Downturn Dilemma: Machine Shops May Hurt Upturn

Alas, Electronic News (the print edition): we hardly knew ye!Machine shops may prove an Achilles heel during the next upturn in terms of capacity, as there may not be enough of them left to crank out parts for semi manufacturing tools.

The industry may be in for a repeat or worse of what happened in the early 1990s, when a number of machine shops closed during tough economic times at the beginning of the decade. Enough capacity was lost that it caused problems as equipment suppliers tried to ramp up tool production.

Given today’s growing reliance on outsourcing, particularly for parts, coupled with the fact that many machine shops around Silicon Valley and throughout the country have closed, it could be a problem again.

“A lot of shops have gone under. … It might happen again,” acknowledged Paul Givens, president of Givmar Precision Machining, recalling the problem in the early 1990s. Givmar has done work for the likes of Novellus, KLA-Tencor and Teradyne, among others.

Because it has a broad customer base that’s not dependent on the chip industry, Givmar is weathering the tough times. But many shops that concentrated on semiconductor-related work have closed their doors, Givens said. “I don’t know what’s going to happen in the next upturn,” he added. In addition to the lost capacity, qualified machinists have left Silicon Valley, looking for jobs elsewhere. To make matters worse, the National Tool and Machining Association (NTMA) recently had to close its training facility in San Francisco.

But the problem is not just a localized phenomenon. Some 500 machine shops have closed in the past year across the United States, according to Matthew Coffey, president of the NTMA. “We’ve lost companies that I thought were very solid companies,” Coffey said.

Compounding the potential ramp problem, some surviving machine shops that have served high-tech equipment suppliers for decades are now turning their backs on the industry, looking instead to aerospace, medical and military business. Having been caught off guard by the extent and depth of the current semiconductor industry downturn – after adding capacity to meet the booming demand in 2000 – the machine shop industry is feeling forsaken as the industry turns to offshore manufacturing, namely in China.

“Even in a ramp, a lot of these suppliers aren’t going to go back and embrace the electronics industry,” noted Richard Wills, owner of D&H Manufacturing Co., a Fremont, Calif., shop that has been around for 45 years. In the past D&H has done work for many of the large OEMs in Silicon Valley, among them Applied Materials, Lam and Novellus. These companies have tried to drive prices so low during the downturn that shops like D&H have turned down their business because it wasn’t worth it, Wills said.

Chinese flagBut these other industries can’t supply the volume of business that companies such as Applied Materials Inc. brought to a business like D&H. “We’ve known they have gone to China. They’ve told us. They’ve found some large [Chinese] suppliers,” Wills said.
Not Everyone Enamored with China

For the machine shop industry, and by association American manufacturing in general, it is China that is the underlying issue in its economic problems. China, which by and large has been embraced with open arms by the semiconductor industry, is drawing jobs and business away under circumstances with which domestic machine shops can’t hope to compete, the industry says.

China has only served to add insult to economic injury, as far as the machine shops are concerned. “Will there be an upturn or will it all be offshore when it turns so it will be irrelevant to domestic manufacturers? ? No one is thinking about that,” said the NTMA’s Coffey.

Both Coffey and Wills pointed out that today much of the machine work is being done on option, being auctioned off via the Internet, and much of it is going to Chinese companies. “The Chinese are always going to bid below us,” Coffey said. “Economic recovery by itself is not going to heal this problem. There is a major systemic shift that is destroying the supply chain infrastructure nationwide.”

Despite the political wrangling over Chinese trade that has taken place in Washington, D.C., over the past several years prior to China’s entry into the World Trade Organization (WTO)—wrangling that has heavily involved the semiconductor industries lobbyists—there are still high tariffs and duties on goods imported into China, according to Coffey and Wills.

“The playing field is not level at all,” Wills noted. In addition to the tariff and duty imbalance, the Chinese and other Asian governments provide a lot of economic support, and environmental regulations are exceptionally lax compared to those in the United States, particularly California, he said.

“I believe the politicians don’t understand that we are losing manufacturing at a tremendous pace. Will we be able to remain a manufacturing power? I doubt it. ? If they are going to use the term, ‘free trade,’ it has to be equal, both ways,” Wills said.

But the NTMA is not taking the issue lying down; it is trying to form a political action committee to bring Washington’s attention to its plight.

