A little technical information does not hurt. If you are not into electronics or technology, this section might seem boring. But I assure you it isn't...
In the beginning, disc drives were extremely big, noisy, low-speed devices that were designed to communicate with a special controller connected to a personal computer or a mainframe. They quickly evolved and decreased in physical size and increased in capacity.
The picture below shows two of my all time favorite disc drives. The Quantum ProDrive LP-52s 50Mb disc drive which is very easy to remember because of its unusual capacity, and the Seagate ST-351A/X 40Mb disc drive, which I consider the best of all the IDE stepper drives.
Other approaches to computer storage media include solid state drives. At some point in the late 1980s and early 1990s, there were some attempts to make them with special static RAM chips. Quantum used to manufacture them for a while, but they were very expensive and they required another expensive piece of hardware: a SCSI host adapter. They didn't make it to common people like me and you. However, the conventional drives that includes moving parts such as platters and heads, was quickly adopted and gained huge popularity mainly due to the low cost per Mo and their enhanced reliability.
Later Edit: Now as we are in 2011, the solid state devices (SSD) are returning under the form of NAND memory. This is new technology and is the future of storage in computing technology.
You probably know that a disc drive contains a number of one or more rigid aluminum or glass platters, the said discs. If there weren't for these discs, they wouldn't be called hard disc drives. These discs are perfectly machined in order to ensure for accurate movement. They are coated with a high-quality magnetic material that has the property of switching states as the head writes individual 0 and 1 states. Each individual platter has the same geometry that divides it into sectors and tracks.
In order to reach for the specific data you need, the read/write heads first need to correctly position on the right track and then wait for the sector that contains your data. This process is amongst other, what defines the drive seek speed.
The platters are coated with an oxide material, most commonly found on pre-1990 year drives, or a thin film, commonly found on all other drives.
Disc drive units read data from the platters by using an array of heads mounted on a mobile actuator. The heads are mechanically connected to the actuator via flexible stainless steel spring tensioning arms that are calibrated to the required pressing force. Each platter has 1 or 2 corresponding heads, depending on the drive configuration and factory design. Early drives used one platter for control and positioning purposes. Modern drives interleave positioning servo information with the actual data.
These heads are electronic devices containing small coils and a very tiny gaps, all located on a special supporting material. The heads can be either traditional metal in gap (MIG), magnetoresistive (MR), giant magnetoresistive (GMR) or tunneling magnetoresistive (TMR).
In write operations, their role is to magnetize the particular portion of the platter(s), defined by track and sector, where the drive controller knows there is empty space available. The electric signals from the drive electronics is converted into a magnetic field that further magnetizes the required platter surface.
In read operations, the heads detect the previously magnetized disc surface and convert the stored magnetic fields into electric currents.
The magnetic heads do not directly touch the platters. Instead they fly over the surface of the platters at a very low distance, measured in only a few nanometers.
In the picture, I illustrate an array of heads mounted on the arm of the actuator of an old Seagate 5.25" full height SCSI disc drive. Note the tiny individual purple cables that connect the individual heads to the drive electronics. The actuator assembly is a high precision electro-mechanical part.
The array of heads are connected to a series of head amplifiers. These are electronic devices built as integrated circuits and their role is to amplify the logic signals that the drive electronics are sending to the heads, or those that are read by the heads from the platters. Head amplifiers also adapt the impedance of the heads to the signal transmit flexible ribbon cable in order to ensure electric stability.
Modern drives contain dedicated integrated circuits specially manufactured for them. In the picture you can see 3 individual head amplifier units. You would think that they are a single amplifier per chip. Instead this particular integrated circuit contains 6 different amplifiers built-in. In our case the drive has 8 platters and 16 heads. So the 3 individual head amplifier integrated circuits give us 18 individual amplifiers. Naturally, 2 of them are probably unused as it is a very common practice.
The head amplifiers are always mounted as close as possible to the heads. So it's very likely that you'll find them under the top cover, mounted on a flexible PCB that extends via flexible ribbon cable to almost half the length of the actuator assembly.
