Upgrading and Repairing PCs (17th Edition)
| Many types of hard disk drives are on the market, but nearly all share the same basic physical components. Some differences might exist in the implementation of these components (and in the quality of the materials used to make them), but the operational characteristics of most drives are similar. The basic components of a typical hard disk drive are as follows (see Figure 9.6):
Figure 9.6. Typical hard disk drive components.
The platters, spindle motor, heads, and head actuator mechanisms usually are contained in a sealed chamber called the head disk assembly (HDA). The HDA is usually treated as a single component; it is rarely opened. Other parts external to the drive's HDA, such as the logic boards, bezel, and other configuration or mounting hardware, can be disassembled from the drive. Hard Disk Platters (Disks)
A hard disk drive has one or more platters, or disks. Hard disks for PC systems have been available in several form factors over the years. Normally, the physical size of a drive is expressed as the size of the platters. Table 9.4 lists the platter sizes that have been associated with PC hard disk drives:
Larger hard disk drives that have 8", 14", or even larger platters are available, but these drives are not used with PC systems. Currently, the 3 1/2" drives are the most popular for desktop and some portable systems, whereas the 2 1/2" and smaller drives are very popular in portable or notebook systems. Most hard disk drives have two or more platters, although some of the smaller drives used in portable systems and some entry-level drives for desktop computers have only one. The number of platters a drive can have is limited by the drive's vertical physical size. The maximum number of platters I have seen in any 3 1/2" drive is 12; however, most drives have 6 or fewer. Platters have traditionally been made from an aluminum/magnesium alloy, which provides both strength and light weight. However, manufacturers' desire for higher and higher densities and smaller drives has led to the use of platters made of glass (or, more technically, a glass-ceramic composite). One such material, produced by the Dow Corning Corporation, is called MemCor. MemCor is composed of glass with ceramic implants, enabling it to resist cracking better than pure glass. Glass platters offer greater rigidity than metal (because metal can be bent and glass can't) and can therefore be machined to one-half the thickness of conventional aluminum diskssometimes less. Glass platters are also much more thermally stable than aluminum platters, which means they do not expand or contract very much with changes in temperature. Several hard disk drives currently use glass or glass-ceramic platters; in fact, Hitachi Global Storage Technologies is designing all its new drives with only glass platters. For most other manufacturers as well, glass disks will probably replace the previously standard aluminum/magnesium substrate over the next few years. Recording Media
No matter which substrate is used, the platters are covered with a thin layer of a magnetically retentive substance, called the medium, on which magnetic information is stored. Three popular types of magnetic media are used on hard disk platters:
Oxide Media
The oxide medium is made of various compounds, containing iron oxide as the active ingredient. The magnetic layer is created on the disk by coating the aluminum platter with a syrup containing iron-oxide particles. This syrup is spread across the disk by spinning the platters at high speed; centrifugal force causes the material to flow from the center of the platter to the outside, creating an even coating of the material on the platter. The surface is then cured and polished. Finally, a layer of material that protects and lubricates the surface is added and burnished smooth. The oxide coating is usually about 30 millionths of an inch thick. If you could peer into a drive with oxide-coated platters, you would see that the platters are brownish or amber. As drive density increases, the magnetic medium needs to be thinner and more perfectly formed. The capabilities of oxide coatings have been exceeded by most higher-capacity drives. Because the oxide medium is very soft, disks that use it are subject to head-crash damage if the drive is jolted during operation. Most older drives, especially those sold as low-end models, use oxide media on the drive platters. Oxide media, which have been used since 1955, remained popular because of their relatively low cost and ease of application. Today, however, very few drives use oxide media. Thin-Film Media
The thin-film medium is thinner, harder, and more perfectly formed than oxide medium. Thin film was developed as a high-performance medium that enabled a new generation of drives to have lower head-floating heights, which in turn made increases in drive density possible. Originally, thin-film media were used only in higher-capacity or higher-quality drive systems, but today, virtually all drives use thin-film media. The thin-film medium is aptly named. The coating is much thinner than can be achieved by the oxide-coating method. Thin-film media are also known as plated, or sputtered, media because of the various processes used to deposit the thin film on the platters. Thin-film plated media are manufactured by depositing the magnetic medium on the disk with an electroplating mechanism, in much the same way that chrome plating is deposited on the bumper of a car. The aluminum/magnesium or glass platter is immersed in a series of chemical baths that coat the platter with several layers of metallic film. The magnetic medium layer itself is a cobalt alloy about 1 m-inch thick. Thin-film sputtered media are created by first coating the aluminum platters with a layer of nickel phosphorus and then applying the cobalt-alloy magnetic material in a continuous vacuum-deposition process called sputtering. This process deposits magnetic layers as thin as 1 m-inch or less on the disk, in a fashion similar to the way that silicon wafers are coated with metallic films in the semiconductor industry. The same sputtering technique is again used to lay down an extremely hard, 1 m-inch protective carbon coating. The need for a near-perfect vacuum makes sputtering the most expensive of the processes described here. The surface of a sputtered platter contains magnetic layers as thin as 1 m-inch. Because this surface also is very smooth, the head can float more closely to the disk surface than was previously possible. Floating heights as small as 10nm (nanometers, or about 0.4 m-inch) above the surface are possible. When the head is closer to the platter, the density of the magnetic flux transitions can be increased to provide greater storage capacity. Additionally, the increased intensity of the magnetic field during a closer-proximity read provides the higher signal amplitudes necessary for good signal-to-noise performance. Both the sputtering and plating processes result in a very thin, hard film of magnetic medium on the platters. Because the thin-film medium is so hard, it has a better chance of surviving contact with the heads at high speed. In fact, modern thin-film media are virtually uncrashable. If you could open a drive to peek at the platters, you would see that platters coated with the thin-film medium look like mirrors. AFC Media
