We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime. Upcoming SlideShare. Like this presentation? Why not share! Embed Size px. Start on. Show related SlideShares at end. WordPress Shortcode. ShirleyGoodwin Follow. Published in: Education. Full Name Comment goes here. Are you sure you want to Yes No. Be the first to like this. No Downloads. Views Total views. Actions Shares. Embeds 0 No embeds. No notes for slide. DETAIL Description The hard disk drive is one of the finest examples of the precision control of mechatronics, with tolerances less than one micrometer achieved while operating at high speed.
Gasket is used to seal the contact between base plate and top cover. The environment inside the enclosure must be maintained clean. Any particle at the head-disk interface can cause abrasion of the disk resulting in loss of data and increase in number of particles. Therefore, assembly of the drive is done in a clean room to ensure particle-free enclosure.
Particles created during the operation of drive by sudden contacts between disk and slider are thrown out of the spinning disk by the centrifugal force and eventually trapped in the filter, placed in the empty space inside the enclosure. A special feature of both the base casting and top cover is the crash stop. These are small mechanical protrusions from the base plate and top cover used to restrict the movement of the actuator beyond the desired space. Electronic components of an HDD can be categorized according to the following functions: 1.
Electronics for controlling various operations such as read data, write data, transfer data between HDD and host etc of the disk or the disk controller 4. Electronics for interface with the host system, and 5. Several of these functional components are often combined in a single chip. Bit is the smallest unit of recorded information on magnetic media.
It is a tiny piece of the disk surface and contains binary information. Since the disk is spinning and the head is held at a point, the write current magnetizes a circular path on the disk with alternating polarity of magnetization. Type of magnetization on the medium depends on the polarity of write current. A transition in the write current waveform creates a transition of magnetization on the disk. The circular pattern of magnetization created on the disk is called a track Figure 1.
A new track can be created by repositioning the write head to a new point on the disk radius. In a typical 3 12 inch HDD, 70, to , tracks exist on each surface of a disk. The polarity of write current is altered according to the binary bits to be recorded. A 1 in the binary data causes the polarity to be reversed, otherwise it is unchanged.
The minimum distance between two magnetic transitions on the. We should not create a new track that may erase significant part of an adjacent track. Similarly, while data is read from a track, the interference from the magnetic transitions recorded on an adjacent track should be as low as possible. The storage density or bit density of a magnetic recording system is the inverse of Abit. Inverse of Lbit and inverse of Wtrk are known as the linear density and track density, respectively. Imperial units are widely used in the HDD industry and bits per inch2 , bits per inch and tracks per inch are the units of areal density, linear density, and track density, respectively.
When the host system sends data for recording, it is recorded in chunks of bytes. Each of these chunks is called a data block. Bits of a data block are recorded sequentially along the track. In order to locate a data block on the surface, they are tagged with an identification number. From system level point of view, each data block is assigned with a Logical Block Address or LBA, starting at 0 and ending at a number appropriate for the capacity of the entire drive.
These block addresses, however, are not suitable for low level access to the data. Access at the low level uses head number , cylinder number , and sector number assigned to each LBA. An HDD may contain one or more disks with data recorded on both surfaces of a disk. Data on each surface is accessed for reading as well as writing using a separate head for that surface. Each used surface of the disk stack is identified by the corresponding head number; for an HDD with 8 usable surfaces, heads are numbered 0 to 7.
There are tens of thousands of tracks on each surface, numbered 0 on the outermost track and increasing inward. If we consider a stack of disks, then track 0 of all disks form a cylinder and is identified as cyl 0. One of the cylinders is shown in Figure 1. Each cylinder is assigned with a unique identification number.
These special patterns are known as servo sectors Figure 1. Number of servo sectors per track is the same on all surfaces in an HDD. There are typically servo sectors in any HDD produced these days. Tracks and sectors are identified using special magnetic patterns written on the disks during the production of HDD. In the earlier generation drives with four or more disks, entire surface of one disk used to be dedicated for recording these special patterns known as the servo pattern.
Since all heads are moved simultaneously by a single actuator, it can be assumed that when the head on the servo surface surface containing the servo patterns is positioned on the N th track, all other heads are also positioned on the N th track of their respective surfaces. This assumption started to fall apart with increasing track density. Moreover, the scheme of dedicated servo surface is not suitable for drives with few disks.
Current state of the art in recording technology allows storage of approximately Gbytes on a single disk; for many applications a drive with one disk meets the storage requirement. For a drive with 4 disks 8 surfaces , servo overhead is Servo overhead is increased if fewer disks are used. Both of these issues, thermal expansion and increasing servo overhead, associated with HDDs with a dedicated servo surface can be resolved using an alternative servo scheme where the servo patterns are written on every track interleaved with the data blocks.
With this scheme in place, the servomechanism can control the position of any head using servo information written on the corresponding surface as the feedback. However, unlike in the scheme with dedicated surface, feedback signal is available only at discrete sampling points. The method used in the earlier generations of drive with position information encoded on a dedicated surface is called the Dedicated Servo scheme, whereas the other scheme having position information encoded on all surfaces is called the Embedded Servo or Sectored Servo.
The segment of the track containing the servo information in an embedded servo drive is known as servo sectors, and the section between two servo sectors is allocated for storing data bits. Servo patterns, both in dedicated and embedded case, are created during manufacturing of the drive and the firmware of the HDD takes care not to overwrite them in any situation.