Editor’s Note: As explained at length elsewhere on this site, this is a news story written by me that originally appeared on the now-defunct Electronic News’ website, which is long gone. It’s former sister pub Electronic Design News (EDN) currently holds the copyright to all Electronic News copy (to the best of my knowledge). You can still see a copy of this story at EDN.

Outsourcing Has Its Dark Side

Nanometrics Opts for Vertical Integration Instead

Alas, Electronic News (the print edition): we hardly knew ye!

The metrology company is bucking the trend and has become more vertically integrated to guarantee its ability to ramp whenever the industry recovers. A key part of its strategy is being able to deliver its metrology tools quickly and at the same time being able to directly ensure quality.

When the next upturn does come, having its own in-house machine shop will be a competitive advantage for Nanometrics over its larger competitors, according to its CEO John Heaton.

John Heaton, Nanometrics CEO“My only advantage as a small company is to do something quickly, better, faster, cheaper,” Heaton explained. In the past, Nanometrics would have to compete during an upturn for time at local machine shops that also served much larger companies, such as KLA-Tencor Inc., a competitor, and Applied Materials, a customer.

Needless to say, the larger companies often took priority. Furthermore, there was such a rush during 2000 among the entire supply chain that the company had quality-control problems with its suppliers. Thus Nanometrics has brought its own machine shop, anodizing shop and related parts of the business in-house.

But there is more to the strategy than guaranteeing quality and the ability to ramp production. Many of the machine shops used by process technology OEMs were hit hard by the two-year economic downturn. A number of them have closed their doors, which could dramatically impact the ability to ramp for the companies that have relied on them in the past, Heaton said.

“No one is talking about that right now. We believe we will have a competitive advantage,” he added.

Nanometrics’ size and its business model lend themselves to vertical integration. The company has been an early proponent of integrated metrology in an industry that is just beginning to seriously come around to the idea. The company has based its technology on integrated metrology modules, and then migrated that technology to its stand-alone platforms, rather than the other way around.

“We start with a clean sheet of paper and say how do we take our technology and redesign it?” said Peter Gise, Nanometrics’ senior marketing manager.

That has meant the key components of its systems are small, compact systems, which are not particularly challenging for an in-house machine shop to produce quickly while maintaining quality. And by bringing the manufacturing in-house, it keeps the cost of the finished integrated metrology modules down by eliminating a step in the supply chain, he said.

“We believe integrated metrology is the way to go,” Gise said. But this is not to suggest that the company has forsaken its off-line tool base. “We haven’t exactly bet the farm because we still have off-line tools,” he added. But the concept is gaining traction in both OEM and end-user customers. “In three to five years ? it’s really going to take hold,” he said.

With chip production lot sizes decreasing along with feature sizes and cycle times, process parameters are becoming increasingly tighter, and defects more critical. Using feed-forward and feedback of metrology data to control process excursions during production runs is becoming an increasingly popular idea among chipmakers, Gise said. As a result, Nanometrics has begun expanding beyond its traditional thin-film measurement to metrology applications for lithography, planarization, deposition, etch, and copper and low-k films.

This past week the company rolled out the Nano CLP-9010, a laser profiler that nondestructively monitors the metal loss between an isolated copper feature and the surrounding dielectric region. The module is designed to integrate within a metal chemical mechanical planarization tool. It is a new technology for Nanometrics, and an application driven by and large by its OEM customers, Heaton said. Interest in the new module was nearly instantaneous, he said, adding that the company started receiving calls from customers within hours of the news release.

Editor’s Note: As explained at length elsewhere on this site, this is a news story written by me that originally appeared on the now-defunct Electronic News’ website, which is long gone. It’s former sister pub Electronic Design News (EDN) currently holds the copyright to all Electronic News copy (to the best of my knowledge). You can still see a copy of this story at EDN.

Open Architecture vs. Open Standard

Editor’s Note: I included this story, along with the other two bearing this date, October 14 (Open Qualifying Technology; Can Chips Make You a Better Person?) not because I thought they were particularly brilliant journalism, but because they bore datelines from three different countries — one from entirely separate continent — all in the same issue. Pretty cool, no? I doubt I’ll ever win a Pulitzer, but at least I can lay claim to this.

ITC Debates One for All, All For One ATE

BALTIMORE — Not only has the dust not settled in the automated test industry, but the open architecture movement has stirred it up even more.

Alas, Electronic News (the print edition): we hardly knew ye!Last year at the International Test Conference (ITC) it was evident that change was blowing in the wind. This year it would seem that the ATE industry is changing in some ways almost as much as the devices it tests.