The latest advancement in drive media is called antiferromagnetically coupled (AFC) media and is designed to allow densities to be pushed beyond previous limits. Anytime density is increased, the magnetic layer on the platters must be made thinner and thinner. Areal density (tracks per inch times bits per inch) has increased in hard drives to the point where the grains in the magnetic layer used to store data are becoming so small that they become unstable over time, causing data storage to become unreliable. This is referred to as the superparamagnetic limit, and it was originally determined to be between 30Gb/sq. in. and 50Gb/sq. in. However, as technology has advanced, this so-called limit has been pushed further and further back, and commercially produced drives now routinely exceed 100Gb/sq.in. Drives exceeding 200Gb/sq. in. are expected to be possible in the future, with several new technologies in the works. AFC media consists of two magnetic layers separated by a very thin 3-atom (6 angstrom) film layer of the element ruthenium. IBM has coined the term "pixie dust" to refer to this ultra-thin ruthenium layer. This sandwich produces an antiferromagnetic coupling of the top and bottom magnetic layers, which causes the apparent magnetic thickness of the entire structure to be the difference between the top and bottom magnetic layers. This allows the use of physically thicker magnetic layers with more stable larger grains, so they can function as if they were really a single layer that was much thinner overall. IBM has introduced AFC media into several drives, starting with the 2 1/2" Travelstar 30GN series of notebook drives introduced in 2001; they were the first drives on the market to use AFC media. In addition, IBM has introduced AFC media in desktop 3 1/2" drives starting with the Deskstar 120 GXP. AFC media is also used by Hitachi Global Storage Technologies, which owns the former IBM hard drive lines. I expect other manufacturers to introduce AFC media into their drives as well. The use of AFC media is expected to allow areal densities to be extended to 100Gb/sq. in. and beyond. It's also notable that, being a form of thin-film media, if you removed the protective casing around the drive platters, you would find that they look like mirrors.
Read/Write Heads
A hard disk drive usually has one read/write head for each platter surface (meaning that each platter has two sets of read/write headsone for the top side and one for the bottom side). These heads are connected, or ganged, on a single movement mechanism. The heads, therefore, move across the platters in unison. Mechanically, read/write heads are simple. Each head is on an actuator arm that is spring-loaded to force the head into contact with a platter. Few people realize that each platter actually is "squeezed" by the heads above and below it. If you could open a drive safely and lift the top head with your finger, the head would snap back down into the platter when you released it. If you could pull down on one of the heads below a platter, the spring tension would cause it to snap back up into the platter when you released it. Figure 9.7 shows a typical hard disk head-actuator assembly from a voice coil drive. Figure 9.7. Read/write heads and rotary voice coil actuator assembly.
When the drive is at rest, the heads are forced into direct contact with the platters by spring tension, but when the drive is spinning at full speed, air pressure develops below the heads and lifts them off the surface of the platter. On a drive spinning at full speed, the distance between the heads and the platter can be anywhere from 0.5 m-inches to 5 m-inches or more in a modern drive. In the early 1960s, hard disk drive recording heads operated at floating heights as large as 200 m-inches300 m-inches; today's drive heads are designed to float as low as 10nm (nanometers) or 0.4 m-inches above the surface of the disk. To support higher densities in future drives, the physical separation between the head and disk is expected to drop even further, such that on some drives there will even be contact with the platter surface. New media and head designs will be required to make full or partial contact recording possible. Caution The small size of the gap between the platters and the heads is why you should never open the disk drive's HDA except in a clean-room environment. Any particle of dust or dirt that gets into this mechanism could cause the heads to read improperly or possibly even to strike the platters while the drive is running at full speed. The latter event could scratch the platter or the head, causing permanent damage. To ensure the cleanliness of the interior of the drive, the HDA is assembled in a class-100 or better clean room. This specification means that a cubic foot of air can't contain more than 100 particles that measure up to 0.5 microns (19.7 m-inches). A single person breathing while standing motionless spews out 500 such particles in a single minute! These rooms contain special air-filtration systems that continuously evacuate and refresh the air. A drive's HDA should not be opened unless it is inside such a room. Although maintaining a clean-room environment might seem to be expensive, many companies manufacture tabletop or bench-size clean rooms that sell for only a few thousand dollars. Some of these devices operate like a glove box; the operator first inserts the drive and any tools required, closes the box, and then turns on the filtration system. Inside the box, a clean-room environment is maintained, and a technician can use the built-in gloves to work on the drive. In other clean-room variations, the operator stands at a bench where a forced-air curtain maintains a clean environment on the bench top. The technician can walk in and out of the clean-room field by walking through the air curtain. This air curtain is very similar to the curtain of air used in some stores and warehouses to prevent heat from escaping in the winter while leaving a passage wide open. Because the clean environment is expensive to produce, few companies except those that manufacture the drives are properly equipped to service hard disk drives.
