CA1268857A

CA1268857A - Digital servo control system for a data recording disk file

Info

Publication number: CA1268857A
Application number: CA000532018A
Authority: CA
Inventors: Michael L. Workman
Original assignee: International Business Machines Corp
Current assignee: HGST Netherlands BV
Priority date: 1986-04-29
Filing date: 1987-03-13
Publication date: 1990-05-08
Also published as: JPS62257682A; DE3782448D1; SG147694G; EP0243821B1; JP2589689B2; EP0243821A2; DE3782448T2; EP0243821A3; US4679103A; HK138694A

Abstract

Abstract of the Disclosure A digital servo control system for a data recording disk file receives as input a digital head position error signal (PES) and a digital signal corresponding to the head actuator input signal, and outputs a digital control signal.
The control signal is integrated and amplified to provide the input signal to the actuator. The control signal is calculated from estimated values of absolute head position, head velocity and the equivalent actuator input signal required to compensate for constant or very low frequency bias forces. The estimated values are functions of respective predicted values, which in turn are functions of, among other things, constants representing the digital sampling time and physical parameters of the actuator. The updating of the estimated values from the respective predicted values is made with the use of an error term, which is the error between the measured absolute position and the predicted absolute head position. The predicted absolute head position is made sufficiently accurate that the error term is always less than a selected number of tracks. This permits the digital servo control system to operate during track seeking without counting track crossings so that the system is operable with servo information recorded either in sectors on a data disk or in continuous tracks on a dedicated servo disk.

Description

- ~L2~3857 1 ~ DIGITAL SERVO CONTROL SYSTEM
FOR A DATA RECORDING DIS~ FIL~

Technical Field This inYention relates to servo control systems for read/write head positioning in data recording disk files.
More particularly, the invention relates to a single digital servo control system for ~oth track seeking and track following which is usable with disk files having servo information on either a dedicated servo disk surface or in sectors on a data disk surface.

Back~round of the Invention .
Disk files are information storage devices which ntilize a rotatable disk with concentric data tracks containing the information, a head for reading or writing data onto the various tracks, and an actuator connected bv a support arm assembly to the head for moving the head to the desired track and maintalning it over the track centerline during read or write operations. The movement of the head to a desired track is referred to as track accessing or ~seeking", while the maintaining of the head over the centerline of the desired track during a read or write operation is referred to as track "following".
The actuator is typically a "voice coil motor" (VCM) which comprises a coil mova~le through the magnetic field of a permanent magnetic stator. The application of current to the VCM causes the coil, and thus the attached head, to move radially. The acceleration of the coil is proportional ~o the applied current, so that ideally i~3~S~

1 there is no current to the coil if the head is perfectly stationary over a desired track.
In disk files which have a relatively high density of data tracks on the disk, it is necessary to incorporate a servo control system to maintain the head precisely over the centerline of the desired track during read or write operations. This is accomplished by utilîzing prerecorded servo information either on a dedicatea servo disk or on sectors angularly spaced and interspersed among the data on a data disk. The servo information sensed by the read/write head (or the dedi-cated servo head if a dedicated servo disk is used) is demodulated to generate a position error signal (PES) which is an indication of the position error of the head awa~ from the nearest track centerline. One type of conventional servo pattern for use with either a sector ser~ro disk or a dedicated servo disk is the quadrature pattern described by Mueller, et al. in IsM Technical Disclosure Bulletin, ~ol. 21, No. 2 (February 1978) pp.
8n4-805. In the quadrature pattern there are four unique track types which repeat so as to form radially repetitive four-track bands of servo information.
During track seeking, when the head is moving across the tracks, the PES is used to gener~te track crossing -pulses. This track crossing information, together with the PES and a signal representing the desired or target track, is used to generate a total error signal. The total error signal is equal to the sum of the PES plus the difference between the position of the target track and the position of the actual track over which the head is located. The total error signal is then used in the 'J~7 ser~o feedback loop ~o compute the desired velocity of the head, via a reference velocity trajectory generator, to assure that the head arrives at the target track according to the optimum velocity trajectory to effect a move to the target track in minimum time. The computed velocity is then compared with the estimated velocity from an electronic tachometer to generate a velocity error signal to a power amplifier which in turn applies a control current to the VCM. An electronic tachometer for generating a velocity estimate from inputs of VCM control current and quadrature PES is described in U.S. patent 4,246,536 to Bradley, et al., which is assigned to the same assignee as this application. In some disk files the estimated velocity is provided by a mechanical tachometer, such as a linear variable differential transformer (LVDT).
During track following, when the head is located within the boundaries of the desired track, the PES alone is used in the servo feedback loop to generate a control signal to the VCM to move the head back to the track centerline.
A description of operation of a general disk file servo control system during track seeking and track following is given by R. K. Oswald in "Design of a Disk File Head-Positioning Servo", IBM Journal of Research and Development, November 1974, ppO 506-512.

~ . .. .
In such conventional disk files, the use of track crossing pulses to determine the total error signal durinq a track seek requires additional costly and complex analog circuitry in the demodulator to generate the track crossing pulses from the PES. More importantly, in the case of disk files which use sector B~

1 servo data, it is not possible to accurately count track crossing pulses directly from the PES since the head will tvpically have crossed numerous tracks between PES
samples.
During operation of the disk file the VCM is subjected to certain generally constant or very low frequency forces which tend to bias the coil, and thus the attached head, away from the desired position, i.e.
the position needed to follow the desired velocity trajectory during track seeking, or the position of the centerline of the desired track during track following.
These bias forces are caused by such things as circulating air flow passing the coil and the arm assembly which connects the head to the coil, the movement of the flexible ribbon conductor connecting the coil and arm assembly to the read/write electronics of the disk file, friction between the bearings and guide rails of the actuator, and the gravitational effect on the coil and arm assembly if the radial access direction of the VC~ is not perfectly aligned horizontally. In addition to bias forces, there is typically an input voltage offset in the power amplifier which supplies the control current to the VCM, so that even in the absence of a velocit~ error signal to the power amplifier, the VCM may e~perience a non-~ero input current.
The bias forces and voltage offset have little effect on the performance of the servo control system during track seeking when the primary contribution to the total force on the actuator is that due to the current in the actuator. However, at ~he completion of a track seek and during track following, when the head is within SA984-03r Page 4 1 approximately one-half track width of the centerline of the target track and when the total error signal is simply the PES, the bias forces and voltage offset are primary contributions to the PES and must be rapidly and effectively compensated by the control signal to the power amplifier to produce current in the VCM to maintain the head over the track centerline.
One technique for rapidly forcing the total error signal to zero during track following is to incorporate an integrator for the PES into the servo feedback loop.
~his is accomplished by the use of a separate "stiffness"
integrator which is switched into the servo feedback loop onl~r during track following. However, use of the switchable stiffness integrator generates an electrical transient which occurs as a result o~ switching of the integrator when large system bias forces and voltage offset is present. In addition, the stiffness integrator must be switched in at precisely the proper moment. Both the transient caused by switchir.g of the stiffness inte~rator and switchin~ at the improper time can cause overshoot or undershoot of the head with respect to the centerline of the target track which thus delays the subsequent read or write operation.
An additional problem caused by bias forces in all disk files is that there is a gain in the feedback path around the power amplifier in the presence of a generall~
constant or very low frequency PES. In the case of disk files which use an electronic tachometcr, this is because the electronic tachometer receives as input the VCM
control current and the measured PES and, through a combination of control current integration and PES

