Semiconductor design companies, particularly in analog and power, have evolved. Previously, engineers at headquarters defined new products in terms of functionality and specifications, and the only remote facilities were sales offices and fabs. Today, designs can come from anywhere, often with collaboration between far-flung and unexpected organizations.
A few years ago, Analog Devices announced the acquisition of Danish company AudioAsics with experience in hearing aids. Now, that company will be a satellite design center. The Danes had a remote design center of their own in Bratislava, Slovakia. The acquisition helped give Analog Devices a big lead in microelectromechanical-systems (MEMS) microphones.
Bratislava offered a smaller cadre of grad students and post-docs who worked on the electronics under a particular professor at the university there. That’s a different picture than the one painted by North American engineers who think of remote design centers in terms of high-pressure, 24/7 design pushes by underpaid, third-world engineers (see “An Elite Analog Roundtable Discusses Design Centers”).
Where Design Centers Are Located
According to Anil Telikepalli, director of business and general management at Maxim Integrated Products, Maxim has design centers that support his business unit in Milan, Italy, and Bangalore, India, as well as in Sunnyvale, Calif. The larger business unit also boasts remote groups in Dallas, Texas. But Bangalore reports to Milan, not the California headquarters. That was the setup when Maxim acquired both groups, and that autonomy is here to stay.
Telikepalli also says the engineers in India come from a range of diverse backgrounds. “In this case, the design leads have done six or so unique IC designs. They are fairly senior guys who were hired by Maxim as new college hires from the Indian Institute of Technology IIT system,” he says. “Some of the institutes, especially in Karapur on the northeast, and Bombay in the west, have good power electronics design programs and good professors with a lot of background.”
Linear Technology also has a unique structure. “Corporate headquarters is here” in Milpitas, Calif., says CTO Bob Dobkin. “But there are more designers in the remotes than there are here.” That doesn’t mean these designers all are overseas, though (see the figure). Linear has remote design centers in the Boston area and in Tucson, Ariz. The Tucson center goes back to the 2000 Texas Instruments acquisition of Burr-Brown, when some Burr-Brown designers were attracted to the smaller company. In that respect, Dobkin also prefers domestic engineering talent.
“Our remote design centers are more difficult to manage, more inefficient, in a way, than our U.S. design centers just because of the time zone differences, the challenges of support in the back end, and the type of engineers that we’re hiring,” Dobkin says. “There’s no cost advantage to being overseas. We aren’t going to India, for example, and hiring new college graduates and training them. We are always looking for seasoned, very high-end, capable design engineers.”
Still, Linear uses some overseas labor. The difference is in the way it handles staffing. “Right now,” Dobkin acknowledges, “we have a design center in Singapore and one in Germany.” Linear has a design center in China as well. Plus, Singapore gave Linear tax incentives to locate its design center there.
“And over the years, we’ve made that a productive design center. For most of the time over there, we had an expat running the design center because he knew the ins and outs of how to get things done at Linear,” Dobkin says. “Singapore’s a reasonably pleasant place to live. And the expats go over there for a couple of years, or as long as we can get them to stay, and they run the design center.”
Analog Devices has centers in Norwood, Mass., San Jose, Calif., Durham, N.C., Limerick, Ireland, and India, China, and Japan. According to Dave Robertson, vice president of technology at the company, these centers often operate cooperatively.
“It’s hard to think of a product that doesn’t benefit from our global design centers,” Robertson explains. “Our most recent big DSP project was co-developed in our design center in India and in our design group in Norwood, Mass. It had big radio chips that are jointly developed in North Carolina, Boston, and China. We’ve got a number of our Class D audio amplifiers that are jointly developed in China, Japan, and San Jose.”
There’s a significant relationship between product definition and the location of the engineers who turn that definition into reality, Robertson adds. “For us, a big emphasis has always been to have the definition done where the development is done. But you [also] obviously want the definition done close to where the customers are. That’s one of the mechanisms that drives these multiple locations,” he says.