Head Actuator Mechanisms
Possibly more important than the heads themselves is the mechanical system that moves them: the head actuator. This mechanism moves the heads across the disk and positions them accurately above the desired cylinder. Many variations on head actuator mechanisms are in use, but all fall into one of two basic categories:
The use of one or the other type of actuator has profound effects on a drive's performance and reliability. The effects are not limited to speed; they also include accuracy, sensitivity to temperature, position, vibration, and overall reliability. The head actuator is the single most important specification in the drive, and the type of head actuator mechanism in a drive tells you a great deal about the drive's performance and reliability characteristics. Table 9.5 shows the two types of hard disk drive head actuators and the affected performance characteristics.
Stepper motor actuators were commonly used on hard drives made during the 1980s and early 1990s with capacities of 100MB or less. All the drives I've seen with greater storage capacity use a voice coil actuator. Floppy disk drives position their heads by using a stepper motor actuator. The accuracy of the stepper mechanism is suited to a floppy disk drive because the track densities usually are nowhere near those of a hard disk. The track density of a 1.44MB floppy disk is 135 tracks per inch, whereas hard disk drives have densities of more than 5,000 tracks per inch. All hard disk drives being manufactured today use voice coil actuators because stepper motors can't achieve the degree of accuracy necessary. Stepper Motor Actuators
A stepper motor is an electrical motor that can "step," or move from position to position, with mechanical detents or click-stop positions. If you were to grip the spindle of one of these motors and spin it manually, you would hear a clicking or buzzing sound as the motor passed each detent position with a soft click. Stepper motors can't position themselves between step positions; they can stop only at the predetermined detent positions. The motors are small (between 1" and 3") and can be square, cylindrical, or flat. Stepper motors are outside the sealed HDA, although the spindle of the motor penetrates the HDA through a sealed hole. Stepper motor mechanisms are affected by a variety of problems, but the greatest problem is temperature. As the drive platters heat and cool, they expand and contract, and the tracks on the platters move in relation to a predetermined track position. The stepper mechanism can't move in increments of less than a single track to correct for these temperature-induced errors. The drive positions the heads to a particular cylinder according to a predetermined number of steps from the stepper motor, with no room for nuance. Figure 9.8 shows a common stepper motor design, in which a split metal band is used to transfer the movement from the rotating motor shaft to the head actuator itself. Figure 9.8. A stepper motor actuator.
Voice Coil Actuators
The voice coil actuators used in virtually all hard disk drives made todayunlike stepper motor actuatorsuse a feedback signal from the drive to accurately determine the head positions and adjust them, if necessary. This arrangement provides significantly greater performance, accuracy, and reliability than traditional stepper motor actuator designs. A voice coil actuator works by pure electromagnetic force. The construction of the mechanism is similar to that of a typical audio speaker, from which the term voice coil is derived. An audio speaker uses a stationary magnet surrounded by a voice coil, which is connected to the speaker's paper cone. Energizing the coil causes it to move relative to the stationary magnet, which produces sound from the cone. In a typical hard disk drive's voice coil system, the electromagnetic coil is attached to the end of the head rack and placed near a stationary magnet. No physical contact occurs between the coil and the magnet; instead, the coil moves by pure magnetic force. As the electromagnetic coils are energized, they attract or repulse the stationary magnet and move the head rack. Systems like these are extremely quick, efficient, and usually much quieter than systems driven by stepper motors. Unlike a stepper motor, a voice coil actuator has no click-stops or detent positions; rather, a special guidance system stops the head rack above a particular cylinder. Because it has no detents, the voice coil actuator can slide the heads in and out smoothly to any position desired. Voice coil actuators use a guidance mechanism called a servo to tell the actuator where the heads are in relation to the cylinders and to place the heads accurately at the desired positions. This positioning system often is called a closed loop feedback mechanism. It works by sending the index (or servo) signal to the positioning electronics, which return a feedback signal that is used to position the heads accurately. The system also is called servo-controlled, which refers to the index or servo information that is used to dictate or control head-positioning accuracy. A voice coil actuator with servo control is not affected by temperature changes, as a stepper motor is. When temperature changes cause the disk platters to expand or contract, the voice coil system compensates automatically because it never positions the heads in predetermined track positions. Rather, the voice coil system searches for the specific track, guided by the prewritten servo information, and then positions the head rack precisely above the desired track, wherever it happens to be. Because of the continuous feedback of servo information, the heads adjust to the current position of the track at all times. For example, as a drive warms up and the platters expand, the servo information enables the heads to "follow" the track. As a result, a voice coil actuator is sometimes called a track following system. The two main types of voice-coil positioner mechanisms are
The two types differ only in the physical arrangement of the magnets and coils. Linear Actuators
A linear actuator moves the heads in and out over the platters in a straight line (see Figure 9.9). The coil moves in and out on a track surrounded by the stationary magnets. The primary advantage of the linear design is that it eliminates the head azimuth variations that occur with rotary positioning systems. (Azimuth refers to the angular measurement of the head position relative to the tangent of a given cylinder.) A linear actuator does not rotate the head as it moves from one cylinder to another, thus eliminating this problem. Figure 9.9. A linear voice coil actuator.