Last year, the question was what to standardize: the test socket, the tester architecture, the tester language, EDA software, and so forth. Design-for-test (DFT) and test costs, a perennial ITC topic, was on everyone’s lips.

This year, all of the large ATE companies now have, or are migrating to, a single platform, and several have injected their platforms into the open architecture fray. NP Test became the latest ATE vendor to do both. That fray grew in scope, of course, last summer when Advantest Corp. announced the creation of the Semiconductor Test Consortium (STC), which has vowed to take the open architecture idea one step further and create an open architecture standard for the entire industry, independent of any one vendor. Teradyne Inc. founder Alex d’Arbeloff made that topic the theme of his opening keynote address, thus setting a theme for this year’s ITC.

Just where all this open architecture dust will settle and which companies will be left standing remains to be seen. In any event, the ATE industry continues to evolve.

d’Arbeloff co-founded Teradyne in 1960 and is now involved in academia at MIT. During his keynote, he took to task the open architecture standard being pushed by Advantest, Intel Corp. and others involved with the STC. This business model will cut ATE company revenue, d’Arbeloff said, and will thereby slow R&D, while it will leave the headache of system integration up to chipmakers.

“Customers, while congratulating themselves on driving test equipment down from 2 [percent] to 1.5 percent of sales, will spend 10 or 20 times that in trying to make everything work. Worst of all, the pace of technology will slow down,” d’Arbeloff said. This will in turn commoditize ATE, making it resemble the PC industry with all of its related problems, he argued.

But by leveraging R&D of all the vendors through an open standard architecture on which various third-party, plug-and-play instrument modules can be utilized, advanced modules can be brought to market much quicker, countered Sergio Perez, VP of sales for Advantest’s U.S. subsidiary.

In any event, the STC is progressing rapidly. It announced several new supply chain vendors as members and plans to unveil its proposed architecture by the middle of next year. At the same time, STC ATE vendors are developing several tester modules designed for the architecture that are being developed concurrently.

As for the major ATE vendors, Perez said there has been dialogue between them and the STC. He suggested that smaller vendors must pave the way for participation in the consortium so that the large vendors feel comfortable with the model.

The big players still have reservations. “Tell us how to do it so you feel comfortable. Don’t tell us you don’t like it, tell us specifically what you don’t like about it,” Perez said.

But customer interest remains high, according to Advantest. “Customers are down this road further than I expected,” he said.

In the meantime, the ATE industry, both vendors and customers, continues to gravitate toward single-platform systems that have an open architecture in the sense that they are open to third-party instrument suppliers; even as the architecture itself is proprietary.

“A massive platform convergence is underway. The revolution is occurring, not across the industry, but across each ATE company,” d’Arbeloff said.

NP Test Inc., formerly Schlumberger Semiconductor Solutions, announced here that it would roll out such a platform during Q1.

At the same time, NP Test announced that it was migrating the capabilities of its high-speed digital tester, the ITS 9000ZX, to its EXA3000 platform. This gives NP Test a platform scaleable from 200MHZ to 3.2Gbits/sec. data rates, with a pin-count of up to 1,280 pins, explained Jean-Luc Pelissier, VP and general manager of NP Test.

“[Currently] some of the chipsets … are coming with fairly high-performance buses,” Pelissier said. Some of these high-end devices also require pin-count beyond the 1,024 pins commonly found on testers designed for this market segment, he added. By upgrading the EXA3000, this provides high-end performance at a relatively low price-point, he said.

But STC proponents question the proliferation of competing open architecture platforms. If a single vendor supplies scaleable testers for the multitude of devices produced by a large chip company, it puts the chipmaker at risk, Perez suggested. With a single open standard, that risk is moved, he said.

Editor’s Note: As explained at length elsewhere on this site, this is a news story written by me that originally appeared on the now-defunct Electronic News’ website, which is long gone. It’s former sister pub Electronic Design News (EDN) currently holds the copyright to all Electronic News copy (to the best of my knowledge). You can still see a copy of this story at EDN.

Qualifying Technology

Editor’s Note: I included this story, along with the other two bearing this date, October 14 (Open Architecture vs. Open Standard; Can Chips Make You a Better Person?) not because I thought they were particularly brilliant journalism, but because they bore datelines from three different countries — one from an entirely separate continent — all in the same issue. Pretty cool, no? I doubt I’ll ever win a Pulitzer, but at least I can lay claim to this.

