LogicSmith

History of Hard Drives

The hard disk drive was invented by some IBM engineers working under Rey Johnson at IBM in San Jose, CA, in about 1952 to 1954. I worked at IBM from 1965 to 1981 and got to meet and work with some of those men - Rey Johnson, John Lynott, Don Cronquist, Bob Schneider and Lou Stevens come to mind right now.

In 1965 (I think that was the year) a number of engineers left IBM (they were known as the "dirty dozen" within IBM) and founded Memorex. Al Shugart, one of them, later left Memorex and founded first Shugart Associates where the 5 1/4" floppy disk drive was a major product, then Seagate Technology, which effectively started today's industry of small hard disk drives.

The early drives almost all had linear actuators, that is, they moved the heads across the disks in a straight line, using a carriage with wheels. It was only later that rotary actuators, where the heads are held at the tips of a comb-like array and they swing back and forth like a gate, became popular. Because the rotary actuator is cheaper, it's now the standard for all hard disk drives, and that's what I'll be talking about.

The first IBM RAMAC disk drive had a couple of dozen disks, each about 2 feet in diameter, and ONE head! The head was moved from disk to disk and back and forth on each disk using a system of cables and pulleys and stepping motors. The added speed of having at least one head for each disk surface, and of using both surfaces of each disk, soon became obvious, and drives began to look pretty "modern" by 1960, although they were vastly larger and more expensive. Whether the heads are moved in a straight line or swung in an arc, something has to provide the force and the control to move them and keep them in the right place.

Stepping motors, hydraulic actuators and voice coil motors have been used to provide the motive force. Stepping motors have a built-in capability to hold in one position. Hydraulic actuators and voice coil motors (VCMs) provide force, but can't hold a position with great accuracy. A rack with detent pawls has been used, but nowadays a servo system is used, with the positioning information recorded on the disks.

So you can have a disk drive with a stepping motor and you don't need a "servo" or you can have a disk drive with "servo data" recorded on the disks. A stepping motor is simpler (at least in concept) and cheaper, but it's slower in seek time because it isn't really very powerful, and it isn't capable of really, really fine precision. A VCM can provide enormous forces, but it needs control, which the servo system provides. However servo feedback systems are complicated and you have to pre-record the positioning data somehow (usually on one or more of the disks), and that takes up space that could be used for "real" data. On the other hand, servo systems can provide incredibly precise positioning.

The style of hard disk drive we use today began to emerge in the early 1980s. I think it was Maxtor, under Frank Gibeau, where the first high-volume 5 1/4" disk drives with a rotary actuator, a VCM and a servo system were produced.

In 1986, Finis Conner left Seagate and founded Conner Peripherals along with John Squires, and they built the first high-volume 3 1/2" disk drives. The first one, 40 MB, was called the "Fat 40". Not only did they popularize the new smaller "form factor", but they were the first to have an "embedded servo" or "sector servo" in volume.

Meanwhile, Quantum Corporation had been building 8" and 5 1/4" disk drives since 1980, and in the mid 1980's they (actually I think it was Joel Harrison and Bill Moon) saw an opportunity with the 3 1/2" form factor and invented the "hard card", a disk drive on an expansion card that you could just plug into your AT. And that's how the IDE interface got started.

By the way, Quantum used a rather odd variant of the servo system for many years, where the servo information was actually very fine lines etched on a piece of glass attached to the actuator, and read with a photocell. It's actually more complicated, but that's a subject for another discussion.

Around 1990, laptops began to appear, and with them came the 2 1/2" form factor. I never worked closely with 2 1/2" drives, and I'm not very well versed in their historical development.

Way back in the old days, when the world was young and all, <g>, a single disk surface was reserved for "servo data" on disk drives that had a voice-coil actuator. Drives with stepper motors didn't need that stuff, and could justly claim that they didn't have to "waste" disk surface on "servo data".

Servo data, by the way, is information that's pre-recorded on the disk and specially formatted to make it possible for the drive to know where its heads are positioned. A stepper motor has, if no errors have occurred, the equivalent information built into its mechanical structure.

But during the late 1970s and early 1980s, techniques were developed that allowed the servo data to be written on the same surfaces that hold the regular user data. There were several schemes proposed and actually implemented, but the one that has taken hold is called "sector servo", where some number of regions on each data track are specially reserved for servo information. Because the sectors are physically coherent on each disk surface, they're commonly called "servo spokes". I've seen drives with as few as 64 spokes per revolution and as many as 128.

I believe the first major manufacturer to use sector servo was Conner--sector servo and the 3 1/2" form factor were their keys to success when they started in (I think) 1986. [At the time, I was one of the majority of old disk drive people who thought they were headed for disaster. Later, I worked there. Shows how smart I am!]

While there may still be drives manufactured with a dedicated servo surface, I think the last major manufacturer to use one was Seagate, up until a couple of years ago. The first Barracuda drive had a dedicated servo surface. Later Barracudas, and I think all current Seagates, use sector servo technology.

Disk drive production has become easier nowadays - there are even cheap hard drive deals which indicate cost-efficient manufacturing of these devices.


 

Copyright 1999 by Albert Dayes and John Treder