This drive's got ZIP

In the mutable, often over-hyped world that is the computer industry, claims of technological "paradigm shifts" occur about as often as Elvis sightings. Yet, when Kim Edwards, CEO of Iomega Corp. (Roy, UT) declared that his company's new Zip Drive had smashed paradigms in the removable-storage business, he wasn't just boasting.

Unveiled at Comdex in November 1994--a scant seven months after its conception--the drive became an instant media and consumer darling. Like other devices for storing and transporting files, it will make engineers and others more mobile. They'll be able to transfer CAD files between different computers at work or at home quickly and cheaply. They could even run real-time video off the Zip Drive for virtual reality applications.

Iomega's stock soared from $2 to $30 a share with the introduction of the drive. "I've been in the storage business for 20 years, and it's probably the most exciting product I've seen," says Stan Corker, director of removable-storage research at International Data Corp. (Framingham, MA).

He estimates that .75 to 1 million drives will reach consumers this year, with 3 to 4 million shipping in 1996. That compares to the Bernoulli Drive, the company's core product, which has sold 100,000 n15% a year since the early 1980's.

Among the engineering breakthroughs that will drive the Zip Drive's success:

25-to-100 MB storage capacity of the cartridges.

Beyond those achievements, the development of the Zip Drive represented a breakthrough in speed of design. Iomega shipped the product 11 months after its conception--blindingly faster than the two- to three-year cycle normal for similar products.

Deceiving appearance. Studying the outside of a Zip Drive yields little of the significance of the product. And that's its secret. Encased in a comfortably curved, medium-blue, injection-molded polycarbonate case, Zip's innovative innards go about their business with disarming familiarity. A slot on the front accepts removable cartridges that appear to be nothing more than 1.44-MB floppy disks on steroids. And little rubber feet, ingeniously placed, allow the drive to lie flat or stand on its side to conserve space.

So what sounds do smashed paradigms make? One is the noise pleasantly surprised consumers emit when they find out that Zip Drives cost less than $200, the floppy-like cartridges hold 25 to 100 MB and cost $15, and performance is, well, zippy. Compatible with DOS, Windows, Mac OS, and OS/2, the drive boasts performance similar to low-end hard disks.

Just as important is the sound of Iomega engineers revolutionizing the company's design process. Described by Edwards as "ordered chaos," the Zip Drive's entire incubation took less than a year. George Krieger, Iomega's senior director of new product development, kept the tight, Skunk-Works-like team committed to the brutal deadlines and blocked outside distractions. "In the past, we had a thousand ways to say we were past schedule," he says. "There could be no excuses this time."

Vitamin C. Edwards joined Iomega as CEO in January of 1994, and he immediately set out to steer the company away from niche products and into the high-volume consumer market. Rather than recommend a specific technology--something better left to Krieger's R&D staff--he put together a document titled "Vitamin C" that described the general type of product needed: something less than $200, between 60 and 100 MB in capacity, and delivered ASAP. "The "C" really stood for Cost and Consumer," he says, "but it also represented something the company needed: a shot of vitamin C."

Finding employees in need of a boost wasn't hard. Some had spent years working on three previous 3.5-inch disk projects, all of which had been canceled. "But this one was different from the beginning," says mechanical engineer John Briggs, "it was a 'supported-by-God' project; the CEO was behind it, and we could be sure of getting what we needed."

Before engineering could begin, though, Edwards needed to see if his gut feelings, spelled out in the Vitamin C document, were correct. Marketing pulled in an outside research firm and engaged thousands of end users of varied experience in focus-group discussions. The verdict was unanimous: If Iomega would build a 100 MB drive for $200, they would come.

Focus group members also said and did a lot of things that surprised knowledgeable computer users. "Many people said they were scared to death that when they inserted a floppy disk into the computer, they'd never see it again," says Edwards. So the new drive would need a window to let users see the disk at all times. And men and women inserted disks differently; many women cocked their hand to avoid bending or breaking long fingernails. The new drive would need a relief notch to accommodate everybody.

From CAD to Comdex. With the broad specifications defined, engineering dove into the project with a sense of urgency. "We all felt this would make us or break us," says Michael Lyon, a staff engineer of 14 years. "It was our last chance to make a big splash."

Within the first two days, R&D had finished cost estimates for every component in the drive and proposed methods for achieving them. In what Krieger calls the "fastest CAD-selection process in history," engineers chose SDRC's (Milford, OH) IDEAS software package in less than two hours. "I called SDRC and told them we needed training and software in the next two days, and we needed to be putting out prototypes in two weeks," says mechanical engineer Carl Nicklos. Fourteen days later, he held the first stereolithography (STL) model.