The first of these modes is known as Track Seek while the second mode is called the Track Following. Besides, there must be a smooth transfer between the two modes. It is impossible to meet the specifications of both modes using a single control law. Two controllers can be made to produce desired performances if each is designed and tuned independent of the other.
However, while designing and implementing such controller, special attention must be paid to ensure that sudden change in the amplitude of control signal does not occur at the time of switching between modes. Sharp discontinuity in the control signal excites the lightly damped resonances of the actuator. Occurrence of such jerk increases the time it takes to settle and, therefore, must be avoided.
Such a scheme is called constant data rate recording. Frequency of the clock signal used during reading or writing of data remains the same on all tracks, i. If the linear density is optimized on the outermost track then the transitions on an inner track are too close to each other and can not produce significant read back voltage.
On the other hand, if the clock frequency is chosen to achieve optimum linear density on the innermost track, the transitions are sparsely created on outer tracks. In the constant data rate recording, same number of data blocks are stored on all tracks.
However, the circumference of the outer track is larger than that of an inner track and, therefore, it makes better sense to store more number of data blocks on outer tracks. It is possible to achieve the ideal solution to this problem if either radiusdependent clock frequency or radius-dependent spindle speed is used. In CDROM, the speed of the spindle motor is continuously adjusted as the head moves from one track to another.
This ensures constant linear density of recording and hence constant areal density. Radius-dependent clock frequency is not used in any storage device. Each zone has its own recording frequency which optimizes the linear density on the innermost track of that zone. Frequency of recording is increased from inner zone to outer zone. All tracks within a zone use constant data rate recording and contain equal number of data blocks.
A schematic illustration of zoned-bit recording in Figure 1. There are more data blocks per track in an outer zone than an inner zone, which is clearly shown in this diagram. More is the number of zones, better is the utilization of storage space. The extreme end is to assign one track per zone and the clock frequency is optimized for each track individually to achieve optimum linear density on all tracks. This also results in constant areal density. The marginal improvement due to increase in number of zones is significant when few zones are used. There is approximately Commercially available drives use zones.
The disk drive industry has evolved through dramatic changes in the five decades of its existence. Demand for larger capacity, need to have smaller dimensions for specific applications, requirement of data transfer rate compatible for fast on-line applications etc are some of the driving forces behind the extraordinary growth of this industry. Desire to have large capacity in a smaller dimension is directly linked to the demand for ever increasing storage density. The most obvious change that took place in the hard disk drive industry over last four decades is the phenomenal increase in the storage capacity of HDD.
As recent as in early s, a typical PC used to be shipped with an HDD capable of storing approximately megabytes of data. Today, even a computer for home or personal use comes with HDD storage of 80 gigabytes or more. Demand of storage capacity caused by larger size of programs and multimedia data has driven the manufacturers to increase the capacity of their products. While the capacity continued to increase, the price of HDDs experienced a continuous fall. This was made possible by increasing the amount of data stored on each surface of the disk, i. Head technology has been continuously improved.
Improved technologies have enabled new generation media to reliably hold magnetic domains of smaller size. Smoother disk surface, better quality of lubricant and better air bearing technologies allow the slider to fly in closer proximity of the disk so that the bit size is reduced. Typical fly heights in were 25 nanometers nm. Today they are about 5 nm. The read-write electronics and data encoding schemes have played their part in improving bit density by enabling detection of information reliably from ever smaller data signals contaminated by the surrounding noise.
Excellence in design and production maintained the steady growth in the areal density which was accelerated time to time by availability of new technologies. Introduction of new technologies has always made an impact on the strive for improving areal density. Drives in those days used inductive heads for both reading and writing. The ever-increasing areal density is sustained by increasing both track density and linear density.
Starting in the later part of s, the track density TPI has been growing at a rate faster than that of the growth in linear density BPI. Table 1. It is evident that both the densities have been increasing. This trend in BAR suggests that the increase in track density is taking place at a rate higher than that of linear density. Capability of sustaining the rate of increase in TPI will play a more important role in pushing continuously the recording density higher.
It is generally believed in the HDD industry that areal density still has a lot of room to grow, and it is expected to reach 50 Terra bytes per squared inch in the future. The growth in track density must be sustained at an appropriate rate to make this projection a reality. Introducing a new form factor standard requires coordination between manufacturers of computers, producers of HDD, and other support industries that produce components for HDD as well as for computers.
There is a natural resistance in the industry to changes in form factor unless there is a compelling reason to do so. At the emergence of laptop computers, the HDD industry created new, smaller drives to save space as well as power, a very important consideration in the world of mobile computing.
This was a necessity that the HDD industry and other support industries eagerly acted to meet. The most. The form factor usually refer to the width of the drive enclosure or the diameter of the disks used. However, in some cases, form factor represents neither of the two. For example, the width of a 3 12 inch HDD enclosure is 4 inch and the disks used in these drives have diameter larger than 3.
This particular form factor got its name from the fact that the size fits well in the space originally allocated for 3. Phenomenal increase in areal density achieved over last few decades allows the manufacturers to increase storage capacity with simultaneous decrease in the size of hard disk drive suitable for applications such as laptop computers, cameras, and other small devices.
