Method and apparatus for characterization of voice-coil motor with eddy current effect

COPYRIGHT STATEMENT

[0002] A portion of the disclosure of this patent document contains material which is subject to copyright or mask work protection. The copyright or mask work owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright or mask work rights whatsoever.

Background of Invention

[0003] 1. Field of Invention

[0004] This invention relates to improvements in methods for modeling voice coil actuator/motors (VCMs) of the type used in mass data storage devices, or the like, and more particularly to improvements in such methods that may be used to determine the modeling parameters of such VCMs.

[0005] 2. Relevant Background

[0006] Typical mass data storage devices include well-known hard disk drive assemblies (HDAs) of the type to which the invention pertains. Generally, an HDA includes one or more rotating disks that carry a magnetic media to which data may be written, and from which previously written data may be read. The data is written to and read from the disk by one or more magnetic heads or transducers that are a part of a voice coil motor (VCM) assembly, which moves the heads to the desired locations at which data is to be written or read.

[0007] An exploded view of a portion of a typical HDA 5 is shown in FIG. 1 to which reference is first made. The HDA 5 includes a VCM apparatus 10 in conjunction with a plurality of rotating disks 12. The VCM assembly 10 includes one or more arms 14 that are pivoted about a bearing point 16 to carry and move the heads or data transducers 18 radially inwardly and outwardly within the stack of data disks 12 to be enabled to selectively read or write data to the magnetic media of the disks 12.

[0008] The outboard end 24 of the arm 14 carries a coil 20 that is selectively energized by currents from VCM positioning circuitry 22. The outwardly extending end 24 of the arm 14 is located between two horizontal magnets 26 and 28, which are mounted to base plates 30 and 32. The base plates 30 and 32 and magnets 26 and 28 are spaced apart by spacers (not shown) to allow the arm and coil portions 24 and 20 to freely swing therebetween. The plates 30 and 32, spacers, and magnets 26 and 28 are securely fastened to the base plates 34. A top cover plate 35 encloses the top side of the base plate 32. Thus, as the currents from the VCM positioning circuitry 22 are applied to the coil 20, magnetic fields are established by the current induced field of coil 20 which interact with the fields of magnets 26 and 28 to precisely position the heads 18 at a desired location under control of the VCM positioning circuitry 22. (It should be understood that references herein to specific directions, for example, up, down, left, and right, relate to illustrations in the drawings and are expressed only for reference and clarity of understanding. It should be noted that actual structural orientations in an actual device represented by the drawings may differ.)

[0009] When the apparatus 5 is powered down, typically the head mechanism is moved to a position (not shown) at which the heads 18 are “parked” or “landed”, often at the inner radius of the disk. In other cases, such as when the head is parked on a ramp, they may be parked along the outer radius of the disk. In order to properly move the heads to the park position, generally a driving current is applied to the coil 20 that is of sufficient magnitude to bring the head assembly to the park position. However, it will be appreciated that if the head mechanism is overdriven, the delicate head mechanism and other parts of the disk assembly may sustain damage. On other hand, if the head is underdriven, the head mechanism may not reach the park position, which may result in loss of the air bearing between the head and disk surface, which may also cause damage both to the head mechanism and to the underlying magnetic media of the disk assembly 12 above which the heads 18 fly.

[0010] The heads are positioned by the positioning circuitry 22, also referred to herein as a servo circuit, which operates in the retraction or parking of the heads to their landing zone or landing ramp. The servo circuit 22 may incorporate a floating-terminal BEMF detection scheme (FLBD) in its design to control the retract of the heads to their parked position. The purpose of FLBD is to extract the BEMF signal from the VCM terminal voltage difference, VPN=VP−V N, at the input nodes of the coil 20. This is done normally by turning off all four driver FETs of the driving circuitry to let Vp and VN float for a short time. After the flyback current in the VCM coil 20 decays to a predetermined level, which is defined to be at or near zero with the rate of change of the current also at or near zero. VPN theoretically will approximate the BEMF voltage, since with no current, there should be no voltage drop across the resistive and inductive elements of the motor.