Over the course of the project, engineers generated dozens of STLs and spent tens of thousands of dollars on them. David Jones is convinced it was worth every cent. "We found all kinds of problems, and we found them early," he says. As the development pace quickened, the task of expediting prototypes fell to Michael Lyon. He claims that the rapid access to prototypes was a key aspect to the project's success: "Engineers could kick out a file during the day and return the next morning to find an STL on their desk."

Price was proving to be as big a challenge as time. Having built Bernoulli drives for 15 years, engineers knew they could achieve the required quality and performance. But could they do it for $200?

Krieger decided a good way to learn would be to study existing consumer products. So he and members of the staff stopped by a local electronics store and bought dozens of CD players, tape players, and tape rewinders to see what made them durable but cheap. Tearing them apart in late evening sessions proved fun. "Probably more important design decisions were done at 8 pm than at any other time," says Krieger. And some general do's and don'ts emerged from the experience. Do use a flip-top lid like a CD player to cut cost; don't use tons of tiny screws to hold it together.

Red Flag raised. In June of 1994, the company committed to displaying a booth of fully functional prototypes at the Comdex trade show the following November. The pace of work, already extreme, quickened. Friendly competitions arose between engineers trying to see who could work the most, with some threatening to stay overnight rather than be beat. Then, in mid-July, marketing raised a red flag.

Having become big believers in market research, the company had run a batch of studies to gauge reception to early prototypes. The response was overwhelmingly negative. "Teachers told us kids would rip the top off in half a day," explains Edwards, "business people complained of contamination." The top-load idea would have to go.

With four months until Comdex, an outside industrial design firm penned a handsome front-load concept, and engineers began anew. The design, however, required several inherently more expensive features, such as a solenoid-driven load/unload mechanism. The clock was ticking, the costs were rising, and the writing was appearing on the wall--literally.

As the manager charged with keeping up spirits, Krieger took out a magic marker and vented his feelings in a note on a hallway wall. Soon, reams of graffiti appeared, everything from quotes to frustrations, encouragements to jokes. "The facilities department wasn't too happy with me," he says, but his overtaxed designers appreciated the outlet.

Firmware engineers, whose task normally falls after the design of the mechanism, suffered especially. They were committed to entering production in early 1995, a date that could only be met by leveraging existing products. But their design had to be much lower cost, a requirement that demanded a completely new firmware design and too much time. So they did both.

"We decided the only way to do this project was to have an architecture we could get to market quickly; that means existing code," says Clark Bruderer, a firmware engineer. "We would then follow on with a cost-reduced version."

Comdex debut. In November, twenty precious, hand-built prototypes graced Iomega's booth at Comdex and stunned showgoers. Their pleasing blue exteriors consisted of silicone-molded plastic parts, almost indistinguishable from production items. Engineers had come in at 2 am the night before to ensure the drives would be there. But the plan had worked. "This was the first product we ever hit our marketing windows with," says Lyon.

The months following Comdex were filled with the practical issues of translating the prototype design to mass production, and thoroughly testing the drive for reliability and performance. Looking back, engineers surprised themselves with their own success. "I think the development process was unique anywhere, not just here," says Nicklos. "There was no fighting for ownership of parts; everybody was responsible for getting the entire project done."

"It's not often that engineers get to participate in a project that changes the complexion of his company, much less have implications for altering an industry," says Krieger. "This time, things came together and magic happened."

The engineering behind the drive

Like a good marriage, the Zip Drive contains a little old, lots new, a few things borrowed, and a bit of blue. Faced with extreme cost and time pressures, engineers pulled ideas for the mechanism and electronics from wherever they found them, including several previously canceled projects.

Both the head-to-disk interface (HDI) and the disk media, for example, trace their origin to the Aspen, a four-disk, 400 MB drive that was never produced. "We used components off the shelf with a few changes and put them together to make an HDI nobody has seen before," says Clark Bruderer, firmware development manager. It features heads similar to those of a hard disk, measuring 10 microns wide by 1 micron long. But whereas Aspen used rotary actuators similar to a hard disk, Zip employs a patented linear actuator that rides on two sapphire bushings over a 1- mm diameter stainless steel rod.

This design eliminates the need for tiny ball-bearing slides, which would cost $7 to $10 each. It also provides the assembly with an extraordinarily high resonance frequency--well beyond the bandwidth of the servo head-positioning system.

A high-precision servo motor spins the disk at 2,941 rpm and achieves a radial runout of less than 10 microns. It couples to the disk via a magnetized hub instead of the less precise pin-drive mechanism found on regular floppies.

