeezed Offtrack Capacity.

[0087] Capacity Optimization Across Capacity Zones

[0088] In this example (Vertical Zoning Recording) TPI (and optionally BPI) are adaptive depending on the head/media pair performance. With variable TPI, each disk surface 23 can be divided into multiple TPI zones or virtual cylinders 29 (e.g., FIG. 11), wherein each TPI zone 29 overlaps multiple data zones. In addition, all disk surface(s) can be further divided into multiple capacity zones 27 with each capacity zone including multiple TPI zones 29. The capacity of each capacity zone 27 is adaptive and is determined by the head/media performance at the capacity zone 27. The formation of the capacity zones 27 allows the drive capacity to be traded off between the capacity zones 27 while still maintaining the required drive capacity. The capacity optimization is performed at nominal TPI/BPI before variable TPI/BPI optimization is performed within each capacity zone.

[0089] Variable TPI/BPI Optimization

[0090] In an example variable TPI/BPI optimization, an algorithm based on 747 geometric model is utilized. This algorithm allows optimization of TPI and BPI at the same time during the self scan test of the drive to meet the OTC performance, and the drive capacity, requirements. The disk drive capacity for a given OTC performance can also be maximized. The algorithm can be applied to drives with multiple heads and disk drives with only one head drive, wherein a drive with a single head can be optimized by trading off TPI against BPI.

[0091] In order to use variable recording density (e.g., TPI), an example technique according to the present invention includes the steps of selecting and using TPI optimally on each disk surface corresponding to each head. The selection process is performed with variable TPI optimization at self-scan test of the disk drive. Within each capacity zone, each head is assigned a TPI, optimally based on the Offtrack Capacity (OTC) performance of the heads within the capacity zone. For a single head drive, this technique also allows the TPI to be traded off against BPI to obtain optimal capacity.

[0092] 747 curves are used to determine performance of the heads as a function of head geometry. A 747 measurement of each head in the drive is obtained, to determine the proper TPI and optionally BPI for a head at each zone. The 747 measurements for each head can be taken at different areas of a corresponding surface (e.g., inner, middle, outer diameter, etc.). Therefore, in manufacturing during a test process, measurement of 747 performance of each head is obtained, and from the 747 curves the TPI and BPI are selected to provide desired capacity format for each head per zone and virtual cylinder. This is performed for each head, and every surface in each disk drive. As such, in an example, five disk drives with four heads each, meet a certain minimum capacity (though disk drives need not have identical capacity), but each disk drive has a different surface format than others. This is because surface format optimization is performed for each head based on measured performance of each head/surface.

[0093] Referring to FIG. 17, in each disk drive, the record/playback capability of each head is determined (step 170). Then, the heads are ranked according to capability (e.g., weak or strong) (step 172). Then a surface format such as TPI per head and zone (or virtual cylinder) is selected for each head in the disk drive (step 174). In one case, there are several predetermined TPI formats, such as one for strong heads and one for weak heads. Ranking of the heads can have different levels, and a corresponding predetermined format for each level. As such, in another example, the heads can be ranked weak, medium and strong, wherein a predetermine format is selected for each head. The total capacity is calculated based on the selected formats for the heads (step 176), to determine if required capacity and performance are satisfied for each disk drive (step 178). If not, TPI is traded off between the heads by changing TPI of the heads until the desired capacity and performance (e.g., error rate) are satisfied (step 180). For example, stronger heads are assigned higher TPIs to increase capacity, and a weaker head are assigned lower TPI to meet error rate requirements.

[0094] Referring to FIG. 18, an example of the step 172 of determining record/playback capability of each head includes the steps of: selecting TPI level per zone (step 190) and data is recorded with the head per zone on the media surface 23 at the selected TPI per zone (step 192). The level of track density (TPI) can be one of fixed number of preselected levels or can be derived from an algorithm that is based on the location of a portion 35 of the media surface 23. Thereafter, the recorded data is read (step 194) and an error rate of the recorded data is measured (step 196). The measured error rate is compared to an acceptable error rate for each zone (step 198), and if the measured error rate is greater than the maximum acceptable error rate for a zone, the previous steps are repeated for those zones another track density value, for example, the originally selected value less a decrement (step 200). This process continues until the measured error rate is less than or equal to the acceptable error rate, to provide a maximum recordable track density of data for each head per zone (step 202) in the disk drive 100. Preferably, in the first iteration, the selected track density is a maximum value for the pair of magnetic head 16 and media surface 23.

[0095] In an example 747 measurement, a nominal BPI value is first used to determine record/playback performance/capability of each head, and then the assigned BPI for each head is adjusted based on head capability Using a geometric 747 model, the performance of a head can be estimated or measured with a 747 profile. Two points on the 747 profile at a fixed error rate, Unsqueezed Offtrack Capacity (UOTC) and Squeezed Offtrack Capacity (SOTC), can be used to uniquely define the 747 profile performance of the heads. The purpose of the optimization in the disk drive is to allow all the heads to have maximum UOTC and SOTC margins (i.e., Highest Offtrack Capability with a maximum disk drive capacity) while meeting the disk drive capacity and performance requirements. This can be achieved by first moving the 747 curve of each head individually (i.e., by changing BPI and/or TPI), to a point of minimum performance margin. A minimum performance margin point is defined by the minimum required SOTC at a pre-defined track squeeze. At this minimum performance point, the disk drive is also at the maximum capacity point. The next step is to trade off capacity for more performance margin by moving 747 curves of all the heads collectively (i.e., by changing BPI and/or TPI), to a point that meets the minimum capacity requirement. By moving the 747 curves of all the heads collectively, the same SOTC performance margin is maintained. An example 747 geometric model and the variable TPI/BPI technique are described in more detail below.