[0011] One technique controlling a VCM is shown in U.S. Pat. No. 6,184,645, issued Feb. 6, 2001 (Attorney docket number TI-29097), incorporated herein by reference. One technique for measuring the BEMF of the coil of the actuator used in said U.S. Pat. Nos. 6,184,645 is shown in 6,204,629, issued Mar. 20, 2001, incorporated herein by reference.

[0012] As mentioned, when the head is to be moved to the park position, one method that may be employed is to tristate the drive transistors, wait a period of time to allow the flyback current to occur and dissipate down to a predetermined magnitude. Thus, after the flyback current has dissipated to the predetermined level, the voltage appearing between the drive nodes is measured, which, at least in theory, should represent the BEMF developed across the coil 20. Since the BEMF has a value almost directly proportional to the speed of the coil of the VCM, knowing the velocity of the coil 20 enables the precise required drive current to be determined that will properly move the heads to the parked position at a controlled velocity.

[0013] However, in practice, it has been found that the BEMF that is measured using the prior art techniques does not always accurately represent the correct velocity of the coil 20, and, consequently, the head assembly controlled thereby. As discussed in said copending patent applications Ser. No. 09/451,697, filed Nov. 30, 1999 (Attorney docket number TI-29500), and Ser. No. 09/464,315 (Attorney docket number TI-29436), filed Dec. 16, 1999, this is due at least in part to the influence of eddy currents induced in the structures adjacent the coil 20 of the VCM on the voltage induced into the coil during its movement at the same time that the BEMF is measured. This can be seen from the Bode plots of FIGS. 2A and 2B, which respectively show differences in the actual curve 40 of the magnitude of the VCM compared to the theoretical magnitude curve 42, and differences in the actual phase 44 of the VCM compared to the theoretical phase curve 46. Thus, although the eddy current phenomenon has been recognized, the manner for modeling the eddy currents has not been fully understood heretofore.

[0014] What is needed, therefore, is a method and circuit for more accurately modeling the eddy currents in determining the BEMF when the VCM drivers are tristated to enable the current needed to be applied to the coil to properly park the heads at a controlled velocity to be more precisely determined.

SUMMARY OF INVENTION

[0015] Thus, in light of the above, it is, therefore, an object of the invention to provide a method to characterize the eddy current parameters of a VCM device.

[0016] It is another object of the invention to provide a method for determining or measuring eddy current parameters using a measured step response, fly-back voltage and Bode-plot, to optimize the match in frequency and time domains.

[0017] Thus, according to a broad aspect of the invention, a method is presented for characterizing a VCM assembly. The method includes measuring a first response of the VCM assembly, removing a first conducting part of the VCM assembly, and measuring a second response of the VCM assembly with the first conducting part removed. Then, the method includes constructing an eddy current model of the VCM assembly from the first and second measured responses.

[0018] According to another broad aspect of the invention, a method is presented for characterizing a VCM assembly in which a desired VCM characteristic curve is developed, and an actual response of the VCM assembly to a predetermined test drive signal, is measured. Using curve fitting techniques, VCM compensation parameters are determined to construct a model of the VCM assembly.

BRIEF DESCRIPTION OF DRAWINGS

[0019] The invention is illustrated in the accompanying drawings, in which: FIG. 1 is a perspective view of a portion of a mass data storage device and associated VCM assembly with which the circuit and method in accordance with a preferred embodiment of the invention may be employed.

[0020]
FIGS. 2A is a Bode plot showing differences in the actual curve of the magnitude of a VCM compared to a theoretical magnitude curve.

[0021]
FIGS. 2B is a Bode plots showing differences in the actual phase of a VCM compared to the theoretical phase curve.

[0022]
FIG. 3 is an electrical schematic diagram of a model of the VCM assembly of FIG. 1, in accordance with a preferred embodiment of the invention.

[0023]
FIG. 4 is an flow chart of the method in accordance with a broad aspect of the invention.

[0024]
FIG. 5 is a flow chart of the method in accordance with another, more detailed, aspect of the invention.

[0025] In the various figures of the drawing, like reference numerals are used to denote like or similar parts.

