I seriously doubt that Stephen Parks, the president of Sipex Corp., is even 40 years old. He might be, though he hardly looks it, and I've been tracking him practically from his graduation from Lehigh University. I knew him at Cherry Semiconductor, where he helped build a PC power-management business, even did some freelance work for him when he was marketing bipolar foundry services for AT&T Microelectronics in Reading, Pa.
His presidency adds an obviously youthful image to a company that has found new strength as a supplier of bipolar and BiCMOS ICs. Sipex's most visible products these days are backlight drivers for LCDs - for example, the Motorola StarTac cell phone uses Sipex drivers (effectively charge pumps) for the electroluminescent (EL) backlighting. Such handheld portables are a key market focus for the company, Parks told me on a recent visit; it is focusing on making its products increasingly useful to manufacturers of cell phones and PDAs. For instance, one of the interesting products just introduced is a combination EL and piezoelectric driver for cell phone applications. The device will help reduce phone size by replacing ringers with piezo transducers that will output a variety of amplitudes and frequencies.
The company's product line includes power conversion and interface products, such as low-power (3.3 V) RS-232 and -485 drivers, increasingly used for digital camera connections as well as data cables. Newer parts of the power product line include precision LDOs and buck converters; older parts include charge pumps and microprocessor supervisors. Overall, the company generated 70 new products in 2000, Parks told me.
A fiber-optics initiative within the company will attempt to build photodiode detectors and gain blocks (transimpedance amplifiers) in CMOS. CMOS will support 2.5-Gbit/second transceivers, Parks believes, though his own bipolar experience suggests there are sizing and scaling problems to be solved.
Sipex currently has about 450 employees, though three-quarters of them are tied to manufacturing. "Analog designers are the absolute cream of the company," said Parks. "Finding and attracting them is very difficult." Perhaps he will have an easier time when engineers get word of his company's success.
Sipex (in Billerica) is one of the Massachusetts high-tech companies - a group I've dubbed "the 128 Circle" (after the highway that circles greater Boston) - that have sought to recast themselves after the military's commercial off-the-shelf (COTS) program pretty much took down the market for ruggedized data-acquisition modules. Until 1986, Sipex was called Hybrid Systems Corp. Ninety-five percent of its sales were to the Pentagon and all its products were hybrid circuits, said Jim Donegan, Sipex's Chairman and CEO in a phone interview. Hybrid circuits were bare-chip multichip-module circuits on ceramic substrates with thick-film interconnects, packaged in hermetically-sealed gold tubs. The users were military customers for whom performance, small size and ruggedness were more significant than cost.
In the mid 1980s the privately held, $20-million Hybrid Systems Corp. made a transition into what Donegan called "monolithics." Because of IC integration, it was only a matter of time before every hybrid circuit became a monolithic device. "It was easier to move into ICs," the chairman recalled. He pointed to a popular Analog Devices part number, the much-duplicated AD574 12-bit A/D converter. "You could live in the hybrid arena for three or four years - until it became a monolithic," Donegan said. It was "the big guys" - IBM and AT&T - that remained focused on multichip modules, and competing with them in commercial markets would be increasingly difficult, he said.
In 1986, Hybrid Systems merged with a Harris spinout, Data Linear, and changed its name to Sipex. Two years later it acquired a BiCMOS supplier, Barvon, and inherited a position in the market for 1488/1489 RS-232 line drivers/receivers - still a significant source of Sipex revenue, Donegan said. Sipex's business is now 95 percent in commercial markets, with a third of that in communications, another third in data communications, and the rest in consumer devices, Donegan said. There is a residue of military customers, he added, and the company has proposed more than one "last-time buy" in hopes of squeezing off this business without jeopardizing government programs that have become entrenched around particular part numbers.
Sipex's ICs for commercial markets are bipolar and BiCMOS, with various geometries - some with dielectric isolation and high-voltage protection (up to 100 V). For precision data converters, it has sputtering and laser trimming capability for silicon chromium resistors. Though it has its own bipolar and BiCMOS fabrication facilities capable of generating 20,000 wafers a month, it will utilize foundries like UMC - especially for CMOS. "The current management is much more savvy to ICs," said a happy Steve Parks, recalling his company's past.
Soul of different machine
Tracking the 128 Circle companies, their successes and failures, the personalities that drove them, would probably make a book or article series as interesting as the family lineage of Silicon Valley. At one time the region was dominated by the minicomputer business - the late Digital Equipment Corp. (DEC) and its competitors and spinoffs like Data General, Prime Computer and Apollo - and their success made up what Michael Dukakis, then the governor and later a presidential candidate, called "the Massachusetts miracle."
