Electronicdesign 5249 74776tab4

Analog Chip Makers Switch Focus To Vertical Markets

Dec. 12, 2012
Expecting flat or negative growth in 2013, top execs at major analog semidonductor companies explain strategies. Most are based on high integration in chips aimed at specific vertical market segments, employing unique IP -- rather than leapfrogging each other in raw specs on building-block chips.

Today’s executives don’t expect an imminent return to boom times. They’re strategically positioning new products that will help them weather what appear to be doldrums more than hurricanes. Companies are focusing on higher levels of integration, tailoring new products to specific applications—preferably as many as possible with similar needs (see Fig. 1 and Fig. 2).

1. Texas Instruments’ AFE7071 is a complete radio transmitter that reduces board space by up to 80% compared to discrete implementations. It integrates a dual digital-to-analog converter (DAC), tunable baseband filters, an IQ modulator, and a digital quadrature modulation correction circuit. The AFE7070 adds a direct digital synthesizer (DDS) with a 32-bit numerically controlled oscillator (NCO) and a low-voltage differential signaling (LVDS) output buffer.
2. The Analog Devices monolithic ADAS1256 digital X-ray analog front end integrates the entire charge-to-digital conversion signal chain, including low-noise programmable-charge amplifiers, correlated double-sampling circuitry, and 16-bit ADCs.

Opening New Markets

Intersil CEO Dave Bell doesn’t expect any industry growth in 2013. “If you look at the last few years, it’s been right around $300 billion (U.S.) for the worldwide semiconductor market. It’s really down a little bit, and I think just due to hyper-competitive pressures, it’s going to stay that way,” he said. With this forecast in mind, the company recently announced specialized products for active cables and picoprojectors and will announce more in 2013.

In the short term, Intersil is focused on copper, but fiber cables could be in the equation too considering the need for active cables in terms of the relative properties of USB versus Apple’s Thunderbolt.

“With USB 1.0, if you shoved a 10-Mbit signal down a 2-meter cable, you got a pretty decent looking signal out the other end. Today, if you try and do that at Thunderbolt speeds—20 Gbits/s—bidirectionally, on the same sized cable, you get nothing but garbage out the other end. You need the electronics to drive and to recover those signals within the cable connectors,” Bell said.

Also, Bell expects active cabling to be a long-term growth driver because data rates will continue to increase, as they have with USB. He also acknowledges a growing need in the datacenter inrastructure.

“If you’re looking at servers and switches and routers, and network-attached storage, clearly, those are all bandwidth hogs. Today, most of those applications use passive copper cables for connections up to a few meters. Beyond that, they have to go to fiber. Yet active copper cables would allow them to go up to something like 20 meters,” Bell said.

Looking further ahead, Bell said that one day, we might see copper and fiber in a hybrid cable, with the copper providing power in a manner similar to today’s Power over Ethernet.

Even in today’s off-the-shelf Thunderbolt cable, Bell noted, the protocol supports power delivery up to approximately 15 W. He added that it is already possible to deliver 3.3 or 15 V over Thunderbird, so users can get three or four times as much power via a Thunderbolt cable as they could get through a USB cable. Bell said he is familiar with at least one manufacturer that is working on much longer distance optical Thunderbolt cables, and he expects there will be some parallel power wires in those cables as well.

Furthermore, Bell sees continuing Internet growth functioning as a technology driver, relating it to Moore’s Law. “People keep saying, ‘Well, Moore’s law is just about out of gas.’ But they were saying that a decade ago and it keeps finding a way. I think you’re going to see the Internet continue to exponentially grow as it’s done from its inception,” he said.

“I remember an Intel PowerPoint slide that said that, around the world, 24 hours a day, there are 22 new mobile subscribers every second. Those users are going to continue to drive data traffic up and up.”

In the last two years, Intersil has made a big move into ICs for video applications as well, first with devices for carrying data from security cameras, and more recently, with devices that interface between video sources and the optical systems in picoprojectors. In fact, Intersil has been making laser diode drivers for more than a decade. They were originally used in DVD writers and later in Blu-ray writers. Those technologies and related ones have translated well for controlling the color balance and the laser power in picoprojectors.

