United States Patent [191
`Sidman
`
`[11] Patent Number:
`[45] Date of Patent:
`
`4,536,809
`Aug. 20, 1985
`
`[73] Assignee:
`
`[54] ADAPTIVE MISPOSITION CORRECTING
`METHOD AND APPARATUS FOR
`MAGNETIC DISK SERVO SYSTEM
`[75] Inventor:
`Michael Sidman, Colorado Springs,
`Colo.
`Digital Equipment Corporation,
`Maynard, Mass.
`[21] Appl. No.: 376,972
`[22] Filed:
`May 10, 1982
`[51] Int. Cl.3 ............................................ ..G11B 21/10
`[52] u.s.c1. .................................................... .. 360/77
`[58] Field of Search ............. .. 360/75, 77, 72; 369/43;
`318/594, 600, 632
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`3,404,392 10/1968 Sordello .............................. .. 360/77
`3,881,184 5/1975 Koepcke et a1.
`360/78
`3,994,016 l1/1976 Moghadam
`360/77
`4,087,842 5/1978 Manly .......... ..
`360/77
`4,135,217 1/1979 Jacques et al. .
`4,136,365 1/1979 Chick et a1.
`
`..
`.. 360/77
`. 360/78
`
`4,208,679 6/1980 Hertrich . . . . _
`
`. . . . . . . . . . .. 360/ 77
`
`4,371,960 2/1983 Kroiss . . . . . . . . . . . .
`
`. . . . . . . . . . .. 360/77
`
`4,419,701 12/1983 Harrison et a1. .................... .. 360/77
`
`FOREIGN PATENT DOCUMENTS
`
`0009087 4/ 1980 European Pat. Off. .
`
`OTHER PUBLICATIONS
`Patent Abstracts of Japan, V. 3, No. 40, Apr. 6, 1979, p.
`152E102 & JP.A—54-l97l8, “Locating System of High
`Accuracy”.
`Patent Abstracts of Japan, V. 4, No. 18, Feb. 13, 1980,
`p. 75E171 & JP.A—54-l58206, “Servo-Information
`Recorder of Magnetic Disc Memory”.
`IBM TDB, vol. 19, No. 6, Self-Calibrating Disk Storage
`Apparatus by Griffths et al.
`
`IBM TDB, vol. 23, No. 2, Track Locating and Follow
`ing Apparatus for a Flexible Disk File by .lahnke.
`Electronics, vol. 55, No. 8, Apr. 21, 1982, D. Sutton,
`“Winchester Cartridge Challenges Other Backup Sys
`tems”, pp. 112-116.
`Primary Examiner——Raymond F. Cardillo, Jr.
`Assistant Examiner—-Steven R. Garland
`Attorney, Agent, or Firm--Cesari and McKenna
`[57]
`ABSTRACT
`A misposition correction system for correcting misposi
`tion errors due to spindle runout and other slowly vary
`ing errors in a servo positioning system of a magnetic
`disk storage device. The system includes means for
`dynamically measuring misposition error with respect
`to a data track centerline using an anti-aliasing analog
`?lter, means for digitizing the measured analog signal
`and for removing selected harmonics of the fundamen
`tal-frequency of the resulting cyclic error signal, means
`for transforming the digitized error signal by a matched
`digital ?lter whose transfer function contains indepen
`dently adjustable DC gain, fundamental-frequency gain,
`and phase lead components thereby to generate a mis
`position error correction signal that matches the electri
`cal and mechanical response characteristics of the servo
`system. Further, the system includes means to itera
`tively re?ne the correction signal by re-applying it to
`the servo controller when measuring misposition errors
`with respect to the data track centerline. The system
`stores separate misposition correction signals for each
`transducer on the disk so that one of several correction
`signals can be selected depending on which transducer
`in the device is selected. The system further includes
`means for generating a bias force correction signal to
`correct for non-linear position errors resulting from
`variations in bias forces acting on the transducer car
`riage over its range of radial displacement.
`
`STORAGE DATA
`
`SERVO SIGNAL
`DEHUDULATUR
`
`PUIER-UPOR
`m not
`
`[so
`
`g 1
`8155311111
`PROCES
`
`._
`
`conarcnon
`W,
`
`so
`
`19 Claims, 10 Drawing Figures
`
`10 DATA
`‘ PROCESSING
`5151 EH
`
`ERROR RAT E
`SENSOR
`
`52
`
`a
`4
`A/o
`
`m
`
`i
`‘I
`
`msoosmou common SIGNAL
`
`POSITION-ERROR SIGNAL
`
`POSITION FEEDBACK
`vroomr FEEDBACK
`
`SERVO
`31~ CONTROL
`cmcun
`1
`TRACK SEEK
`39“ AND COUNT -mci no,
`cmcunmr
`
`“m 44
`4|
`"no, /
`E5 on
`0R TRANSDUCER
`
`1
`
`Micro Motion 1042
`
`

`
`U.S. Patent Aug;20% 1985
`
`Sheet1of1O
`
`4,536,809
`
`53S
`
`2.3.8....
