# Computer Hard Disks

The computer hard disk is a sealed unit in a computer that stores data on a magnetic coating applied to the surface of circular platters that spin like a record. The platters are constructed on a base of aluminum or glass (both nonferromagnetic), which makes them rigid - hence the term hard disk. Since the unit is sealed, the internal platters and components are inaccessible, and a "crash" (a term applied to the loss of data from a disk or the malfunction thereof) usually requires that the entire unit be replaced.
Hard disks are currently available with diameters from less than $1$ in. to $5 {1 \over 4}$ in., with the $3 {1 \over 2}$ in. the most popular for today's desktop units. Lap-top units typically use $2 {1 \over 2}$ in. All hard disk drives are often referred to as Winchester drives, a term first applied in the 1960s to an IBM drive that had 30 MB [a byte is a series of binary bits (0s and 1s) representing a number, letter, or symbol] of fixed (non-accessible) data storage and 30 MB of accessible data storage. The term Winchester was applied because the 30-30 data capacity matched the name of the popular 30-30 Winchester rifle.
The magnetic coating on the platters is called the media and is of either the oxide or the thin-film variety. The oxide coating is formed by first coating the platter with a gel containing iron-oxide (ferromagnetic) particles. The disk is then spun at a very high speed to spread the material evenly across the surface of the platter. The resulting surface is then covered with a protective coating that is made as smooth as possible. The thin-film coating is very thin, but durable, with a surface that is smooth and consistent throughout the disk area. In recent years the trend has been toward the thin-film coating because the read/write heads (to be described shortly) must travel closer to the surface of the platter, requiring a consistent coating thickness. Recent techniques have resulted in thin-film magnetic coatings as thin as one-millionth of an inch.
The information on a disk is stored around the disk in circular paths called tracks or cylinders, with each track containing so many bits of information per inch. The product of the number of bits per inch and the number of tracks per inch is the Areal density of the disk, which provides an excellent quantity for comparison with early systems and reveals how far the field has progressed in recent years. In the 1950s the first drives had an Areal density of about 2 kbits/in.2 compared to today's typical 4 Gbits/in., an incredible achievement; consider 4,000,000,000,000 bits of information on an area the size of the face of your watch. Electromagnetism is the key element in the writing of information on the disk and the reading of information off the disk.
Fig. 1: Hard disk storage using a U-shaped electromagnet write head
In its simplest form the write/read head of a hard disk (or floppy disk) is a U-shaped electromagnet with an air gap that rides just above the surface of the disk, as shown in Fig. 1. As the disk rotates, information in the form of a voltage with changing polarities is applied to the winding of the electromagnet. For our purposes we will associate a positive voltage level with a 1 level (of binary arithmetic) and a negative voltage level with a 0 level. Combinations of these 0 and 1 levels can be used to represent letters, numbers, or symbols. If energized as shown in Fig. 1 with a 1 level (positive voltage), the resulting magnetic flux pattern will have the direction shown in the core. When the flux pattern encounters the air gap of the core, it jumps to the magnetic material (since magnetic flux always seeks the path of least reluctance and air has a high reluctance) and establishes a flux pattern, as shown on the disk, until it reaches the other end of the core air gap, where it returns to the electromagnet and completes the path. As the head then moves to the next bit sector, it leaves behind the magnetic flux pattern just established from the left to the right. The next bit sector has a 0 level input (negative voltage) that reverses the polarity of the applied voltage and the direction of the magnetic flux in the core of the head. The result is a flux pattern in the disk opposite that associated with a 1 level. The next bit of information is also a 0 level, resulting in the same pattern just generated. In total, therefore, information is stored on the disk in the form of small magnets whose polarity defines whether they are representing a 0 or a 1. Now that the data have been stored, we must have some method to retrieve the information when desired. The first few hard disks used the same head for both the write and the read functions. In Fig. 2(a), the U-shaped electromagnet in the read mode simply picks up the flux pattern of the current bit of information. Faraday's law of electromagnet induction states that a voltage is induced across a coil if exposed to a changing magnetic field. The change in flux for the core in Fig. 2(a) is minimal as it passes over the induced bar magnet on the surface of the disk. A flux pattern is established in the core because of the bar magnet on the disk, but the lack of a significant change in flux level results in an induced voltage at the output terminals of the pickup of approximately 0 V, as shown in Fig. 2(b) for the readout waveform.
(a)
(b)
Fig. 2: Reading the information off a hard disk using a U-shaped electromagnet.
A significant change in flux occurs when the head passes over the transition region so marked in Fig. 2(a). In region a the flux pattern changes from one direction to the other-a significant change in flux occurs in the core as it reverses direction, causing a measurable voltage to be generated across the terminals of the pickup coil as dictated by Faraday's law and indicated in Fig. 2(b). In region b there is no significant change in the flux pattern from one bit area to the next, and a voltage is not generated, as also revealed in Fig. 2(b). However, when region c is reached, the change in flux is significant but opposite that occurring in region a, resulting in another pulse but of opposite polarity. In total, therefore, the output bits of information are in the form of pulses that have a shape totally different from the read signals but that are certainly representative of the information being stored. In addition, note that the output is generated at the transition regions and not in the constant flux region of the bit storage.