French Component Supplier Relies on Technology to Set It Apart

MONTPELIER, France — Equipment component supplier Qualiflow SA is taking a decidedly different tack than its larger and longer established competitors.

Alas, Electronic News (the print edition): we hardly knew ye!Outsourcing is ruling business models, and component suppliers today are positioning themselves as critical subsystems suppliers, building complete modules for their OEM and chipmaker customers and taking advantage of prevailing trends. But Qualiflow, a smaller newcomer, is instead staking its claim based on its technology. It is depending on technological innovation to gain market share, while at the same time outsourcing manufacturing while keeping final assembly in house.

This is something the company couldn’t continue to do if it were to go into the subsystem module business, explained Chairman and CEO Claude Jacquemin. “We could make more profit, maybe, but it would also require more [capital] investment,” he said. “We feel we can bring value to our shareholders through advanced technology.”

The mass flow controller (MFC) market, for example, is a niche market that hasn’t changed much; the same big players that were there 10 years ago are there today, Jacquemin noted. The challenge is finding ways to improve a relatively simple idea and product with new technology.

Qualiflow had hockey-stick growth through 2001.Of course the pressure to innovate is doubled by today’s business climate. Qualiflow, spun out of ASM International in 1997, took off at a dead run. Between 1998 and 2001, its revenues grew from about $2.5 million to about $16.8 million (converted from euros to U.S. dollars), split between the optical fiber and semiconductor markets. Of course like everyone else, that dead run has hit a brick wall—the company estimates 2002 revenue to be between $5.8 million and $7.85 million.

But the company has maintained a strong cash position and R&D spending. At current industry spending levels, Qualiflow can maintain its investment levels and continue on for the next two years, according to its CEO. “We believe we will come out [of the downturn] very well positioned … we still have a lot of money in our bank account,” Jacquemin said.

In its brief history Qualiflow has looked for partners outside of Europe to supply advanced technology, notably in Korea and Japan. In Japan it has taken a 33 percent equity stake in liquid MFC supplier Lintec. Liquid mass flow control is a key technology for both next-generation IC and optical fiber markets, Jacquemin said.

There are still market niches such as fluid control where there is room for innovation and a market insertion point for Qualiflow, he said. MFC sensors is another area where Qualiflow sees an opportunity to grow market share.

“Everybody is bringing new technology. We have our solid-state sensor. Do we have the best solution? I don’t know. But the market is demanding … new ways of sensing,” Jacquemin said.

Qualiflow has four product lines: MFCs, valves, gas and vacuum cabinets, and vaporization systems. While it has 20 percent market share in gas cabinets—primarily from the optical fiber market and French telecomm provider Alcatel—Qualiflow has only begun to scratch the MFC market with a 2 percent share of the market.

The company thinks it can increase that share to about 8 percent within three to four years with its next-generation MFC that it plans to introduce next year, said Pascale Garnier, Qualiflow’s marketing director.

Dubbed Helotis, Qualiflow’s next-generation MFC will contain an improved version of the solid-state sensor first unveiled in its AFC 90MD component last year. Solid state sensors have proved troublesome, as MFC suppliers have experimented with stainless steel tubes and struggled with materials and cleanliness requirements, said Jean Frey, Qualiflow’s director of R&D.

“We are correcting these problems, and we should come out with a reliable product next year,” Frey said. Utilizing thin-film technology, the successful trick has been to make the sensor smaller with more stable materials, such as platinum, he said.

In terms of valves, Qualiflow is developing a high-flow, high-temperature diaphragm valve with a piezoelectric activator base.

Editor’s Note: As explained at length elsewhere on this site, this is a news story written by me that originally appeared on the now-defunct Electronic News’ website, which is long gone. It’s former sister pub Electronic Design News (EDN) currently holds the copyright to all Electronic News copy (to the best of my knowledge). You can still see a copy of this story at EDN.

Can Chips Make You A Better Person?

Editor’s Note: I included this story, along with the other two bearing this date, October 14 (Open Architecture vs. Open Standard; Qualifying Technology) not because I thought they were particularly brilliant journalism, but because they bore datelines from three different countries — one from entirely separate continent — all in the same issue. Pretty cool, no? I doubt I’ll ever win a Pulitzer, but at least I can lay claim to this.

