Modeling a MEMS probe-based storage device

Modeling a MEMS probe-based storage device
Maria
1
Varsamou ,
1Department
2
Pantazi
Angeliki
of Electrical and Computer Engineering, University of Patras, Greece
e-mail: mtvars, [email protected]
2IBM
Research - Zurich, 8803 Rüschlikon, Switzerland
e-mail: [email protected]
Introduction
 Exact simulator that enables the reliability study of a probe-based MEMS
storage device even under extreme noise conditions and various kinds of
external disturbances.
Probe-based Storage System Simulator
Movement
Parameters
Read-back Signal
Evaluation GUI
Channel
Model 1
Movement in
X,Y axis
Scanner
Movement
Model
External
Disturbance
Probe-based Storage Device
Read-back
Signal
Stored Data
Channel
Model 2
 Ultra-high density storage device based on AFM techniques (> 1Τbit/in2) [1].
 Thermo-mechanical recording in thin polymer films.
 Parallel operation of multiple probes to compensate for low data rate of
individual probes.
 The medium is moved underneath the probes on Χ/Υ axes via an
electromechanical microscanner.
Channel
Statistics
Data Detection
Α
0
-0.5
-1
40
...
Magnet
Stored Data
20
0 0
Decoding/
Error Correction
Channel
Model Ν
10
20
Error Statistics
Electronics and
Media Noise
Multiplexer
Ν Storage Fields
Coil
User Data
Thermal Position
Sensors
– Movement using two voice-coil actuators, one for each
direction X,Y – Movement area: 120 x 120 μm2
– Mass balancing for disturbance rejection -> 100 times
acceleration reduction.
Tip radius ~ 3-5nm
 The simulator incorporates all system functionalities, i.e. the microscanner
movement and the sensing capabilities, the read-back signal of multiple
storage fields and the complete data mapping and coding scheme.
 Based on Matlab/Simulink standard and custom functions/models.
 The reference movement signal is generated according to the line offset from
the beginning of the storage field.
 External disturbances in the form of acceleration measurements over time
can be applied as an input.
– Two pairs of thermal sensors provide X/Y position
information of the microscanner to the servo controller
– Sensitivity 1 - 2 nm
Thermo-mechanical write/read
Line _Offset
Line Offset
X
[ShockTime ShockSigx ]
x-disturbance
Y
[ShockTime ShockSigy ]
y -disturbance
LineScan
Write pulse: 1ms – 5ms
Resistive heater temperature: 350oC –500oC  Tip temperature ~ 200oC –300oC
Write Force: 50nN – 300nN (Electrostatic force pulse ~ 3V – 10V)
External
Disturbance
ReadOut1x16
Resistive
heater
1
Readback Signal
X
substrate
substrate
Writing “0” does not alter the
polymer surface
Sensing
current
Positioning system
 Exact models of the microscanner and the thermal position sensors based on
measurements on a actual prototype are included.
 The exact LQG algorithms that control the microscanner movement on both
X,Y axes are implemented.
 The medium-derived PES that assists the control algorithms is generated.
More cooling by substrate
=> T => R
Less cooling
by substrate
Storage fields layout
 Each probe performs write/read/erase operations on a dedicated storage field
~ 100μm x 100μm.
 The data are stored on constant symbol distance on X-axis, forming sequences
of indentations which are stored on constant line distance on Y-axis.
Field size: 100μm x 100 μm
Microscanner velocity: 2.5nm/μsec
Symbol distance: 20nm
Χ-Axis reference movement
Servo field
Data field
Thermal position sensor
10
20
 For every (X,Y) value of the probe movement, the current line and the
current symbol inside the line is calculated. Depending on the next stored
symbol, the A,B,C pattern is decided. The (X,Y) depth value in the pattern
gives the read-back sample.
 The model also includes the various noise sources that affect the read-back
signal [2], such as the electronics noise and the media noise (due to the
anomalies on the polymer surface), based on measurements on an actual
prototype.
A) Simulated vs. Experimental read-back signal
Storage field representation
 Distinct models to generate the read-back signal from each storage field –
Can be parameterized for any number of fields.
 A 3D indentation model (b) produced by experimental data regarding the
actual indentation (a) is used.
Experimental
B) Complete read operation simulation when the system
is affected by a specific external disturbance
•Read procedure of a single line assuming no errors
during recording
•Microscanner velocity: 2.5nm/μsec
•Symbol distance: 14nm
•Electronics + Media NoiseSNR: 12 dB
Movement in a storage field
...