SA984-03'~ Page 5 `` 12~

1 differentiation, estimates the velocit~ of the head.
Thus, even if the head is stationary, but the coil and arm assembly is subjected to generally constant or very low frequency bias forces, the electronic tachometer will output a non-zero velocity estimate. In the case of disk files which use a mechanical tachometer, electrical offsets cause a non-zero output of estimated velocity even when the head is stationary.

Sum ary of the Invention The digital servo control system of the present invention receives as input, at discrete sample times, the PES and a digital value representative oE the VCM
current, and outputs, at discrete sample times, a digital control signal. The digital control signal is converted to an analog control signal and input to an integrating pot~er amplifier which prov~des the control current input to the VCM.
The digital control signal is generated by a microprocessor attached by a data bus to associated memory devices. As part of the computation of the control signal for each PES sample, the microprocessor computes the measured absolute head position from the PES
and an estimate of VCM current equivalent to the amount of current required to compensate for bias forces acting on the actuator. In addition, the microprocessor computes an estimate of the absolute head position relative to a fixed reference and an estimate of the head velocity. A digital velocity trajectory generator provides a value representative of the optimum velocity 68~5t7 .

l of the head as a function of the difference between the absolute position of the target track and the estimated absolute head position. A velocity error is then computed as the difference between the computed estimate of head velocity and the optimum velocity from the velocity trajectory generator. The microprocessor also computes an estimate of head acceleration from the measured VCM current and the estimated bias force eauivalent VCM current. The estimated head acceleration, together with the velocity error and a value representative of the control siqnal for the prior PES
sample, are used to form the digital control signal.
The estimated absolute head position, head velocity, and bias force equivalent VCM current are computed from respective predicted values of those ter~s. The predicted values are updated with each PES sample and are functions of, among other things, constants representing the digital sampling time and physical parameters of the VC~5, such as frictional effects and force constants. The 20 - estimated values of these terms are computed with the use of an error term which is the difference between the computed measurement of absolute head position and the predicted a~solute head position.
The estimation of head acceleration is made such that, durinq track following when the head is yenerally ~rithin one-half track of the centerline of the tar~et track, the average value of the estima-ted head acceleration is zero. Thus when the head is relatively stationarv and the VCM is subjected to bias forces, there is no feedback path around the integrating power SA984-03~ Pag~ 7 ~ 5t~

1 amplifier which has a gain in the presence of bias forces.
With the use of the digital servo control system according to the present invention, it is not necessary to measure or estimate track crossings so that the system is fullv operable with either a dedicated servo disk or a data disk with sector servo information. The prediction of the absolute head position is made with sufficient accuracv such that the error between the measurement of the absolute head position and the predicted absolute head position is always less than a specific number of tracks. For example, when the servo control system is utilized with a quadrature pattern in which the servo information is radially repetitive in four-track bands, it is only necessary to measure the position of the head within a band without knowing in which particular band the head is located.
The digital servo control system also includes means for computing a value representative of a correction signal to the digital control signal in order to compensate for voltage offsets to the integrating power amplifier. The correction signal is applied to the control signal and updated during track following.
For a further understanding of the nature and advanta~es of the present invention, rererence should be made to the following detailed aescription taken in conjunction with the accompanying drawings.

SA9~4-~36 Page 8 35~

1 Brief Description of the Drawings Fig. 1 is a block diagram of a conventional analog servo control system with a stiffness integrator switchable into the system during track following;
Fig~ 2 is a block diagram of the digital servo control system of the present invention;
Fig. 3 is an illustration of servo signals from a quadrature pattern and the manner in which the signals are compared to identify a specific track in one of the four-track bands in the quadrature pattern;
Fig. ~ is a graphical representation illustrating the function of the state estimator and control signal algorithm of the digital servo control system;
Fig. 5 is a flow chart of the state estimator and control signal algorithm;
Fig. 6 is an illustration of head arrival trajectories for the digital servo control system with and without a switchable stiffness integrator in the presence of a large frictional bias force; and Fig. 7 is an illustration of a digital integrator in the digital servo control system to compensate for an input voltage offset to the integrating power amplifier~

Description A. Prior Art The block diagram of Fig. 1 is an illustration of a conventional analog disk file servo control system. A
pair of disks 10, 12 are supported on a spindle 1~ of the disk file drive motor 1~. Each of the disks 10, 12 has . . , . - . ., . . , . .. .. ~ . . . .

8~35~7 1 two surfaces 20, 22 and 24, 26, respectively. For purposes of this descr.iption, surface 20 on disk 10 and surfaces 24, 26 on disk 12 are data recording surfaces.
Surface 22 on aisk 10 is a aedicated servo surface and contains only prerecorded servo information. The servo information is recorded in concentric ~racks and is t~pically written in such a manner that the intersections of adjacent servo tracks on servo surface 22 are radially aligned with the centerlines of the data tracks on surfaces 20, 24, and 26. The servo informatîon on surface 20 may be the quadrature pattern, as described in the previouslv cited reference bv Mueller, et al.
The specific tracks on the data disks and the servo disk are accessed by heads 30, 32, 34, 36, each o which is associated with a respective disk surface and supported by an associated arm assembl~. The heads 30, 32, 3~, 36 are attached to a common accessing means or actuator, such as VCM 40. Thus the heads 30, 32, 34, 36 are all maintained in a fixed relationship with one another relative to the radial position on their respectlve disk surfaces.
The output of the dedicated servo head 32 is supplied to an amplifier ~2 and then to a demodulator 44.
Demodulator 44 processes the servo information signal from disk surface 22 and demodulates it to generate an analog PES. The PES from demodulator 44 is an indication of the position of servo head 32 away from the nearest servo track centerline, and thus the position of the data heads 30, 34, 36 awav from the centerlines of the nearest data tracks on the respective disk surfaces 20, 24, 26.
Demodulator 44 also provides a track crossing pulse which i8~35~

