EE Times is the leading weekly newspaper for design
engineers and technical managers who are major decision-makers in the electronics
industry. The publication is circulated to professionals working in the computer,
communications, industrial, military/aerospace, components, test &
measurement, and other industrial OEM markets.
Visit EE Times Online for more technology
news and features. |
Coverification "Stars" Coverification takes
center stage at conference
Welcome 12-Inch Wafer Industry readies fabs for
next-generation wafers
EDA Rumble in Japan Cadence and Avant! figure in Japan
tangle, too
OnTrak On the Lam Lam buys OnTrak to secure foothold
in wafer polishers
Battery Talk Portable debate turns to batteries
Oracle NCs Oracle rolls out StrongArm X86-based network
computers |
Coverification "Stars"
Coverification takes center stage at conference
By Richard Goering, EE Times
Santa Clara, Calif. -- With major introductions from Simulation
Technologies and Ikos Systems Inc., hardware/software coverification promises
to be a dominant theme at the combined International Verilog Conference-VHDL
International Users Forum (IVC-VIUF) here. Meanwhile, several new EDA companies
will make their first public appearance, and Viewlogic and Cadence will renew
their Verilog-simulation speed race.
Jointly located for the second time, IVC-VIUF has become the second biggest
EDA conference of the year, with roughly 60 exhibitors and a two-day technical
program.
Simulation Technologies (St. Paul, Minn.) will announce V-CPU, a product
that lets embedded systems designers develop software on a hardware "virtual
prototype" through a link to logic simulation. Ikos Systems (Cupertino, Calif.
) is taking a different route with Avatar, a lower-priced version of its
VirtuaLogic emulation system focused on software designers.
With Chronologic VCS version 4.0, Viewlogic Systems Inc. (Marlboro, Mass.)
is claiming a two- to threefold speed increase due to its "Roadrunner"
technology. Cadence Design Systems Inc. (San Jose, Calif.) is taking its
fastest Verilog simulator, NC Verilog, and adding timing features needed
for ASIC-vendor sign-off.
New EDA companies at IVC-VIUF include Novas Software and Interra Inc. (see
March 24, page 16). Also making introductions are the Advanced Technology
Center's CoverMeter division (Laguna Hills, Calif.), which provides a new
Verilog coverage tool; FTL Systems (Rochester, Minn.), which offers
hardware-description-language (HDL) analyzers, compilers and simulators;
and Simpod Inc. (Santa Clara), which makes a desktop verification tool that
can serve as a low-cost functional tester.
Simulation Technologies' V-CPU is based on the same general concept as the
Seamless CVE product from Mentor Graphics and Eaglei from Viewlogic. Such
products link software debuggers or instruction-set simulators to HDL simulation
and provide fast processor models. Software designers can thus start writing
code before hardware exists, and ASIC designers can use embedded software
code as stimulus for simulation.
Since Simulation Technologies already offers Verilog debugging and code-coverage
tools, V-CPU is a natural extension of its mission, said president Rich
Davenport. The product was developed and used initially by Cisco Systems
Inc., and has been purchased and upgraded by Simulation Technologies. It
was described in a paper given by Cisco engineers at last year's Design
Automation Conference.
On the software side, V-CPU supports host-mode execution, in which software
is compiled for a host platform and analyzed with a debugger, or target-mode
execution, in which software is cross-compiled for the target and analyzed
with an instruction-set simulator. The host mode yields faster performance
but less detail.
V-CPU also provides C language CPU resource models and high-level functional
models. The idea here is to model as much as possible outside the HDL simulator,
which slows the process considerably. On the hardware side, V-CPU uses
bus-functional rather than full-functional HDL processor models.
At present, V-CPU supports MIPS, PowerPC, Motorola 68000, X86, ARM and Intel
i960 CPUs, as well as the PCI bus. Simulators supported are Cadence's Verilog-XL,
Viewlogic's VCS and Avant!'s Polaris. V-CPU is now available for $40,000
on Unix platforms.
Compared with Mentor and Viewlogic products, Davenport said, V-CPU requires
fewer changes to embedded software, and has had longer use in production
due to its Cisco heritage.
Though V-CPU promises speeds many times faster than logic simulation, hardware
emulation gets even closer to real-time speeds. With Avatar, Ikos is seeking
to propagate "clones" of its VirtuaLogic emulation system, without hardware
debugging features and at a lower cost, to embedded software designers.
"We eliminate anything that has a hardware look and feel," said Larry Melling,
vice president of marketing at Ikos. "This should look just like a workstation
to the software designers." Otherwise, all the specifications are the same,
including the 500-kHz to 2-MHz speed range and capacities of over a million
gates.