I remember a time before DEC stumbled, before COTS, before deep submicron whatchamacallit (before portable computers and e-mail tied up your evenings on the road) when it was as important for an electronics trade magazine to keep a correspondent on the 128 Circle as it was to keep one in Silicon Valley. An analog reporter, 128 was a stomping ground for me. I spent days at a time visiting data converter makers - Datel, Micro Networks, Analog Devices, Data Translation, Hybrid Systems and Analog Devices - even the venerable Teledyne Philbrick. (Vicor, a dc-dc power supply manufacturer, and Unitrode in Merrimack, N.H., were in the same orbit.) Evenings I'd sit in the bar at the Newton Marriott, flirt with women who spoke of "Mayn" or "Nu Hamshah" and pretend I was a Celtics fan. Hell, when Larry Bird was their center, you didn't have to pretend: The excitement of the crowd was infectious.
Engineers for the Massachusetts Focus Groups on Planet Analog (the selection process was an extremely tight sieve) were drawn from the 128 Circle. Like the engineer from Bose, some were working on newer applications for analog and mixed-signal technology. Others, like those from Datel and Charles Stark Draper Laboratories, were tweaking some of the older control-oriented applications. You could easily picture these engineers with needle-point pencils in their shirt pocket protectors and HP calculators strapped to their belts. I won't repeat what they said about Planet Analog here, except to say that they were a predictably crusty group.
What we knew as the analog and mixed-signal business seemed at one time to live in the orbit of the science-oriented minicomputer companies. Apart from the hybrid circuits produced for military customers, the analog signal-conditioning circuits, data converters and integrated data-acquisition systems were all board-level products, sometimes potted in epoxy, for industrial control manufacturers that would use a DEC minicomputer as a host. This was what you'd call now "a small but significant business." The entire 1980 data-converter market (including chips, boards and modules), according to a Frost & Sullivan report I used, was little more than $240 million.
At one time, Analog Devices (Norwood) and Tucson, Ariz.-based Burr-Brown appeared neck-and-neck in this business. Both companies perceived the utility of single-chip integration, though Analog Devices cultivated partnerships that would lead to competency in CMOS. The big difference between the two, however, was that Analog Devices was quicker to see and respond to commercial market potentials; it conscientiously worked down the percentage of military business it did. (Analog Devices is currently on its way to $2 billion in annual revenues.) Burr-Brown held onto military business a little longer than perhaps it should have, responding to commercial opportunities by setting up little companies, ultimately poorly supported, but refusing to do anything that would take away from what it perceived to be its core business. This was not a good place to be when the Department of Defense starting buying commercial parts under COTS. Burr-Brown president Syrus Madavi is credited with steering the company into more commercial activities, but even the early forays into xDSL were done at the behest of high-end companies like Pairgain. Burr-Brown's annual revenues were in the range of $300 million to 400 million when it was taken over by Texas Instruments
Still, even as a takeover target, Burr-Brown fetched a pretty penny (a stock value of over $6 billion). Not everyone in the business of designing and packaging data-acquisition systems - competitors on the 128 Circle - made out as well. Datel, acquired by General Electric and spit out again, retained its focus on specialized boards and modules. It is privately held, and doing about $100 million a year. Its biggest source of revenue is dc-dc converters - low noise, fast response, basically unkillable devices - for telecom and network system builders.
Data Translation (Marlboro) attempted for a time to parlay its data-acquisition expertise into graphics and imaging boards. It's back in the data-acquisition business, but much of its revenue comes from contract manufacturing. It is public and profitable, but doing about $16 million a year. Micro Networks Corp. (Worcester) has expertise in delay lines and SAW filters as well as data converters to peddle, but similarly makes money from contract manufacturing. Once owned by Unitrode (Merrimack, N.H.), Micro Networks received several infusions of venture money in the last few years, but, because the company is privately held, we don't see the results.
Analogic Corp. (Peabody), made famous by the colorful conservatism of its president and technology guru, Bernie Gordon, took its data-acquisition business deeper into high-end systems for medical and telecommunications equipment. The parent of Sky Computers, Analogic supports computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound as well as high-speed data acquisition and digital-imaging boards. The company made headlines a couple of years ago for applying tomography to airport bomb sniffing. Analogic is highly profitable, but its annual revenues are about $300 million.
Sipex is now doing about $32 million a quarter. The company closed fiscal 1999 at about $90 million, but Parks can see a $200 million year with current growth rates. Sipex's mid-1996 initial stock offering went out at less than $10 a share, but its stock is now up around $37; it's been as high as $49.50. Steve Parks can take that to the bank, I'm sure.
I'm bemused, a little jealous, I confess, when young guys like Steve whiz by me on their way to the corporate boardroom. I made press agents laugh recounting the number of now-famous CEOs I swaddled and bounced on my knee. I remember Joe Costello, the charismatic onetime president of Cadence Design Systems, as an almost timid voice on the telephone doing the marketing legwork for a speech synthesis company in Berkeley, Calif. The very next day, he was marketing manager of Jim Solomon's startup, SDA Systems, and president of Cadence the day after that. By the end of the week his picture was in Business Week, and the following Monday he was courted by Hollywood moguls. Everything, it seems to me, happens that fast. Or am I just getting old?