DETAILED DESCRIPTION

[0026] A model 50 of the voice control actuator/motor (VCM), according to a preferred embodiment of the invention, is shown in the electrical schematic diagram of FIG. 3.

[0027] The model is more fully described in said copending application Ser. No. 09/451,697, filed Nov. 30,1999 (Attorney docket number TI-29500), and 09/464,315 (Attorney Docket number TI-29436), filed Dec. 16, 1999, which are incorporated herein by reference. The model 50 takes known VCM effects into account, including eddy current effects in all the structures in the neighborhood of the actuator coil, which have been unrecognized heretofore, and which, therefore, have not been modeled. The model 50, therefore, is believed to be a more accurate representation of a physical VCM assembly and its associated electrical components than models used heretofore.

[0028] The model 50 includes a number of ideal model parts between the input terminals 52 and 54, which model or represent the actual VCM terminals of a physical VCM, such as that shown in FIG. 1. A capacitor 56 is connected between the input terminals 52 and 54, to represent the input capacitance of the system. The inductance of the motor coil 20 is modeled by inductors 64 and inductor 58, connected together at point 67. The inductor 64 represents a winding leakage inductance of the VCM coil 20. The inductor 58 represents the mutual inductance between VCM coil and its metallic neighborhood including at least parts 26, 30, 28, and 32. The low end of inductor 58 is connected to input terminal 54 through a voltage source 60, which represents the back EMF (BEMF) of the coil 20. The BEMF, of course, is a time varying quantity; therefore, the voltage source 60 is likewise a varying voltage source that follows the BEMF waveform generated in the physical device.

[0029] A resistor 62 is connected in series between the top input terminal 52 and the left end of the inductor 64. The resistor 62 represents the resistance of the physical VCM coil 20. A resistor 66 is connected from a node 67 between the inductor 64 and inductor 58 to the bottom end of the inductor 58. The resistor 66 representing the magnetic hysteresis loss is in parallel with the VCM inductor 58, and would typically be of very high value. Consequently, in many applications, the resistor 66 may be ignored.

[0030] As mentioned, one of the reasons that the physical VCM does not behave as predicted by prior art models is that the coil 20 of the VCM creates eddy currents in the adjacent magnets and other structures of the VCM assembly. The eddy currents do not self-extinguish as rapidly as the flyback current, and consequently result in the creation of a voltage across the coil when the excitation voltage has been removed. Thus, top and bottom current loops 68 and 70 are included in the model 50 to consider the eddy current effects.

[0031] The loop 68 includes a mutual inductor 75, having an inductance equal to the value of the VCM mutual inductor 58, an inductor 74, and a resistor 76, connected in series. The inductor 75 represents the mutual inductance between VCM coil and the top VCM magnetic plate. The magnet plate includes the top VCM magnet 28 and the surrounding structures, including the mounting plate 32 and top cover plate 35, into which eddy currents may be induced. The inductor 74 represents the leakage inductance of the top VCM magnet plate, and the resistor 76 represents the resistance of the top VCM magnet plate.

[0032] Likewise, the bottom loop 70 includes a mutual inductor 78 having a value equal to the mutual inductance of the VCM inductor 58, and inductor 80, and a resistor 72, all connected in series. The inductor 78 represents the mutual inductance between VCM coil and the bottom VCM magnetic plate, which includes the bottom VCM magnet 26, and the surrounding structures, including the mounting plate 30 and base plate 34, into which eddy currents are induced. The inductor 80 represents the leakage inductance of the bottom VCM magnet plate, and the resistor 72 represents the resistance of the bottom VCM magnet plate. The first and second loops 68 and 70 are interconnected, as shown, at one side of the inductors 72 and 78.

[0033] First and second parasitic capacitors 86 and 88 are connected between the top and bottom ends of coil 58 and the interconnection nodes of inductors 75 and 74 and inductors 78 and 80, respectively. The values of capacitors 86 and 88 may be very small. Consequently, they may be ignored in many applications.