Although the linear actuator seems to be a good design, it has one fatal flaw: The devices are much too heavy. As drive performance has increased, the desire for lightweight actuator mechanisms has become very important. The lighter the mechanism, the faster it can accelerate and decelerate from one cylinder to another. Because they are much heavier than rotary actuators, linear actuators were popular only for a short time; they are virtually nonexistent in drives manufactured today. Rotary actuators also use stationary magnets and a movable coil, but the coil is attached to the end of an actuator arm. As the coil moves relative to the stationary magnet, it swings the head arms in and out over the surface of the disk. The primary advantage of this mechanism is its light weight, which means the heads can accelerate and decelerate very quickly, resulting in very fast average seek times. Because of the lever effect on the head arm, the heads move faster than the actuator, which also helps to improve access times. (Refer to Figure 9.7, which shows a rotary voice coil actuator.) The disadvantage of a rotary system is that as the heads move from the outer to the inner cylinders, they rotate slightly with respect to the tangent of the cylinders. This rotation results in an azimuth error and is one reason the area of the platter in which the cylinders are located is somewhat limited. By limiting the total motion of the actuator, the azimuth error is contained to within reasonable specifications. Virtually all voice coil drives today use rotary actuator systems. Servo Mechanisms
Three servo mechanism designs have been used to control voice coil positioners over the years:
The three designs are slightly different, but they accomplish the same basic task: They enable the head positioner to adjust continuously so it is precisely positioned above a given cylinder on the disk. The main difference between these servo designs is where the gray code information is actually written on the drive. All servo mechanisms rely on special information that is written to the disk when it is manufactured. This information is usually in the form of a special code called a gray code. A gray code is a special binary notational system in which any two adjacent numbers are represented by a code that differs in only one bit place or column position. This system enables the head to easily read the information and quickly determine its precise position. At the time of manufacture, a special machine called a servo-writer writes the servo gray code on the disk. The servowriter is basically a jig that mechanically moves the heads to a given reference position and then writes the servo information at that position. Many servowriters are themselves guided by a laser-beam reference that calculates its own position by calculating distances in wavelengths of light. Because the servowriter must be capable of moving the heads mechanically, the process requires either that the lid of the drive be removed or that access be available through special access ports in the HDA. After the servowriting is complete, these ports are usually covered with sealing tape. You often see these tape-covered holes on the HDA, usually accompanied by warnings that you will void the warranty if you remove the tape. Because servowriting exposes the interior of the HDA, it requires a clean-room environment. A servowriter is an expensive piece of machinery, costing up to $50,000 or more, and often must be custom-made for a particular make or model of drive. Some drive-repair companies have servowriting capability, which means they can rewrite the servo information on a drive if it becomes damaged. If a servowriter is not available, a drive with servo-code damage must be sent back to the drive manufacturer for the servo information to be rewritten. Fortunately, damaging the servo information through disk read and write processes is impossible. Drives are designed so the heads can't overwrite the servo information, even during a low-level format. One myth that has been circulating (especially with respect to ATA drives) is that you can damage the servo information by improper low-level formatting. This is not true. An improper low-level format can compromise the performance of the drive, but the servo information is totally protected and can't be overwritten. Even so, the servo information on some drives can be damaged by a strong adjacent magnetic field or by jarring the drive while it is writing, which causes the heads to move off track. The track-following capabilities of a servo-controlled voice coil actuator eliminate the positioning errors that occur over time with stepper motor drives. Voice coil drives are not affected by conditions such as thermal expansion and contraction of the platters. In fact, many voice coil drives today perform a special thermal-recalibration procedure at predetermined intervals while they run. This procedure usually involves seeking the heads from cylinder 0 to some other cylinder one time for every head on the drive. As this sequence occurs, the control circuitry in the drive monitors how much the track positions have moved since the last time the sequence was performed, and a thermal-recalibration adjustment is calculated and stored in the drive's memory. This information is then used every time the drive positions the heads to ensure the most accurate positioning possible. At one time, most drives had to perform the thermal-recalibration sequence every 5 minutes for the first 30 minutes that the drive was powered on and then once every 25 minutes after that. With some drives, this thermal-recalibration sequence was very noticeable because the drive essentially stopped