“So if the big customer is in Europe, we’ll typically have an engineering team there that’s based in Europe, but that may also connect with a product development group back here in the Boston area. The definition will spread across both sites. You do have, then, part of the product development team that’s in the same time zone and, hopefully, [thinking in the] same language as the key customers.”
Jack Olson, vice president and general manager of power management dc solutions at Texas Instruments, says his team has “10 or 11” design centers in New Hampshire, Rhode Island, North and South Carolina, Illinois, Arizona, and Texas. He also acknowledges the 24/7 advantages of globally dispersed design centers working on the same project, noting one particular project for Intel Core Power.
“You can have one team working on it in the Eastern time zone or California time zone and then another team working in, say, in Asia, China, or India or Japan. So you get kind of 24/7 coverage in terms of keeping the project moving and making progress from hour-to-hour versus day-to-day,” Olson says.
“I think a secondary benefit is the combination of expertise and skill sets of people at the different sites that makes them able to work together and poke at designs to find ways to do things better, more creatively, and to really increase the level of rigor in design reviews and evaluations, “ he adds.
Intersil has several design centers in the U.S., including locations in California, Texas, and the East Coast. Its overseas locations include several in China, one in India, and one in Japan. “Why and where we develop these design centers, or have built them, is driven by the availability of talent tied to the products that we develop,” says Ramanujam Thodur, managing director of the company’s India design center.
“For example, Dallas comes from a strong legacy of battery management,” he says. “We have taken that from simple battery backup protection to ATM and that type of equipment to consumer electronics and phones.”
Managing Maxim’s Design Centers
The difficulties in managing these remote design centers initially inhibited their use. Back before the Web, when the only practical way to exchange design data was via a T1 line and a VAX, it made sense to keep the design work at headquarters. Today, it’s a different story. A variety of solutions is available, including e-mail, Skype, telepresence, and even getting on an airplane and flying to those other locations. Yet managing the cooperation between multiple centers remains a challenge.
“It’s easier to concentrate the development in a single design center. At least with respect to the ownership and the accountability of the overall design, it’s immensely beneficial from a project management standpoint to have it all in one place,” says Telikepalli.
On the other hand, if it’s possible to reduce a project to separate building blocks, the management is easier. “If we look at the project as IP [intellectual property] blocks, it can be done in multiple locations and integrated, but the integration ownership has to be at one location,” Telikepalli says, citing the interaction between the design operation in India and its parent in Italy as an example.
“We recently introduced a high-voltage (4.5- to 65-V) buck regulator. Our differentiator is that it’s synchronous. [Synchronous regulators at higher voltages are still somewhat rare.] We eliminated the Schottky on the low side and made it a MOSFET,” he says.
“We gave the project to our team in India, which has done a good number of power regulators, although not necessarily at very high voltage. They have also done several of our regular PWM [pulse-width modulation] controllers and sometimes ac-dc supplies, which have to deal with high voltage although not in the controller space. The project was completed in India with some support from here in Sunnyvale,” he explains.
In this case, Italy and India drove the product definition. “The India design center reports into a design director who is based in Italy. The design director [interacts] with the design lead and the managers to write the spec. Initial product definition came from our marketing and product definition team based here in Sunnyvale, but when it got down to the details of the spec, our product definer was based in India,” Telikepalli says.
ADI Management Insights
Robertson also sees the link between the customers’ location and the design centers. “For us, a big emphasis has always been to have the definition done where the development is done. But you [also] obviously want the definition done close to where the customers are. That’s one of the mechanisms that drives these multiple locations,” he says.
If there’s a big customer is in Europe, Robertson says, Analog Devices typically will have an engineering team there that’s based in Europe. This team also may connect with a product development group in the Boston area. The definition will spread across both sites. Part of the product development team will be in the same time zone and, hopefully, the same language as the key customers, he adds.