I is input to a difference counter 46 and, if the servo pattern is a quadrature pattern, a signal (not shown) which identifies one of the four specific track types in the quadrature pattern.
The difference counter 46 also receives a command from the disk file control unit (not shown) indicating the target or desired track, denoted as td~ With each track crossing by the servo head 32, the track crossing pulse causes the value of td stored in difference counter 46 to be decremented. The output from difference counter 46 is passed to a digital-to-analog converter (D/A) 48 which provides an analog value representative of the number of tracks to go to reach the desired trac~, td ~ ta, where ta is the nearest actual track position of servo head 32 at the time of the PES measurement.
The generation of track crossing pulses from the PES
requires relativelv costl~ and complex circuitr~ (not sllo~m) in demodulator 44. The servo control system shown in Fig. 1 requires a dedicated servo disk and is not operable with sector servo information since during a track seek the head will typically cross numerous tracks between PES measurements. Thus, if the basic servo control svstem shown in Fig. l is to be operable with sector servo information, then an alternative to direct counting of track crossings must be used to determine the number of tracks to go~ td ~ ta The value td ~ ta from D/A converter 48 is then summed in junction 50 with the PES, the output of summing ~unction 50 being the total error signal representative of the distance-to-go, xtg, to -thc centerline of the desired track.

~ -t7 1 The distance-to-~o is fed to a velocity trajectory generator (VTG) 5~ after addition in summing junction 55 with the output from switch 52. During a trac~ seek, switch 52 is opened and the output of stiffness integrator 53 is reset to zero. The output of summing junction 55 is then only xtg. Within some specified distance from the target track, for example less than one-quarter of a track width, the switch 52 is closed, the reset signal is removed from stiffness integrator 53 and stiffness integrator 53 is allowed to operate. When switch 52 is closed the input to VTG 54 consists of xtg and the integral of xtg. During track following the effect of stiffness integrator 53 on the servo control svstem is to force the a~terage value of xtg (which is eqnal to the PES) to zero.
VTG 54 computes a commanded velocity, vc, based upon the value of ~tg. Conventional analog type velocity trajectory generators are typicallv either diode breakpoint or multiplier circuit implementations, both o~
which are described and shown as prior art in Figs. lA
and lB of U. S. Patent 4,486,797 to ~Iorkman, which is assigned to the same assignee as this application.
Located between VTG 54 and VC~ 40, as shown in Fig. 1, is a compensator 56 and an integrating power amplifier (IPA) 58. The compensator 56 contains circuitry having specific gain and phase shifting characteristics to assure stability and performance o~
the overall closed loop s~stem. The ou-tput o compensator 56 is the control signal, u, which is directed to IPA 58 where the compensated control signal is integrated and amplified to generate a control SA984-03~ Page 12 5~7 1 current, i, to VC~ 40. The control current to VCM 40 is a specific current level t~ move the coil so that the servo head 32, and thus the ~ata heads 30, 34, 36, are moved with the desired velocity tottard the centerline of the desired track.
An electronic tachometer (ETACHJ 60 is located in a feedback pa~h around the compensator 56 and IPA 58.
ETACH 60 receives as inputs the ~C~I control current, i, from IPA 58 and the P~S from demodulator ~. The output of ETACH 60 is an estimate of the velocity of the head.
ETACH 60 provides a signal, ve, based upon a combination of the differentiated PES and integrated control current i. A more thorough aescription of ETACH 60 is set forth in the previously cited patent to Bradley, et al. The result of the incorporation of ETACH 60 in the servo feedback loop is that the input to compensator 56 is a velocity error signal, v 5, which is the difference bet~een the commanded velocity, v , output by VTG 54 and the estimated velocity, ve, output by ETAC~I 60.
2n During track following, if the servo head 32 is stationarv but the VC~ 40 is subjected to bias forces, ETACH 6n continues to provide a feedback path around the IPA 58 and thus contlnues to generate a non-zero velocity estimate, ve.
During a track seek~ stJitch 52 is opened so that stlffness integrator 53 is bypass~d~ VTG 54 computes the velocity signal, vc, as a non-linear function of xtg by a reference curve, as explained in the prior art description of the previously cited patent to Workman.
The total error signal, xt , is the sum of the PES and the computed number of tracks to go/ td ~ ta. However, SA984-03~ Page 13 ~ :, . ...

~` ~Z~ 35'~

1 near the completion of a seek, when the head has approached to within approximately one-quarter track of the centerline of the desired track, switch 52 is closed, thus switching stiffness integrator 53 into the servo feedback loop. At this time the total error signal~ xtg, is simplv the PES since the head is now within the track houndaries of the desired track. Stiffness integrator 53 now rapidlv dri~res the total error signal to ~ero. VTG
54 now provides a velocity command, vc, which is a linear function of the output from summing junction 55.
The switching of stiffness integrator 53 into the servo loop near the termination of a track seek causes a transient in the velocity command, vc, which causes overshoot or undershoot of the head with respect to the centerline of the desired track, thus delaying the subsequent read or write operation.

B. Preferred Emhodiment A simplified hlock diagram of the digital servo control system of the present invention is shown in Fig. ?.. A microprocessor 80 is connected to a random access memory (RAM) 82 and a programmable read only memor~ (PROM) 83 via a data bus 84. The disk ~ile control unit 86 is also connected to data bus 8~.
Control unit 86 provides numerous commands to microprocessor 80, including a signal, tdr representing the target track, and a signal, RZ, representing a "re-zero" to initialize the servo control system. Not sho~m in Fig. 2 are address and control lines for micFoprocessor 80. The analog portion of the servo SA984-03~ ~age 1~

- . - . .. . . ~ .