Producing reliable replicates is harder than it sounds, said Melling. "Our
system is 100 percent reproducible," he said. "Once we load it into a VirtuaLogic
emulator we can load it onto 100 Avatars and it will work, guaranteed." Other
emulator vendors can't match that pledge, Melling said, because their software
doesn't account for timing differences that occur even between identical
FPGA devices. Further, he noted, Avatar has the same performance and capacity
as VirtuaLogic.
Companies can buy any number of Avatars but must have at least one VirtuaLogic
for the initial download. Avatar incremental pricing is around 40 cents/gate
beyond the base configuration, compared with 56 cents/gate for VirtuaLogic
and more than $1/gate for many competing systems.
Avatar is available now starting at $100,000 for 200,000 usable gates. The
product has already been used in the design of a 3-D multimedia accelerator
from Nvidia Corp. (Sunnyvale, Calif.).
In addition to new coverification products, IVC-VIUF attendees will hear
competing performance claims from the two primary vendors of Verilog simulators,
Viewlogic and Cadence. Viewlogic's VCS 4.0 is the first version of that product
to incorporate Roadrunner, previously announced as a standalone Verilog
simulator.
Designed to compete with cycle-based simulation, Roadrunner offers a two-state
simulation option and claims to provide efficient evaluation of sequential
logic, event reduction between sequential logic blocks, global data-flow
optimization and more comprehensive module and function in-lining. The result,
said Ghulam Nurie, Viewlogic's director of ASIC-verification marketing, is
a two- to threefold increase in speed over the current VCS 3.1.
Also new with VCS 4.0 are negative timing checks, which are important for
deep-submicron designs. VCS 4.0 will ship in June on Unix platforms for $40,000,
with no extra charge for Roadrunner.
Viewlogic and Cadence have an intense rivalry over simulation speed, and
Nurie said VCS 3.1 is already three times faster than Cadence's NC Verilog,
and uses four times less memory. But Dave Kelf, Verilog product manager at
Cadence, has customer benchmarks that claim NC Verilog 1.0 runs as fast as
VCS 3.1, and the new 1.1 release claims to double simulation speed.
More significantly, NC Verilog is adding timing features required for ASIC-vendor
sign-off, such as Standard Delay Format (SDF) annotation. That means the
simulator now has all the timing features of Cadence's widely used Verilog-XL
product. NC Verilog costs $40,000 on Unix platforms; version 1.1 is available
now, and version 1.2 is due to arrive in June.
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
Welcome 12-Inch Wafer
Industry readies fabs for next-generation wafers
By Brian Santo, EE Times
Portland, Ore. -- Despite tepid support in 1996 for its emergence,
the shift to 300-mm (12-inch) wafers is moving slowly but inexorably forward.
As many as eight 300-mm production fabs are tentatively planned to come online
by the end of 1999, according to Semiconductor Equipment and Materials
International (SEMI).
Meanwhile, a "philosophical" difference between Japan and the rest of the
world on the subject of wafer-handling could become a serious snag once
manufacturing-equipment suppliers start developing products.
Causing a problem right now is the lack of device-grade 300-mm wafers, which
is holding back development, reported SEMI's Global 300-mm Initiative director
George Lee last week.
Lee said that SEMI compiled confidential information from IC manufacturers
that indicates as many as nine pilot lines might be in operation by the end
of 1998 and another five by the end of 1999. Two low-volume production lines
based on 300-mm wafers and equipment might be ramped by the end of 1998 and
five low-volume and one medium-volume fab by the end of 1999.
Despite the intention of IC companies to shift from 200-mm (8-inch) wafers
to 300-mm wafers, several problems could impede an orderly transition. Japanese
IC manufacturers are investigating a wafer-handling scheme that differs from
the one favored by chip makers elsewhere. Japanese companies are looking
at using an open cassette, while their competitors want to keep wafers in
an enclosed environment.
The ramifications are serious enough to prompt the Japanese to be willing
to set up their own 300-mm evaluation and development support corporation,
Selete (Yokohama), rather than join the I300I (International 300 Initiative;
a Sematech spin-off in Austin, Texas) in which most of the other IC companies
are participating.
In essence, Lee explained, the different approaches are the result of beliefs
about the major sources of microcontamination. It is unlikely that there
will be any effect on big-ticket machinery such as lithography and deposition
systems, but the rift might result in an inefficient split in the equipment
industry in terms of carriers, transportation systems such as overhead loaders,
and the configurations of the vacuum load locks on manufacturing equipment.
Lee said that another point of international contention--where to put load
locks--was recently settled. Semiconductor manufacturing equipment will be
side-loading.
Mechanical-grade 300-mm wafers are available but still very expensive, Lee
reported. Used primarily for equipment evaluation, the quantities that may
be obtained have doubled from 100 to 200 per month to 200 to 400 per month.
There is a small number of device-quality wafers available anywhere.
Furnace-quality wafers are just now being delivered to some equipment companies.