Plethora of plastic. To keep cost down, engineers used as much plastic as possible. "The trick was to take all these plastic parts--which have tolerances orders of magnitude higher than needed--and have them be unimportant to the drive," says Briggs. For example, the cartridge case is merely a protective shell and has nothing to do with precisely locating the disk during use.

Some seemingly trivial details play a large part in the design. For example, the flex cable that carries information to and from the head exerts a light but measurable force on the actuator. To limit this effect, engineers specified an unusual design made from 0.5-mil Mylar(TM) instead of 2-mil.

And Briggs spent weeks tracking down several frustrating electrostatic discharge problems. "Just walking up and touching the cartridge would blow out the heads," he explains. To prevent this, the case now includes an extra wall of insulating plastic along the front edge.

In another instance, a spark anywhere in the room would cause errors with the drive. The solution: adding a small ground screw.

Cost drove firmware engineers to ultimately use an 8-bit processor instead of the more usual 16-bit device. "This was a risky, because we hadn't done any development with it recently," says Electrical Engineer Rich Penman. Further increasing the risk was the decision to use the four-zone recording process found in hard drives, even though the 8-bit controller wasn't designed for any zones at all.

Frugality extends to the solenoid-triggered eject mechanism, as well. Engineers originally wanted two solenoids, one to eject the disk and a second one to trigger a switch that registers the position of the moveable platform. But solenoids cost a buck each, and one would have to do.

Engineers solved this problem by making a single solenoid perform double duty. A small amount of current moves a plastic pick, triggering the switch. A large current moves both the first pick and a second one that ejects the disk.

The electronics and firmware also benefited from the increased volumes expected of the Zip Drive. Systems that cover several circuit boards on the Bernoulli drive were reduced to just a few chips on a partial board. Says Penman, "With high volumes, you can take advantage of further engineering development that simply cost too much to justify before."

Vendor helps mold a breakthrough design

No material plays a larger part in the Zip Drive than plastic. And no vendor played a larger part in making those plastic parts than Complex Tooling and Molding (Boulder, CO). A close partner of Iomega's for more than a decade, Complex Tooling became an integral part of the design/production process.

The company produces more than a dozen parts for the Zip Drive, including the base, cover, front panel, rear panel, platform, guide rails, load ramp, retainer plate, feet, window, cartridge case, and disk jewel cases. Most exterior components are formed from a polycarbonate ABS blend. The internal components consist of glass-filled polycarbonate.

Engineers from Iomega approached the company early in the design process, months before the usual first contact is made. "We got a call from them in July (1994) when the project was still at the concept stage," recalls David Thomsen, Complex Tooling's project manager.

With the design so immature--and changing complexion faster than a chameleon--Thomsen relied on CAD data translated to a neutral IGES file from Iomega's SDRC system and transmitted to Colorado via modem. "It was completely paperless," he says. "On some parts we still don't have prints yet."

The reliance on CAD data instead of prints saved Complex Tooling 25 to 40% in time, Thomsen estimates. "We just told them we needed everything within five thousandths," says Carl Nicklos, a mechanical engineer at Iomega. "If parts came in outside that spec, we'd live with it, or work with them to get it changed." Such a process resulted in parts that adhered more closely to the original design with less room for manually introduced errors--at least in theory.

One of the biggest headaches manufacturing engineers faced was the appearance of "false geometry." Unbeknownst to them, the IGES translator sometimes produced subtle errors. Bosses, for example, differed from Iomega's data by tenths or hundredths of an inch. The problem gave Thomsen fits until a third-party IGES translator eliminated the issue.

Unlike many customers, Iomega accepted or even solicited Complex Tooling's design suggestions. An early platform design, for instance, involved geometry that was not possible to make via injection molding. "That resulted in the platform being broken into two pieces, a platform and a load ramp," says Thomsen.

To account for the higher volumes expected of the Zip Drive, Complex Tooling & Molding added a new EDM machine, a new CNC machine center, and seven new machines at the molding center in Loveland, CO. Additionally, the company has built and begun operating stack tooling for the first time.

Stack tooling differs from conventional tooling in that three plates are used instead of two. Two sets of parts are formed on top of each other. And since the area of the mold cavities is not increased, twice as many parts can be made from the same capacity press.

Thomsen indicates that the Zip Drive project has been quite a challenge. Constant tooling changes can result in less than optimal gates and features for molding. But as in most businesses, the customer is (usually) right. Iomega's Nicklos agrees. "They really worked with us. I spent several evenings at 2 a.m. shooting parts at their facility."