...D ~

1 control system is shown essentially to the right of data bus 84 in Fig. 2.
The signal read by servo head 32 is input to amplifier 42 and then demodulator 44. While the invention is operable with any of numerous types o~ servo patterns and servo signal demodulation techniques; the ser~jro control system will be e~plained with reference to a quadrature servo pattern. The auadrature pattern on servo surface 22 is demodulated by demodulator ~4 in the following manner. First, demodulator 44 receives the ~uadrature servo signal from amplifier 42 and ~enerates two separate analog waveforms, which are designated primary (PESP) and quadrature (PESQ), as shown in Fig. 2.
The analo~ PESP and PESQ signals from demodulator 44 are sent to analog-to-digital (A/D) converters 88, 89, respectively. The discrete values of PESP and PESQ at any sample time are designated PESP(n) and PESQ(n), where n represents a time index for each digital sample.
PESP(n) and PESQ(n) are then used by microprocessor 80 to determine on which of the ~our tracks in one of the four-track bands of the quadrature pattern the servo head 32 is located. Once that has been determined, then the selection of the correct signal, i.e. either PESP(n) or PES~(n), is made to determine PES(n). The determination of track type, denoted t (n), and PES(n) is made accordinq to the following truth table.
PESP to PESQ Relationship Track T~rpe, t (n) PES~n~
.. . ... ~
rPESP(n) C PESQ(n)~ and [PESP(n) > -PESQ(n)] 0 PESP~n) rPESP(n) C PESQ(n)j and ePESP(n) < -PESQ(n)3 1 PESQ(n3 30 ~PESP(n) > PESQ(n)~ and [PESP(n) < -PESQ(n)¦ 2 -PESPln) [PFSP(n) > PESQ(n)~ and ePF.SP(n) > -PESQ[n~] 3 -PESQ(n) Table 1, SA9~-036 Page 15 ~L2~ 5~

1 The truth tabl~ in Table 1 can also be understood by reference to Fig. 3. Fig. 3A repres~nts the P~SP and PESQ signals output by demodulator 44 as a function of radial position of servo head 32. Figs. 3B and 3C are the digital values of the two comparisons of PESP(n~ to ~/- PESQ(n). ~ig. 3D is the digital track type signal, tq(n), computed from the logic of Table 1. In the preferred embodiment t (n) and PES(n) are computed by microprocessor 80. Alternatively, Table 1 can be implemented in analog hardware and the resulting track type number, t (n), input to data bus 84. Similarly, PESP and PESQ can be selected thru analog switches to qenerate, according to the logic of Table 1, a PES which is then sent to a single A/D converter.
Referring again to Fig. 2, demodulator ~4 also provides directlv to data bus 84 a one-bit digital signal, GBOD, to indicate that servo head 32 is located o~er the "guard band outside diameter", i.e. the radially outermost head position in the disk file. This signal is generated from a special code recorded on the radially outermost track on servo surface 22.
IPA 58 provides the analog control current, itt), to VCM 40 and as feedback to A/D converter 90. A/D
converter 90 provides to data bus 84 a digital current sample, i(n), corresponding to the sample of analog current, i(t).
Thus, as shown in Fig. 2, the data inputs to microprocessor 80 of the digital servo control system for purposes of this explanation are control unit commands for target track td and re-z~ro RZ, head position relative to nearest track centerline PES(n), which is . - . . . . , . . , .. ,.. i :. . , . . . . . ., .~ , ;. .

BBS~
1 determined from PESP(n) and PESQ(n~ according to Table 1, VCM control current i(n), and GBOD.
A di~ital control signal, u(n), is output bv microprocessor 80 to a digital-to-analog converter (DAC) 92. As shown in Fig. 2, A/D converters 88, 89, 90 are driven by the same clock input so that the digital sampling of PESP, PESQ and i, all occur simultaneously.
The output of the control signal u(n) computed from the digital sample of PESP(n), PESQ(n) and i(n) occurs after a fixed calculation delay time and before the input of the next digital sample of PFSP(n+l), PESQ(n~l) and l(n+l). DAC 92 provides the analog control signal, u(t), to IPA 58.
Microprocessor 80 also sends a special inhibit signal to IPA 58. IPA 58 is inhihited when the servo control svstem is first turned on to pre~ent any initial siqnal conditions from causing undesired movement of VCM
~0 .
The simplified block diagram of Fig. 2 is intended onlv to represent possible hardware implementation of the diqital servo control system. An important feature of the digital servo control system which is not shown in Fig. 2, but which will be explained below, is the manner in which the microprocessor 80 co~putes the control signal u(n) so as to eliminate the necessity of a stiffness integrator and the adverse effects of a feedback path around the IPA 58 which has a gain in the presence of bias forces.
The microprocessor 80 uses an algorithm, referred to as a state estimator, to estimate absolute head position xe, head velocity ve, and h~ad acceleration ae. The SA984-03~ Page 17 ~l 3BS7 1 estimated head position, velocity and acceleration for each digital sample n are functions of the predicted head position, velocity and acceleration, the measured head position and the measured VCM current. The predicted head position, velocity and acceleration are updated with each sample and are derived from functions based upon a physical model of the VCM and arm assembly.
The operation of the digital servo control system, and in particular the state estimator, will be explained with reference to the graphical representation of Fig. 4.
It should be noted that the items in the non-analog portions of Fig. 4, such a~s the summin~ junctions, are not discrete hardware components but merely graphical aids to assist in understanding the computations performed by microprocessor 80.
As shown in Fig. 4, the state estimator 100 generates outputs of xe(n), ve(n) and ae(n). The difference ~et~een the tarqet track td from control unit 86 (Fig. 2) and the estimated head position xe(n) is the "distance-to-go", xtg(n), which is represented as the output of summin~ junction 101 and is input to VTG 102.
The input signal representing xtg(n) is operated on in digital VTG 102 to form an output address which is then input to a lookup table stored in PRO.~ 83. The output from PROM 83 is a digitizecl value for the velocity curve which is used directlv as a diqital velocity command output, r (n), from VTG 102. A more thorouqh description of diqital VTG 102 is given in the previously cited Workman patent. Alternatively, VTG 102 may utilize a diqital cur~e implementation, as shown in Fig. lC of the Workman patent. The output of ~TG 102 is a velocity SA984-03~ Page 18 ., . ~ . , . ~, B~,~7 1 command, vC(n), which is a function of the distance-to-go, xtg(n). If the head is within approximately plus or minus one-half track of the centerline of the target track tdr then vC(n) is essentlallv a linear function of the input, xtg(n), and can be represented as klxt (n), where kl is a constant.
The output vC(n) of VTG 102 is a signal representative of the optimum velocity of the head for the particular sample n and distance to the target ~tg(n). This value is then compared, as shown by summing junction 10~, with the estimated velocity v (n) generated by state estimator 100. The difference between vC(n) and ve(n) is the velocity error signal~ veSIn), from summing junction 104.
Located between VTG 10~ and summing junction 104 is a switch 103 for switchinq VTG 102 out of the servo control system. ~hen switch 103 is "switched on"
microprocessor 80 provides a constant velocity, vCOast~
to summing junction 104. Switch 103 is switched on and 'he constant velocity input vcOast is provided during initialization of the servo control system in order to cause the servo head 32 to "coast" at constant velocity from the inside diameter to the guard hand outside diameter. Thus, during the coast phase veS(n) from summing junction 10~ is the difference between vCOaSt and the estimated head velocity vetn).
The value of veS(n) from summing junction 104 is operated on by a lead-lag algorithm, ~hich is represented graphically as a lead-lag network 106, and then input to summing junction 110.