Converters, OP Amps, And More

Linear Technology’s data converters are moving toward more bits and faster speeds. “We’re just about ready to release 20 bits at 2 MHz, and that’s a very clean 20 bits, in terms of integral and differential nonlinearity and the size of bit-jumps at the transitions,” said Linear CTO Robert Dobkin.

The new analog-to-digital converters (ADCs) are intended primarily for instrumentation applications. Steppers would be a typical example. “At 22-nm, steppers need as many bits as fast as they can get it,” Dobkin said.

For communications, Linear’s latest ADCs are in the 200- to 250-Mbit/s range at 16-bit resolution. “We’re going to duals, and we’re also going to lower power. I think it’s really important that dual and the quad versions of these 100-plus-MHz A-to-D converters run at low power, because they’re going into very dense systems where heat is a problem,” Dobkin said.

“We’re doing high-speed, high-precision, low-distortion op amps for the 16- and 20-bit converters. Without a good op amp to drive them, they’re not 16 bits and they’re not 20 bits. We are pushing precision as well as speed, settling time in the op amps as well, specifically mated to the data converters,” Dobkin said.

A few years ago, Linear was still relying heavily on its own 0.6-μm technology for the sake of its ability to handle large input-voltage swings. The process technologies for the company’s newest parts are foundry-based, 0.18- to 0.25-μm geometries.

Design Services And The Fabless Market

Talk of foundries leads to the analog considerations of fabless companies. Today, independent device makers (IDMs) such as Intersil and Linear Technology aren’t the only corporations that combine analog and power product lines.

The S3 Group, an engineering services company, is headquartered in Ireland. But semiconductors aren’t geographically limited. In fact, S3 typically facilitates design reviews and coordinates with companies a half a world away. Recently, the company has awakened to the value of maintaining its intellectual property (IP).

In terms of what S3 calls its “Silicon Business,” the company made a conscious decision to monetize its IP through licensing with a focus on mixed-signal IP, while continuing its traditional engineering services (see “The Role Of Design Services In Analog Design”). S3’s product lines comprise converters along with low-jitter, low-phase-noise phase-locked loops (PLLs) and analog front ends (AFEs) that use both.

Today, IP sales are responsible for a great deal of S3’s bottom line. In the present business structure, some 80% of the company’s efforts support some communications application, whether it’s wireless communications like cellular and Wi-Fi or power-line communications. Yet a few customers are using S3 IP in solar applications such as maximum power-point tracking as well.

S3’s IP tends to start around 10 Msamples/s and go up to about 250 Msamples/s for its ADCs and between 500 and 800 Msamples/s for its digital-to-analog converters (DACs). Resolution varies from 8 to 14 bits. The ones that customers are using most are in the 10- to 12-bit range, operating from 60 to 250 Msamples/s. Also, S3’s PLLs are designed to complement the company’s data converters, emphasizing low phase noise and jitter.

Dermot Barry, vice president of consumer silicon at S3, said that the competition comes from companies such as Synopsys, exploiting the ChipIdea capabilities that came with the MIPS acquisition, and Cosmic Circuits, an Indian company. He also said that S3’s design’s are generally faster and more precise.

Vertical Management

Among the established analog and power IDMs, Maxim Integrated has taken some bold steps in redefining itself for the new business realities. In a way, it is doing what S3 is doing (on a much larger scale) in terms of developing resuable IP and putting it into highly integrated products for specific end uses.

How that affects day-to-day operations depends on what end of the telescope you look through. At the management end, president Tunç Doluca has shortened the company name to Maxim Integrated, with a new logo, a new headquarters, and a team of senior vice presidents whose responsibilities focus on vertical markets, rather than on product lines organized by vertical market segments.

At the dedication of the new HQ, Doluca said the company was making the external changes “to better reflect the big internal changes we have made, and are making, to help customers tackle complex system and architectural issues.” Doluca also said that the re-imagined Maxim Integrated would accelerate the development of highly integrated products that complement the company’s broad selection of single-function devices.

Distribution channels were strengthened, product delivery lead times were trimmed to an average of six weeks, and the company has intensified its customer focus. “Taken together, these initiatives are going to change the way you think about analog and about us,” he said.

“Times have changed, and so have we. That’s why we’re leading the way with game-changing, high-integration products that set you apart. And we’re sharpening our focus on customer support by providing you with skilled technical support and systems know-how,” he said.