what it was doing, and you heard rapid ticking for a second or so. This was often misinterpreted as the drive having a problem reading data and having to reread it, but this was not true. As multimedia applications grew in popularity, thermal recalibration became a problem with some manufacturers' drives. The thermal-recalibration sequence sometimes interrupted the transfer of a large data file, such as an audio or a video file, which resulted in audio or video playback jitter. Consequently, some companies released special A/V (audio visual) drives that hide the thermal-recalibration sequences so they never interrupt a file transfer. Most of today's ATA and SCSI drives are A/V capable, which means the thermal-recalibration sequences do not interrupt a data transfer. A/V-capable ATA drives are also used in set-top boxes that are utilized for digital recording, such as the popular TiVo and ReplayTV devices. While we are on the subject of automatic drive functions, most of the drives that perform thermal-recalibration sequences also automatically perform a function called a disk sweep. Also called wear leveling by some manufacturers, this procedure is an automatic head seek that occurs after the drive has been idle for a period of time. The disk-sweep function moves the heads to a cylinder in the outer portion of the platters, which is where the head float-height is highest (because the head-to-platter velocity is highest). Then, if the drive continues to remain idle for another period, the heads move to another cylinder in this area, and the process continues indefinitely as long as the drive is powered on. The disk-sweep function is designed to prevent the head from remaining stationary above one cylinder in the drive for too long, where friction between the head and platter eventually would dig a trench in the medium. Although the heads are not in direct contact with the medium, they are so close that the constant air pressure from the head floating above a single cylinder could cause friction and excessive wear. Figure 9.10 shows both a wedge and an embedded servo. Figure 9.10. A wedge and an embedded servo.
Wedge Servo
Early servo-controlled drives used a technique called a wedge servo. In these drives, the gray-code guidance information is contained in a "wedge" slice of the drive in each cylinder immediately preceding the index mark. The index mark indicates the beginning of each track, so the wedge-servo information was written in the PRE-INDEX GAP, which is at the end of each track. This area is provided for speed tolerance and normally is not used by the controller. Some controllers had to be notified that the drive was using a wedge servo so they could shorten the sector timing to allow for the wedge-servo area. If they were not correctly configured, these controllers would not work properly with the drive. Another problem was that the servo information appears only one time every revolution, which means that the drive often needed several revolutions before it could accurately determine and adjust the head position. Because of these problems, the wedge servo never was a popular design; it no longer is used in drives. Embedded Servo
An embedded servo is an enhancement of the wedge servo. Instead of placing the servo code before the beginning of each cylinder, an embedded servo design writes the servo information before the start of each sector. This arrangement enables the positioner circuits to receive feedback many times in a single revolution, making the head positioning much faster and more precise. Another advantage is that every track on the drive has its own positioning information, so each head can quickly and efficiently adjust position to compensate for any changes in the platter or head dimensions, especially for changes due to thermal expansion or physical stress. Most drives today use an embedded servo to control the positioning system. As in the wedge servo design, the embedded servo information is protected by the drive circuits and any write operations are blocked whenever the heads are above the servo information. Thus, it is impossible to overwrite the servo information with a low-level format, as some people incorrectly believe. Although the embedded servo works much better than the wedge servo because the servo feedback information is made available several times in a single disk revolution, a system that offered continuous servo feedback information would be better. Dedicated Servo
A dedicated servo is a design in which the servo information is written continuously throughout the entire track, rather than just once per track or at the beginning of each sector. Unfortunately, if this procedure were used on the entire drive, no room would be left for data. For this reason, a dedicated servo uses one side of one of the platters exclusively for the servo-positioning information. The term "dedicated" comes from the fact that this platter side is completely dedicated to the servo information and can't contain any data. When building a dedicated servo drive, the manufacturer deducts one side of one platter from normal read/write usage and records a special set of gray-code data there that indicates the proper track positions. Because the head that rests above this surface can't be used for normal reading and writing, the gray code can never be erased and the servo information is protectedas in the other servo designs. No low-level format or other procedure can possibly overwrite the servo information. Figure 9.11 shows a dedicated servo mechanism. Typically, the head on top or one in the center is dedicated for servo use. Figure 9.11. A dedicated servo, showing one entire head/side used for servo reading.