The trend in form factors is downward: to smaller and smaller drives. The first form factor used in a PC 5 14 inch have now all but disappeared from the mainstream PC market, and the 3 12 inch form factor dominates the desktop and server segment.
- Life Is Like Music Play On;
- Hard disk drive : mechatronics and control;
- Hard disk drive: mechatronics and control;
- While I Was Sleeping: A true story about love, hope and determination.
For laptop market, the dominant form factor is 2 12 inch. HDDs of smaller form factor is the most desirable choice for the emerging market of digital entertainment with devices such as digital camera, MP3 etc. The micro drive of IBM is less than 0. Continuous growth of areal density will initiate soon a transition to the 2 12 inch form factor for the desktop and server drives. The reasons for this shrinking trend include the enhanced rigidity of smaller platters, reduction of mass to enable faster spin speeds, and improved reliability due to enhanced ease of manufacturing. The capacity of this drive was 2 Gbyte in and 4 Gbytes in Toshiba announced their plan to introduce 8 Gbyte micro drives in which will use perpendicular recording technology.
Smaller form factor drives usually come with lower performance than a larger drive. However, because of the small size of disk s , it can be spun up very fast. A small drive can spin up to full speed in less than half a second. This makes it possible to spin down the drive frequently, which is an essential feature for portable computers.
Another significant change that the HDD industry experienced in the past is the trend in the access time and data transfer rate. For applications that require faster data rates, speeding up the disk rotational speed has reduced the latency component of access time and increased the speed of data flow from the heads.
There has been a steady progression over the years from. The higher data rates coming into the head due to the higher rotational speed and bit density have introduced challenges in drive electronics to be able to reliably process the data. The time required to move the head to a new track position and get it ready for reading or writing is called access time. It is the sum of the time required to find the new track seek time , time required to settle on it settling time , and latency.
One-third stroke seek times are around. Low access time is very important in computer applications because the number of data transfers is so high that a small increase in the time required for each transfer causes considerable overall delays in processing data or running programs. Hard disk drives are expected to maintain their position as the primary on-line, non-volatile storage device for computing systems in the foreseeable future because of their advantages of large capacity with fast access but at low cost. Though access time is faster in semiconductor memory, its higher cost per stored bit makes it less attractive for mass storage.
This is expected to be continued in the future. However, the semiconductor devices will have their usage for low capacity functions such as on-board RAM and flash memory for small and portable applications such as digital camera and MP3 player. The optical storage devices cannot compete with hard drive technology in either storage capacity or data access speed and will continue to fill the niche functions of high capacity data portability and program distribution.
In addition, with the proliferation of consumer-oriented devices for which data storage is a critical capability, non-computer applications are expected to contribute significantly to future growth in overall disk drive demand. There still exists wide opportunity for the amazing technological development of hard drives to continue. A consortium of industry, academic, and government participants Information Storage Industry Consortium, NSIC has recently targeted gigabits per square inch for their new magnetic storage demonstration development project.
As a result, magnetic north and south poles of a grain suddenly and spontaneously reverse corrupting the stored data and therefore, making the storage device unreliable. Alternative technologies to overcome this problem include perpendicular recording, heat assisted magnetic recording and recording on patterned media. Perpendicular Magnetic Recording: At present, the HDDs employ Longitudinal recording which, as its name indicates, aligns the data bits horizontally, parallel to the surface of the disk. This accomplishment represents a doubling of todays highest data densities on longitudinal recording technology.
Such products with perpendicular recording is expected to greet the market as early as in Projection made by Hitachi suggests the availability of 1-inch micro drive with 20 gigabytes capacity and 3 12 inch products with terabyte capacity. Heat Assisted Magnetic Recording HAMR : HAMR shows the promises to be the key enabling technology that will increase the areal density to a level breaking through the so-called super paramagnetic limit of magnetic recording. This technology is expected to deliver storage densities as high as 50 terabits per square inch.
If disk drives are produced to have such a great areal density, one can store the entire printed contents of the Library of Congress on a single disk drive. If the phenomenal growth rate of bit density continues, the size of an individual bit will soon reach such a small dimension that the bits become magnetically unstable. This phenomenon is known as super paramagnetism.
This problem can be overcome by heating the medium with a laser beam at the precise spot where a data bit is being recorded and subsequently cooling the spot rapidly to stabilize the written bit. Heating makes it easier to write on the medium. This heat assisted recording can increase the recorded density dramatically.
Conventionally, the disk is coated with a thin layer of magnetic alloy. If the disk surface is examined at high magnification, it becomes apparent that within each bit cell there are many tiny magnetic grains. These grains are randomly created during the deposition of the magnetic film. Each grain behaves like an independent magnet whose magnetization can be flipped by the write head during the data writing process. In patterned media, the magnetic alloy is not coated on the entire disk surface.
The layer is created as an ordered array of highly uniform tiny islands, each island capable of storing an individual bit. Each bit is stored in a single deliberately formed magnetic switching volume. This may be one grain, or several exchange coupled grains, rather than a collection of random. Single switching volume magnetic islands are formed along circular tracks with regular spacing. Magnetic transitions no longer meander between random grains, but form perfectly distinct boundaries between precisely located islands.