S~98~-036 Page 19 ~Z~:~8~ 7 1 As shown between state estimator 100 and summing junction 108, the estimated head acceleration, a2(n), from state estimator 100 is multiplied by a constant scale factor k2, and the prior control signal u~n-l~ is multiplied bv a constant scale factor k3. These two products are then sum~ed, as shown by summing junction 108, and the sum subtracted from the output of lead-lag network 106, as shown bv summing junction 110. The result is the digital control signal, u(n), which is input to DAC 92.
Located between summinq junction 110 and DAC 92 is a switch 111 for switching the normal control signal out of the servo control system. When switch 111 is n switched on~, microprocessor 80 provides a control signal to DAC
92 which is equal to i(n) - ipark, where ;park is a constant current level. The purpose of switch 111 is to "park" the actuator by moving VCM 40 up against the inside diameter crash stop of the disk file. When the actuator is parked, servo head 32 is located over the radiall~ innermost track.
As described and illustrated by Fig. 4, the control sisnal u(n) is expressed as follows:

u~nJ = -k3u(n~ k2ae(n) + (kS3 + kS2 54 sl ks2 ves(n), (1) ~there dsl(nJ is the state variable of the lead-lag network 106 and is determined bv d (n) = v (n-l) + k 4d l(n-1). ~2) SA984-03~ Page 20 ~6B~S7 , 1 The last two terms in equation (1) are the output ~rom lead-lag network 106 and are strictly a function o~
constant scale factors, k5i, and the present and prior ~elocit~r error signals, veS(n) and veS(n-1), respectively.
The algorithm for computing xe(n), ve~n) and ae~n), and ultimately the control signal u(n), will now be explained with reference to the flow chart of Fig. 5 and, where indicated, the hlock diagram of Fig. 4. This e~planation will begin with the initial "power on~ of the disk file, after which the servo control system is initialized, and then with operation during track seeking and following.
When the disk file is first turned on, the inhibit signal to IPA 58 is on, thus preventing any initial signal conditions to IPA 58 from causing movement of VCM
40. The servo control system then undergoes a "re-zero"
routine and receives the re-zero command, RZ, Erom control unit 86.
If a re-zero command has been received by microprocessor 80, then the servo control svstem's initial function is to park the actuator. The park phase is commenced b~ turning on switch 111. This prevents the otherwise normal control sianal from reaching DAC 92.
The control signal to DAC 92 is no~ equal to the current to VCM 40 less a constant value, i . The current to park VCM ~0 will increase until it reaches the value ipark.
VCM 40 will accelerate until it reaches the crash stop and will be maintained at the crash stop as long as switch 111 is on. The park phase continues for a i8~35~

1 predetermined time delay which is sufficient to assure that VCM 40 has been parked.
After the time delay, the state estimator 100 initializes the estimated values by setting xe equal to xmax, the largest possible absolute head position, and ve and ae to zero. Switch 111 is then turned off, so that the normal control signal u(n~ is provided to DAC 92.
Following the park phase, the servo control system commences the coast phase, the purpose of which is to mo~e VCM ~0 from the inside diameter crash stop to the radiall~r outermost track, i.e. the guard band outside diameter. This is commenced by turning switch 103 on to switch VTG 102 out of the servo control system.
Since the servo head 32 is now moving across the tracks, n is set equal to zero and the sampling of PESP(n) and PESQ(n) begins. GBOD is also input to microprocessor 80 and has an initial value of zero. The coast phase continues so long as both switch 103 is on and GBOD e~uals zero, indicatina that servo head 32 has not yet reached the guard band outside diameter.
Durin~ the first digital sample of PESP(n) and PESQ(n), microprocessor 80 computes the absolute head position, xa, relative to some fixed reference, such as track ~ero, which is typicallv the guard band outside diameter. The term xa(n) is computed according to the ollowing equations ~3) an~

~m(n) = TW tq(n~ + PES(n) (3) xa(n) = J(~TW) f xm(n) I~) SA98~-03f. Page 22 ~~3B5~7 l As previously described, PES(n) is the distance of the head from the nearest track centerline and is computed from the logic of Table l. The term "TW tq(n)~ is the 'ltrack ~Tidth" times the digital value corresponding to S the track tvpe. Thus in e~uation (3~ the term xm(n) is the measured position of the head within some four-track ~and of the quadrature pattern. In equation (4) the term x~(n), i.e. the measured absolute position of the head relati~e to a fixed reference, is equal to xm(n) plus some integral multiple, denoted "J", times the width of a four-track band (4~1). In the operation of the servo control s~stem, only xm(n) is actually measured. The expression "J(4TH)" is determined based upon the predicted absolute position of the head from an expression for xp(n) to be described later. This is possible because the state estimator is designed such that the error bet~een the measured absolute head position, x , and the predicted absolute head position, xp, is alt7ays less than two track widths. Thus xa is computed bv determining J, which is the number of integral multiples of four-track bands, in the predicted a~solute head position xp. ~The predicted position xp can be considered as a distance equal to some integral multiple J of four-track bands plus a remainder or fraction of a four-track band). In this manner the term J is used to compute xa in equation (4).
Referring again to the flo~ chart of Fig. 5, after the comptltation of xa, the term e(n) is computed accordinq to the follo~ing equation ~5).
e(n) = xa(nj - xp(n) (5) SA98~-n3~ Page 23 12;8~

1 The term e(n) is the error between the measured absolute head position, xa(n), and the predicted absolute head position, xp(n), for that sample.
The estimated head position and velocity and a term we(n) are then computed according to the following equations (~)-(8).