When the drive moves the heads to a specific cylinder, the internal drive electronics use the signals received by the servo head to determine the position of the read/write heads. As the heads move, the track counters are read from the dedicated servo surface. When the servo head detects the requested track, the actuator stops. The servo electronics then fine-tune the position so the heads are precisely above the desired cylinder before any writing is permitted. Although only one head is used for servo tracking, the other heads are attached to the same rack so that if one head is above the desired cylinder, all the others are as well. One way of telling whether a drive uses a dedicated servo platter is if it has an odd number of heads. For example, the Toshiba MK-538FB 1.2GB drive that I used to have in one of my systems had eight platters, but only 15 read/write heads. That drive uses a dedicated servo positioning system, and the 16th head is the servo head. The advantage of the dedicated servo concept is that the servo information is continuously available to the drive, making the head positioning process faster and more precise. The drawback to a dedicated servo is that dedicating an entire platter surface for servo information is wasteful. Virtually all drives today use a variation on the embedded servo technique instead. Some drives combined a dedicated servo with an embedded servo, but this type of hybrid design is rare. Regardless of whether the servo mechanism is dedicated or embedded, it is far more accurate than the stepper motor mechanisms of the past. Of course, as mentioned earlier, today's ATA and SCSI drives have head, track, and sector-per-track parameters that are translated from the actual physical numbers. Therefore, you usually can't tell from the published numbers exactly how many heads or platters are contained within a drive. Automatic Head Parking
When you power off a hard disk drive using CSS (contact start stop) design, the spring tension in each head arm pulls the heads into contact with the platters. The drive is designed to sustain thousands of takeoffs and landings, but it is wise to ensure that the landing occurs at a spot on the platter that contains no data. Older drives required manual head parking; you had to run a program that positioned the drive heads to a landing zoneusually the innermost cylinderbefore turning off the system. Modern drives automatically park the heads, so park programs are no longer necessary. Some amount of abrasion occurs during the landing and takeoff process, removing just a "micro puff" from the magnetic medium, but if the drive is jarred during the landing or takeoff process, real damage can occur. Newer drives that use load/unload designs incorporate a ramp positioned outside the outer surface of the platters to prevent any contact between the heads and platters, even if the drive is powered off. Load/unload drives automatically park the heads on the ramp when the drive is powered off. One benefit of using a voice coil actuator is automatic head parking. In a drive that has a voice coil actuator, the heads are positioned and held by magnetic force. When the power to the drive is removed, the magnetic field that holds the heads stationary over a particular cylinder dissipates, enabling the head rack to skitter across the drive surface and potentially cause damage. In the voice coil design, the head rack is attached to a weak spring at one end and a head stop at the other end. When the system is powered on, the spring is overcome by the magnetic force of the positioner. When the drive is powered off, however, the spring gently drags the head rack to a park-and-lock position before the drive slows down and the heads land. On some drives, you could actually hear the "ting...ting...ting...ting" sound as the heads literally bounce-parked themselves, driven by this spring. On a drive with a voice coil actuator, you activate the parking mechanism by turning off the computer; you do not need to run a program to park or retract the heads, as was necessary with early hard disk designs. In the event of a power outage, the heads park themselves automatically. (The drives unpark automatically when the system is powered on.) Air Filters
Nearly all hard disk drives have two air filters. One is called the recirculating filter, and the other is called either a barometric or breather filter. These filters are permanently sealed inside the drive and are designed never to be changed for the life of the drive, unlike many older mainframe hard disks that had changeable filters. A hard disk on a PC system does not circulate air from inside to outside the HDA or vice versa. The recirculating filter permanently installed inside the HDA is designed to filter only the small particles scraped off the platters during head takeoffs and landings (and possibly any other small particles dislodged inside the drive). Because PC hard disk drives are permanently sealed and do not circulate outside air, they can run in extremely dirty environments (see Figure 9.12). Figure 9.12. Air circulation in a hard disk.
The HDA in a hard disk drive is sealed but not airtight. The HDA is vented through a barometric or breather filter element that enables pressure equalization (breathing) between the inside and outside of the drive. For this reason, most hard drives are rated by the drive's manufacturer to run in a specific range of altitudes, usually from 1,000 feet below to 10,000 feet above sea level. In fact, some hard drives are not rated to exceed 7,000 feet while operating because the air pressure would be too low inside the drive to float the heads properly. As the environmental air pressure changes, air bleeds into or out of the drive so internal and external pressures are identical. Although air does bleed through a vent, contamination usually is not a concern because the barometric filter on this vent is designed to filter out all particles larger than 0.3 microns (about 12 m-inches) to meet the specifications for