`
`.55
`
`5.88585
`
`.a_=:§
`
`§:__§
`
`.s_§.a_=
`
`_a_::.._..8
`
`§m:2_._
`
`\.u.Q
`
`2
`
`
`
`

`
`U.S. Patent Aug. 20, 1985
`
`Sheet2of10
`
`4,536,809
`
`zoEmoa
`
`
`
`em:.o_m
`
`«:5.29:
`
`uu<nE:mmm
`
`
`
`k._<zu_mm.Izo_.5mEoo
`
`
`
`zo_:mon_m_s__
`
`
`
`mo»<s__5m>:uo.m>
`
`mmxmmm545
`
`HZDOUoz<
`
`>m.::um.u
`
`96mo;
`
`
`
`
`
`I._.<n_xu<moHr_zoEmoaE
`
`
`
`4<zo_mmommm_IzoEmoa
`
`mE.<.5oosm_o
`
`o>mmm
`
`N.9...
`
`
`
`on<25zofiommmou
`
`
`
`
`
`zoEmon_m_2ommohmsoc...
`
`<\o
`
`mm.Em>zoo
`
`3
`
`
`
`
`
`

`
`US. Patent Aug. 20, 1985
`
`Sheet3ofl0
`
`4,536,809
`
`Fig. ,3
`
`4
`
`

`
`U.S. Patent Aug. 20,1985
`
`Sheet40fl0
`
`4,536,809
`
`f——+— SERVO FIELD s4——+—-— DATA FIELD _->2
`
`BURST A
`
`BURST B
`
`-——-—
`
`~2
`
`- - '
`
`as
`
`1
`
`LM 1
`
`j
`
`- — — —-
`
`~b+1
`
`_ _
`
`m-
`
`—_
`
`___
`
`83
`
`87
`
`)
`
`_ _ __
`
`1
`
`__
`
`I/\ i
`
`b
`
`1-1
`- _ _ i __
`
`-
`\b-1
`
`k
`
`V
`
`3
`1
`
`1'2
`
`>
`
`_ __ _. l _ 1AM
`
`_.._
`
`Z
`
`5
`
`

`
`US. Patent Aug. 20,1985
`
`SheetSoflO
`
`4,536,809
`
`22.582:
`
`225518
`
`3 E2 EEEA E .335
`EEWZSQ
`
`__5
`
`:55
`
`$5650
`
`555425
`
`, mom
`
`n 9k
`
`.3:
`i
`
`:3 swag:
`
`E2 :55
`
`2.:
`
`5::
`
`_
`
`
`
`MON mON
`
`I
`
`52
`2,
`
`5:528
`
`Ease
`
`2:: w wv
`5::
`magi: A|_ N2
`
`:22”
`
`$5252..
`
`6
`
`

`
`US. Patent Aug. 20,1985
`
`Sheet60fl0
`
`4,536,809
`
`D/A'58 OUT PUT-MlSPOSlTION CORRECTION SIGNAL
`
`+
`
`m
`
`,,
`
`Fig. 6
`
`7
`
`

`
`US. Patent Aug. 20, 1985
`
`Sheet7ofl0
`
`4,536,809
`
`‘Fig. 7A
`
`A~C
`
`PONER ON. NEw CARTRIOOE?
`
`)
`
`INITIALIzE MATRIx Y[m.n] = O_
`‘
`NUERE Ym - MIs-POsITION ERROR CORRECTION
`INFORMATION F
`TRANsOUCER "m" ANO sAMPLE POINT "n"
`
`MEAsURE BIAs FORCE ERROR ANO COMPUTE CORRECTION
`TABLE PM As A FUNCTION OF RAOIAL TRACK POsTIOT-L"'b"
`
`r,
`I
`|
`I
`
`l
`_,
`
`5“
`
`SELECT ANO MOvE TRANSDUCER
`"m" TO POSITION MEAsUREMENT TRACK
`i
`APPLY MIs-POsITION CORRECTION SIGNAL
`Em - Y[rn.n] ANO FLO) T0 sERvO CONTROLLER
`i
`MEASURE LOw-PAss FILTEPEO ANALOG MIs-POsITION ERROR
`FN/ SIGNAL Xm AT N SAMPLE POINTs ABOUT DISK. AVERAGE OYER
`sEvERAL REvOLUTIONs. DIGITIZE Xm. sTORE IN
`TEMPORARY ARRAY Xm[n].