Since each island is a single magnetic domain, patterned media is thermally stable, even at densities far higher than can be achieved with conventional media. Though the concept of patterned media looks simple, realization of this to achieve high recording density is immensely challenging. Creating islands of such dimension is beyond the capabilities of optical lithography. E-beam lithography and nano imprint replication are considered to be two approaches that can be used to realize patterned media commercially. The HDD industry will soon embrace these and other technologies to manufacture commercially hard disk drives with extremely high areal density.
This makes the design of the head positioning servomechanism more challenging. Shrinking bit size also means narrower track pitch. Many disturbances ignored today will ask for special attention at such high track density. Ultra high areal density will also require the head to fly very low such that occasional contact between head and disk will become inevitable. The servomechanism must be robust enough to withstand these unpredictable disturbances.
Head Positioning Servomechanism When an HDD is powered up, the disks are spun to a precisely regulated speed and the heads are allowed to move radially over the disk surfaces. Limited vertical movement within a very small range, self-regulating by the formation of an air bearing surface ABS between the head and slider is also allowed. Accurate and precise control of radial position of the head slider is done by the head positioning servomechanism. This servomechanism is a feedback system consisting of a sensing element that measures the displacement of the head, a servo motor and actuator, an amplifier, and a controller controlling the movement of the actuator.
In the early generations of HDD, the controller used to be implemented using analog electronics but all modern drives come with digital controller. Binary bits are stored in an HDD by setting a small area of magnetic material coated on the disk to one of two possible polarities. This tiny area, called a bit cell , consists of several grains of the magnetic material alloy. The bit cells are created by the write head while the disk spins causing the bits to be arranged in concentric Figure 2.
There can be as many as , tracks on each surface of a disk used in a 3 12 inch HDD. Recording or writing of the bits and playback reading is performed with a write head and a read head, respectively. The disks are spun at a precisely controlled speed when the operation of writing or reading is performed. The two heads are fabricated on a single slider, which is epoxy-bonded to a stainless-steel or aluminium gimbal at the end of a long and thin structure known as the suspension arm.
Each surface of an HDD is accessed by a dedicated head slider mounted at the tip of a suspension arm. It also regulates the position of the head over the center of a track while data is being written on or read from that track. When the disk spins at high speed, an ABS is formed between the slider and the spinning disk that makes the slider float above the disk surface.
The suspension arm is designed such that it produces precise load force and damping required by the slider to interact with the ABS formed. The movement of the slider perpendicular to the disk surface is self-regulated by the interaction between ABS, load force and damping. The movement of the slider in direction.
This is the motion a sliders go through during repositioning of the head over a new track, as well as during the track following, i. Both of these operations, track seek and track following, uses the same actuator to create the motion parallel to disk surface. Error tolerance during the track-following is in the scales of nanometers, and it must be achieved in presence of various disturbances acting on the slider, suspension and actuator arm.
On the other hand, the transfer of head from one track to another is expected to be performed in few milliseconds. The HDD servomechanism is a unique example of practical applications that demonstrate the degree of precision achieved in a mechatronics system. Deviation of the head from this desired position increases the probability of occurrence of erroneous bits by either accidental overwriting on adjacent track or unwanted interference from the adjacent track, and hampers the reliability of the disk drive. Projections suggest that track density will reach , TPI in laboratory demonstration by the year and in production by , particularly for small form factor drives, i.
Desired error tolerance of the head positioning servomechanism for such drives will be 0. The HDD market is now dominated by 3 12 inch form factor drives; but the smaller form factors have shown a growth potential comparable to those of the 3 12 inch drives in the early years of s. Starting in , the growths of 2 12 inch, 1. Global shipment of small form factor drives 2 12 inch and below was 50 million units in and is expected to reach million units in Insatiable demand of notebook PC and application in consumer electronics, e. As HDDs are being used in new applications, they are expected to meet more stringent performance specifications.
For example, drives to be used in PDAs, camcorders, or automobiles must be able to withstand much larger vibration than those experienced by drives used in PC. A comprehensive illustration of the closed loop head positioning servomechanism of HDD is shown in Figure 2.
The VCM actuator moves the read-write head between tracks track seek mode and regulates the position of the head track following mode. The servo sectors are created at the time of manufacturing and are never overwritten or erased. The closed loop servomechanism uses the feedback signal generated by decoding the information written in these sectors. The servo sectors and the demodulation of the written information are explained later in section 2.
The servo information and user data are multiplexed in space around the track. When the disks spin, this spatial multiplexing becomes temporal multiplexing. The PES signal is proportional to the radial distance between the track-center created during servo track writing and the actual position of the read head. The composite feedback signal is used by the control algorithm which is implemented on a digital processor. There are separate heads for reading and writing, but the two heads are fabricated on a single slider. The suspension provides a preload to press the slider down towards the surface of the disk.
Once the head reaches the target track, it is regulated precisely over the track so that the PES is minimized Track Following. Smooth settling, i. The VCM is the torque producing component of the head positioning servomechanism. When current is passed through the coil of VCM suspended in the magnetic field produced by permanent magnets, a force torque is generated. The force torque , proportional to coil current, can be controlled by changing the amplitude and polarity of the current.
These are shown in Figure 2. In the first of these types, the coil is wound around a central yoke placed between two permanent magnets. The coil, when energized, is free to move forward and backward. As a result, the actuator arm attached to the coil structure moves in and out of the yoke.