[0034] With the recognition that the effects of the induced eddy currents affects the accuracy of the measurement of the BEMF, according to the above described model, their effects can now be taken into account in measuring the BEMF and characterizing the VCM. Thus, generally, with reference additionally now to FIG. 4, a theoretical VCM model, like the model of FIG. 3, may be developed 90. The desired response may include, for example, one or more Bode plots following the magnitude and phase response of the VCM if it were to respond without the eddy current effects, such as curves 42 and 44 in the curves of FIGS. 2A and 2B.

[0035] Next, in accordance with a preferred embodiment of the invention, a current or other predetermined signal may be injected into the coil 20 and an actual response of the VCM system may be measured 92. Finally, VCM parameters of the model of FIG. 3 are determined by curve fitting of the data to the measured VCM response 94.

[0036] The process may be refined by determining the actual response of various parts of the VCM system to enable the curve fitting to be more accurately determined, for instance, to forth order or higher equations. This is illustrated in FIG. 5, to which reference is now additionally made. Thus, more particularly, a first response of the VCM in its original configuration to a first signal stimulus is first measured 100. The signal stimulus may be, for example, a signal having a step function waveshape.

[0037] The next step illustrated is to remove a portion of the structure of the VCM on one side of the coil 20. For example, a magnet 28 on the top side of the VCM coil 20 may be removed 102. (It should be noted that the removal of magnet 28 preferably will include the concomitant removal of the plate 32 that carries the magnet 28, since the eddy current effects described herein exist in both structures. Consequently, the phrase “magnet is removed” used herein should be understood to include the removal of at least the magnet and optionally, and preferably, removal of supporting or other surrounding structures as well.)

[0038] Then, a second response is measured of the VCM to a predetermined signal 104. The next step illustrated is to remove a structure on the other side of the VCM coil 20, for example the magnet 26, step 106. (In reality, the entire VCM assembly may be removed from proximity of the magnet 26, since the magnet 26 may be affixed to the box or container of the VCM and may not be removable by itself.) Then a third response is measured of the VCM to a predetermined signal 108.

[0039] Using the three measurements, and using curve fitting techniques, such as the curve fitting programs set forth in Attachment A, the eddy current parameters are determined, step 100. The eddy current parameters are used to fit the curve of the model of FIG. 3 to overlap the actual response curve to enable an accurate VCM model to be realized 112. To assure accuracy of the model, a verification step, shown in step 112 may be performed.

[0040] It should be noted that since the measurement information is used to perform a curve fitting function, the order of the measurements is not an important consideration. Also, the accuracy of the final parameter determination is dependent upon the amount of data that is collected. As a result, for example, it may be possible to omit one or more of the steps, if a less accurate result is acceptable for a particular purpose. For example, in the embodiment of the method illustrated FIG. 5, both magnets on top and bottom are removed in steps 102 and 106. If a less accurate curve fit is acceptable, for example, one of the magnet removal steps may be omitted.

[0041] More particularly, the order of the equations that may be developed decreases with the removal of each part. When everything is intact, the equation has the highest order. It can be seen with reference to FIG. 3 that the circuit model has three branches in parallel. Roughly speaking, the top and bottom portions 68 and 70 represent top and bottom plates. When removing a plate, the corresponding branch disappears, resulting in the decrease of equation order.

[0042] It should also be noted that three kinds of curves may exist: time domain curves, frequency domain curves, or curves in both domains. Preferably, curve fitting is performed on Bode Plots obtained in step 100 (FIG. 5) with all parts intact, because this is the object curve that includes all of the known current effects (i.e., eddy effect plus any unknown effects). By performing the magnet removal steps 102 and 106, the curves derived therefrom may be used to roughly determine the equation parameters R0 , (L1+LM ), Lp1, and Rp1(or Lp2and Rp2, but not both, since one branch does not exist when the other magnet is absent). The DC measurements can be used to determine R0 the parameters (LM +L1) can be determined from the Bode Plot or time response. After putting one plate on, Rp2 and Lp2 can be estimated. (The term “estimated” is used because the LM parameter may have tiny differences with the magnet on or off, and, additionally, other unknown factors in addition to the eddy current may exist, as well.)