`
`6.‘,
`
`H‘\-
`
`CORRECT ELEMENTS OF X [n]
`FOR AID CONVERTER OFF ET.
`Y
`AOJUsT XmLNJ FOR OEsIREO PHASE LEAO TO
`COMPENsATE FOR CONTROL sYsTEM LAG. ANTI-ALIAsINO
`FILTER LAO. ANO sAMPLE MEASUREMENT LAG.
`i
`UPOATE X [n] BY
`ADDING YEmm'T TO Xm[n]'.
`i
`CONvOLvE X,?[n] HITH h(n)'T0 COMPUTE NEN MIs-POsITION
`CORRECTION SIGNAL Em ANO sTORE REsULTs IN Y[m.n].
`
`\
`
`J-
`
`C
`
`ITERATION COMPLETE FOR TRANsOUCER "m"?
`
`)‘N2
`
`r
`
`(
`
`.E
`
`has
`.
`ITERATION COMPLETE FOR ALL M TRANsOUCERs? JEL '
`has
`
`8
`
`

`
`US. Patent Aug. 20,1985
`
`Sheet80fl0 I 4,536,809
`
`T
`READ/WRITE OPERATION?
`NNO
`
`YES
`
`1
`
`,
`'
`HAvE CONDITIONS CNANCED? E.G. TIME.
`TEMPERATURE. ERROR RATE sENsOR. OFF TRACK SENSOR?
`NO
`SELECT TRANsDuCER "m".
`
`MOVE SELECTED TRANSDUCER
`"m" TO SELETED DATA TRACK.
`
`T
`
`7L
`
`]/’'M
`
`/N
`
`_/'0
`ENTER TRACK FOLLONINO MODE WHILE APPLYING
`STORED CORRECTION SIGNAL Em AND i-‘(bHo sERvO CONTROLLER.
`
`PERFORM READ/WRITE OPERATION.
`
`F 1'9. 78
`
`9
`
`

`
`US. Patent Aug.20, 1985
`
`ShOet9ofl0
`
`4,536,809
`
`Fig. 8A
`
`FIRsT TIME
`BIAS-FORCE
`COMPUTATION
`ENTRY POINT
`
`MEASURE AID CONvERTER OFrsET USING‘ GROUND
`REFERENCE SIGNAL: sTORE AS VA/D.
`i
`INITIALIZE BIAS-FORCE CORRECTION ARRAY FtbJ-O
`e.q.. F(b) - 0. FOR b - I TO B.
`
`SUBSEOUENT
`ITERATION
`ENTRY
`POINT /
`B
`
`CN
`
`FORHARD DIRECTION COHPUTATIONS
`SUD-1
`SEEK TO OUTER CYLINDER POSITION.
`PRIOR TO BAND NO. I .
`1‘
`SET Ffw) - HO)
`
`_
`
`D"\.
`
`,
`E '\.
`
`G\
`
`‘
`
`H
`
`I
`
`SEEK FORWARD TO AND FOLLOW CENTERLINE OF BAND "b"
`HHILE APPLYING BIAS FORCE CORRECTING SIGNAL Ff (b)
`TO D/A CONVERTER OF SERVO CONTROLLER 37
`I L
`APPLY ANTI'STATIC'FRICTION CHIRP'DITHER SIGNAL
`TO SERVO CONTROLLER 37 AND HAIT FOR SETTLING
`i
`MEASURE. WITH A/D CONVERTERK. LOrhJ'PASS FILTERED POSITION
`ERROR Xnk . COMPUTE X - g “2:; Xnk EOR N DIscRE'TE
`CIRC’UNEERENTIAL POSITIONS AND K REvOLuTIONs 0F DIsN
`TO REMOVE A.C. COMPONENT FROM ERROR SIGNAL Xnk.
`i
`NORMALIZE X AND MORE IN F?a)
`e.q. Ff(b) - Ff(b) + X'INK
`CORRECT Ff (b) EOR AID CONvERTER oEEsET
`e.q. Ff(b) " F?bTs VA/D
`v
`TEST IXINK - vmlsa '2
`IF NOT. ITERATE AGAIN.
`has
`Is LAST BAND 8 IN EORNNRD DIRECTION COMPLETED?