The VCM is fixed rigidly to the base plate outside the area of the disk, and the movement of the arm takes place along a radius of the disk Figure 2. With this arrangement, the orientation of the slider with respect to track remains the same at all radial position of the slider. In a rotary VCM, the actuator arm is pivoted at a point between the coil structure and the suspension arm. The coil is attached using epoxy glue to one end of the arm and the suspensions carrying sliders to the other end.
- Monosaccharide Sugars: Chemical Synthesis by Chain Elongation, Degradation, and Epimerization;
- Enquete sur la formation des élites (French Edition).
- La fiancée du viking (Les Historiques) (French Edition).
- Best Hospitals 2013.
- Hard Disk Drive Mechatronics and Control by Hicham Meknassi - Issuu.
- The Good Fight of Faith.
- Mechatronics, BEng (Hons).
The pivot point is nearer to the coil which is suspended in the magnetic field of permanent magnets. Force is generated whenever the coil is energized by allowing current to pass through it. This force makes the coil move in a way that generates torque around the pivot point and causes the sliders to move on an arc. We can use the diagram in Figure 2. As a result, there always exists a physical gap between the read sensor and the write head.
The read head is also used to sense the servo patterns from the servo sectors which is used to derive the position feedback signal. The read head is used as the position sensor. During the operation of data reading, it is the read head whose position is regulated by the servomechanism, making the sensor and point of control collocated.
On the contrary, during write operation, the point of control is the write head but position feedback comes from the read head. The micro-. However, had there been any drive with MR head and linear VCM actuator, the micro-jog distance would be constant for all radial position and would require simpler calibration algorithm.
If we let a current flow through the coil of the VCM, it experiences an electromagnetic force as shown in Figure 2. The electromagnetic force acting on the coil is produced by the interaction between the magnetic field of the permanent magnet and the field produced by the coil current.
If we assume the field from the permanent magnet constant for the entire range of operation, then the magnitude of the torque depends on the field produced by the coil current and the actuator geometry. Since the. In practice, the torque constant may vary with the position of the coil, i. However, the change in the magnitude of torque constant is usually very small and insignificant. We shall assume throughout this book a constant value for this parameter of VCM. It is very common in the HDD industry to express the displacement of read head in units of track, i. If the VCM is driven by a voltage amplifier, shown on the left of Figure 2.
On the other hand, if a current amplifier shown on the right of Figure 2. A sensing resistor of very low ohm is connected in series with the VCM coil. Voltage across the sensing resistor is proportional to the coil current, which is then used as feedback to control the coil current. Circuit representation of a typical VCM driver is shown in Figure 2. The current in the VCM driver is proportional to the input voltage as long as the amplifier operates in the unsaturated mode.
If the amplifier is saturated, the output current can not be increased anymore. So one can model the VCM driver as a current amplifier with an upper limit bounding the amplitude of the current. The fact that the amplitude of current is upper bounded must be taken into consideration while designing the closed loop feedback controller.
Assuming that the output current of the amplifier operating in the linear region is proportional to the input, setting an upper bound on the input u is equivalent to setting an upper bound on the current I. This model represents only the rigid body dynamics of the actuator. It may need a transfer function of order as high as 40 to accurately model the dynamic behavior of the head positioning actuator .
Frequency response of a typical HDD actuator is shown in Figure 2. The response of an identified model that includes a double integrator plus 10 poles and 10 zeros is also drawn on the same figure . In the head positioning servomechanism of HDD, position feedback is obtained from the readback signal produced by sensing special magnetic patterns written on the disks. These patterns, which are explained later in section 2. This position signal is available only in an assembled and servo-written HDD.
However, to obtain a dynamic model of the head positioning actuator, one may use other means to measure the displacement. The use of interferometry to measure changes in position is well known ,. The interferometer optics split the laser light into a reference path and a measurement path. These lights are reflected using two retroreflectors - the reference beam from a stationary reflector but the measurement beam from a retroreflector attached to the object whose change in position is to be measured.
Recombination of the two reflected beams creates an interference signal. The measuring electronics measures and accumulates the phase and provides a position output. This method has some drawbacks for application in the measurement of displacement of HDD actuator. When a retroreflector, which is usually quite heavy, is attached to the actuator, dynamics of the actuator arm is significantly modified.
Besides, our aim is to measure the displacement of the head slider, which is too small to carry the load of the retroreflector. One possibility is to attach the retroreflector on the E-beam of the actuator, but then the measured displacement does not reflect the dynamics of either the suspension or the slider-gimbal assembly. Laser Doppler Vibrometers LDV are optical instruments for accurately measuring velocity and displacement of vibrating structures completely without contact.
A rugged laser head is mounted on a large vibration-free platform with the laser beam pointing to the object whose displacement or velocity is to be measured. Christian Doppler was the first to describe the frequency shift that occurs when sound or light is emitted from a moving source. The velocity of the moving reflector can be measured by measuring the change in frequency between the incident and reflected beams.
The LDV measures the velocity according to this principle and then integrates it to provide displacement measurement. Since the measurement with LDV requires only a reflecting surface on the moving object, we can attach a tiny. Frequency response shown in Figure 2. The schematic diagram of the setup is shown in Figure 2. The excitation signal, which is the input to the VCM driver, and the displacement measurement from the LDV are fed to two channels of the signal analyzer.