cleanliness inside the drive. You can see the vent holes on most drives, which are covered internally by this breather filter. Some drives use even finer grade filter elements to keep out even smaller particles. I conducted a seminar in Hawaii several years ago, and several of the students were from one of the astronomical observatories atop Mauna Kea. They indicated that virtually all the hard disk drives they had tried to use at the observatory site had failed very quickly, if they worked at all. This was no surprise because the observatories are at the 13,796-foot peak of the mountain, and at that altitude, even people don't function very well! At the time, they had to resort to solid-state (RAM) disks, tape drives, or even floppy disk drives as their primary storage medium. IBM's Adstar division, which made all IBM hard drives before the creation of the Hitachi Global Storage Technologies joint venture, developed a line of rugged 3 1/2" drives that are hermetically sealed (airtight), although they do have air inside the HDA. Because they carry their own internal air under pressure, these drives can operate at any altitude and can withstand extremes of shock and temperature. The drives are designed for military and industrial applications, such as systems used aboard aircraft and in extremely harsh environments. They are, of course, more expensive than typical hard drives that operate under ambient pressures. Other vendors, such as EDO MBM Rugged Systems Ltd. (www.edombmrugged.co.uk/), have also introduced hermetically sealed drives. Hard Disk Temperature Acclimation
Because most hard drives have a filtered port to bleed air into or out of the HDA, moisture can enter the drive, and after some period of time, it must be assumed that the humidity inside any hard disk is similar to that outside the drive. Humidity can become a serious problem if it is allowed to condenseand especially if you power up the drive while this condensation is present. Most hard disk manufacturers have specified procedures for acclimating a hard drive to a new environment with different temperature and humidity ranges, and especially for bringing a drive into a warmer environment in which condensation can form. This situation should be of special concern to users of laptop or portable systems. If you leave a portable system in an automobile trunk during the winter, for example, it could be catastrophic to bring the machine inside and power it up without allowing it to acclimate to the temperature indoors. The following text and Table 9.6 are taken from the factory packaging that Control Data Corporation (later Imprimis and eventually Seagate) used to ship with its hard drives:
If you have just received or removed this unit from a climate with temperatures at or below 50°F (10°C) do not open this container until the following conditions are met, otherwise condensation could occur and damage to the device and/or media may result. Place this package in the operating environment for the time duration according to the temperature chart. As you can see from this table, you must place a hard disk drive that has been stored in a colder-than-normal environment into its normal operating environment for a specified amount of time to allow it to acclimate before you power it on. Spindle Motors
The motor that spins the platters is called the spindle motor because it is connected to the spindle around which the platters revolve. Spindle motors in hard disk drives are always connected directly; no belts or gears are involved. The motor must be free of noise and vibration; otherwise, it can transmit a rumble to the platters, which can disrupt reading and writing operations. The spindle motor also must be precisely controlled for speed. The platters in hard disk drives revolve at speeds ranging from 3,600rpm to 15,000rpm (60250 revolutions per second) or more, and the motor has a control circuit with a feedback loop to monitor and control this speed precisely. Because the speed control must be automatic, hard drives do not have a motor-speed adjustment. Some diagnostics programs claim to measure hard drive rotation speed, but all these programs do is estimate the rotational speed by the timing at which sectors pass under the heads. There is actually no way for a program to measure the hard disk drive's rotational speed; this measurement can be made only with sophisticated test equipment. Don't be alarmed if some diagnostics program tells you that your drive is spinning at an incorrect speed; most likely, the program is wrong, not the drive. Platter rotation and timing information is not provided through the hard disk controller interface. In the past, software could give approximate rotational speed estimates by performing multiple sector read requests and timing them, but this was valid only when all drives had the same number of sectors per track and spun at the same speed. Zoned-bit recordingcombined with the many various rotational speeds used by modern drives, not to mention built-in buffers and cachesmeans that these calculation estimates can't be performed accurately by software. On most drives, the spindle motor is on the bottom of the drive, just below the sealed HDA. Many drives today, however, have the spindle motor built directly into the platter hub inside the HDA. By using an internal hub spindle motor, the manufacturer can stack more platters in the drive because the spindle motor takes up no vertical space. Note Spindle motors, particularly on the larger form-factor drives, can consume a great deal of 12-volt power. Most drives require two to three times the normal operating power when the motor first spins the platters. This heavy draw lasts only a few seconds or until the drive platters reach operating speed. If you have more than one drive, you should try to sequence the start of the spindle motors so the power supply does not have to provide such a large load to all the drives at the same time. Most SCSI and some ATA drives have a delayed spindle-motor start feature.
