Control of mechanical systems

steamgloomyΗλεκτρονική - Συσκευές

15 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

99 εμφανίσεις

Control of mechanical systems
in data storage

Collaborations in Data Storage

STMicroelectronics


Agrate


MI

Computer Mechanics Lab
-

UC Berkeley

Prof. Masayoshi Tomizuka


prof. Roberto Horowitz

Center for Magnetic Recording Research (CMRR)
-
UC San Diego


Prof. Frank Talke

Data Storage Center


Carnegie
-
Mellon University

Prof. William Messner

HITACHI
-
IBM Almaden Research Center

Data stored on concentric
circular tracks

Operating modes


Track Following when
performing R/W operations


Track Seek when changing
track

Actuators:


Spindle motor at a constant
speed, between 3600 and
18000 rpm


Brushless


Provides the rotation of
the disk stack assembly


Voice Coil Motor (VCM)


DC motor


Moves the heads over
the disk surface

Hard Disk Drives

Capacity indexes:


TPI: Tracks Per Inch from
25k to 60k


BPI: Bits Per Inch up to
500kb


BPIxTPI: 30Gb per in
2

Spindle

VCM

R/W Heads

Channel


Microelectronics
(analog and digital)


Code theory


Digital filters

BER < 10
-
9

Spindle


Electric Drives


Mechanics

Elasticity

Eccentricity

Torque ripple

Bearings

Disk modes

Technologies, problems and objectives

Subsystem

Technologies

Problems and objectives

Suspension


Mechanics


Materials


Aerodynamics

Resonant
modes

Fly height:
15 nm

Speed: 120
km/h

Heads


Electric Drives


Digital control

Seek Time:
5

ㄲms

PES: 5

㜥 Tr.

(i.e. 50 nm
precision in
servo
-
positioning)

Servopositioning

Head Servo
-
positioning

Servo sector

Sector

Servo sector



synchronization signals



Track number



Head postion w.r.t. track center (PES)

Sampling

F = N x rpm / 60 ; N = #Servo sectors

F = 8

㌰3k


The head servopositioning system

N
RRO
Non
Repetitive Run Outs

Windage

Effect of air turbulence on head
support

(wind speed may reach up to 100 km/h)

-
4
-
3
-
2
-
1
0
1
2
3
4
-
3
-
2
-
1
0
1
2
3
-
4
-
3
-
2
-
1
0
1
2
3
4
-
3
-
2
-
1
0
1
2
3
-
4
-
3
-
2
-
1
0
1
2
3
4
-
3
-
2
-
1
0
1
2
3
Ideal track center

Actual track center

n

n
-
1

RRO Repetitive Run Outs

Track deformation


Due to initialization, heating, bearing
imperfections


Track pitch: <1


⡈䑄⁌潷 䕮dF


RRO: repetitive disturbance, locked in
phase with disk rotation


Amplitude may be more than track pitch


Frequencies: harmonics of rotational
frequency (5400 rpm


㤰9䡺)

Hard Disk

Interesting, multi
-
disciplinary case of
study:

Modeling of complex mechanical systems

Identification and control

Power electronics and electric drives

Vibration suppression

Data coding, magnetic materials,
aerodynamics, signal processing …


Research Activities in HDD Servo

Modeling and Simulation

Digital control algorithms design and
test

Active vibration suppression

VCM voltage command

Modeling and Simulation

Experimentally tuned simulator:


suspension

e
-
block

l
s

θ
vs

l
x

K
vs
, B
vs

[
rad
]
Arm
Resonances
Arm
Resonances
1/s
1/s
1/s
1/s
1/J
1/J
..
[
Nm
]
[
rad
/s
2
]
[
rad
/s]
t
F
F
F
.
[
rad
]
Non
Linear
Friction
Model
Non
Linear
Friction
Model
TPR
TPR
RRO
RRO
NRRO
NRRO
PES
[
Tr
]
[
Tr
]
Windage
Windage
Bias
&
Flat Cable
Bias
&
Flat Cable
[
rad
]
Arm
Resonances
Arm
Resonances
1/s
1/s
1/s
1/s
1/J
1/J
..
[
Nm
]
[
rad
/s
2
]
[
rad
/s]
t
F
F
F
.
[
rad
]
Non
Linear
Friction
Model
Non
Linear
Friction
Model
TPR
TPR
RRO
RRO
NRRO
NRRO
PES
[
Tr
]
[
Tr
]
Windage
Windage
Bias
&
Flat Cable
Bias
&
Flat Cable
0
10
20
30
40
50
60
70
80
90
-0.06
-0.04
-0.02
0
0.02
0.04
0.06
0.08
[Samples]
[Tracks]
PESrro
Average
data1
data2
data3
data4
data5
data6
data7
data8
data9
data10
data11
data12
data13
data14
data15
data16
Dual Stage Actuation
-

Piezo

LDV

suspension

spindle

gold
-
coated

slider

Dual stage actuator can be simulated

Piezo (experimentally tuned)

Mems (multi
-
body mechanical system)