`IE NOT. sET b - b ~ 1
`
`I
`NO
`
`No
`
`‘YES
`
`10
`
`

`
`U.S. Patent Aug. 20, 1985
`
`Sheet 10 oflO 4,536,809
`
`Fig. 8B
`J/
`
`+
`REARNARD DIRECTION COMP uTA‘TIONs
`SET b' B
`
`'
`
`SET Fr(b) - m)
`
`K\
`
`L\
`
`M
`W
`
`SEEK REARNARD To AND EOLLON CENTERLINE OF BAND "1;"
`NHILE APPLYING BIAs FORCE CORRECTING SIGNAL Frl(b) TO
`D/A CONVERTER 0F SERVO CONTROLLER 37
`I
`APPLY ANTI-STATIC FRICTION CHIRP'DITHER SIGNAL
`TO SERVO CONTROLLER 37 AND WAIT FOR SETTLING
`I
`MEASURE. WITH A/D CONVERTER. LON'PASS FILTERED POSITION
`K I N
`
`\ ERROR X"k . ?oMPuTE X - g! T; Xnk FOR N DISCRETE
`
`CIRCuME ERENTIAL POSITIONS AND K REvOLuTIONs 0F DISK
`TO REMOVE A.C. COMPONENT F ROM ERROR SIGNAL Xnk .
`
`0
`\
`
`p
`{
`
`Q{
`
`NORMALIzE X AND STORE IN F, ('s)
`E .G. F,
`= F (b) + X INK.
`CORRECT Fr (b) FOR Alb CONVERTER OFF sET
`E .6. Fr (b) = Fr (b) — VA,D
`
`'
`
`TEST lX/NK -
`s1 ?
`IF NOT. ITERA?/DA'GAIN.
`lYES
`Is LAsT BAND b = 1 COMPLETED?
`IF NOT. sET b= b- 1'
`
`No
`
`NO
`>'_
`
`R~[
`
`5 ~
`
`1
`
`‘YES
`E
`COMP uTE BIAs FORCE ERROR CORRECTION SIGNAL F(b)
`+
`HO) - I l2(F (b) + F' (b) )
`FOR b- 110 B ‘10 AVERAGE Rh)
`i
`R TO EACH F(b) IN ARRAY F[b]
`APPLY 3-POINT SMOOTHIN
`TE
`~ F (b4 ) IN. FOR b - 2 TO (B-I )
`E .G . F(b ) - F (b-l )IH P F
`AND MORE RESULTS IN BIAs FORCE STORAGE ARRAY F [b] 206
`
`J
`
`U4
`
`BIAS FORCE COMPUTATION COMPLETE
`
`J
`
`11
`
`

`
`ADAPTIVE MISPOSITION CORRECTING
`METHOD AND APPARATUS FOR MAGNETIC
`DISK SERVO SYSTEM
`
`25
`
`CROSS REFERENCE TO RELATED PATENT
`APPLICATIONS
`This invention is related to commonly assigned US.
`patent application Ser. No. 06/376,971 entitled CON
`TINUOUS-PLUS-EMBEDDED SERVO DATA
`POSITION CONTROL SYSTEM FOR MAGNETIC
`DISK DEVICE, ?led on May 10, 1982, by the same
`inventor hereof, the application being incorporated
`herein by reference.
`BACKGROUND OF THE INVENTION
`This invention pertains to a servo control system for
`use in a magnetic disk storage device. More speci?cally,
`the invention concerns a servo control system that
`adaptively corrects dynamic and static alignment errors
`between a data head and a data track centerline on a
`data disk in the storage device.
`Disk storage devices are used in data processing sys
`tem for storing relatively large amounts of information
`which can generally be accessed within milliseconds.
`Structurally, a typical storage device comprises a rotat
`ing magnetizable disk medium having several surfaces,
`in the form of an asssembly of one or more stacked
`platters, on which data is magnetically sensed and/or
`recorded in addressable sectors located on circular data
`track centerlines. The disk assembly is mounted upon a
`drive spindle in the storage device that rotates it at a
`constant speed, about 3600 revolutions per minute. The
`storage device also includes one or more transducers, or
`read/write heads, associated with each surface of the
`disk. The transducers are mounted in spaced relation on
`an arm of a movable transducer carriage. A servo con
`troller actuates the carriage in a controlled fashion to
`move all the data heads in unison radially over the disk
`surfaces thereby to position any one of the data heads
`over a selected track centerline. Since all of the data
`heads on the carriage move together, the device also
`includes control circuitry that selects one of the read/
`write heads to perform a data transfer operation.
