Designers' wish list: more tools: need software for RF, microwave circuits for cellular com, wireless computer networks

Electronic News, June 15, 1992 by Peter Dunn

Need Software for RF, Microwave Cicrcuits For Cellular Com, Wireless Computer Nets

CAMBRIDGE, MASS.--High-performance radio frequency (RF), microwave and millimeter-wave circuits and systems, once the province of military contractors, have become commonplace in cellular communications, wireless computer networks and a host of other commercial applications.

While good quality software tools exist for many aspects of the design process, engineers must still use archaic iterative techniques for a number of essential tasks along the way. Automation of these tasks could provide a new market niche for existing computer-aided engineering (CAE) software firms, or an opportunity for start-ups.

Among the "wish list" items: tools to assist in the development of phased array antennas, needed to receive gigahertz-range frequencies; better simulation of non-linear circuitry like oscillators and mixers, and a top-down integrated circuit (IC) design methodology.

Raytheon Co.'s Andover, Mass., Advanced Device Center has been seeking to broaden its scope into the commercial market from its traditional strengths in defense ICs. The center recently won its first high-volume commercial contract, to produce a gallium arsenide monolithic microwave IC (MMIC) for satellite television converters being produced by New Japan Radio Co. Ltd.

While Raytheon was able to use primarily third-party software (from Cadence Design Systems Inc. and EEsof Inc.) in the circuit's development, said center general manager James DiLorenzo, about 30 percent of the tool-set was written internally. "It gets complicated," he explained. "A lot of third-party software is designed for a specific task, and we have to interface it with our tools for things like checking layout parasitics and back annotating. We use 'glue' software to put commercial programs together; you've got to take discrete pieces and put them together into one package.

"We've been promised an integrated workstation with layout and back annotation for the better part of two years," added Mr. DiLorenzo, "but software is notoriously late."

David Layton, manager of design and business development for the center, cited "non-linear simulation tools" as tops on his list of products he'd like to see. "Oscillators and mixer circuits are very difficult to simulate now with the tools that are available," he said.

One critical link in an RF system is the antenna. Despite this, however, antenna design is still primarily an iterative process in which a rough design is calculated, then fabricated, tested and redesigned as needed.

Vorad Safety Systems Inc., San Diego, faced a daunting task when it set out to make a phased-array antenna designed to mount on the front of a vehicle for an automotive collision avoidance system (EN, May 11). A commercial antenna layout simulator would have been a welcome sight at Vorad, according to vice president of engineering Jerry Woll.

There is some software designed for military applications, but it is not useful for the 24GHz frequency used by Vorad, Mr. Woll added. "The military does nothing at 24GHz," he said. "It's all custom-designed for this frequency. But software for getting 400 elements to make one beam has been developed for the military."

"There's a lot of (antenna) work done on an empirical basis," added Raytheon's Mr. Layton. "Numerical calculations can take you to a certain point, but it's very hard to simulate."

Perhaps the most fundamental need, however, is something that digital and analog IC designers have come to take for granted: the ability to design a circuit from the top down, beginning with a set of specifications and using simulation and models to generate a final design.

"That's a major hole in microwave simulation, characterization and design," said Warren Seely, member of the technical staff at Motorola's Semiconductor Product Sector in Tempe, Ariz. "Nothing ties the component level to the virtual level for top-down design."

Currently, said Mr. Seely, a circuit designer must work from model libraries supplied by the various microwave circuit houses. "Folks like TriQuint, Raytheon, Motorola and TI have layouts and models in their manuals," he explained. "If you need a 6.5-picofarad capacitor, well, maybe there's only a 6.4 or a 6.6. But you make your choice, and manually key in the model and go from there. But it's time-consuming, and you don't always get what you want. Maybe it's a square capacitor, and you wanted a rectangular design."

Ideally, an engineer would have access to high-level specifications of inductors, field-effect transistors and other building blocks. This in turn would provide a foundation for automated generation of components. "If you want to build a medium-drive, low-noise amplifier, there are specific algorithms you can go through," noted Mr. Seely. "There's nothing that says you can't put that in a program--have the user fill out a spec sheet and have the program generate the amplifier design." Ultimately the same techniques could be used for system-level design.