The signal analyzer computes the gain and phase at each of the frequencies of swept-sine signal.
ukyrelazok.tk: Control Engineering series
Care must be taken while setting the amplitude of the swept sine signal. Small amplitude of input signal results in small displacement of the slider and, therefore, low signal-to-noise ratio SNR in the output of LDV. Too large an amplitude, on the other hand, may cause the actuator to move beyond the range of LDV. Since the head slider moves on an arc, the reflected beam is not in line with the incident beam. If the angle between the two beams is large, the reflected beam is not received well by the measurement electronics.
A Tutorial on Control Design of Hard Disk Drive Self-Servo Track Writing
The gain of the actuator is higher in the lower frequencies and is expected to decrease with increasing frequency. That means the amplitude of the input excitation should be lower for in the low frequency range and should be increased as the frequency of the input goes higher. For the result shown above, input amplitude is kept reasonably small below mV for the lowest range of frequency, and was increased to almost 1 V for high frequency.
Dynamic signal analyzers available in the market these days come with the capability of automatic adjustment of the input amplitude. Frequency response in the low frequency range Figure 2. The measurement was carried out with the LDV resolution set to 0. This double integrator model or, as used in chapter 3, a second order model with poles on the left hand side of the complex plane is often used as the nominal model for the sake of controller design.
However, knowledge of the flexible mode dynamics is also crucial. There are well established methods for identification of a transfer function from frequency response data , . The frequency response data include two vectors: 1. The frequency response of a system is equal to its transfer function evaluated at the points along the positive imaginary axis of the complex plane, i. This is a nonlinear least squares problem and can be solved iteratively. Commercial softwares are available for solving such problems, e. The estimation algorithm puts higher weights on the measurements in the higher frequency.
This becomes problematic particularly in cases where measurement data span several decades of frequencies. The feedback signal for the head positioning servo loop is obtained by decoding spatially coded magnetic patterns written on the disks. These patterns, known as servo pattern, are created at the time of manufacturing HDD after the spindle, disks, actuator and heads have been assembled inside the drive enclosure. The process of writing servo patterns is known as Servo Track Writing STW and is carried out using a very high precision equipment that controls the position of the actuator of HDD and writes the servo patterns on the disks.
For all HDDs manufactured these days, the enclosure is covered and sealed after servowriting and, as a result, the same head-suspension-actuator assembly and disk-spindle assembly are used for normal operation of HDD. Writing of servo patterns were performed in bulk, several tens of disks at a time. One servo-written disk and several other blank disks were then assembled inside the drive enclosure.
Mathematical Problems in Engineering
In an HDD assembled in such fashion, there is one surface of a disk containing servo pattern. This scheme of creating servo information is known as dedicated servo, where only one side of one disk in a multi-disk HDD contain the spatial servo patterns. In such drive, the signal necessary for position feedback came from the head accessing the servo surface.
Since a single VCM actuator controls the motion of sliders on all surfaces simultaneously, moving the slider on the servo surface to a desired track is equivalent to moving any other head to the same track. The main assumption for proper functioning of this scheme is that sliders on all surfaces are displaced by precisely the same amount which, in reality, is impossible. The dedicated servo scheme worked well for drives of the past, but its limitations started to surface with the trend of increasing track density when the disparity between thermal expansions became comparable to error tolerance.
The need to overcome the thermal expansion related problems to pave the way for higher track density gave birth to a new scheme of servo encoding - embedded servo or sectored servo, in which servo patterns are created on all surfaces and same head is used for accessing both servo and data.
Instead of one dedicated servo surface, the embedded scheme puts servo patterns on all surfaces interleaved with the data blocks. The servo sector, a small segment of the track containing servo patterns, are created at regular intervals, and the space between two servo sectors is designated for storing data. The embedded servo scheme with interleaved data blocks and servo sectors is illustrated in Figure 2. Besides the problem of thermal expansion, large servo overhead is another drawback of dedicated servo scheme when only a few disks are used in an HDD. Servo overhead is the percentage of available area that is consumed by.
For an HDD with only 2 disks, which is quite common nowadays because of large areal density, one out of four surfaces are used for servo information if dedicated servo is employed, i. In an embedded servo HDD, the servo sectors are placed on all tracks interleaved with data blocks. Small section of a track is illustrated in Figure 2. Each of these fields in the servo sector has specific function.
Only two of them are directly related to the generation of position feedback signal and are explained here. These are the track number field and burst pattern field. The grey coded track number is the identification of a track; the outermost track is tagged as T rk0 and it increases inward.
So each track has a unique grey coded track number field, and the same pattern is repeated in the track number fields of all sectors of a track. The schematic layout of the magnetic patterns in the servo burst field is illustrated in Figure 2. This represents only a tiny segment of the disk surface. Moving from left to right in this figure or vice versa is equivalent to moving in the direction along the track, i.
This illustration shows the track-centers of 4 consecutive tracks with track numbers increasing upward. Definition of the track-centers will become clear after the following analysis of the signals obtained from these patterns. When the read sensor scans a magnetic transition, a voltage pulse is produced.
The polarity of the pulse depends on the type of transition. An example is shown in Figure 2. The readback waveform shows two similar pulses of opposite polarity. The amplitudes of the pulses depend on the magnetic flux linking the read head and hence on the distance of the head with respect to the transitions, both in the vertical plane as well as on the plane parallel to the disk surface.