`The servo controller responds to commands from the
`data processing system. The controller does this by
`transforming these commands into an analog servo
`signal which ultimately drives, usually through a power
`ampli?er, an electro-mechanical actuator that connects
`to the transducer carriage.
`Typically, the disk device operates in one of two
`different modes. The ?rst is a “seeking” mode in which
`the magnitude of the servo signal is used in a controlled
`fashion to drive the carriage, and thus the selected data
`head, to travel to the vicinity of a desired circular track
`centerline; once the data head reaches that vicinity, the
`system is then switched to a‘ second or “track-follow
`ing” mode. In the track-following mode, the position of
`the actuator is controlled to cause the center of the
`selected data head to align itself with the centerline of
`the data track. However, even in this mode there exists
`a ?nite alignment error between the center of the data
`head and the selected track centerline. The magnitude
`of this alignment error places an upper limit on the data
`track density and thus, on the data storage capacity of
`the storage device.
`To minimize alignment error, servo systems typically
`employ formatting information prerecorded on the data
`
`45
`
`65
`
`1
`
`4,536,809
`
`2
`disk to allow the controller to detect the displacement
`between the data head and the track centerline. A pre
`ferred format might include servo data that is continu
`ously prerecorded along servo tracks on a dedicated
`surface of the disk assembly (“dedicated” servo data),
`together with servo data that is prerecorded in circum
`ferentially spaced servo sectors interspersed, or embed
`ded, between adjacent pairs of storage data sectors on a
`data surface of the disk assembly (“embedded” servo
`data). Dedicated servo data is read by a read-only servo
`head, while embedded servo is read along with the data
`by a read/ write head and thereafter separated from the
`data by servo data processing circuitry.
`The servo data from both the dedicated and data
`surfaces is decoded by the disk controller, thereby en
`abling it to modify a servo control signal, if necessary,
`and thus continuously maintain the position of the data
`head in alignment with a selected data track centerline.
`Several factors, however, limit the alignment accuracy,
`and thus the maximum attainable data track density, of
`a disk storage device. The most common of these fac
`tors stem from electrical and mechanical disturbances
`or noise. D.C. Bias forces and electrical offsets are ex~
`amples of some disturbances. A most notable mechani
`cal disturbance is spindle “runout”, or “wobble”, which
`is the difference between the actual centerline of a track
`and the effective centerline presented to a head posi
`tioned a ?xed distance from the mounting center of the
`disk. It is typically caused by slight eccentricity in the
`mounting of the disk on its drive spindle. Runout is
`more prevalent in disk systems using exchangable disk
`cartridges and results from even the slightest off-center
`mounting (e.g., a fraction of a thousandth of an inch), as
`well as from slippage or tilt in seating of the disk car
`tridge after mounting. Carriage play between the trans
`ducer carriage and its guide rods, as well as misalign
`ment due to uneven thermal expansion of the carriage,
`arms, disk, or transducers, further contribute to the
`mechanical disturbances. Generally, positioning toler
`ances should be within i 10% maximum of track pitch
`(e. g., spacing between adjacent track centerlines). Thus,
`for example, a 1000 track-per-inch servo system should
`maintain a data head within ilOO micro-inches of a
`data track centerline. With typical currently available
`exchangeable disk systems, such alignment accuracy is
`not readily attainable.
`Control system lag is another factor that affects posi
`tioning accuracy. Lag is the time delay between the
`time that the controller detects an off-track condition
`and the time that the actuator begins to move the trans
`ducer into alignment with the data track centerline.
`Some of this delay is attributable to the electrical re
`sponse characteristics of the servo control system, such
`as, for example, that resulting from a low sampling rate;
`the remaining delay is attributable to the mechanical
`response characteristics of the electromechanical actua
`tor. These delays characterize the “bandwidth” of the
`servo control system. The greater the bandwidth, the
`faster the positioning system can respond to an off-track
`condition thereby providing tightly controlled position
`ing of the data head. A positioning system having high
`bandwidth provides increased data track density be
`cause centerlines can be followed within a smaller toler
`ance. There are other factors, as well, that contribute to
`misalignment during track following operations.
`Conventional methods of increasing servo bandwidth
`include increasing the frequency of structural mechani
`
`12
`
`

`
`4,536,809
`4
`3
`noise reduction and iteration are not used in the Jacque
`cal resonances, providing continuous position feedback
`system.
`from a dedicated servo surface, and providing a higher
`sample-rate position feedback emanating from the data
`Moreover, none of the aforementoned schemes at
`tempt to correct for do bias-force positioning offsets
`surfaces, among others.
`that may even be a non-linear function of the radial
`position of the carriage.
