|
In our device, the polymer medium is positioned on a MEMS scanner
with x/y-motion capabilities of about 120 µm. Because actuation
distances are typically so small and components have small masses,
the positioning delays are much smaller than in disk drives. The
device also includes thermal position sensors that provide x/y-position
information to the servo controller. MEMS-based storage devices
require a closed-loop servo system to write uniformly spaced data
tracks and to read them back with sufficient accuracy to ensure
a low error rate. As the areal density of such a system is being
increased to the Tb/in² regime and beyond, the performance
requirements for the servo system become severe. In general, the
servo system in such a storage device has two functions. First,
to locate the target track to which information is to be written
or read from, starting from an arbitrary initial position of the
scan table carrying the storage medium. This is achieved by the
so-called seek-and-settle procedure. During seek, the scan table
is moved rapidly to position the read/write probes close to the
beginning of the target track. This is followed by the settle mode,
which consists of a smaller additional displacement in the cross-track
direction to position the probes on the center of the target track.
In our prototype MEMS-based storage device, the thermal position
sensors provide x/y-position information to the servo controller
during the seek/settle mode of operation.
The second function of the servo system is to maintain the position
of the read/write probes on the center of the target track as they
are being scanned along the length of this track during normal read/write
operation. This is achieved by the track-follow procedure, which
controls the fine positioning of the read/write probes in the cross-track
direction and is critical for reliable storage and retrieval of
user data. It is typically performed in a feedback loop driven by
a medium-derived position-error signal (PES) that indicates the
deviation of the current position from the track-center line. There
are two types of track-follow servo architectures in practical use.
In embedded servo, segments of position information are interspersed
with the data of a track. In dedicated servo, certain probe tips
and corresponding storage fields are dedicated solely to providing
position information to the servo system.
In our approach to MEMS-based storage, dedicated servo fields are
employed to achieve both timing synchronization and servo control,
i.e., a small number of storage fields is reserved exclusively for
timing recovery and servo-control purposes. This approach is based
on the concept of vertically displaced bursts, arranged in such
a way as to produce two signals that guarantee a uniquely decodable
PES. The servo marks for the in-phase signal are labeled A and B
bursts, those for the quadrature signal C and D bursts. Each of
the four types of bursts is pre-written in a separate servo field.
These four servo fields are identical, except for the position of
the servo marks in the cross-track direction. The A, B, C, and D
servo fields are placed in the 2D array in such a way that they
can always be accessed in parallel, irrespective of the addressing
scheme. Similarly, the same strategy can be used for obtaining timing
information by implementing dedicated clock fields. The basic concept
is to have continuous access to a pilot signal for synchronization
purposes. Servo and timing functions can also be combined in the
same dedicated fields. Because of the large number of levers in
our proposed storage system, this solution appears to be advantageous
in terms of overhead compared to the alternative dedicated servo
architecture.
Our research focuses on the design and characterization of servomechanisms
to achieve very accurate positioning, track-following, and short
access times in our probe-based storage device. Based on a discrete
state-space model of the scanner dynamics, a controller designed
using the linear quadratic Gaussian approach with state estimation
achieves seek times of about 4 ms in a ±50 µm range.
Moreover, feasibility experiments on closed-loop track following
using solely the thermal position-sensor signals yield typical position-error
standard deviation of approximately 2 nm.
|
IBM Research