[0043] These parameters provide the initial parameters used in an optimized curve fitting step in the frequency domain. A computer program, such as the program set forth below, an iterative type optimization program, may be used to determine the best curves that match the known Bode Plots, with a target accuracy. Preferably the program starts from a reasonable set of initial parameters rather than from a wild guess to insure convergence. The more accurate the initial parameters are, the quicker and better result that can be expected. Once the final program parameters are determined, they can be used to construct a circuit simulator to check the fly-back time domain response (the time it takes for the current to fall to zero) against those obtained in steps 100-112 to verify the result. If they do not provide the required or desired accuracy, the parameters may be adjusted, as necessary. It should be also noted that weighting may be added to critical sections of the curve to increase the localized accuracy in that area with respect to the rest, if desired. For example, the middle frequency range may be of more importance that the high and low frequency ranges, and increased accuracy thereat may be desired.

[0044] Although the invention has been described and illustrated with a certain degree of particularity, it is understood that the present disclosure has been made only by way of example, and that numerous changes in the combination and arrangement of parts can be resorted to by those skilled in the art without departing from the spirit and scope of the invention, as hereinafter claimed.

[0045] Program Listing Deposit

1Copyright Texas Instruments Incorporated% This program finds lm, M, l1, R1, l2, R2 using 4-order system model% It need calling function edd3_par.m%%Tan Du 11/01/99%% Idea: Using 4-order curve fitting to get target. Then using optimizatio%The objective function is to minimize the magnitude and phase errors bet%% Hrl --- Ideal VCM Bode plot% Hrml--- Practical VCM Bode plot% H_modl0-Initial VCM for iteration% H_idf - Bode obtained from curve fitting which is the target of optimiz% H_modl1-Final Model Bode, whos parameters are worked out throughoptimclear;global A0 B0 Lm Rm M lm l1 R1 l2 R2 Z_S P_S Hrlm w C N DVm=11; Vt=2;V0=(Vm+2*Vt); Ipk=1e−3; M=1.525e−3; % M is the only parameterpresetRm=12; lm=0.07e−3; C=1e−12; l1=1e−3; R1=100; l2=1.6e−3; R2=3e3;%Nova% optimized:% lm=1.977612315966087e−004% M=8.022387684033913e−004% l1=2.732007293077334e−004% R1=1.153813695860884e+002% l2=7.560818869356664e−004% R2=30.06080925428558Lm=lm+M;%VCM 4-order transfer functionN=[0, M*R1*R2, M*(R1*l2+R2*l1), M*l1*12];D=[R1*R2, M*(R1+R2)+(R1*l2+R2*l1), l1*l2+M*(l1+l2)];Z_S=[lm*C*D(3)+C*N(4);lm*C*D(2)+C*N(3); lm*C*D(1)+D(3)+C*N(2); D(2); D(1P_S=[Rm*lm*C*D(3)+Rm*C*N(4);Rm*lm*C*D(2)+lm*D(3)+N(4)+Rm*C*N(3);Rm*lm*C*D(1)+lm*D(2)+Rm*D(3)+N(3)+Rm*C*N(2);lm*D(3)+Rm*D(2)+N(2);Rm*D(1)];Z_S=Z_S/abs(Z_S(5));P_S=P_S/abs(P_S(5)); % make abs(Z_S(S=0))=1; abs(P_S(S=0))=1;% Get measurement dataload darlvcm.mat; Hrl=o2i1; %vcm with no magnet, no H bridgeload