~e(n) = ~p(n) - L1 e(n) (6) '~e(n) = vp(n) - L2 e(n) (7) e(n) wpln) - L3 ~ e(n) (8 The term xp(n) in equation (6), vp(n) in equation (7), and wp(n) in equation (8) are the predicted position, velocity and bias force equivalent VCl~1 current, respectively. These terms have been computed durin~ the prediction update phase from the prior sample, and will be e~plained with reference to that portion of the flow chart of Fig. 5. The term we(n) in equation (8) is the estimated hias force component of the total VC~I current reauired to give the esti~ated acceleration of servo head 32. The Li terms in equations (6)-(8) are constants.
Referring again to the flow chart of Fig. 5, the measured VCM current sample, i, is input to microprocessor 80. After the current has been input to microprocessor 80, then a decision is made as to whether the ser~!o control system is still in coast, i.e. whether the head is still moving from the insiae diameter crash stop towards the guard band outside diameter. If switch t n3 is still on, then the coast phase is still in SA984-03~ Page 2~

8~7 1 progress, the commanded velocity, vc, is set equal to "~oast' and VTG 102 is bypassed.
The following equations (9) and (10) are then computed.
s es(n) = Vc(n) ~ ve(n) a~(n) = (Kf/M)[i(n) + we(n)] (10) The velocity error signal from equation (9) is the difference between the commanded velocity, in this case v~OaSt, and the estimated velocity, ve(n). The estimated acceleration ae(n) in equation ~10) is proportional to the sum of the measured VCM current i(n) and the term we(n), the estimated bias force equivalent current. The term ~Je(n) is an estimate of the VCM current required to compensate for bias forces. The proportional constant Kf/~l in equation (10) is the actuator force constant divided b'l the actuator mass.
After veS(n) and a~(n~ have been computed in equations (9) and (10) the control signal u(n) is computed accordin~ to e~uation ~1) and output to DAC 92.
The microprocessor 80 next computes the term d51(n~
from equation (2) ar.d the predicted head position xp(n), velocit~ vp(n) and bias force equivalent currsnt wp(n~
from the following equations (11)-(13).

x (n) = Xe(n-l) + ~ll u(n-2) + g21 u( g31 i(n-l) + Pl2 ' ve(n-l) ~ p ' w (n 1) (11) SA9R~-03~ Pa~e 25 12~B~ 7 1 v (n~ = ve(n~ gl2 U(n-2) + g22 u(n 1) g32 i(n-l~ + P23 We(n-l) (12) T~p(n) = we(n-l) (13) As is apparent, the ~redicted terms xp(n), vp(n) and ~rp ~n) are functions of prior control signals, the prior measured VCM current and the prior estimated head position, velocity and bias force equi~ralent current.
As sho~rn in the flow chart of Fig. 5, the counter is th~n incremented by one, to n+l. If there i5 no change in the tar~et track, then control is returned for receipt of the next PESP(n7 and PESQ(n) input.
The above-described routine for the computation of the control signal u during the coast phase continues so long as G~OD e~uals 2ero. When the servo head 32 reaches the guard band outside diameter, then GBOD equals 1 and control transfers to connector 3 (Fig. 5). At this point the coast phase has been completed and the estimated 2n values are re-initialized. The estimated head position Xe is set e~ual to ~GBOD (the ahsolute position of the guard band outside diameterl, the estima-ted velocity v is set equal to vcOast~ and the estimated acceleration ae is set equal to zero. The target track td is then set equal to xGBoD
The coast phase is terminated by "switching off"
s~itch 103. Control is tr~n~erred back to connector (Fig. 5) for receipt of a target track commandu If control unit 86 sends a command different than td = XGBoD/ then a track seek ~lill begin. The counter is res~t to n=0, and control is transferred back to 5~

1 connector 1 (Fig. 5). The same computations as previously described occur with the exception that when the second decision point for switch 103 is reached, the distance-to-go, xt , and new commanded velocity, ve, are computed according to the following equations (14) and (15).

Y~tg(n~ = td ~ xe(n) (14) ~r~(n) = F[xtg(n)] (15) The total error signal, xtg, computed according to equation (14) is merely the difference between the a~solute position of the target track certerline and the estimated absolute head position. The function term in eauation (15) represents the output from VTG 102 and is a nonlinear function of ~tg(n) during track seeking and a linear function, klxt~(n), auring track following.
During the seek the control signal u continues to be calculated according to equation (1) for each sample and output to DAC 92. The seek continues and the counter is incremented to n+l until a new target track command, tdr is received. Even at the completion of a seek, the above-described sampling of PESP(n), PESQ(n) and i(n), and the computation of u(n), con~inue. This is not~ the track following mode, which, from the perspective of the state estimator and control signal algorithm, is indistinguishable from track seeking.
The ~ij and Pi; in equations (11)-(13~ are constants determined from a physlcal model of the actuator, including such parameters as the mass of the coil and arm .. .

SA984-03~; Page 27 5'7 1 assembly, the frictional effects, the force constants, and the digital sampling time. A description of the manner in which the gij and Pi; constants are generated for a physical svstem is given by Franklin and Powell in Di~ital Control of Dynamic Systems, Addison-Wesley Publishing Co. (19~3), Chapter fi, pp. 131-139.
The kSi, Li, and ki constants, as shown in Fig. 2 and used in control signal equation (13 and equations ~6)-(8), are selected constants which ~ield the desired overall servo control system requirements for servo bandwidth, closed-loop pole locations and transient response. The specific values of these constants are a direct function of system design objectives and the specific hardware to be controlled.
With the use of this state estimator and control signal algorithm in the digital servo control svstem there is but one servo control system, which is identical in function during both track seeking and track following. There is no switching of a stiffness integrator into the servo control system during track following. The only difference between track seeking and track following is that during track following, when the head has approached to within approximately one-half track of the centerline of the target track, the function of distance-to-go output from VTG 102 is now a linear function represented as klxtg.
The control signal represented by equation (1) is a function of the prior control signal; the velocity error signal, which in turn is a function of the estimated head velocity and the total error signal; and the estimated head acceleration, which in turn is a function of the - SA98~-03~ Page 28

