RF Meeting In San Diego>
3/06/99 - EDA firms, phone makers square off on RF-tool issues
Mar. 05, 1999 (Electronic Engineering Times - CMP via COMTEX) -- San Diego - Cellular-handset makers and EDA companies
locked horns last week, turning a meeting of the International Wireless Packaging Consortium (IWPC) here into a dispute over who
should shoulder responsibility for easing the design of next-generation RF systems.
To the cell-phone makers' calls for transportable models, interoperable tools, richer component data and more system-level
simulators, EDA tool vendors countered that the RF world does not have consistent design flows that lend themselves well to
automation. Moreover, they said, the EDA community should not be charged with writing everybody's RF component models. And,
they added, development of a useful geometry-interchange format must get a bigger influx of financial resources than seems visible
at the moment.
The IWPC was meeting to determine the requirements of a next-generation RF tool set, or data-interchange format, called
Electronic Product Design System (EPDeS). EPDeS, targeted to roll out by the fourth quarter of 2000, arises from a tacit
agreement that entirely too much design time is spent collecting and verifying the mechanical, thermal and electrical
characteristics of RF components and materials, and in transferring design files from one tool to another, said IWPC chairman and
EPDeS driver Don Brown.
But the meeting, hosted by Qualcomm Inc. (San Diego), kept turning to the issue of RF tools, and Brown had to step up to the
podium again and again to keep the audience's eye on the ball. Phone makers like Nokia, Qualcomm and Lucent Technologies
confronted tool vendors HP EEsof, Mentor Graphics, Ansoft, Zuken-Redac and others, demanding better software.
Indeed, the event disclosed significant differences between what cellular and RF system builders said they needed and what EDA
tool vendors seemed willing to provide.
"We've acknowledged that there is an issue here," said Brown, "but we're coming together to talk about it. . . . Think of it this way:
The Mideast peace talks couldn't have been easy."
The outcome of the sometimes tense event was the establishment of a number of subcommittees and working groups, each
charged with moving a different aspect of the EPDeS project forward. Brown suggested that the most significant work would lie in
defining the interface-the information and data that needs to transcend boundaries-among various parts of the design food chain:
between phone designers and manufacturers, mechanical and electronic component suppliers, and materials and process
suppliers.
Opening-day discussions provided sometimes searing insights into the obstacles the group must overcome to develop a useful tool
set or exchange format. Paul Draxler, senior project chief at Qualcomm, worried in his presentation that device models and tools
are poorly integrated, keeping RF designs from moving between physical layout tools and electromagnetic simulators. "We have to
reenter the data," Draxler complained.
Markku Lindell, project chief in Nokia's Research Center in Helsinki, Finland, said he wanted behavioral-level models that would
allow him to explore RF, IF and baseband partitioning issues on the systems level. He called for "preverified design objects-virtual
components" he could use in his simulations.
Lindell advised EDA vendors to pack more value into the data exchanged, not in the tools themselves. Nokia wants reusable
analog, analog/mixed-signal (AMS) and RF intellectual property, he said.
Lucent's Ron Barnet shared his vision of mechanical/physical and electrical-engineering teams using linked tools to design a
portable radio transceiver. His team is working on wireless local-loop designs, in which the loop serves as the phone infrastructure
for selected areas of a developing country. In Barnet's description of the design process, the RF system designer ordinarily pulls up
a substrate on his layout screen and populates it with active devices. He fills in the conductive traces between the devices, and
then performs electrical, thermal and mechanical analysis on the circuit.
The trouble with this process, Barnet said, is that the designer seldom has accurate geometry or leakage information on the
components and materials used. Barnet's wish list includes a physical-design input tool that could drive all simulators without data
reentry, and better electrical and physical characterization data for components and materials.
But materials suppliers and EDA tool vendors came back with both guns blazing: HP EEsof's HFSS project chief Dave Wilson
criticized the IWPC's first meeting on EPDeS, held in Helsinki last October, claiming the goals of the project are too aggressive.
Wilson claimed that the utility of RF tools was based on "domain optimization"-their ability to analyze data on a time or frequency
axis.
But point-tool optimization would encourage irresolvable differences between libraries and symbols, simulation libraries, file formats
and data representations, Wilson said. He recommended the IWPC concentrate primarily on a materials database standard, and
narrow its focus until a viable source of funding is identified.
Proprietary protection
Ariel Kao, AMS product manager for Mentor Graphics Corp. (Wilsonville, Ore.), reminded manufacturers that while RF models must
be widely available and distributed, much of the intellectual property will need to conceal proprietary manufacturing-process
information. "At the end of the day, a neutral data format doesn't work," agreed Michael Heimlich, president of SmartLinx.
Mentor will support interfaces to other popular tool sets, but not all of them, Kao said. "It's a money sink."
"This is about money," agreed HP EEsof's Wilson. "When customers put their money on the table, then we'll have something that
works."
But such sentiments clearly offended many IWPC members. "We expect leadership from you," said Nozad Karim,
electrical-characterization manager with Amkor Technology, a packaging house. "Don't come to me for an investment in something
that may be a product that you'll sell back to me."
"There is a mismatch between tool capability and user needs," said Mafet's Donny Barton, who heads the Microwave Design
System consortium. But Barton-project chief at Raytheon Systems Co. (formerly TI's Defense Electronics Group in Dallas),
program director of the Mafet tool-development effort and a veteran mediator of intercompany squabbles-served as the voice of
reason. By agreeing to steer the EDA Tool File Transportability working group, he'll take the lead in finding points where diverse
EDA tools intersect and can benefit from a standard file format.
The working groups established last week will explore useful design primitives that could be exchanged, design flows and a
materials database. The working group on a Materials Database will fall under the leadership of Dupont's Sam Horowitz, while the
group on RF Models and Primitives will be co-chaired by Draxler of Qualcomm and Paul Collander of Nokia.
The formation of a funding group, to explore alternatives for government grants and external sources, was put off for another time.
-0-
By: Stephan Ohr
Copyright 1999 CMP Media Inc.
|