`SUMMARY
`In view of the foregoing, an objective of this inven—
`tion is to improve track following accuracy and thereby
`increase the maximum attainable data track density in a
`magnetic disk storage device by dynamically correcting
`transducer alignment errors resulting from cyclic mis
`position errors and electrical and mechanical disturb
`ances in the servo system.
`Another objective of this invention is to provide a
`misposition error correcting system that is less suscepti
`ble to noise when iterating the misposition error signal.
`Another objective of this invention is to provide a
`misposition correcting system that iteratively and rap
`idly converges to an optimum misposition error correc
`tion signal.
`In furtherance of these objectives, one aspect of this
`invention comprises a servo control system that periodi
`cally and iteratively accesses embedded servo informa
`tion prerecorded on circular data track centerlines on a
`data surface of the magnetic disk for measuring a mispo
`sition error thereby to generate a set of misposition
`correcting signals that are subsequently applied to the
`servo actuator to dynamically correct transducer mis
`alignment during a subsequent read/ write operation.
`In the preferred structure, the data disk contains ?rst
`and second sets of high-frequency bursts of servo sig
`nals in servo sectors on each data surface thereof, the
`bursts in each set being recorded in alternate track loca
`tions at centerlines shifted radially by the width of one—
`half track with respect to the centerlines of the storage
`data tracks in the data sectors. A demodulator detects
`the bursts, and compares their magnitudes to generate a
`position-error signal. The sum of the magnitudes of the
`bursts is normalized to insure an accurate re?ection of
`the positioning error. The misposition error signal is
`measured through an optional low-pass anti-aliasing
`?lter during at least one complete rotation of the disk
`and then converted to a digital signal. Measurements
`are made at a plurality of discrete equally-spaced cir
`cumferential positions about the disk.
`A matched digital ?lter adjusts the phase of the digi
`tized misposition error signal to compensate for known
`servo control system lag and known low-pass ?lter lag;
`and also adjusts phase lead and gain terms of the funda
`mental-frequency and selected harmonics thereof, thus
`allowing for the rejection of high-frequency harmonics.
`Thereafter, the digital ?lter generates a correction sig
`nal from the phase-corrected misposition error signal.
`The system iterates the aforementioned process on one
`or more subsequent disk rotations by re-applying the
`correction signal to the servo system while again mea
`suring misposition errors to generate a subsequent opti
`mum correction signal. It then stores these optimum
`correction signals for later use during subsequent read/
`write operations.
`Misposition error information can be stored for each
`data head in a multi-transducer, multi-platter, disk sys
`tem. Thus, when subsequently accessing a storage data
`track, the system selects an associated set of correcting
`signals. Moreover, any number of iterations can be
`performed to enable the system to converge to an opti
`
`DESCRIPTION OF PRIOR ART
`One approach for overcoming some of the effects of
`the electrical and mechanical disturbances has been to
`improve the tolerances of the mechanical and electrical
`circuit components of the servo system, but this is an
`expensive proposition and is only marginal at best in
`solving the problem. Thermal compensation networks
`have also been used to reduce head misalignment result
`ing from uneven thermally induced dimension or posi
`tion changes of the mechanical components. This ap
`proach only partially corrects misalignment errors of
`the transducer because it is based on a model that at
`tempts to correct only some of the average offset errors,
`but not the runout errors.
`Another approach for improving head alignment has
`been to provide sectorized, i.e. embedded, servo posi
`tioning data directly in the storage data track. This
`approach has been used as an alternative to, as well as a
`supplement to, providing servo positioning information
`on a dedicated surface of a disk. The controller uses the
`sectorized servo data to update its servo signal between
`the passage of successive data sectors. US. Pat. No.
`4,208,679 assigned to the assignee hereof, describes such
`a system. This approach, however, does not overcome
`bandwidth limitations of the servo system and thus
`cannot compensate for control system lag. In fact, a
`system employing solely embedded sectorized servo
`data has a slightly reduced bandwidth due to a time-lag
`between sampling of the servo data in the servo sectors.
`A more recent approach to improving track follow
`ing operations has been to dynamically modify the
`servo control signal with a supplementary correction
`signal during a read/write operation. A “misposition”
`error correction servo signal counters a previously
`measured, cyclic off-center transducer alignment error.
`This signal can be derived with the aid of one or more
`prerecorded position-reference tracks on the disk. It
`operates by measuring off-center track misalignment
`when a transducer is positioned at a radially ?xed, sta—
`tionary reference point over the position-reference
`track on the rotating disk. Misposition errors associated
`with various circumferential positions are sampled and
`stored, and later recalled and applied to the servo con
`' troller during a subsequent read/write operation. US.