The actuator is moved by either a stepper motor or a voice coil. The stepper motor technology belongs to the 1980s and the early 1990s while the voice coil technology is what all modern drives use. In the picture you can see a very large and powerful voice coil that powers the movement of the 16 head actuator that I was talking about above. Below the voice coil you can see a very strong neodymium magnet. I removed the top one for obvious photographic reasons.
The voice coil is just a coil built from enamel insulated copper wire. It is called like that because its principle is the same as that of the speakers. The voice coil receives electric currents with a specific intensity calculated by the drive electronics in order to produce a precise movement of the actuator to the required track. These electric currents generate a magnetic field in the coil which forces it to move to destination in the strong permanent magnetic field caused by the 2 neodymium magnets.
The voice coil servo positioning information is often read by the heads from one of the platters or from interleave encoded position data. When power is applied to a disc drive, it first calibrates. That is what the crazy random start-up squealing sounds mean. The calibration process ensures the heads will always correctly track servo positioning data and also regular data on the platters.
Each disc drive has at least some sort of drive electronics that handles the spindle motor and the actuator positioning commands. The drive electronics is also responsible of converting and interpreting the signals that come from the disc controller or from the read/write heads.
These days, drive electronics are often very complex microsystems (as in microcomputers) that have a central processing unit (CPU), individual RAM, and a specific ROM (firmware). All these in addition to the various motor and head controllers. Pictured is a portion of an over-engineered drive electronics from the same disc drive I was talking about above. It is powered by an Intel 80C188 CPU. Now hold on a second, there used to be computers based on 80C186 and 80C188 CPUs. You can see at least a dozen of other very large scale integrated (VLSI) circuits and an army of passive components.
As technology matured, specific integrated circuits and microcontrollers were built in order to reduce the number of electronic parts thus production costs. Nowadays it's quite common to see drive logic boards with only one big ULSI integrated circuit and a few other supporting parts.
Hard disc drives need a way to communicate to the computer. This specific piece of hardware that connects the drive to the computer is called a controller. The controller has a specific electrical interface. Interfaces are of different types, the oldest one being ST-506 from 1981.
The ST-506 interface and its flavors such as ST-412 and ST-412 RLL, were the first that allowed drives of the era to connect to the personal computer. These interfaces use either modified frequency modulation (MFM) or run length limited (RLL) technology in order to transmit and encode data on the platters.
Both MFM and RLL interfaces are electrically identical but RLL uses a different data transfer format. It squeezes the data onto the platters in order to increase the storage capacity by up to 50%. Common practices back then were to format a MFM drive with a RLL controller to gain more storage space and a lower media quality or format a RLL drive with a MFM controller to gain a more reliable media but loosing some storage space. These drives connect to the controller via 2 cables per drive.
IDE is the next big thing after the ST-506. The IDE interface means that the controller itself is located on the disc drive instead of the separate extension logical board. This means that the IDE extension logical board controller is not really a controller but a host interface. The host interface translates the IDE commands and adapts them to the ISA or PCI bus of the computer. The drive is connected to the adapter via a single 40-pin or 80-pin ribbon cable.
As IDE technology matured, it evolved into ATA (AT Attachment), ATA-2, ATA-3 and so on. All these new technologies built on the IDE basis, introduced new features such as ATA Identify Command, S.M.A.R.T. monitoring, and so on, while constantly increasing the communication bandwidth.
IDE was by far the most common drive interface throughout the 1990s and the early 2000s. As we are in 2011 now, the dominant technology is SATA-3 while IDE is on its way to extinction, becoming obsolete.
ESDI was another type of interface that was the performance equivalent of SCSI at the time. However the technology died soon and we never heard of it again. ESDI interfaced the drive to the computer via 2 connection cables.
SCSI stands for Small Computer System Interface. It was and still is the ultimate high-speed communication technology available to connect disc drives to a computer. Back then, top-of-the line systems such as the Apple Macintosh, almost exclusively used a SCSI bus. The physical connection between the drive and the host adapter is done through a single 50-pin or 68-pin ribbon cable.