“In certain cases, we may have a heavy RF team in one site and a heavy digital team in another, and the overall architecture of the chip makes it relatively straightforward to separate those things, hierarchically. Then the two groups can run fairly independently, coordinating on a weekly basis, but not necessarily on an hourly basis. There are other situations where co-design is intermingled and the teams stay in lock-step on an hour-to-hour or day-to-day basis,” Robertson says.
As some projects get bigger, particularly SoCs, a team of program managers and system architects can be found constantly performing verification. These designers run up and down the hierarchical chain to ensure that the interfaces are evolving properly for each of the blocks as they’re being developed and that it all will play together when it’s time for assembly. So, one of the teams on the big chips is primarily in charge of making sure all the pieces are going to play together, he says.
“That becomes a big, big job, and I suspect that such a team had to exist even when everybody was all in the same physical location, because, even then, the digital guys were working in different tool sets and speaking a different language than the analog guys, who are speaking a different language than the crew who were doing the power management, who are speaking yet a different language from the crew that was doing the RF part of the design.”
This is where electronic communication falls down. “There are communication tools, but we spend a lot of time on airplanes. We get our technical communities together for internal technical conferences. At project kickoff, we will bring a lot of people into the room together,” Robertson says.
The Personal Touch
Robertson says that management involves more than electronic paperwork. The idea, he says, is to make sure you have some personal connection. “We do a lot of expatriate assignments or people from the remote design centers will do a tour of duty back in one of the mothership offices just to keep that personal connection very high,” he says.
“In addition to the formal structure, you definitely want an informal structure to complement that so that people, even though the document says, ‘X, Y, Z,’ if that doesn’t feel right they need to be able to pick up the phone or send an e-mail to somebody saying, ‘Hey, is this really okay?’ or ‘I’m kind of nervous about this’ without formalizing everything,” Robertson says.
“We just had our big internal technical conference here last month. We do that every year and have, I think, well over a thousand engineers all gathered together. For us, that’s a very important part of making that formula go,” he says.
Linear Focuses Less on Collaboration
“As a rule, our company generally is a ‘one designer, one project,’ kind of company, certainly historically,” says Chris Mann, director of satellite design centers at Linear Technology. “Recently [in the U.S.], we’ve gotten into situations where we’ve done more collaborative design projects where we have people from multiple sites working together. But we have not attempted that with our international design centers yet. So for the most part, a designer in a remote might be working with someone else, but it’s somebody in the next cubicle and not somebody at another location.”
Mann also notes that Linear has people at its headquarters who are good at defining products. “They’re the managers that go out, talk to customers, and see what customers need. And come up with the idea that we need to make this product: ‘Here is what I saw and why it could be used.’ And they get with their people,” he says.
“That said, there are areas of specialty within the company and some of them are remote locations. For instance, battery charging comes out of Boston. And whether the manager is there or the manager is here, once there’s a germ of an idea for a product, they talk together and they come up with what they’re going to do for the product,” Dobkin adds.
“When it comes to a spec, the spec is half a dozen features on the product and it gets fleshed out from there. Then after that they get to the point when they want to make a product. It gets approved internally, which is a one-day item, and then they flesh out the product with pins and specs and block diagrams and all the rest. And it could happen here, it could happen at the [U.S.] remote site,” Dobkin says.
“The product then gets assigned to a person at the remote and the manager at the remote oversees it. And during the course of the design, the designer may come here several times for design reviews, test-equipment reviews,” Mann says.
“As some of the design centers have grown—Boston, for example, is nearly the size of what we have here [in Milpitas] now—there’s been a trend over the last five years towards greater autonomy in the design centers. Instead of just having chip designers like we used to, today we have test engineers and applications engineers. The remotes are able to take the product from beginning to end almost without Milpitas’ assistance,” Mann says.
And Then There’s China
“Our design center in China started up because there were several Chinese people working at Linear who left and went back to China. They all liked working for Linear so we got back to them and asked if they’d like to join a Linear design center over there, they were all enthusiastic,” Mann says.
“And we were able to hire all of them back. They told us that the reasons that they had gone had to do with family and proximity to family, so they were happy to be back,” Dobkin notes.