In the new Maxim Integrated, Matt Murphy is responsible for the communications and automotive business, Chae Lee is responsible for the mobility market, and Chris Neil is responsible for industrial and medical. This approach works its way down to the director level.

Of course, some things don’t change all that much. The company’s design engineers still behave as if they are specialists in data converters, or class-D amplifiers, voltage regulators, and other components, and of course they are. The goal now is to synthesize the results of those engineers’ skills into application-specific end products.

The Role Of Design Services In Analog Design

By adding intellectual property (IP) licensing to its portfolio, engineering services company S3 Group found new “stickiness” in its business relationships, according to Dermot Barry, S3’s vice president of consumer silicon. Yet the company continues its chip design business, which will account for 70% of its revenue this year. S3 also is trying to focus on areas where there is a shortage of skills, including the whole mixed-signal space.

“Over the last 10 years, 90% of our customers have been semiconductor companies, whether they’d be the old IBM-type organizations or fabless companies. With the new business model, we’re moving away from those guys, not entirely, but to a degree,” Barry said.

“They’re small to medium-sized system companies that have existing products in the market. Their products contain a reasonable quantity of chips, and discrete components, and usually, their product sales volumes would be in the 100,000 to, maybe, a million, 2 million per annum, not the very high-volume consumer-product companies,” he added.

These companies typically don’t have any IC design capabilities and don’t realize the benefits that integration can deliver. Their product lines, then, use standard parts and discrete components.

“They could achieve significant cost reductions were they to do an integration exercise where they took the functionality of a lot of those components and integrated that into one or two or three chips, even, maybe even in a multi-chip module-type approach or a system-in-package approach,” Barry said.

“They don’t necessarily need to go to a single chip at 28 or 40 nm. That’s probably far too expensive for them and represents a kind of overkill. But doing an integration of, let’s say, a whole lot of analog functionality onto an analog chip, and a whole lot of digital functionality onto a digital chip, they could realize product cost savings, of, in some cases, more than $100 per unit,” he said.

“Now, of course, there’s a large engineering effort to realize that benefit, so you have to do the return on investment analysis over what’s the product, what’s the remaining product lifetime, and how long is it going to take to actually hit the break-even point. In many cases that can be in less than 18, or up to 24 months once the chip or chips are available,” Barry said.

For example, S3 has worked with Iridium, which operates the communications business originally created by Motorola. A separate company purchased those assets and has been operating that system for the last 10 years. Now, Iridium is launching a series of satellites to enable next-generation services. A portion of the business is military, but a larger segment deals with logistics, tracking assets or fleet movements in situations that aren’t accessible via standard terestrial communications systems.

Iridium’s existing handset and modem were based on standard components that the company was buying through distributors. That sales relationship meant Iridium didn’t have direct relationships with semiconductor vendors that could have helped it with new designs. That’s why Iridium deals with a design services company.

“S3 was able to engineer a different partitioning of their system that eliminated roughly 250 components through an integration of analog functions on 0.18-μm CMOS and digital functions on a 0.13-μm process,” Barry said.

The program took a little over a year before actual chips were available and then about another three months to ramp those into volume production. The rapid prototyping was largely based on previously developed IP, which accelerated the design and reduced risks, Barry said.

Sponsored Recommendations

Design AI / ML Applications the Easy Way

March 29, 2024
The AI engineering team provides an overview and project examples of the complete reference solutions based on RA MCUs that are designed for easy integration of AI/ML technology...

Ultra-low Power 48 MHz MCU with Renesas RISC-V CPU Core

March 29, 2024
The industrys first general purpose 32-bit RISC-V MCUs are built with an internally developed CPU core and let embedded system designers develop a wide range of power-conscious...

Asset Management Recognition Demo AI / ML Kit

March 29, 2024
See how to use the scalable Renesas AI Kits to evaluate and test the application examples and develop your own solutions using Reality AI Tools or other available ecosystem and...

RISC-V Unleashes Your Imagination

March 29, 2024
Learn how the R9A02G021 general-purpose MCU with a RISC-V CPU core is designed to address a broad spectrum of energy-efficient, mixed-signal applications.


To join the conversation, and become an exclusive member of Electronic Design, create an account today!