R&D Consortium Seeks Cross-Disciplinary Approach to Realize the Future

LEUVEN, Belgium — Here in Flanders, the crossroads of Europe, researchers envision electronics improving human existence in a tailored, non-invasive way — far more than a few steps beyond ergonomic keyboards.

Alas, Electronic News (the print edition): we hardly knew ye!“Ambient intelligence … where people don’t have to adapt to the electronics, the electronics adapt to us,” said Rudy Lauwerens, VP of design technology for integrating information and communications systems for IMEC.

The propeller heads here at IMEC are taking a cross-disciplinary approach that includes MEMS, software, and semiconductor technology to make wearable—and in some cases implantable—wireless computing devices with virtual interfaces and distributed transducers a reality. The Belgians have dubbed this approach the M4 initiative.

Forget about your PDA. Think WDA, as in wearable digital assistant. Forget about LANs or even WLANs. IMEC wants to give you WBANS, or wireless body-area networks utilizing implanted sensors that can take direct measurements of body chemistry. IMEC’s R&D wizards have dubbed this initiative Human++.

These may sound like lofty goals, and granted, these technologies were the stuff of science fiction a few short years ago. But the scientists here at this R&D consortium believe they, along with their partners in industry, can manifest these dreams by the end of the decade.

Of course it’s going to take a lot of hard science and hard work in the fab to make implants that can monitor a diabetic person’s blood and inject insulin or glucose automatically when needed as well as periodically upload data for a doctor’s review. But these implants also could allow coaches of Olympic athletes to monitor blood oxygen and lactate content while an athlete trains on the playing field—not in the lab.

“We are far from the power we need for ambient intelligence,” said Lauwerens, who spoke to members of the technology trade press gathered here for IMEC’s annual update on its avenues of research. “Today is the time to start solving the roadblocks for these kinds of systems.”

These roadblocks are more familiar to those in the trenches concerned with the day-to-day mundane task of getting working devices out the fab doors. To make these dreams reality is going to require gigahertz RF and mixed-signal bandwidths on small, inexpensive chips or packages.

Thus, IMEC says it is turning to its semiconductor and related technology partners to get these ideas out of the think tank and into the R&D fab, and eventually onto the production fab floor. It means bringing companies into several avenues of research at once. “We allow companies to join us in cross-disciplinary research,” Lauwerens said.

It is a role that IMEC says it is suited to play. Unlike other collaborative R&D efforts, IMEC’s private funding, which accounts for about two-thirds of its budget, is not intended for specific programs designed to benefit certain member companies and market segments — rather it is consortium-based, meaning participating companies’ funds are pooled for group efforts.

This makes IMEC the ideal forum to bridge the gap between applied microelectronics and complex systems design, said Liesbeth Van der Perre, program director of multimode multimedia for IMEC. “This is somewhere we can take the risk and have the advantage in IMEC,” she said.

Having devices that can adapt to the user means developing reconfigurable hardware and software systems that make today’s embedded software and ASICs look like Tinker Toys.

Of course the future doesn’t invent itself. It requires a lot of hard work and R&D, which eventually translate into applicable, everyday process technology. IMEC appears not to be neglecting this even as it dreams big.

The same day it unveiled its M4 and Human++ cross-disciplinary research initiatives, IMEC announced plans for a 300mm research foundry that it wants to bring online in 2004. IMEC is looking to stay two steps ahead of the industry and work on the 45nm node and beyond.

IMEC's research food chain.

“The problem we’re faced with is we have to make this affordable,” said Luc Van den hove, VP of silicon process and device technology at IMEC. And not only are ASPs going down, but the cost of development is going up, a la the several billion dollars it costs to construct a 300mm fab.

The obvious conclusion for chipmakers and suppliers alike is to partner with one another to benefit from mutually conducted research, Van den hove said. “We believe by combining the interests … we can create required synergy that is needed … and share the costs,” he said.

The 300mm R&D foundry will focus on developing specific process modules for advanced technology nodes and will be a single-wafer fab. But it also will compare experiments between certain batch and single-wafer processes. Its goal is to develop complete single-wafer processes, and it will follow IMEC’s current model of specializing strictly in R&D.

There currently are no plans to produce any production wafers for its member companies, Van den hove said.

Editor’s Note: As explained at length elsewhere on this site, this is a news story written by me that originally appeared on the now-defunct Electronic News’ website, which is long gone. It’s former sister pub Electronic Design News (EDN) currently holds the copyright to all Electronic News copy (to the best of my knowledge). You can still see a copy of this story at EDN.