`Pat. No. 4,136,365 issued to Chick et al. describes such
`a system. This system, however, lacks the use of phase
`compensation and noise reduction techniques. It also
`does not employ an iterative procedure to re?ne the
`misposition error measurements. Without noise reduc
`tion, successive iteration of measured runout, if per
`formed, cannot be employed to attain an optimal runout
`rejection correcting signal.
`US. Pat. No. 4,135,217 issued to Jacques et al. de
`scribes another system that modi?es a servo position
`signal with a misposition error correction signal.
`Jacques, et al. derive their misposition error signal from
`a coarse positioner on the transducer carriage, rather
`than from the disk medium itself. It does not allow
`measurement and correction of the alignment errors
`experienced at the data head, and could give false cor
`rection information as static and dynamic errors can be
`quite different for each data head in the system. Again,
`
`60
`
`65
`
`35
`
`45
`
`55
`
`13
`
`

`
`6
`The above and further objectives and'advantages of
`this invention will become apparent by reference to the
`following detailed description taken in conjunction
`with the accompanying drawings in which:
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`FIG. 1 is a block diagram of an illustrative servo
`positioning system helpful in explaining the invention;
`FIG. 2 is a block diagram depicting a closed-loop
`position control system which characterizes the inven
`tion;
`FIG. 3 depicts a portion of a data disk from which
`servo positioning information is derived;
`FIG. 4 is an enlarged illustration of a portion of the
`disk of FIG. 3;
`FIG. 5 is a block diagram showing the misposition
`measurement and computation circuitry of the inven
`tion in detail;
`FIG. 6 shows a control signal wave form generated
`by the chirp-dither signal generator of FIG. 7.
`FIGS. 7A and 7B are flow charts of operations per
`formed by the servo control system of FIG. 1 in gener
`ating a misposition correction signal; and
`FIGS. 8A and 8B are flow charts of operations per
`formed by the system of FIG. 1 in generating a bias
`force error correcting signal.
`
`5
`
`20
`
`25
`
`4,536,809
`5
`mum misposition error correction signal and allow
`more rapid optimization of a misposition correcting
`signal. This gives the system the ability to rapidly adapt
`to slow-time-varying changes in runout.
`In accordance with yet another aspect of this inven
`tion, the positioning system additionally includes cir
`cuitry for correcting d.c. offset errors resulting from
`electronic offsets and bias forces (or torques in a rotary
`actuator) which act to radially offset the movable trans
`ducer carriage from the desired track location. In this
`regard, the system measures misposition, as stated pre
`viously, and then calculates offset corrections for bias
`force and dc. errors at various, preferably at equally
`spaced, radial positions of the carriage. The system then
`stores them in a memory. Whenn a storage data track
`centerline is subsequently accessed during a read/write
`operation, the stored bias force or do. offset correction
`information is recalled and applied to the actuator, just
`as the other misposition error correction information, so
`that both the misposition and the offset correction sig
`nals contribute to the correction of alignment errors.
`To generate the bias force error correction signal,
`again in the preferred embodiment, offset errors are
`iteratively measured in the forward and reverse direc
`tion for each data track centerline by positioning a
`transducer thereover, average offset error over several
`revolutions of the disk to reject A.C. disturbances, reap
`plying the offset error to the transducer positioner and
`again measuring offset error until a predetermined mini
`mum offset is measured, and then storing the accumu
`lated offset error for each small group of adjacent track
`centerlines or “bands”. The forwardly and rearwardly
`measured offset error can further be averaged, and the
`resulting correction table that corrects for even non-lin
`ear radial variations of do. offset errors for each band
`can be digitally-?ltered and thus smoothed. This proce
`dure permits compensation for striction and friction
`effects, and also reduces measuring and computing er
`rors. Because offset errors equally affect each data head,
`the offset error of only one head, preferably a dedicat
`ed-servo head, need be measured. The correction data is
`thus based on radial position and is independent of he
`selected data head.
`Certain other advantages, not possible with the Chick
`et a1. or Jacques et al. systems, can be gained by process
`ing the measured position error signals to produce the
`dynamic correction signals. Speci?cally, the phase of
`the stored misposition information can be shifted to
`compensate for control system lag and ?lter lag. Once
`obtained, the correction signal is re-applied to the servo
`system to obtain an iterative misposition correcting
`signal. This and yet further iterations enable the system
`to converge to a more perfect misposition correction
`signal.
`The fact that position error is cyclic with disk rota
`tion is utilized to advantage by employing a matched
`?lter to better reject noise and undesired harmonics.