Fluid Dynamic Bearings
Traditionally, spindle motors have used ball bearings in their design, but limitations in their performance have now caused drive manufacturers to look for alternatives. The main problem with ball bearings is that they have approximately 0.1 micro-inch (millionths of an inch) of runout, which is lateral side-to-side play in the bearings. Even though that might seem small, with the ever increasing density of modern drives, it has become a problem. This runout allows the platters to move randomly that distance from side to side, which causes the tracks to wobble under the heads. Additionally, the runout plus the metal-to-metal contact nature of ball bearings allows an excessive amount of mechanical noise and vibration to be generated, and that is becoming a problem for drives that spin at higher speeds. The solution is a new type of bearing called a fluid dynamic bearing, which uses a highly viscous lubricating fluid between the spindle and sleeve in the motor. This fluid serves to dampen vibrations and movement, allowing runout to be reduced to 0.01 micro-inches or less. Fluid dynamic bearings also allow for better shock resistance, improved speed control, and reduced noise generation. Several of the more advanced drives on the market today already incorporate fluid dynamic bearings, especially those designed for very high spindle speeds, high areal densities, or low noise. Over the next few years, I expect to see fluid dynamic bearings become standard issue in most hard drives. Logic Boards
All hard disk drives have one or more logic boards mounted on them. The logic boards contain the electronics that control the drive's spindle and head actuator systems and present data to the controller in some agreed-upon form. On ATA drives, the boards include the controller itself, whereas SCSI drives include the controller and the SCSI bus adapter circuit. Many disk drive failures occur in the logic board, not in the mechanical assembly. (This statement does not seem logical, but it is true.) Therefore, you sometimes can repair a failed drive by replacing the logic board rather than the entire drive. Replacing the logic board, moreover, enables you to regain access to the data on the drivesomething that replacing the entire drive does not provide. Unfortunately, none of the drive manufacturers sell logic boards separately. The only way to obtain a replacement logic board for a given drive is to purchase a functioning identical drive and then cannibalize it for parts. Of course, it doesn't make sense to purchase an entire new drive just to repair an existing one except in cases in which data recovery from the old drive is necessary. If you have an existing drive that contains important data, and the logic board fails, you will be unable to retrieve the data from the drive unless the board is replaced. Because the value of the data in most cases far exceeds the cost of the drive, a new drive that is identical to the failed drive can be purchased and cannibalized for parts such as the logic board, which can be swapped onto the failed drive. This method is common among companies that offer data recovery services. They stock a large number of popular drives they can use for parts to allow data recovery from defective customer drives they receive. Most of the time the boards are fairly easy to change with nothing more than a screwdriver. Merely removing and reinstalling a few screws as well as unplugging and reconnecting a cable or two are all that is required to remove and replace a typical logic board. Cables and Connectors
Hard disk drives typically have several connectors for interfacing to the computer, receiving power, and sometimes grounding to the system chassis. Most drives have at least these three types of connectors:
Of these, the interface connectors are the most important because they carry the data and command signals between the system and the drive. In most cases, the drive interface cables can be connected in a daisy-chain or bus-type configuration. Most interfaces support at least two devices, and SCSI (Small Computer System Interface) can support up to seven (Wide SCSI can support up to fifteen) devices in the chain, in addition to the host adapter. Older interfaces, such as ST-506/412 or ESDI (Enhanced Small Device Interface), used separate cables for data and control signals, but today's SCSI, ATA (AT Attachment), and Serial ATA drives have a single data connector on each drive.
The power is supplied via the larger four-pin peripheral power connector found on all PC power supplies. Most hard disk drives use both 5- and 12-volt power, although some of the smaller drives designed for portable applications use only 5-volt power. In most cases, the 12-volt power runs the spindle motor and head actuator, and the 5-volt power runs the circuitry. Make sure your power supply can supply adequate power for the hard disk drives installed in your system. The 12-volt power consumption of a drive usually varies with the physical size of the unit. The larger the drive is, the faster it spins. In addition, the more platters there are to spin, the more power it requires. For example, most of the 3 1/2" drives on the market today use roughly one-half to one-fourth the power (in watts) of the older 5 1/4" drives. Some of the very small (2 1/2" or 1.8") hard disks barely sip electrical power and actually use 1 watt or less! A grounding tab provides an optional ground connection between the drive and the system's chassis. In most computers, the hard disk drive is mounted directly to the chassis using screws, or the drive is grounded via the ground wires in the power connector, so an extra ground wire is unnecessary. Configuration Items
To configure a hard disk drive for installation in a system, you usually must set several jumpers (and, possibly, terminating resistors) properly. These items typically vary according to the type of interface the drive supports but can vary somewhat from drive to drive as well.
The Faceplate or Bezel
At one time, hard disk drive vendors offered a front faceplate, or bezel, as an option (see Figure 9.13). In most systems today, the bezel is a part of the case and not the drive itself. Figure 9.13. Typical 5 1/4" and 3 1/2" hard drive bezel shown from the front (as seen on the outside of the PC case) (top) and from the back (bottomthe inside mounting and LED wiring).
Older systems had the drive installed so it was visible outside the system case. To cover the hole in the case, you would use an optional bezel or faceplate. Because today's drives are almost always mounted in an internal bay or an external bay behind a blank faceplate, you seldom need to buy a drive bezel. If you need to use one, you typically use one included with your system or case. However, drive bezels are still available from several vendors in several sizes and colors to match various PC systems. Many faceplate configurations for 3 1/2" drives are available, including bezels that fit 3 1/2" drive bays as well as 5 1/4" drive bays. You even have a choice of colors (usually black, cream, or white). Some bezels feature a light-emitting diode (LED) that flickers when your hard disk is in use. The LED is mounted in the bezel; the wire hanging off the back of the LED plugs into the drive. In some drives, the LED is permanently mounted on the drive, and the bezel has a clear or colored window so you can see the LED flicker while the drive is being accessed. Most systems today have an LED indicating drive access on the case's front panel, which is connected to a hard drive LED connector on the motherboard. In systems in which the hard disk is hidden by the unit's cover, a bezel is unnecessary. In fact, using a bezel can prevent the cover from resting on the chassis properly, in which case the bezel must be removed. If you are installing a drive that does not have a proper bezel, frame, or rails to attach to the system, check the Vendor List on the disc accompanying this book; several listed vendors offer these accessories for a variety of drives. |
Категории