mille.mat; Hrlm=o2i1; %vcm with magnet, no H bridge%--Identification of original VCM --w= o2i1x*2*pi;f= o2i1x;% B -- numerator; A= denominatorHrl=(Hrl/abs(Hrl(1))); %correct to make dc gain=1Hrlm=(Hrlm/abs(Hrlm(1))); %correct to make dc gain=1nb=4; na=4;Wt=ones(size(w)); Wt(200:600,1)=800*ones(401,1); % Weight ofcurve fitti[B,A]=INVFREQS(Hrlm,w,nb,na, Wt); %Body Curve Fitting, B0,A0 contains alB=B/abs(B(nb+1)); % make abs(A0(S=0))=1; abs(B0(S=0))=1;A=A/abs(A(na+1));H_idf=freqs(B, A, w); %Get freq response of curve fitting% --- Model calculation ---H_modl0=freqs(Z_S, P_S, w); % Initial guess of the modelfigure(1); clgsubplot(211);semilogx(w/2/pi, [20*log10(abs(Hrl)),20*log10(abs(Hrlm)),20*log10(abylabel(‘Gain (dB)’)axis([1e0, 1e5, −40, 10]);title(‘Magnitude of measured VCM vs assumed VCM’);text(2,−25, ‘Yellow=Ideal, Purple=Lab Measured’)text(2,−35, ‘Blue=Curve Fit, Red=Initial Model’)grid onzoom onsubplot(212);semilogx(w/2/pi,180/pi*[angle(Hrl), angle(Hrlm), angle(H_idf),angleaxis([1e0, 1e5, −100, 0]);title(‘Phase of mesured VCM vs assumed VCM’)xlabel(‘Frequency (Hz)’)ylabel(‘Phase (deg)’)text(2,−70, ‘Yellow=Ideal, Purple=Lab Measured’)text(2,−90, ‘Blue=Curve Fit, Red=Initial Model’)grid onzoom onformat long;options=foptions; % use default first%options(14)=5000;% then change: increase iteration number%options(1)=1;options(2)=1e−15;options(3)=1e−15;vcm_par=[lm, l1, R1, l2, R2, C]; %<===TransErr=fmins(‘edd3_par’, vcm_par, options);disp(‘your [lm, M, l1, R1, l2, R2, C] = ’); [lm;M;l1;R1;l2;R2;C]%Recalculat VCM transfer function using l1, R1, l2, R2, C obtainedfrom ‘N=[0, M*R1*R2, M*(R1*l2+R2*l1), M*l1*l2];D=[R1*R2, M*(R1+R2)+(R1*l2+R2*l1), l1*l2+M*(l1+l2)];Z_S=[lm*C*D(3)+C*N(4); lm*C*D(2)+C*N(3);lm*C*D(1)+D(3)+C*N(2); D(2); D(1P_S=[Rm*lm*C*D(3)+Rm*C*N(4);Rm*lm*C*D(2)+lm*D(3)+N(4)+Rm*C*N(3);Rm*lm*C*D(1)+lm*D(2)+Rm*D(3)+N(3)+Rm*C*N(2);lm*D(3)+Rm*D(2)+N(2);Rm*D(1)];Z_S=Z_S/abs(Z_S(5)); P_S=P_S/abs(P_S(5));% --- Model optimized---H_modl1=freqs(Z_S, P_S, w);figure(2)%semilogx(w/2/pi, [20*log10(abs(Hrl)),20*log10(abs(Hrlm)),20*log10(abs(H—semilogx(w/2/pi, [20*log10(abs(Hrl)),20*log10(abs(Hrlm)),20*log10(abs(H_mylabel(‘Gain (dB)’)axis([1e0, 1e5, −40, 0]);title(‘Magnitude of mesured VCM vs assumed VCM’);text(2,−35, ‘Yellow=Ideal, Red=Lab measured,Blue=Model’)grid onzoom onfigure(3)%semilogx(w/2/pi,180/pi*[angle(Hrl), angle(Hrlm), angle(H_idf),angle(H_msemilogx(w/2/pi,180/pi*[angle(Hrl), angle(Hrlm), angle(H_modl1)]);axis([1e0, 1e5, −100, 0]);title(‘Phase of mesured VCM vs assumed VCM’)xlabel(‘Frequency (Hz)’)ylabel(‘Phase (deg)’)text(2,−90, ‘Yellow=Ideal, Red=Lab measured,Blue=Model’)grid onzoom on

	Number	Date	Country
Parent	09451697	Nov 1999	US
Child	10063174	Mar 2002	US
Parent	09464315	Dec 1999	US
Child	10063174	Mar 2002	US

Method and apparatus for characterization of voice-coil motor with eddy current effect

Information

Publication Number

Date Filed

Date Published

Inventors

Original Assignees

CPC

US Classifications

International Classifications

Abstract

Description

Claims

CROSS REFERENCE TO RELATED APPLICATIONS

Continuation in Parts (2)