`Digital signal processing enables the fundamental-fre
`quency and selected harmonic frequencies of the cyclic
`error signal to be independently rejected and/or ad
`justed in gain and phase. Without employing a matched
`?lter technique for noise reduction, iteration cannnot be
`conveniently performed because certain of the high-fre
`quency components of the error signal begin to add
`without limit with each iteration.
`This invention is pointed out with particularlity in the
`appended claims.
`
`30
`
`35
`
`45
`
`55
`
`60
`
`DESCRIPTION OF A PREFERRED
`EMBODIMENT
`The servo control circuit diagram of FIG. 1 uses
`embedded servo data recorded on a data surface of a
`disk to determine alignment errors. Generally, the sys
`tem includes a rotating disk assembly in the form of
`stacked platters 14, 16 and 18 which rotate on a spindle
`20. Each platter 14 through 18 may store data on each
`side thereof in circular data track centerlines. A trans
`ducer carriage 22 carries a plurality of read/ write heads
`24, 26, and 28 for the upper disk surfaces (and corre
`sponding heads, not shown, for the lower disk surfaces)
`on respective carriage arms, 25, 27 and 29 which radi
`ally position these read/write heads over circular track
`centerlines on upper or lower surfaces of the disk plat
`ters 14 through 18. One of the data heads is preferably
`a read-only servo head positioned over a dedicated
`servo surface of the disk to supply supplementary high
`frequency positioning information to the servo control
`system.
`For purposes of illustration, the transducer carriage
`22 is shown as supported by guide rods, one of which is
`shown as guide rode 30, and is actuated by a linear
`actuator 31 comprising a ?xed stator or ”?eld” coil 32
`which drives a lightweight moving coil 34 that is con
`nected to carriage 22. In response to signals applied to
`coil 34 from a power ampli?er 36, carriage 22 moves
`along guides 30 to position the heads 25-29 over a de
`sired track. In the “seeking” mode of operation, track
`seeking circuitry 39 counts crossings of data track cen
`terlines in order to advance, slow down, and stop the
`carriage when it arrives in the vicinity of the selected
`data track centerline.
`A velocity estimator 41 responds to carriage motion
`by providing to a servo control circuit 37 signals indica
`tive of the rate at which the heads are moving across the
`disk. Velocity control in the seek mode is used for pro
`?ling the speed of the carriage in a conventional manner
`during a seeking operation so as to decrease the carriage
`speed as it approaches the desired track, thereby to
`
`14
`
`

`
`5-1 5
`
`25
`
`4,536,809
`8
`7
`avoid overshoot and thus to expedite settling of the
`56 itself comprises a storage matrix 200 and a storage
`array 206 (see FIG. 5). Each entry is tagged with its
`read/ write head over the centerline of the track.
`associated data head, track number and sector number
`When the selected data head reaches the vicinity of
`corresponding to the HEAD SELECT, TRACK NO.,
`the desired data track (e. g., one or two tracks away), the
`and SECTOR NO. signals from the data processing
`control system enters a “following” mode and develops
`system so that when they are recalled during a subse
`a “?ne” error signal corresponding to the distance of
`quent read/write operation, the corresponding correc
`the data head from the track centerline. It uses the sig
`tion signal is combined with the POSITION-ERROR
`nal in a negative feedback arrangement to minimize the
`SIGNAL in the adder 44 prior to being supplied to the
`error and thus maintain the data head in alignment with
`the centerline. It is these aspects of control in the head
`control circuit 37. A D/A converter 58 converts to an
`analog signal the stored digital correction data con
`following mode to which this invention relates.
`In order to perform this servo control function, the
`tained in the array 56. ’
`The adaptation sequencer (which may comprise sim
`circuitry of FIG. 1 may use embedded servo data that is
`ply a periodic timer, which periodically emits a pulse to
`interleaved with data sectors and prerecorded in spaced
`cause recomputations in processor 50) activates the
`servo sectors in the data tracks. To extract the servo
`information, a head select and ampli?er circuitry 38
`computation processor 50 in response to a variety of
`conditions to compute desired misposition correction
`responds to a head select signal from a data processing
`device to select one of the possible data heads that is
`data, e. g., in response to detection of an off-track condi
`tion by a sensor 52 in the track following mode during
`carried by the transducer carriage 22. The read/write
`circuitry 40 and servo signal demodulator 42 separate
`a read/write operation, or in response to detection of
`errors by a data (e.g., a cyclic redundancy check error
`the read/write storage (“user”) data and servo data in
`response to a sector synchronizing signal that is gener
`detection cir