Dual Stage Actuation
-

Piezo

10

3

10

4

10

-
3

10

-
2

10

-
1

10

0

milli
-
actuator data

f [Hz]

mag [

mu m/V]

10

3

10

4

-
200

-
100

0

100

200

f [Hz]

phase [

deg]

Characterization of piezo suspension (also vs. fly
height)




“Modeling Product Variabilities of Dual
-
Stage Suspensions for Robust Control”
-

M. Rotunno,
R. Oboe, R.A. de Callafon
-

ISPS 2002


Santa Clara (USA)


June 2002


“LQG / LTR control of a dual stage actuator hard disk drive with piezoelectric secondary
actuator”


A.Beghi, R.Oboe


European Control Conference ECC 2001


Porto (Portugal)


September 2001

Windage modelling

Head position measured with LDV

Closed loop and open loop identification

R(s)

P(s)

P(s)

w

w

y

y

Bias

Digital servo control design and test

VCM

Preamp

Channel

Controller

DAC

Power

Drive

External Board

Controller

DAC

Estimated state feedback controller with
disturbance observer

Xs1: estim. position

Xs2: estim. velocity

Xs3: estim. disturbance

Xs4: u(k
-
1)



“Loop shaping issues in hard disk drive servo system design”
-

A.Beghi, R.Oboe, P.Capretta,
F.Chrappan Soldavini
-

Advanced Intelligent Mechatronics AIM 2001


Como (Italy)


July 2001



“Optimal Estimation for Disk Drive Head Positioning System”
-

D.Ciscato, R.Oboe, G.Picci,
E.Colecchia, G.P.Maccone, G.Traversa
-

The 2nd Annual Magnetic Recording Conference on Recording
Systems
-

Hidden Valley, Pittsburgh PA (USA), June 12
-
15 1991

Servo algorithms

Repetitive control



0
90
180
270
360
450
540
630
720
810
900
990
1080
1170
1260
1350
1440
1530
1620
1710
1800
1890
1980
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Tracce^2/H
z
Frequency (Hz)
0
90
180
270
360
450
540
630
720
810
900
990
1080
1170
1260
1350
1440
1530
1620
1710
1800
1890
1980
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0
90
180
270
360
450
540
630
720
810
900
990
1080
1170
1260
1350
1440
1530
1620
1710
1800
1890
1980
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Tracce^2/H
z
Frequency (Hz)
FFT PES

Standard Controller

Repetitive Controller

0

500

1000

1500

2000

2500

3000

3500

0

0.5

1

1.5

2

2.5

3

FFT PES

Frequency [Hz]

Tracks/Hz

10

2

10

3

-

50

-

40

-

30

-

20

-

10

0

10

Frequency (Hz)

Magnitude (dB)

Larger Estimator Bandwidth

Larger Estimator Bandwidth

Estimator Bandwidth : 500 to 900 Hz

Estimator Bandwidth : 500 to 900 Hz

Controller Bandwidth : 500 Hz

Controller Bandwidth : 500 Hz

10

2

10

3

-

50

-

40

-

30

-

20

-

10

0

10

Frequency (Hz)

Magnitude (dB)

Larger Estimator Bandwidth

Larger Estimator Bandwidth

Estimator Bandwidth : 500 to 900 Hz

Estimator Bandwidth : 500 to 900 Hz

Controller Bandwidth : 500 Hz

Controller Bandwidth : 500 Hz

Sensitivity Function

Ts
=m

Tc
Tc
u (k,0)
u (k,1)
u (k,i)
u (k,m
-
1)
x (k,0)
x (k,1)
x (k,i)
x (k,m
-
1)
k

T
s
(k+1)

T
s
i

Tc
(i+1)Tc
Ts
=m

Tc
Tc
u (k,0)
u (k,1)
u (k,i)
u (k,m
-
1)
x (k,0)
x (k,1)
x (k,i)
x (k,m
-
1)
k

T
s
(k+1)

T
s
i

Tc
(i+1)Tc
Pubblications:



“D
isturbance

rejection

in

hard disk drives

with
multi
-
rate

estimated state feedback


R. Oboe, F. Marcassa
-

To appear in
Mechatronics
2002



Berkeley

(
USA
)


December

2002

Servo algorithms

Multirate control



Objectives (from literature):


Reduce command discontinuities


Reduce phase delay


Enlarge control BW


Reduce TMR

Results:


Short Seek improvement


Analytical evaluation of closed
-
loop
sensitivity function


Worsening in sensitivity observed
(analytically and experimentally)

K

Zoh

plant

Estimator

H

Tc

RRO&NRRO

u(k,i)

x(k,i)

)

Plant

H

Ts

PES

Position

Target

Ts

= mTc

2

3

4

5

6

7

8

x 10

-
3

1.2288

1.2288

1.2288

1.2288

1.2289

1.2289

1.2289

1.2289

1.2289

1.229

x 10

4

short seek (1 track): sr vs mr

time [sec]

head position [tracks]

multi
-
rate

single
-
rate

2

3

4

5

6

7

8

x 10

-
3

1.2287

1.2288

1.2289

1.229

1.2291

1.2292

1.2293

1.2294

x 10

4

short seek (5 tracks)

time [t]

head position [tracks]

single
-
rate

multi
-
rate

“DISTURBANCE

REJECTION

IN

HARD

DISK

DRIVES


WITH

MULTI
-
RATE

ESTIMATED

STATE

FEEDBACK”