[0096] 747 Geometric Model

[0097] The use of SOTC and UOTC as performance metrics is based on an example geometric 747 model. The UOTC and SOTC can be defined as a function of write width (WW), read width (RW), erase width (E), track pitch (TP), amount of squeeze (SQZ) and on-track bit error rate (BER) as shown in equations (1) and (2) below:

[0098] For BPI optimization, UOTC is used as the performance metric. For any given head, WW, RW and E are all constant. Therefore, UOTC is directly a function of BER or BPI as shown in equation (3) below. In addition, SOTC is also a function of BPI if TP and SQZ are constant as shown in equation (4) below:

[0099] Whereby,

[0100] For TPI optimization, SOTC is used as the performance metric. For a given BPI, SOTC is a function of TP and SQZ. Therefore, the track pitch or TP can be determined from the parameter SOTC once SQZ is defined according to equations (5) and (6) below:

[0101] Variable TPI/BPI Optimization Algorithm

[0102] Prior to Variable TPI/BPI (vTPI/BPI) optimization, all the heads assume the nominal TPI/BPI and capacity, and each head can be positioned on a 747 profile according to its OTC capabilities. In one example, the final goal of the optimization is such that all the heads have similar UOTC and SOTC capabilities while meeting the overall drive capacity requirement. This can be achieved by first moving the heads (i.e., moving 747 curves of the heads by changing BPI and/or TPI) individually to a point of the minimum performance margin, and then moving 747 curves of the heads collectively to meet the capacity requirement.

[0103] The optimization algorithm can be divided into two major parts. First, move the 747 profile (curves) of the heads individually to the minimum performance margin point defined by the drive requirements of UOTC, SOTC, and SQZ. The point of minimum performance margin on all heads is also the point of maximum capacity for the drive. The drive capacity is determined, and if the drive does not meet the minimum capacity requirement at this point, the drive is either set back to the default condition or a best estimate is used to meet the capacity requirement. Second, for the point of minimum performance and maximum capacity, if the drive has excess capacity, the TPI and/or BPI for the heads can be relaxed by moving to a 747 profile with higher OTC margin to meet the capacity requirement. If the drive has less than the required capacity, the TPI and/or BPI for the heads can be increased, by moving to a 747 profile with lower OTC margin to meet the capacity requirement. By adjusting all the heads by the same amount, the same margin can be gained by all the heads, satisfying the requirement of maximizing the performance of the drive.

[0104] The basic steps of the example vTPI/BPI optimization process according to the present invention are listed below. The minimum performance point is defined by the following test limits:

	UOTC1:	minimum required UOTC margin
	SOTC1:	minimum required SOTC at SQZ1
	SQZ1:	SQZ test point for SOTC1

[0105] The example optimization process includes the steps of:

[0106] 1. Find the minimum acceptable performance point for each head by first optimizing BPI: (a) run channel optimization for new BPI (for every different data rate, there is channel optimization) and (b) optimize BPI within the allowed range of formats or data rates, such that the difference (UOTCUOTC1) is minimized while satisfying the requirement of UOTC1<UOTC.

[0107] 2. Find the minimum acceptable performance point for each head by optimizing TPI: optimize track pitch within an allowed ATP (Adjacent Track Pitch) range such that difference (SOTCSOTC1) is minimized while satisfying the minimum performance requirement of SOTC1<SOTC.

[0108] 3. Optimize BPIs for all the heads to meet the capacity requirement: (a) calculate: delta_capacity(current_capacityminimum_capacity), and (b) if delta_capacity< >n*BPl_step_size, then increase/decrease BPI by n*x %, within the allowed BPI formats for each of the heads if possible.

[0109] 4. Optimize TPIs for all heads to meet the capacity requirement: (a) calculate the new capacity, determine the new delta_capacity, (b) calculate the delta_ATP allowed for each head, and if delta_capacity< >n*ATP_step_size, decrease/increase track pitch by delta_ATP within the allowed ATP range for each of the heads if possible.

[0110] A data storage format and storage device according to the present invention provides many advantages over conventional disk drives. Because not all heads in disk drives perform the same way, in conventional disk drives, if one of multiple heads has a weak performance and therefore can read/write at lower than expected storage capacity, the overall disk drive capacity is lower than expected and the disk drive is wastefully discarded as a failed drive. However, according to an embodiment of the present invention, by making the storage density adaptable to the head capability, the storage format for a better performing head is adjusted such that the better performing head can compensate for the weak head, and achieve the expected disk drive storage capacity. This improves the disk drive yield and disk drive performance, and reduces overall disk drive costs by allowing use of disk drives with weak heads. Further, by making the storage density adaptable to the head capability, the storage format can be adjusted to obtain maximum capacity per disk drive depending on the performance of the heads in each disk drive.

[0111] The present invention has been described in considerable detail with reference to certain preferred versions thereof; however, other versions are possible. Therefore, the spirit and scope of the appended claims should not be limited to the description of the preferred versions contained herein.