2~ 5~7 1 measured VCM current and the estimated bias force equivalent VCM curren~. Thus, the control signal can be represented in functional terms as follows:

u(n) = Fnlu(n-l), ve(n), xtg(n), i(n), we(n)}, (16) where Fn merelv represents the function expressed by equation (1). During track following the mean or e~pected value of absolute head position xe is a constant and the mean or expected value of the head velocity ve is zero. Similarlv, during track ollowing the prior control signal u(n-l) has a mean or expected value of zero. The state estimator estimates the bias force equivalent VCM current, we, so that the total acceleration estimate, which is (Kf/M) ' [i(n) ~ we(n~, has a mean or expected value of zero. In this manner the only term remaining in equation (16) which has a non-zero mean or expected value is the total error signal, xtg, ~hich durin~ track following is simply the PES. The PES
is driven to zero by IPA 58 without the need for a separate switchable stiffness integrator. With the use of the state estimator in the digital servo control s~rstem as describea herein to estimate w , ~here is thus no feedback path around IPA 58 that provides a gain in the presence of generally constant or very low frequency hias forces.
In addition, the use of the state estimator eliminates the necessity for counting or otherwise estimating track crossings, and thus permits the servo control system to be fully operable with sector servo information, The state estimator is designed, through SA984-~3fi Page 29 26~3~5t7 1 the selection of constants L1, ~2 and L3 in equation5 (6)-(8~, so that the measured absolute head position, xa(n), is always within less than two track widths of the predicted head position, xp(n~. Thus, since it is only necessary to actually measure the position of the head within a four-track band, through the use of PES(n) and t~(n), it does not matter how many tracks have been crossed by servo head 32 between PES samples.
Referrinq now to Fig. 6, the improved arrival of servo head 32 to the centerline of the target track is sho~m with the use of the above-described digital servo control system. Curve A of Fig. 6 illustrates an arrival trajectory of head 32 to the centerline of a target track for a tvpical multiple track seek. The discontinuity at approximately 6.0 ms. is caused by ~he switching of a stiffness integrator into the servo feedback loop. The arrival trajectory of Curve A was simulated by the use of a digital servo control system identical to that shown in Fiq. 4 but in which a digltal stiffness integrator (not sho~m) was located between summing junction 101 and VTG
ln2. This stiffness integrator was switched into the loop to drive xtg to zero when the servo head 32 was within approximately one-half track of the centerline o the target track. In comparison with the arrival trajectorv of Curve A, the trajectory of Curve B in Fig. 6 sho~s a smooth, continuous arrival of servo head

3? to the centerline of the target track. This arrival as reached with the disital servo control system of the present invention, i.e. without a switchable stiffness integrator.

SA984-036 Pa~e 30 , , .,, ... . .... , , .~, . . ........

--` 12ti8B57 1 Referring now to Fig. 7, there is illustrated a modification to the above-described digital servo control s~stem which compensates for an offset input voltage to IPA 58. Even in the presence of no input to DAC 92, there can be an output voltage offset on DAC 92 and an input voltage offset to IPA 58. The value of this offset can vary with time and temperature. In order to compensate for this offset duxing track following, the control current u(n~ is digitally integrated by digital integrator 120. This results in a correction factor btn) summed with the control signal u(n) to compensate for an inp~lt signal offset to IPA S8. The equation computed by microprocessor 80 is expressed as follows:
b(n) = b(n-1) + cO(n) u(n-l), (17) where cO is a non-zero positive constant. During track seeking the microprocessor sets cO(n) equal to zero.
Thus the digital integration and updating of h(n) for voltage offset only occurs during track following. The use of integrator 120, bv se~ting cO(n) equal to a non-zero value, causes no adverse effects because the variation in the value of b is extremely slow relative to any bias force variations which can occllr over the time it takes to do any track seek.
While the preferred e~odiments o thè present invention have been illustrated in detail, it should be apparent that modifications and adaptations to those embodiments may occur to one skilled in the art without departing from the scope oE the present invention as set forth in the following claims.

SA9~-03~ Page 31

Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A method for controlling the data track seeking and following of a head in a data recording disk file of the type having a disk with servo information thereon defining data track centerlines, a head for reading samples of the servo information during rotation of the disk, an actuator connected to the head and responsive to an input signal for positioning the head to the centerline of a target track during track seeking and maintaining the head over the centerline of the target track during track following, the method comprising the steps of:
generating from the servo information a sampled head position error signal (PES) representative of the position of the head relative to the centerline of the nearest track;
estimating, for each PES sample, the absolute head position, relative to a fixed reference, and the head velocity;
estimating, for each PES sample, the equivalent actuator input signal required to compensate for bias forces acting on the actuator;
measuring, for each PES sample, the input signal to the actuator;
computing, for each PES sample, a commanded head velocity as a function of the distance from the estimated absolute head position to the target track;
computing, for each PES sample, the head velocity error as a function of the difference between the commanded head velocity and the estimated head velocity;
estimating, for each PES sample, the head acceleration as a function of the measured actuator input signal and estimated bias force equivalent actuator input signal;
generating, for each PES sample, a control signal as a function of the head velocity error, the estimated head acceleration, and the control signal genera-ted for the prior PES sample;
integrating the control signal;
providing the integrated control signal as the input signal to the actuator;
storing, for each PES sample, the estimated absolute head position, the estimated head velocity, the estimated bias force equivalent actuator input signal, the measured actuator input signal and the control signal;
predicting, for each PES sample, the absolute head position and the head velocity as functions of prior control signals, the prior estimated absolute head position, the prior estimated head velocity, the prior estimated bias force equivalent actuator input signal and constants representative of physical parameters of the actuator, the predicted absolute head position and velocity for each PES sample being used during the step of estimating the absolute head position and head velocity for the subsequent PES sample; and.
predicting, for each PES sample, the bias force equivalent actuator input signal as a function of the prior estimated bias force actuator input signal, the predicted bias force equivalent actuator input signal for each PES sample being used during the step of estimating the bias force equivalent actuator input signal for the subsequent PES sample.

2. The method according to claim 1 wherein the servo information is of the type providing a primary signal and a quadrature signal from a quadrature pattern recorded in radially repetitive four-track bands, each track in the band being a unique track type, and wherein the step of generating the PES includes the step of determining, from the primary and quadrature signals, which of the four track types the head is on.

3. The method according to claim 1 wherein the servo information is recorded in radially repetitive bands, each band containing a fixed number of uniquely identifiable tracks, and wherein the step of estimating the absolute head position includes the step of measuring the position of the head within a band without detecting or estimating track crossings by the head.

4. The method according to claim 2 wherein the step of measuring the position of the head within a band includes the steps of identifying the specific track number over which the head is located within the band, multiplying the identified track number by the track width, and summing the resultant product with the PES.

5. The method according to claim 3 wherein the step of estimating the absolute head position further comprises the step of determining a distance equivalent to the integral number of bands comprised by the predicted absolute head position and summing that distance with the measured head position within a band, thereby computing a measured absolute head position.

6. The method according to claim 1 wherein the actuator is a voice coil motor (VCM), wherein the step of measuring the actuator input signal further comprises the step of measuring the input current to the VCM, and wherein the step of estimating the bias force equivalent actuator input signal further comprises the step of estimating the equivalent VCM current to compensate for bias forces.

