One of the big differences between floppy disks and hard disks is that the heads do not (normally) touch the surface of a hard disk except when at rest. The hard disk is, as the name suggests, made up of a number of rigid hardplattersand thehead–disk assemblyis enclosed in a dust-free environment.
All the heads are fixed to the sameactuatorand fly free just above the surface of the disk, held up by aerodynamic pressure (Fig. 5.5). At the time of writing, a common size of disk platter, usually referred to as theform factor, is 3.5 inches, although 2.5 inches is often used for notebook computers and is becoming more common for standalone drives. The number of heads is likely to be between two and six and the rotational speeds are typically from 3,600 to 7,200 revolutions per minute12. Servo control circuits are used to position the head assembly and to reduce
9 Note that this figure is calculated using powers of ten units for kilo, mega and giga, for which we divide by 1000 × 1000 × 1000. If we were to use powers of two units, more normal for computer work, we would divide by 1024 × 1024 × 1024 and get a result of 7.87 Gbyte, a number which is also sometimes used.
10 Throughout we have used “INT 13h” to refer to the legacy disk interrupt mechanism and “INT 13h extensions” to refer to the extended system.
11 IBM have designed a new low-level format which does not use ID bytes at all. It is called the No-ID™ sector format; see IBM (1995).
12 At the time of writing, disks from a number of manufacturers now have rotational speeds of 15,000 rpm.
any rotational variations in speed to approximately ±0.1%. A nice analogy for the standard of engineering required is quoted here from Seagate (1995a): “Today’s new generation of disc drives achieve the engineering equivalent of a Boeing 747 flying at Mach 4 just two meters above the ground, counting each blade of grass as it flies over.”
As in the case of floppy disks, except for high-performance systems and RAID (see later section), only one head is active at any one time for reading or writing data. Having said that, in some hard disk systems, one of the disk surfaces13may be used for holding pre-formatted control information, and thus its associated head is also in use at the same time as the data head for servo control purposes. Tracks on this surface are calledservo tracksorindex tracksand the head itself is known as the servo heador theindex head. Disks that use this technique often appear to have an odd number of real physical heads, since one has been reserved for servo use. Other systems may instead embed the servo information into the data tracks asservo sectors(IBM, 1995). The head assembly, instead of being positioned by a stepper motor actuator14is controlled by a linear motor system which is often referred to as a voice coil actuator. In this system, the head reads the preset data on the servo tracks and uses this with a feedback loop to position the voice coil actuator and the head
Fig. 5.5 Western Digital Caviar (with kind permission of Western Digital Corporation).
13 Usually the topmost.
14 An outdated mechanical head positioning system that was used with early hard disks drives and is still used for floppy disk drives.
assembly very accurately to any particular cylinder position. It is this approach that has led to a rapid expansion in the capacity of hard disks by radically increasing the number of real physical cylinders15. This development was apparently not foreseen by the early designers who, if the low-level format ID data structure sizes are any indication, expected instead the number of heads to increase towards 256. We recall from our discussion above of the 8.4 Gbyte barrier that the ID field in a sector only has scope to specify up to 1024 cylinders, rather than the many thousands that we may find in modern hard disks. This is yet another limitation to which we will return shortly.
When the drive is switched off, the heads rest on the surface. Most modern disks “park” their heads automatically onto an unused track as they power down, though earlier disks had a specialPARKprogram designed for this purpose. It is certainly unwise practice to use any version of this program with a modern self-parking disk! Typical distances for heads flying free are 0.2 to 0.5 microns above the surface, or some 12 millionths of an inch. By comparison, a fingerprint and a smoke particle are some five to ten times thicker at 3 microns and a hair is some thirty times thicker at 10 microns. Contamination can therefore be a serious potential problem.
Hard disks have an internal air filter which is used for filtering the air already in the case. To avoid the head–disk assembly exploding as a result of low external air pressure, such as might occur in an aircraft cabin at high altitude, the case has a venti- lation slit which is also protected by an air filter.
Note, however, that it is not a good idea to place your laptop in the hold baggage compartment when flying without first removing the hard disk(s) and carrying them in the cabin with your hand baggage. It is not that the decompression of the hold will itself damage the disks; air will simply be forced out of the ventilation slits. The problem occurs on returning to normal atmospheric pressure. Air will then be forced back into the disk, and this air may contain contaminants which the simple filters cannot handle. The result can be a catastrophic head crash. Another problem that can arise is that of thermal shock. The temperature may change from about –25 °C in the hold at cruising height to +20 °C on the ground over a period of about half an hour, and this may result in severe physical damage to the disk.