Analog expands its indispensable role in 2007

Danish physicist and Nobel Laureate Niels Bohr reputedly noted that "prediction is very difficult, especially about the future," and that's still true. Despite our industry's extreme reliance on technology road maps to shape plans, investments, and products, we still have difficulty predicting what will happen, when it will happen, if it will happen at all, or be "just around the corner" over and over again. After all, you need to get both the "what" and "when" right, and in the same prediction, for that crystal ball to pass inspection.

But we can look ahead to 2007 and have some reasonable assessments. Most striking is that "analog" circuitry and functions, in the broadest sense of the term, will continue to grow in importance to successful products. From mass-media products in multimedia, cell phones, and HDTV, to wired and wireless networks, to industrial and medical applications, anytime you have some sort of real-world I/O or transducer, or a transition between physical media, you have an analog requirement.

We also see some developments that have been in the works for a while, and are now close to critical mass as the necessary pieces of technology and market come together. These developments both require and advance analog technologies:

High-performance mixed-signal ICs: mixed-signal tools and processes and now advanced enough that for many applications, it's possible to get both high-performance analog and digital circuitry onto a single IC. Even if the analog performance is not quite as good as you could get by partitioning into an analog-only IC (article: "Smart partitioning in WiMAX radios illustrates design challenges, Part 1"), it's often good enough and cost-effective enough that it meets the technical and market requirements. Recently announced modeling tools, such as Virtuoso from Cadence Design Systems, make these types of high-performance mixed-signal designs much more predictable and less of a development risk, which is also critical.

Software-defined radio goes mass: with all the formats which today's multimedia and wireless systems must support, the virtues of mostly hardwired designs are also their weakness: they do one or a few formats very well, but can't handle new or modified formats without major upgrade headaches. But SDR lets the bulk of the channel's function be defined by software-based algorithms. A practical SDR design requires low-distortion, broadband amplifiers and A/D-D/A converters, with fairly stringent performance specifications. Although SDR is now in use in some base stations, it hasn't reached consumer products—yet. But the requisite parts, at acceptable cost and power consumption, are now appearing from the leading analog vendors, along with the design and development tools for these algorithms.

For example, the SDR Forum announced finalists for its SDR Challenge, while Texas Instruments, in collaboration with Xilinx, Lyrtech, and other third parties, today unveiled a Small Form Factor Software Defined Radio (SDR) Development Platform.

Silicon-based optics and electro-optics: as Moore's law, the laws of physics, and production issues make further electronic-only speed and bandwidth advances more difficult, optical-based systems for interconnect and even computing become attractive. But even a basic LED laser requires exotic substrates such as lithium niobate. However, if you could use silicon (which is difficult due to quantum-level constraints), the vast industrial and scientific silicon infrastructure is available.

This year, Intel and the University of California at Santa Barbara announced they had developed a hybrid, silicon-based laser diode, (article: "Intel, UCSB tout hybrid Si-laser chip advance"), and a Japanese company has developed a blue LED on a silicon wafer (article: "Japan's Shimei grows blue LED on silicon wafer"). As these developments advance, we will see integrated electro-optics used for intrachip, then interchip links, as first steps.

MEMS moves ahead, into many niches: Silicon-based MEMS devices are already well established as acceleration sensors in airbags and game controllers, but we'll see more MEMS devices and sensors in mass-market consumer products.

Examples include MEMS temperature sensors, microphones (Akustica), even used in a clever way as the dirt sensor in the Roomba self-managed vacuum cleaner from iRobot!), and oscillators/resonators such as from Ecliptek and SiTime. In addition to the potential size, performance, and reliability of silicon-based transducers, designers can incorporate signal-conditioning and pre-processing on the same die, for further cost and space savings.