One inch no cinch for IBM storage gurus
When two IBM designers started tinkering around in a motor-research project five years ago, they weren't planning to create a new form factor around 1-inch disks. After all, the industry hadn't embraced a smaller disk-drive format in more than a decade.
But by last month several digital-camera makers had lined up with vendors of MP3 devices and handheld computers to endorse the fruits of that research program: a tiny 1-inch disk drive made by IBM's Storage System Division.
The large show of support for a single-source product surprises its developers at IBM's Almaden research lab almost as much as engineers who marvel at a 340-Mbyte disk drive that's about the size of a matchbook.
The two researchers were simply looking for a way to read increasingly smaller tracks on a disk, using tiny motors to finely position a drive's heads.
"This started as a funded research project to look at micromechanics since, with smaller bit sizes and finer track widths, sensitivity to movement and vibration is more critical," said Timothy Reiley, research staff member in IBM's Research Division. "Our view is that MEMS [microelectromechanical-system] devices will be a key part of a disk drive in the future, so we were looking at micromotors on the head for finer positioning. As we went along, we found it was more suited to a miniaturized drive than to use micromotors in larger drives."
The result is a 1.6 x 1.4 drive so small that marketers took the unusual tack of calling it the microdrive, using only small letters, with no attention-grabbing capital letters. Once the program's goal shifted to creating a drive tiny enough to fit into cameras and other small products, the challenges were daunting. It was tough to develop motors and head assemblies small enough to fit in a drive standing only 0.19 inches tall, and getting all the necessary ICs into that little package confounded even those who drew sketches.
"One of the breakthroughs came when we could see the electronics people were going to succeed in fitting all the chips on that little card," said Thomas Albrecht, manager of mechanical technologies for advanced files. "Our original prototype only allowed components on one side, and a clever engineer came up with a way to use direct chip attach and get components on both sides. Before this, our drawings always showed components on one side and the leftovers somehow floating in air."
While electronics played an important role in the development, the mechanical aspects of the drive were also crucial. After all, disk drives are electromechanical devices with platters that must spin reliably at precise speeds for thousands of hours with a minimum of vibration.
"Bearings were important in defining the geometries of this thing, since bearings have not evolved the way motors have," Reiley said. "Ball bearings probably drove more of the mechanical design than anything else."
The drive's compact form factor made it tough to develop prototypes. But with the proper motivation, the engineers found they could work in that small space.
"The night before one of the top execs was coming for his semiannual visit to Almaden, Tom and I were up gluing things together. A lot of microscopic wiring got done overnight," Reiley said. "Getting it running for him was fortuitous."
Overcoming these technology challenges was critical, but a number of marketing issues were equally vexing. Many vendors had tried to create a smaller form factor, but the 2.5-inch drive was the last successful move to smaller disk drives. The 1.8-inch market saw the demise of several startups, and Hewlett-Packard's attempt to make a 1.3-inch drive, the Kittyhawk, nose-dived.
"The Kittyhawk was a spectacular failure," Albrecht said. "HP left the disk-drive market completely after it failed. The fact that it went down in flames made everyone at IBM cautious.
"There wasn't a lot of overt opposition to our project, but there was much reluctance for any help for it," he said. "One reason is that the 2.5-inch drives were very successful and to divert research to this new thing was distracting."
To get that support, negating the Kittyhawk's ominous presence was critical. Any ploy was brought in, even a simplistic evolutionary chart implying that the Kittyhawk had flopped because it didn't follow the straight size reduction line from 14 down to 2.5 inches. They also used honest technical and market analysis.
"Kittyhawk preceded the rapid advance in a real density over the past few years, and the active market for it back then was 100 percent hype and 0 percent market. That was the time of the [Apple] Newton," Albrecht said. "Now you've got digital cameras, and handheld systems are far more popular. One clever thing we did that HP didn't do is to pick a form factor compatible with flash memory."
Working with the standards body that was creating a new version of the widely used CompactFlash format was another challenge. The drive mechanics wouldn't fit in the low-profile, first-generation specification, so IBM and others who wanted to put modems and other products in the form factor were pushing for a larger size. While IBM had some partners, formidable forces on the other side side were pushing specifications that didn't meet the needs of the microdrive, which required a taller package.
"The effort to standardize the form factor was an exciting process. It was not an automatic success," Albrecht said. "Microsoft had worked with a team of companies to develop what they needed, growing the form factor in two of the three dimensions. They thought they had the votes to pass their proposal, so they were in no mood to talk to us about getting a shape that worked for a disk drive.
"We really had to work to show people our approach was smarter," Albrecht recalled, "and it was very necessary to have a secret ballot, which is rare in these committees, because it's hard for companies to publicly oppose Microsoft."
Not holding a grudge, Microsoft was among those who endorsed the drive in a recent statement announcing the microdrive's shipment. Others were camera makers Eastman Kodak, Minolta and Nikon, along with Compaq, Casio and an MP3 specialist, Diamond MultiMedia's RioPort Division.
That bevy of endorsements is a far cry from 1996, when Albrecht and Reiley turned to marketing and took the first prototype to Comdex. They expected camera people, who had to deal with small capacities of expensive flash, to get excited over the 170 to 340 Mbytes that the tiny drive provided. That was a mistake.
"The most-common reaction seemed to be that the camera people had worked hard to build an architecture based on very little storage space," Albrecht said. "After they had struggled to work with 4 Mbytes, they were not necessarily happy to get 340. Many were not even prepared to even discuss it."
While there was little initial interest from those camera makers, a research partner at IBM's Fujisawa, Japan, facility got very interested. An amateur photographer, he became a missionary for the drives, promoting it to camera makers and others in Japan who shared the country's zeal for gadgetry. Much of the development was done in Fujisawa.
"I traveled very extensively to Japan for three years," Albrecht said. "For a while either Tom or I was going every other month. E-mail was also very convenient; it bridges the time gap. You finish up your work, send several e-mails and with any luck they're all answered by the morning."
He added that there are definite cultural differences between the Japanese and American engineers. "In Japan, e-mail is used for less; it's something that's not natural to them. It's more common for them to meet face to face. They didn't even have voicemail until last year."
Now that the drive is starting to ship, the researchers are glad they talked their marketing counterparts into another deviation from IBM's usual development process. Drivemakers typically don't formally unveil a drive until it's ready to ship. But IBM went public last year, one of the longest prereleases in disk-drive history.
"For IBM Disk Drives, this was an uncharacteristically early announcement. We normally would never announce a product a year in advance of shipments," Albrecht said. "Everyone agreed that it was necessary. People needed to design Type II slots, and there were also questions whether we were serious about this."