When the disk spins, the entire track containing all data block and servo sectors is scanned by the head. The read head positioned at the point marked 0 in Figure 2. For this case, the read head senses maximum flux from burst C and no flux from burst D. For bursts A and B, the flux linkage is smaller than that for burst C. As a result, the amplitude will be maximum for the burst C, zero for burst D, and non-zero but less than maximum for bursts A and B. Corresponding waveform is shown in Figure 2. It is revealed from the observation of waveforms shown in Figure 2.
The information contained in the amplitudes of these bursts is used to measure the displacement of the read head with respect to the burst patterns. A down-track line that is line along the track is designated as the center of a track if the amplitudes of burst A and burst B are equal when the head scans the burst along this line. The dimension of the read head is usually smaller than the width of a track, and if the head moves far from the center of the track it senses flux emanating from only one of the two bursts either burst A or burst B but not both.
Appropriate manipulation of in-phase and quadrature PES signals produces an error signal proportional to the distance between read head and the center of a track. The feedback signal used by the servo loop is the combination of PES signal and track number obtained from the grey code field. In disk drives using embedded servo, the position feedback is available only at discrete points in time and the servo control is also implemented in discrete-time.
However, the designer of servo controller is not at the liberty of selecting the sampling frequency arbitrarily. The position signal is available disk at a frequency S N60 , where S and Ndisk are the number of servo sectors per track and rotational speed of disks in units of revolutions per minute or RPM, respectively. The sampling frequency can be increased either by spinning the. Each of these options comes with its disadvantages. Larger is the number of servo sector, more is the servo overhead. Increase in spindle speed, on the other hand, generates more heat which in turn requires better cooling mechanism.
- Training the Disaster Search Dog!
- Timothée lécureuil (French Edition)!
- Adios Desperada.
- Hard Disk Drive: Mechatronics and Control - CRC Press Book.
Increased speed also translates into larger rate of data transfer between media and electronics that demands for expensive electronics in the read-write channel. Besides, higher spindle speed pushes the spectrum of disturbances related to disk rotation to higher frequency and, as a consequence, higher bandwidth is required for actuator servo. As a result, the disk drive servomechanism remains as an example of control system which demands for as large a bandwidth as possible but comes with severe restrictions on the sampling frequency.
A good estimate of the burst amplitude is the most important consideration for reliable generation of the PES signal. Until very recently, the servo demodulation used non-coherent analog method. In this method, the burst waveform is first processed through a full-wave rectifier. In a method called the peakdetected servo demodulation, a circuit that can detect and hold the peak amplitude of the rectified burst waveform is used.
Another method, known as the area demodulation, finds the area under the rectified waveform. The analog area detection method uses several precautionary measures to minimize detection error. The zero-crossings of the burst waveform are first detected. Once a zero crossing is found in a burst, a demodulation window is opened. The window is closed after a pre-defined number of cycles of the burst waveform have elapsed. The rectified waveform that falls inside the window controls a charge pump that, in turn, charges a capacitor.
The capacitor voltage is proportional to the area of the burst falling inside the demodulation window. The process is illustrated in Figure 2. Once the area of one burst is obtained, the amplitude of the capacitor voltage is converted into a digital number and latched to a register. The capacitor is then discharged to zero voltage before the demodulation window for the next burst is opened. The first few cycles in the burst are not used for charging the capacitor.
During this period, the zero-crossing detector synchronizes the charging process. Each burst is, therefore, made longer than the number of cycles actually used for meaningful area detection. The peak-detected servo demodulator uses simpler circuit, but it is prone to error in presence of noise. Area detection, on the other hand, is an averaging process and provide better immunity to broadband noise. Availability of higher processing power in modern disk drives encouraged the use of digital algorithms to find the burst amplitude.
The first approach. The result of the summation is proportional to the burst area. The number of samples N is selected such that the summing window is equal to an integer number of periods of the burst waveform. Two other methods proposed for estimating amplitude or area of a burst waveform from its samples are digital maximum likelihood detection , and coherent detection with selective harmonics . The estimated burst amplitude using these two methods are shown below for the samples yA kTS.
The vector yA contains the samples of the burst waveform and the vector d0 consists of the samples from a ideal model signal of the burst. This is a special case of the maximum likelihood ML detection. The model signal d0 in the ML detector contains all harmonics of the nominal noise-free signal. For an HDD operating at 10, rpm with servo sectors per track, the sampling frequency is 20 kHz. This sampling frequency is good enough to meet the specifications for the head positioning servomechanism during trackfollowing.
Sampling frequency can be increased in the HDD head-positioning servomechanism in two possible ways: 1. Spinning the disk at higher RPM also reduces average latency. However, the speed can not be increased arbitrarily. Increased RPM comes at the cost of higher power consumption and generation of excessive heat inside drive enclosure. Many of the disturbances discussed later in this chapter are synchronous to the spindle motor. Spinning the disk at higher RPM shifts the spectrum of these disturbances to higher frequency.
As a result, higher bandwidth is required for satisfactory rejection of those disturbances. Increasing sampling frequency with higher spindle speed is therefore not a solution that. Neither it generates more heat. However, it takes up more space of the disk that can otherwise be used for storing data. If the length of the servo sectors can be reduced without compromising the quality of demodulation of bursts then more servo sectors can be added with no additional servo overhead. Dual Frequency Burst Pattern In the conventional method, the four servo bursts use the same pattern of magnetization, i.
