San Jose, CA--Two decades ago, Silicon Valley icon Gordon Moore predicted that microprocessors would double in performance while shrinking in cost every 18 months.
Good thing Moore's Law didn't apply to disk drives. Because if it did, engineers and researchers from IBM's Almaden Research Center would be up on felony charges right about now. Their new disk drive, known as the Microdrive, breaks all the laws of electronic innovation. With a diameter smaller than that of a half-dollar, it packs 340 Mbytes of data. That's about three times larger than today's biggest flash memories. All in a package about one-ninth the size of a laptop hard drive.
How is that possible? IBM researchers will tell you that it was no easy task. Starting in 1996, the computer giant threw some of its best resources at the project. Along with the Almaden Research Center, IBM drew upon its Storage Systems Divisions in San Jose, and Rochester, MN. It also tapped the expertise of engineers at the Thomas J. Watson Research Center in Yorktown Heights, NY; at facilities in Fujisawa and Yasu, Japan; and at IBM-Thailand.
The result of their effort holds tremendous promise for the electronics market. Engineers say the Microdrive could serve in palm top computers, digital cameras, camcorders, music players, and a host of other applications. There, it could hold more data than anyone has dared dream of up to now.
The key question, however, is: How'd they do it? Almost from the outset, their goal was to build a tiny disk drive that would fit in a slot now used by so-called flash memories. That served as the primary technical challenge--because a Compact Flash Type II memory slot measures a scant 42.8 X 36.4 X 5 mm. To accomplish that, they needed to pack an actuator, spindle drive, bearings, and electronic controls into a package 5 mm deep--that's about one-fifth of an inch. "Our job was to take everything that's inside a 2.5-inch laptop drive and shrink it down to one-ninth the volume," notes Tom Albrecht, manager of mechanical technologies at the Almaden Research Center.
In bringing their drive down to the flash memory form factor, electronics were a key challenge. The first step was to squeeze the device's main chips onto a board with only 10 cm(super 2) of real estate. The main chips, which include the microprocessor, read/write head signal processor, buffer, and spindle drivers, are typically used on boards with about six times as much area. To put all of them on Microdrive's smaller board, IBM's disk drive design team in Fujisawa called on their colleagues in Yasu, who employed a technique called direct chip attach. Direct chip attach, which involves aligning the chips and attaching them to little bonding pads on the board, eliminates the need for lead frames that take up valuable space. "We didn't have room for all of the plastic packaging and lead frames," says Tim Reiley, research staff member at Almaden. "So we matched the bonding pads on the back of the silicon chip to the pads on the board." Using that technique, Reiley says a single chip can attach to as many as a hundred bonding pads.
IBM team members also worked with component suppliers to custom design miniature connectors for the spindle motor, flex cable, and host socket, all as a means of saving space. "We fought for microns," Reiley says.
On the mechanical side, engineers needed to work even harder to meet the 5-mm depth requirement. Their stickiest problem: to get the spindle motor and bearings to fit. To accomplish that, they collaborated with motor and bearing manufacturers to find the best configuration. Ultimately, they settled on a miniature wound motor with so-called 2 X 5 bearings (meaning that the bearing balls have a 2-mm ID and 5-mm OD). To minimize the motor's height, they worked to develop a 12-pole, 9-slot motor that uses an inner rotor design. That's a departure from most motors, which use an inner stator to maximize torque radius. But analysis of small motors showed that an inner rotor would have better shock resistance and more winding space.
With the height constraints looming as a challenge, no detail was too small for consideration, engineers say. They further minimized height by employing a stacked arm rotary actuator and designing a special asymmetric load/unload ramp. The load/unload ramp serves as a place to park the drive's read/write head when it's not in use--a feature that's important for devices, such as cameras, which are subjected to higher
Palm top computers
impact loads than laptops. By designing an asymmetric ramp, rather than a conventional symmetric one, they saved 200 microns in height.
The engineering team also cut height by reducing the thickness of the disk, circuit board, and base plate as much as possible. They even cut a hole in the base plate to make room for the drive's actuator yoke, thus saving an additional half a millimeter, they say.
The small size and weight of the unit yielded some unexpected advantages, engineers say. Prime among those is a fast "spin-up" time. Unlike conventional desktop hard drives, which take several seconds to reach full speed, the lightweight, low-inertia Microdrive spins up in less than half a second. That capability would serve digital cameras well because the disk would be ready by the time the camera's auto-focus feature finished working, engineers say.
The Microdrive's most profound impact, however, will reportedly be in applications, such as palm top computers, requiring more than 100 Mbytes of storage. There, its small size and high density gives it an edge on lower capacity flash memories.
For mechanical engineers, the Microdrive may provide the ultimate lesson in integration. "Everything that exists in a laptop hard drive is contained in this little unit," Albrecht says. "There's a lot of miniaturization here."
Additional details...Contact Mike Ross, IBM, 650 Harry Rd., San Jose, CA 95120; Tel: (408) 927-1283; Fax: (408) 927-3011; E-mail: [email protected].
Tom Albrecht has been a research staff member and manager at IBM's Almaden Research Center in San Jose, CA since 1989. He spearheaded much of the mechanical design of the Microdrive, including the effort to squeeze its spindle motor and bearings into a 5 mm depth. During his years at IBM, Albrecht's work has centered primarily on data storage technologies for magnetic tape and disk drives. In tape technology, he contributed to a novel track-following servo technology, known as "timing-based servo," which enables higher track density and better reliability. Throughout his career, he has co-authored 32 publications and has been issued 32 patents. Albrecht's background includes a B.A. in physics from Carleton College in 1985 and a Ph.D. in applied physics from Stanford University in 1989. On the Microdrive project, Albrecht worked with fellow IBM engineers Mistushiko Aoyagi, David Albrecht, Kenji Kuroki, and Tim Reiley.