7. The method according to claim 1 wherein the step of estimating the head acceleration further comprises the step of estimating the head acceleration to have a mean value of zero when the expected value of the head position is a constant.

8. The method according to claim 1 wherein the disk file is of the type also having integration means coupled to the actuator, and further comprising the steps of, prior to integrating the control signal, adding to the control signal a correction signal to compensate for an input voltage offset in the integrating means, and updating the correction signal during track following.

9. A method for controlling the data track seeking and following of a head in a data recording disk file of the type having servo information recorded as radially repetitive bands of a fixed number of uniquely identifiable tracks either as continuous tracks on a dedicated servo disk surface or as sectors of tracks on a data disk surface, a head for reading the servo information during rotation of the disk, means for demodulating the servo information read by the head to provide an analog position error signal (PES) representative of the position of the head relative to the centerline of the nearest track, a voice coil motor (VCM) connected to the head for positioning the head to the centerline of a target track during track seeking and maintaining the head over the centerline of the target track during track following, an integrating power amplifier for integrating and amplifying an analog control signal so as to provide an input current to the VCM, the method comprising the steps of:
generating, from the analog PES, a digital PES
sample;
identifying, from the analog PES, the track within a band corresponding to the track from which the PES sample was generated;
measuring, from the PES sample and the identified track, the position of the head within a band;
estimating, from the measured position of the head within a band, the absolute head position relative to a fixed reference;
estimating the head velocity;
estimating the equivalent VCM input current required to compensate for bias forces;

measuring the digital value of input current to the VCM at the time of the PES sample;
computing a commanded head velocity as a function of the difference between the estimated absolute head position and the absolute position of the target track;
computing the head velocity error as a function of the difference between the commanded head velocity and the estimated head velocity;
estimating the head acceleration as a function of the measured VCM input current and estimated bias force equivalent VCM current, the step of estimating the head acceleration including the step of estimating the average head acceleration to be zero during track following;
generating a digital control signal as a function of the head velocity error, the estimated head acceleration, and the control signal generated for the prior PES
sample;
converting the digital control signal to an analog control signal;
directing the analog control signal to the integrating power amplifier to thereby position the head toward the centerline of the target track in response to the received digital PES sample;
predicting the absolute head position relative to a fixed reference and the head velocity as functions of prior digital control signals, the prior estimated absolute head position relative to the fixed reference, the prior estimated head velocity, the prior estimated bias force equivalent VCM current and constants representative of physical parameters of the actuator;
and predicting the bias force equivalent VCM current as a function of the estimated bias force equivalent VCM
current for the prior PES sample, wherein the steps of estimating the absolute head position, the head velocity and the bias force equivalent VCM current include the step of computing the error between the measured absolute head position relative to a fixed reference and the predicted absolute head position for each PES sample.

10. The method according to claim 9 wherein the servo information is a quadrature pattern, wherein the demodulator provides an analog primary PES and an analog quadrature PES from the quadrature pattern servo information, and wherein the step of generating a digital PES sample includes the steps of generating a primary PES
sample (PESP) and a quadrature PES sample (PESQ) and selecting either PESP or PESQ as the PES sample.

11. The method according to claim 9 further comprising the steps of providing a digital correction signal to compensate for voltage offsets in the integrating power amplifier, summing the correction signal with the digital control signal prior to converting the digital control signal to an analog control signal, and updating the correction signal during track following.

12. An improved data recording disk file of the type having means for processing commands received from a disk file control unit, at least one rotatable disk with generally concentric data tracks thereon, the data disk or a separate servo disk having servo information recorded thereon, at least one head for reading the servo information during rotation of the disk and for providing an analog servo signal, a demodulator for processing from the analog servo signal a head position error signal (PES) representative of the distance of the head from the nearest track centerline, an actuator responsive to an input signal and attached to the head for positioning the head to the centerline of a target track in response to a command from the control unit and for maintaining the head over the centerline of the target track during a read or write operation, the actuator having predetermined physical parameters and being subjected to generally constant or low frequency bias forces during operation of the disk file, the improvement comprising:
analog-to-digital conversion means for the analog PES received from the demodulator;
means for measuring the analog actuator input signal;
analog-to-digital conversion means for the measured actuator input signal;
means for computing, from the digital PES, the measured absolute head position relative to a fixed reference and for computing an estimate of the head velocity and an estimate of the actuator input signal required to compensate for bias forces acting on the actuator;

a velocity trajectory generator for generating a value representative of the optimum velocity of the head as a function of the difference between the absolute position of the target track and the estimated absolute head position;
means for computing a value representative of the velocity error as a function of the difference between the optimum velocity and the estimated head velocity;
means for computing an estimate of the head acceleration as a function of the digital actuator input signal and the estimated bias force equivalent actuator input signal;
means for computing a digital control signal as a function of prior digital control signals, the value representative of the velocity error and the estimated head acceleration;
means for converting the digital control signal to an analog input signal to the actuator;
means for storing the digital values representative of the control signals, the measured actuator input signals, the estimated head position and head velocity, the bias force equivalent actuator input signals, and constants representing predetermined physical parameters of the actuator; and means for predicting the absolute head position, head velocity, and bias force equivalent actuator input signal from the estimated absolute head position, head velocity and bias force equivalent actuator input signal, respectively, computed from the prior digital PES sample, whereby the error between the measured absolute head position and the predicted absolute head position is used to generate the estimated absolute head position, the estimated head velocity and the estimated bias force equivalent actuator input signal.

13. The improvement according to claim 12 wherein the servo information is recorded in radially repetitive bands of a fixed number of tracks per band, and wherein the demodulator provides a primary analog servo signal and a quadrature analog servo signal, wherein the analog-to-digital conversion means for the PES received from the demodulator includes means for generating a digital primary PES (PESP) and a digital quadrature PES
(PESQ), and further comprising means for identifying from the digital PESP and PESQ the specific track within a band over which the head is located.

14. The improvement according to claim 12 wherein the means for computing an estimate of head acceleration includes means for estimating the average head acceleration during track following to be zero.

15. The improvement according to claim 12 wherein the means for converting the digital control signal to an analog actuator input signal includes an integrating power amplifier.

16. The improvement according to claim 15 further comprising means for computing a value representative of a correction signal as compensation for a voltage offset to the integrating power amplifier, means for adding the correction signal value to the digital control signal prior to conversion of the digital control signal to analog, and means for updating the correction signal value during track following.