Federico

Marcassa

and

Roberto

Oboe


Control

Engineering

Practice

2003



In

press


IVC
D
D
x
x
S
S
IVC
D
D
x
x
S
S
Mode Switching Control

y
R
x
S1
u
Disturbance
PROCESS
ESTIMATOR
-
a
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
+
-
y
a
>Y
D
x
Sn
x
Si
y
R
x
S1
u
Disturbance
PROCESS
ESTIMATOR
-
a
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
+
-
y
>Y
D
x
x
y
R
x
S1
u
Disturbance
PLANT
ESTIMATOR
-
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
SWITCH
+
-
y
>Y
D
x
e
Threshold
y
R
x
S1
u
Disturbance
PROCESS
ESTIMATOR
-
a
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
+
-
y
a
>Y
D
x
Sn
x
Si
y
R
x
S1
u
Disturbance
PROCESS
ESTIMATOR
-
a
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
+
-
y
a
>Y
D
x
Sn
x
Si
y
R
x
S1
u
Disturbance
PROCESS
ESTIMATOR
-
a
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
+
-
y
>Y
D
x
x
y
R
x
S1
u
Disturbance
PLANT
ESTIMATOR
-
y
y
n
TRACK FOLL.
CONTROLLER
SEEK
&
SETTLING
CONTROLLER
u
t
u
s
SWITCH
+
-
y
>Y
D
x
e
Threshold
140
160
180
200
220
240
260
280
300
-
0.3
-
0.25
-
0.2
-
0.15
-
0.1
-
0.05
0
0.05
0.1
Samples
Tracks
MSC+IVC
MSC
140
160
180
200
220
240
260
280
300
-
0.3
-
0.25
-
0.2
-
0.15
-
0.1
-
0.05
0
0.05
0.1
Samples
Tracks
140
160
180
200
220
240
260
280
300
-
0.3
-
0.25
-
0.2
-
0.15
-
0.1
-
0.05
0
0.05
0.1
Samples
Tracks
MSC+IVC
MSC
140
160
180
200
220
240
260
280
300
-
0.3
-
0.25
-
0.2
-
0.15
-
0.1
-
0.05
0
0.05
0.1
Samples
Tracks
MSC

MSC+IVC

Seek: Settling time phase

Servo algorithms

Initial Value Compensation



“Initial value compensation applied to disturbance observer
-
based servo control in HDD”


R. Oboe, F.
Marcassa
-

Advanced Motion Control 2002 (AMC2002)


Maribor (Slovenia)


July 2002


Objectives:



Bumpless switching



Limitation of transients
during settling phase


Solution:


Act on estimated states in
order not to have a transient in
both command and position
estimate

Vibration compensation

MEMS

rotational

accelerometer

Feedforward

compensator

controller

PES

13

kTPI

, 5400 RPM HDD

mounted

on a shaker

VCM

Amp

i

ff

MEMS

rotational

accelerometer

Feedforward

compensator

controller

PES

13

kTPI

, 5400 RPM HDD

mounted

on a shaker

VCM

Amp

i

ff

Standard Controller

FF Filter

FF Gain

Compensation up to 600 Hz

Active vibration damping

Active suspension with two
piezo strips:

Actuation

Sensing

Active damping of resonant
modes

Pros



Bandwidth 50kHz


䤽GU


Robustness against variations in R
T

L
VCM

Cons



Dissipation: R shunt


Linear Amplifier


Silicon area: Linear Amplifier


Digital current loop


數灥湳楶攠䄯A

1
R
T
+sL
Vcm
[V]
[A]
J*s
Kt
Kt
Bemf
-
+
I(s)
U(s)
R
Linear Power Amplifier
&
Phase Shaping
[V]
Command
VCM control: Current Mode

Controllo VCM: Voltage Mode

Pros



Good performance without current meas.


Cost reduction


Migration toward SOC

1

R

T

+sL

Vcm

[V]

[A]

J*s

Kt

Kt

Bemf

-

+

I(s)

U(s)

Power

PSM

[V]

Command

Digital

Prefilter

Multi

-

rate

1

R

T

+sL

Vcm

[V]

[A]

J*s

Kt

Kt

Bemf

-

+

I(s)

U(s)

Power

PSM

[V]

Command

Digital

Prefilter

Multi

-

rate

Cons



Pre
-
filter cancels out the electrical dynamic
of VCM


R varies
±
30%


On
-
line estimation of R required

Solution: Extended Kalman Filter to estimate R

On
-
line pre
-
filter adaptation






Presently developed for Seagate, IBM and STM

VCM control: Voltage Mode

From servo controller
H1
Ho
VCM+
voltage dr.
EKF
Rt
Tc
Tc
Ts
Head
position
Ts = 8 Tc
Gain matrix and

ff compensator

Pos, vel, current…

x
ref

x
est

+

-