Since these bursts have identical patterns, they can not be distinguished from one another if two of them are aligned in the radial direction cross-track direction. This makes it absolutely essential to place the bursts shifted circumferentially from one another, leaving many voids, i. Then the voids otherwise found between the bursts of conventional patterns are eliminated. Schematic representation of the dual frequency burst pattern is shown in Figure 2. In the conventional method, waveforms of all four bursts are identical. If the read head senses both.
The area detection method can not be applied directly for estimating the amplitude from the dual-frequency burst waveform. One possible solution is to use two band-pass filters to separate the two frequencies, and then applying area detection method to the samples at the outputs of two filters. Maximum likelihood detection and coherent detection using selective harmonics can be used to estimate individual amplitude from the samples of the dual-frequency burst waveform. Both of these methods are sensitive to jitter in sampling clock. If the clock is not synchronized, the phase error between the sampled signal and model signal contributes to error in the estimate of amplitude.
Such error can be eliminated if both sine and cosine are used in the model signal. This method, which extracts the amplitude of the fundamental frequency of the burst signal, is equivalent to discrete Fourier Transform DFT with interest in one frequency only. The coherent detection using DFT can be employed to estimate the amplitude of each of these frequencies present in the burst signal. However, in reality, the servo burst signals y1 and y2 contain not only the fundamental frequencies but also other odd harmonics and noise .
It should be noted that the burst signals are odd signals and therefore contain sine waves only. Applying the above mentioned method for estimating amplitude of the fundamental frequencies, we get As. The number N is chosen such that integer number of full cycles of the fundamental frequency are sampled.
Moreover, if the noise n k is zero-mean AWGN then the following ensemble averages are also equal to zero, E. This method can be applied to estimate the amplitudes of both frequency components of the dual-frequency burst. Another method of obtaining information on the position of the read head scanning the data block was proposed in the patent . Samples from the read waveform are processed by a discrete Fourier transform type algorithm to determine the magnitude of the frequency component associated with the track being scanned.
The output of this process provides an indicator to the position of the head. This result is further smoothened using a simple first order filter. One drawback of these methods of estimating head position from data block is that it can be used only during read operation. The readback waveform is available when a data block is being read and, therefore, can be further processed to estimate PES at high sampling rate. However, during a write operation, the write head is enabled and read sensor is disabled. The readback signal is not available for any kind of processing and the data track is continuously being modified with the new data overwriting the old pattern of magnetization.
Realization of either of the two methods would require major change in the head-slider configuration by inserting an additional read sensor for servo only. The rigid body model of equation 2. However, the frequency response of a practical VCM actuator shown in Figure 2. The frequency response measured in frequencies above 1.
These flexible modes of the actuator are contributions from various bending modes, torsional modes and sway modes of the suspension, the VCM coil, and the gimbal with which the slider is attached to the suspension. The torsional mode of the suspension twists it along the center line of the load beam causing a small amount of in-plane head motion.
The first torsional mode of commercially available suspensions lies typically around 3 kHz. There is a second torsional mode in the frequency range between 5 kHz and 8 kHz. The sway mode is caused by the in-plane deformation of the load beam; it is the result of in-plane bending of the suspension. It was explained earlier that part of the load beam is left without edge so that the suspension arm has necessary compliance to accommodate vertical disk runout.
This section of the load beam is the weakest part of the suspension. Typical frequency of the sway mode lies in the range of 8 kHz - 12 kHz. The sway mode is a greater problem in the rotary. Modes other than the torsion and sway modes have small amplitude or lie in frequencies very far away. It is hard to control precisely the position of the head using an actuator with lightly damped resonant modes, which are subject to variation from drive to drive and in a single drive over time.
As a result, these resonances limit achievable bandwidth of the servomechanism. Application of digital filters is suitable for adaptation to such variations, but implementation of digital notch filter is restricted by relatively low sampling frequency in hard disk drives. This led to the design of multi-rate notch filter for HDD servomechanism.
Design of notch filter as well as multi-rate compensator are explained in details in chapter 3. The relatively heavier part of the actuator to which all the suspension arms are attached is known as the E-block. If it is rigid then the dynamics of one suspension is not coupled to that of another. In such case, each suspension resonates by itself and does not interfere with other arms. The in-phase mode further limits the bandwidth of the servo system.
The external disturbances are typically in the form of shock and vibration that come from the environment. For example, a moving vehicle or a machine running in a factory or an accidental hit by the user of desktop computer or carrying a laptop computer causes the servo loop to be subjected under external forces. Besides these external disturbances, drive components and their interconnections also give rise to several disturbing forces acting on the control loop.
Some internal sources give rise to disturbing forces that are repetitive in nature. Most of the lateral and vertical movements of the fast-spinning disk platters appear as repetitive in the sense that they have definite temporal pattern and they remain the same every time the disks are spun. These disturbances are contributed mostly by mechanical factors. For example, disturbances caused by misalignment between center of the spindle shaft and the disk center, wobbling of the disk platters, vibration modes of the disk platters, defects in the inner and outer races of the ball-bearing etc, repeat with revolution.
The misalignment between the disk-center and shaft center makes the shape of a track elliptical and not circular. A defect in the inner or the outer race of bearing produces a lateral force whenever a ball hits that defect.