0 0
Simulation Results
Read-back Signal
Generation
ΔR/R ~ 10 per nm
...
20
16 Data Fields
-4
... ... ...
-1
40
“1”
Resistive heater temperature : 100oC –200oC
...
...
-0.5
substrate
polymer
READ
0
 Based on the read-back signals, the procedures of symbol detection, data
decoding and error correction can be perfomed to recover the initial user
data.
 Statistics regarding the bit errors that appear in each storage field, the
symbol errors that affect the ECC codewords, the total number of codewords
that cannot be decoded are produced.
Scanner Movement
Y
write
current
01
B
0.5
Dataflow Graphical User Interface
Scope
Positioning System
Scan direction
Υ-Axis
00
0.5
Scan Table
Multiplexer
Storage medium
on X/Y scanner
WRITE
Read-back signal generation system
System Simulator
Atomic Force Microscopy (AFM) techniques use nanometer-sharp tips for
imaging the surface of materials down to the nanometer scale. Such tips are
exploited for creating storage devices capable of storing information with
much higher density than conventional devices. This work presents a very
accurate simulator of such a device and verifies its accuracy using experimental
data of a prototype platform.
2D cantilever chip array
and Theodore
1
Antonakopoulos
Simulator
Seek
Time (sec)
Density
1.2 Tbit/in2
3.0 Tbit/in2
4.0 Tbit/in2
Preamble
Data
0.4
Normalized Depth
Βάθος
Κανονικοποιημένο
Χ-Axis
Symbol
distance
27 nm
18 nm
15 nm
Y-Axis reference movement
Initial
position
Time (sec)
Χ-Axis
0.2
0
Reference Movement
-0.2
-0.4
-0.6
-0.8
-1
40
30
Y-Axis
10
0
0
Απόσταση (nm)
1 μm
All-‘1’ sequence
Sync
Track ID
 A preamble is used at the beginning of each line for synchronization purposes.
 Dedicated servo fields with predefined indentations sequences are used for
generating a medium-derived positioning error signal (PES).
 During write/read operation the microscanner is moved with a constant velocity.
Data Controller Architecture
Demux
ECC
External Disturbance
(b)
00
Mux
01
1
AFE
Interleaving
Read-back signal from one storage field
X-Axis
 Normally, 3D huge arrays with depth values at nanometer-level accuracy
for all stored lines for every storage field would be necessary for the
read-back signal generation.
 Due to the (1,k)-constrained code, there are only three allowable
combinations of successive symbols -> (00) , (01), (10)
 Three 3D patterns (A,B,C) can be used along with the data sequences of
‘1’ and ‘0’ stored on the device.
(1,k) Line Coding
CRC
40
30
Distance (nm)
(a)
Multiple Sectors Data
20
10
Απόσταση (nm)
Distance (nm)
Υ-Axis
User
Data
(Sector)
20
0.5
0
0
-0.5
-0.5
-1
40
-1
40
0.5
0
(1,k) Line Coding
(RS Encoder)
-0.5
Write Operation
-1
40
20
0
Read Operation
Mux
User
Data
(Sector)
CRC
ECC
(1,k) Line
Decoding
5
0
10
0
A
10
5
0
15
20
20
0
B
0
5
20
15
10
C
Detection
40
0
1
0
0
0
0
20
40
60
80
0
1
0
1
0
1
0
0
0
1
240
260
280
0
30
AFE
20
(RS Decoder)
(1,k) Line
Decoding
 The simulator is a flexible tool that can be used to determine the reliability
of a probe-based storage device under various noise conditions, evaluate
new microscanner technologies and control architectures, as well as other
parameters that affect the device functionality and performance.
 Although it is based on the thermo-mechanical recording mechanism, it can
be easily modified to simulate any probe-storage technology.
Example sequence of A,B,C patterns
Demux
De-interleaving
20
20
15
All sectors are corrected
successfully
Conclusions and References
10
0.5
Errors in one storage field
Detection
10
Y
Storage
0
B
X
C
A
A
A
100
B
120
C
140
160
B
C
180
B
200
C
220
A
A
B
C
300
[1] A. Pantazi, A.Sebastian, et al, “Probe-based ultrahigh-density data storage
technology,” IBM J. Res. and Dev., vol. 52, no. 4/5, pp. 493–511, 2008.
[2] A.Sebastian, A. Pantazi, H. Pozidis, and E. Eleftheriou, “Nanopositioning for
Probe-Based Storage Device,” IEEE Control Systems Magazine, pp. 26–35,
August 2008.