TI Manages Complexity
Texas Instruments, a much larger company than Linear, is more rigidly structured. That includes its management of designs across remote design centers in different time zones.
The company has design tools and layout tools that make it easy for designs to be checked in and checked out and worked on in multiple locations at the same time. But before that stage, there’s a preliminary process for evaluating and prioritizing business spaces for its new products, says Olson. This process matches specific business cases with the capabilities and expertise of the team that they’re going to be assigned to.
When it comes to managing the actual day-to-day development work, the process varies based on the level of maturity of the team in the design center, Olson says. “At some remote sites, we have designers with 20-plus years of experience that have worked at other companies or worked at TI in other locations. In those cases, we can count on leveraging their expertise to run a project and not having an excessively high level of involvement on a day-to-day basis,” he says.
“In other cases, we may be starting off at new locations with designers that are either new to TI, or simply less experienced overall in terms of design, and we’d want to be more involved, day to day,” he says.
Texas Instruments takes different approaches when it opens up and manages its individual design centers.
“We’re adding new design centers pretty regularly, and the way they are established varies. In some cases, where we are starting up a new team, we might transplant someone from another location as a lead to get things started and to build up the team. In other cases, we may hire someone from outside, in a particular location where there is a strong pool of talent, and try to recruit talent around that leader,” Olson says.
Some design centers may be around universities that have good relationships with Texas Instruments and professors with strong expertise in areas that interest the company. It helps if those areas also have pools of people who want to stay there when they finish up with school, Olson says.
“In particular, there are certain schools that we have long-standing relationships with and we end up hiring a lot of students from over the course of time,” he says, noting that these universities include schools in China and the U.S. “Also in Japan, we have design centers around universities that we work with.”
A strong leader is important for managing the teams in these design centers, Olson says. “It is a little bit different than just being a design manager in a main site. It requires a bit of recruiting and site management and a broader scope of responsibilities. And we’re also looking for leaders who can really inspire that team, people who have a kind of energy and passion around a particular market area or product segment.”
Texas Instruments uses a variety of processes for facilitating product evolution, especially e-mail and phone calls. It also conducts WebEx meetings and teleconferencing, sometimes using technologies like telepresence and other video conferencing, Olson says.
“And then, of course, face-to-face meetings where folks actually get together in one location and work together, where we might bring representatives from different sites, if they’re working on a project in one of the sites. That helps them to spend time face-to-face and really dig in and build those relationships as well as resolving immediate problems,” he says.
Intersil’s Global Reach
With management split between its headquarters in Milpitas and the original Harris Semi offices in Palm Bay, Fla., and its aggressive growth by acquisition, Intersil certainly understands long-distance management.
“We have some products that are done solely within specific centers, and others that are collaborative designs across multiple centers. The centers deal with the circuit design aspect, evaluation, qualification, automatic test development, and production release—the entire flow to get from beginning to end. And, obviously, you have marketing involved at the beginning and throughout the design phase. I don’t think we have any product development done in a single site exclusively,” says Thodur.
“Everything is virtual, now. E-mail is the way everyone communicates. We have database systems for all our documentation, and our CAD systems are tolerant of multiple-site usage, so all of our design databases are updated across multiple sites,” he says. “Nonetheless, using airplanes is inevitable. It helps if, from time to time, you can get people to travel between suites.”
Thodur prefers conference calls and Web meetings. “We also have started the use of smart boards combined with video conferencing. That way, you can have a video going, you can be editing documents on a board, all at the same time.”
Smart managers also acknowledge regional specialties, as some skill sets have developed in different areas. India is known for its digital design and back-end chip assembly. But Intersil’s India design center has concentrated on analog since its inauguration, Thodur says, which is unusual because the university system in India historically hasn’t focused on analog electronics.
“Many of the Indian engineers who are doing analog for Intersil have bachelor and graduate degrees from the U.S., have worked in the U.S., and have moved back to India. China’s similar, but I think that their education system has focused more on analog than India has,” he says.
