As I mentioned in the introductory page to my coverage of this interface, its common
name, "IDE", also tells you quite a bit about its history. IDE drives were the
first ones to popularize integrating the logic controller onto the hard disk itself. This
change corrected many of the problems that had been associated with hard disks up to that
point, such as poor signal integrity, complexity and the need for every controller to be
"generically" capable of dealing with any hard disk. You can read more about the
logic board and its significance on this page. Here's a
(brief) bit of history of how integrated disk controllers, and the IDE/ATA interface
itself, came about.
The very first hard disks to have integrated controllers weren't technically using the
IDE/ATA interface as we currently know it. They were in fact so-called
"hardcards", which were designed and sold by the "Plus Development"
division of Quantum. These devices were simply 3.5" hard disks that were mounted
directly to a controller card that plugged into an ISA expansion slot. They were popular
amongst those who wanted to add additional storage to existing systems, because of their
simplicity: with the disk and controller integrated, one could just slap the package into
an existing PC like any add-in peripheral, without affecting the existing hard disk and
controller.
While an ingenious idea--one that basically put Quantum "on the map" in the
hard disk world--these hardcards had several problems. Hard disks are heavy, and are
mechanical devices. They just aren't well-suited to being mounted in expansion slots like
a video card or modem. Physical support for expansion cards is poor, because they are held
in place using just one screw; this resulted in vibration issues. The width of the drive
would typically block off at least one additional bus slot, meaning one of these devices
took two expansion bus slots. Cooling was also sometimes a problem. Of course, all of
these are physical implementation issues, and don't detract from the significance of the
advance in integrating the controller and disk assembly.
It didn't take long until manufacturers realized that there was really no reason to
keep the hard disk physically on the controller at all. They decided to put the controller
on the bottom of the hard disk and move the entire hard disk and controller assembly to a
regular drive bay. The connection to the system bus was maintained through the use of a
cable that ran either directly to a system bus slot, or to a small interfacing card that
plugged into a system bus slot. In many ways, then, these drives were connected directly
to the ISA system bus. The official name for the interface, "AT Attachment",
reflects this, as the IBM PC/AT was the first PC to use the now-standard 16-bit ISA bus.
The origins of the actual IDE/ATA interface go back to around the same time of the
hardcard. Compaq Computer, an up-and-coming competitor to IBM at the time (and still a
major market player today) wanted to integrate the hard disk controller into the hard disk
itself to improve flexibility and save a bus slot. They worked to create the precursors of
the modern IDE interface electronics with Western Digital (which was then an interface
logic company and not a hard disk maker) and a division of Control Data Corporation that
is now part of Seagate--isn't this fun? 
Compaq began selling PCs with integrated hard
disks using Western Digital controllers starting with their IBM-compatible Deskpro 386 in
1986. Other manufacturers quickly caught on to the idea and the IDE concept grew in
popularity rather quickly.
As system and hard disk performance improved, the slow speed of the ISA bus became an
issue, so interface cards--often called controller cards, despite the confusion
that this causes--were created for the higher speed VESA local bus, and then the PCI bus. Today, all modern PCs have their
IDE/ATA interface attached directly to the PC bus.
The next evolution of how IDE/ATA drives interface to the system occurred when it
became obvious that every PC was going to have a hard disk, and it was therefore silly to
waste an expansion slot even on a hard disk interface card. Chipset manufacturers began
integrating IDE/ATA hard disk controllers into their chipsets, so that instead of
connecting the drives to a controller card, they were connected directly to the
motherboard. With this change, integration of the interface was complete, with all the
logic either on the motherboard, or the hard disk itself. This is the technique that is
used today (though one can still add a physical controller card if necessary.)
In terms of its basic operation, the IDE/ATA interface is fairly straight-forward, and
also reflects its origins as an extension of the ISA system bus. The connection between
the system and the hard disks is 16 bits wide, so two bytes of data are passed at a time
between the system and any hard disk. This is true regardless of the width of the system
bus, and persists even today with high-performance enhancements like Ultra DMA. Two drives are supported on each IDE/ATA channel,
with special signaling used to ensure that commands sent for one drive don't interfere
with the other. Over time, many performance and reliability features have evolved as the
interface has matured, which I will go into in detail as we discuss the various aspects of
the IDE/ATA interface.
Note: In the early days
of IDE, there were two other variants on the IDE/ATA interface, that were not compatible
with regular IDE/ATA. One was an 8-bit version intended for use on the ancient 8-bit ISA
bus of the first IBM PC/XT machines and clones. Another was a 16-bit version designed for
IBM's MCA (MicroChannel) systems.
Both of these have been obsolete for a decade or more and are no longer seen in modern
systems; I do not cover them on this site (other than this note. )
Next: Official
IDE/ATA Standards and Feature Sets
