第7回電気学会モーションコントロールの高機能化...

ハードディスクドライブの磁気ヘッド位置決め制御

千葉工業大学 機械工学科 熱海武憲

２０１６年５月２６日＠千葉大学

第7回 電気学会 モーションコントロールの高機能化に関する協同研究委員会

磁気ヘッドの位置をより精密に制御する

ＨＤＤの磁気ヘッド位置決め制御系

Voice-coil motor(VCM)

Disk Spindle motor

Magnetichead

PZT actuator

Digital controller

Position signal

Voice-coil motor(VCM)

PZT actuator

磁気ヘッドの位置決め誤差

日本機械学会誌 2016. 4 Vol. 119 No.1169 より引用

https://www.hgst.com/sites/default/files/resources/HGST-Micro-Actuator-TB.pdf より引用

https://www.hgst.com/sites/default/files/resources/HGST-Micro-Actuator-TB.pdf より引用

さらに先の技術

Takenori Atsumi,``Emerging Technology for Head-Positioning System in HDDs'',IEEJ Journal of Industry Applications, Vol. 5, No. 2, pp. 117-122, (2016-3)

Case 1: Film-coil actuator

Case 2: Thermal actuator

Case 1: Film-coil actuator

Case 2: Thermal actuator

Easy Excited Resonant Mode in HDDs

By VCM actuator(Butterfly mode: 5.2 kHz)

By external vibration(Torsion mode: 3.5 kHz)

Direction of coil current

Pads for wire bonding

Film Actuator

Thin film coil (Film actuator)

VCM with film actuator

VCM actuator Film actuator

Frequency Response of Actuator

w/ Film actuator(VCM +PZT+ Film)

w/o Film actuator(VCM+PZT)

Comparison of Sensitivity Function

Case 1: Film-coil actuator

Case 2: Thermal actuator

Single-stage actuator (VCM only)

Dual-stage actuator (VCM + PZT)

103

104

-30

-25

-20

-15

-10

-5

0

5

10

15

Frequency [Hz]

Ga

in [d

B]

Voice-coil motor(VCM)

PZT actuator

Sensitivity function

Heater

Magnetic head

Slider

Thermal Actuator

Disk for mobile HDDs

Thermalactuator

PZT actuator

VCM

Frequency Response of Thermal Actuator

Measurement data Mathematical model

Design results Experimental results

Comparison of Sensitivity Function

Single-stage actuator (VCM only) Dual-stage actuator (VCM + PZT)Triple-stage actuator (VCM + PZT + Thermal actuator) )

Single-stage actuator (VCM only), SD: 3.12 nm Dual-stage actuator (VCM + PZT), SD: 1.57 nmTriple-stage actuator (VCM + PZT + Thermal actuator) ), SD: 0.96 nm

Time domain Frequency domain

Comparison of Positioning Error

ここ最近の学会発表

Estimation Method ofUnobservable Oscillations in Sampled-Data Positioning Systems

Takenori Atsumi

Chiba Institute of Technology

The 2nd IEEJ international workshop on Sensing, Actuation, Motion Control, and OptimizationMarch 8th, 2016, Seikei University, Tokyo

Sampled-data Control System

Continuous-timePlant

Discrete-time

Controller

Hold(D/A)

Sampler(A/D)

Error

Controlled variable

Ref. Control input

Continuous-time systemDiscrete-time system

Example: Tank System

Block Diagram of Tank System

Tank System

Human Being

Hold(D/A)

Sampler(A/D)

Error

Water level

Target level

Turn fauceton/off

Continuous-time systemDiscrete-time system

If the sampling time is too long…

Voice-coil motor(VCM)

Disk Spindle motor

Magnetichead

PZT actuator

Digital controller

Head-Positioning Control System in HDDPosition signal

The control system cannot “see” intersamplingvibrations which cause destruction of users data.

The timing of reading or writing user data corresponds to the time between samples.