In the early years, the use of the same head for the read and write functions was acceptable; but as the tracks became narrower and the seek time (the average time required to move from one track to another a random distance away) had to be reduced, it became increasingly difficult to construct the coil or core configuration in a manner that was sufficiently thin with minimum weight. In the late 1970s IBM introduced the thin film inductive head, which was manufactured in much the same way as the small integrated circuits of today. The result is a head having a length typically less than 1/10 in., a height less than 1/50 in., and minimum mass and high durability. The average seek time has dropped from a few hundred milliseconds to 6 ms to 8 ms for very fast units and 8 ms to 10 ms for average units. In addition, production methods have improved to the point that the head can "float" above the surface (to minimize damage to the disk) at a height of only 5 micro-inches or 0.000005 in. Using a typical lap-top hard disk speed of 3600 rpm (as high as 7200 rpm for desktops) and an average diameter of 1.75 in. for a 3.5-in. disk, the speed of the head over the track is about 38 mph. Scaling the floating height up to 1/4 in. (multiplying by a factor of 50,000), the speed would increase to about $1.9 \times 10^6$ mph. In other words, the speed of the head over the surface of the platter is analogous to a mass traveling 1/4 in. above a surface at 1.9 million miles per hour, all the while ensuring that the head never touches the surface of the disk-quite a technical achievement and amazingly enough one that perhaps will be improved by a factor of 10 in the next decade. Incidentally, the speed of rotation of floppy disks is about 1/10 that of the hard disk, or 360 rpm. In addition, the head touches the magnetic surface of the floppy disk, limiting the storage life of the unit. The typical magnetizing force needed to lay down the magnetic orientation is 400 mA-turn (peak-to-peak). The result is a write current of only 40 mA for a 10-turn, thin-film inductive head.
Although the thin-film inductive head could also be used as a read head, the magneto-resistive (MR) head has improved reading characteristics. The MR head depends on the fact that the resistance of a soft ferromagnetic conductor such as permolloy is sensitive to changes in external magnetic fields. As the hard disk rotates, the changes in magnetic flux from the induced magnetized regions of the platter change the terminal resistance of the head. A constant current passed through the sensor displays a terminal voltage sensitive to the magnitude of the resistance. The result is output voltages with peak values in excess of 300 V, which exceeds that of typical inductive read heads by a factor of 2 or 3:1.
Further investigation will reveal that the best write head is of the thin-film inductive variety and that the optimum read head is of the MR variety. Each has particular design criteria for maximum performance, resulting in the increasingly common dual-element head, with each head containing separate conductive paths and different gap widths. The Areal density of the new hard disks will essentially require the dual-head assembly for optimum performance.
As the density of the disk increases, the width of the tracks or cylinders will decrease accordingly. The net result will be smaller heads for the read/write function, an arm supporting the head that must be able to move into and out of the rotating disk in smaller increments, and an increased sensitivity to temperature effects which can cause the disk itself to contract or expand. At one time the mechanical system with its gears and pulleys was sensitive enough to perform the task. However, today's density requires a system with less play and with less sensitivity to outside factors such as temperature and vibration. A number of modern drives use a voice coil and ferromagnetic arm as shown in Fig. 3. The current through the coil will determine the magnetic field strength within the coil and will cause the supporting arm for the head to move in and out, thereby establishing a rough setting for the extension of the arm over the disk.
Fig. 3: Disk drive with voice coil and ferromagnetic arm.
It would certainly be possible to relate the position of the arm to the applied voltage to the coil, but this would lack the level of accuracy required for high-density disks. For the desired accuracy, a laser beam has been added as an integral part of the head. Circular strips placed around the disk (called track indicators) ensure that the laser beam homes in and keeps the head in the right position. Assuming that the track is a smooth surface and the surrounding area a rough texture, a laser beam will be reflected back to the head if it's on the track, whereas the beam will be scattered if it hits the adjoining areas. This type of system permits continuous recalibration of the arm by simply comparing its position with the desired location-a maneuver referred to as "recalibration on the fly."
As with everything, there are limits to any design. However, in this case, it is not because larger disks cannot be made or that more tracks cannot be put on the disk. The limit to the size of hard drives in PCs is set by the BIOS (Basic Input Output System) drive that is built into all PCs. When first developed years ago, it was designed around a maximum storage possibility of 8.4 gigabytes. At that time this number seemed sufficiently large to withstand any new developments for many years to come. However, 8-gigabyte drives and larger are now becoming commonplace, with lap-tops averaging 20 gigabytes and desktops averaging 40 gigabytes. The result is that mathematical methods had to be developed to circumvent the designed maximum for each component of the BIOS system. Fundamentally, the maximum values for the BIOS drive are the following:
• Cylinders (or tracks) 1024