Still, Thodur says U.S. exposure is essential to gaining practical analog experience, which is difficult to gain otherwise. “Silicon Valley has a certain perspective on how to get projects done, and it has an incredibly aggressive culture and a tendency toward risk taking,” he adds.
Managers, then, need to understand how to modify cultural patterns. “If you go to India, where engineers come to us straight out of school, they don’t have any experience with an aggressive, risk-taking culture, so they tend to take the low-risk path, and they can have schedules that take two years to do something that might take six months if you would do that all in Silicon Valley,” he says. “That takes exposure and education.”
If you want to know how today’s companies are dealing with design centers, you have to talk to the folks in charge. We were fortunate to speak with a cross-section of managers at five of the bigger analog and power semiconductor companies. These executives shared how they created or acquired remote design centers and how they integrate them into their companies.
Bob Dobkin is the founder and chief technical officer of Linear Technology Corp. Prior to 1999, he was responsible for all new product development at Linear. Before founding the company in 1981, he was director of advanced circuit development at National Semiconductor for 11 years. He has been intimately involved in the development of high-performance linear integrated circuits for more than 30 years and has generated many industry-standard circuits. He holds over 100 patents pertaining to linear ICs and has authored more than 50 articles and papers. And, he attended the Massachusetts Institute of Technology.
Chris Mann has served as the director of satellite design centers at Linear Technology since 2003. Previously, he was the vice president of engineering, Networking Products Division, at LSI Logic and director of engineering at Philips Semiconductors. He was an IC design engineer for 13 years at Siliconix, Seagate Technology, and Motorola Semiconductor as well. He has a master’s degree in engineering economic systems from Stanford University and a BS in electrical engineering from Iowa State University.
Jack Olson is the vice president and general manager of power management dc solutions at Texas Instruments. He is responsible for dc-dc power converters used in computing, communications, consumer, and industrial applications. He held various engineering and management positions with Motorola and Sipex Corp. in MEMS pressure sensors and power management products before joining TI in 2002. Most recently, he worked as general manager of the linear power business unit. He has more than 20 years experience working in the semiconductor industry. He holds a bachelor of science degree in materials science and engineering from the Massachusetts Institute of Technology and an MBA from Arizona State University.
Dave Robertson is vice president of analog technology at Analog Devices. He joined ADI in 1985 and has worked on high-speed converter product design in a variety of process technologies, including complementary bipolar, biCMOS, and submicron CMOS. Previously, as product line director for the High-Speed Signal Processing Business Unit, he directed both core converter development and the mixed-signal product strategy focusing on broadband communications applications.
Also, Robertson has served on the Technical Program Committee for the ISSCC (International Solid-State Circuits Conference) since 1999 and has served as chair of the Analog/Data Converter Subcommittee from 2003 through 2009. He is an active member of ISTAC (Information Systems Technical Advisory Committee), a joint committee of industry and government members working to advise on technology export controls.
Additionally, he has authored numerous technical papers and award-winning presentations for venues such as the ISSCC and the VLSI Symposium. He has 16 patents on converters and mixed-signal circuits. He received a bachelor’s degree in economics and a bachelor’s degree in engineering from Dartmouth College.
Anil Telikepalli is the director of business and general management at Maxim Integrated Products. Previously, he was a strategy and marketing consultant at AT Business Consulting and senior director of marketing at Vsolay. At Xilinx, he was director of platform solutions and business development, senior technical marketing engineer of IP partnerships, and an R&D/apps engineer, design services. He attended the University of Kentucky and Osmania University. And, he holds U.S. Patent 5754459, “Multiplier Circuit Design for a Programmable Logic Device.”
Ramanujam Thodur is the managing director of the India design center at Intersil. Prior to joining the company, he held multiple positions with Texas Instruments, including product line manager/sector manager, strategic marketing & engineering manager, general manager, and design lead/engineer. He holds an M tech degree in integrated circuits & systems engineering from the Indian Institute of Technology, Kharagpur, and a BE in electronics and communication from Osmania University.