Position signal area Data area

Measurement signal

High-risk area for data destruction

Head movement

Continuous-timePlant

Discrete-time

Controller

Hold(D/A)

Sampler(A/D)

Actualcontrolled variable

Ref.

Observed controlled variable

Continuous-time systemDiscrete-time system

We have to know difference between observed and actualcontrolled variables

Actual variable

Ns: Number of samples for estimation, t : Sampling time

Unobserved amplitude

Ns t

Am

plit

ude

Time

Unobservable Amplitude of Oscillation

Observed variable

: Natural frequency,

: Damping ratio,

: Damped natural frequency,

: Sample number, : Sampling time: Initial phase,

Oscillation at Sampling Frequency

Event probability

Worst case Best case

Dependence of on 𝑚𝑢 𝜙0

Oscillation at Nyquist Frequency

Event probability

Worst case Best case

Dependence of on 𝑚𝑢 𝜙0

Event probabilityDependence of on 𝑚𝑢 𝜙0

Oscillation at One-third of Sampling Freq.

Worst case Best case

サンプリング定理と矛盾する？

時間領域 周波数領域

Two Index Values for Evaluation

Maximum value based

Root-Mean-Square (RMS) value based

T. Atsumi and W. Messner, "Analysis of Unobservable Oscillations in Sampled-Data Positioning Systems,” The IEEE Transactions on Industrial Electronics, vol. 59, no. 10, pp. 3951-3960, (2012-10)

Maximum value RMS valuefixed initial phase (old)variable initial phase (new)

fixed initial phase (old)variable initial phase (new)

Comparison with Previous Work

Evaluation usingIndex value

Calculation ofIndex value

Set modalparameters

Modal parameters of mechanical system are decided by using measurement results

Index values are calculated by using resonance frequency, damping ratio and number of samples for estimation

Unacceptable sampling frequencies are estimated by using index values and target of reliability

Estimation Procedure

Example: Hard Disk Drive

Evaluation usingIndex value

Calculation ofIndex value

Set modalparameters

Modal parameters of mechanical system are decided by using measurement data

Index values are calculated by using resonance frequency, damping ratio and number of samples for estimation

Unacceptable sampling frequencies are estimated by using index values and target of reliability

Estimation Procedure

Mechanical Characteristics of HDDs

Primary mechanical resonance• Resonance frequency: 4100 Hz • Damping ratio: 0.01

Frequency response Mathematical model

Measurement data Mathematical model

Evaluation usingIndex value

Calculation ofIndex value

Set modalparameters

Modal parameters of mechanical system are decided by using measurement data

Index values are calculated by using resonance frequency, damping ratio and number of samples for estimation

Unacceptable sampling frequencies are estimated by using index values and target of reliability

Estimation Procedure

Dependence of Unobservable Amplitude on Sampling Frequency

• Resonance frequency: 4100Hz• Damping ratio: 0.01• Number of sample: 5(blue), 10(green), 20(red)

Maximum value RMS value

Evaluation usingIndex value

Calculation ofIndex value

Set modalparameters

Modal parameters of mechanical system are decided by using measurement data

Index values are calculated by using resonance frequency, damping ratio and number of samples for estimation

Unacceptable sampling frequencies are estimated by using index values and target of reliability

Estimation Procedure

• Mechanical Resonance: 4100Hz/ Damping ratio: 0.01• Number of sample for estimation: 5• Acceptable unobservable amplitude (RMS): Less 10%

Acceptable level

Unacceptable Sampling Frequency

Unacceptable frequency

• Mechanical Resonance: 4100Hz/ Damping ratio: 0.01• Number of sample for estimation: 5• Acceptable unobservable amplitude (RMS): Less 10%

Acceptable level

Unacceptable Sampling Frequency

Conclusion

Most of positioning control systems are sampled-data control systems with mechanical resonances.

We should check unobservable amplitude on our experimental systems.

ACC2016@BostonでWorkshop