"The problem for the transition from 200-mm to 300-mm is now affordability,"
Lee said, noting that I300I is actually reprocessing used evaluation wafers
in an attempt to reduce per-disk evaluation costs.
Another possible impediment is the question of whether suitable lithography
tools for evaluation purposes can be delivered on time. Selete is reported
to have ordered a deep-UV stepper from Canon. I300I expects to have a similar
machine (from an as-yet unidentified manufacturer) installed by the end of
this year.
The industry slowdown is also having a negative effect. Individual companies
have been forced to reassess their capacity needs and long-term plans. Though
confidential SEMI reports indicate that several pilot lines will be up and
running in 1998, and a number of fabs will be online in 1999, Lee said pilot
production is more likely for late 1998, "and full fab production is likely
to be pushed back to 2000." Nevertheless, the total number of 300-mm pilot
lines and 300-mm fabs seems to be holding steady as some companies call off
plans to build new 300-mm fabs, and others decide to go ahead with them.
Siemens, Intel, NEC and Mosel-Vitelic are among the companies that have publicly
discussed moving to 300-mm wafer fabs.
Ron Dornsief, principal analyst at Dataquest, said that the number of new
fabs being planned--200-mm and 300-mm-- is too high. Plans call for 47 new
manufacturing facilities, according to Dataquest, but the sustainable number
is 33. Meanwhile, the industry faces a continuing downturn and an increasing
excess in capacity. Dataquest projects it will take up to two years for the
capacity situation to stabilize.
Dornsief described 1997 and 1998 as "pause" years. The slowdown in the chip
business does seem to have one salutary side effect on the move to 300-mm
wafers, however, in that it is allowing equipment manufacturers to redirect
resources from 200-mm products to 300-mm development.
Lee said that equipment manufacturers expect that the price of individual
production systems for 300-mm wafers may go up 25 percent to 40 percent in
comparison with systems that handle 200-mm wafers. Fab size is likely to
increase, along with fab power and water consumption.
Applied Materials earlier this year announced a rapid thermal processing
system capable of handling 300-mm wafers, but Lee said he was not aware of
any orders for machinery capable of handling wafers that size.
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
EDA Rumble in Japan
Cadence and Avant! figure in Japan tangle, too
By David Lammers, EE Times
Yokohama, Japan -- While Avant! and Cadence Design have battled bitterly
in the U.S. courts, they have also been entangled in Japan in a bizarre business
relationship.
It is a story--told by knowledgeable sources still working at the companies
involved who wish to remain unnamed--that begins in 1986, when the semiconductor
market crashed and Minoru (Larry) Yoshida was ousted from the presidency
of Tokyo Electron Ltd. (TEL), the second-largest semiconductor equipment
company worldwide and a distributor for dozens of foreign semiconductor and
technology companies.
Seeking vindication, Yoshida set up Innotech Corp., a Yokohama-based trading
company, and vowed to create a company bigger than TEL. Well-connected with
the top Japanese semiconductor managers, Yoshida began selling the IC design
software developed by SDA Corp., the progenitor of Cadence.
After mergers with Ecad Corp. and then Tangent, fast-growing Cadence stuck
with Innotech, and for good reason. Yoshida, tapping his contacts, took Cadence's
sales from $4.5 million in 1987 to $29 million in 1989. More importantly,
he created a team of EDA-savvy sales and support engineers which now exceeds
70 people.
As Cadence's sales in Japan grew, Yoshida began to worry. Distributors who
do poorly are replaced. But distributors who do well, in Japan and elsewhere,
also put themselves in a precarious position, because the OEM may "go direct,"
set up its own sales force, and leave the distributor pounding sand.
In Japan, an employer has social responsibilities to employees, and firing
a large number of people is unacceptable. Though Innotech handled all of
Cadence's IC CAD tools in the Japan market (leaving Cadence's Japan subsidiary
with the system-design tool sales and other responsibilities), people wondered
how long Cadence would stand by Innotech.
Yoshida, according to several sources, in September 1994 opted to protect
himself by secretly setting up a separate company, named Design Solutions
Inc. (Yokohama), which immediately took on Avant! as its main customer. A
core group of Innotech's technical sales managers moved from Innotech to
Design Solutions. Tsuneo Suzuki, who ran the Japan operations of Mentor Graphics
for about 10 years, was hired to head Design Solutions.
Those who asked Innotech managers about the Innotech relationship to Design
Solutions were answered with a wink, a wave of the hand and the smiling response,
"Don't ask."
It is almost certain, these sources say, that Design Solutions was the creation
of Innotech, which was prevented from taking on the Avant! line because it
conflicted directly with Cadence.
Suzuki said "there have been a lot of rumors like that going around." He
said Naoshi Sugiyama, the second-ranking executive at Innotech until recently,
helped Design Solutions find clients "because he was my friend for a long
time." Suzuki said several technical people did join Design Solutions at
the outset. He said the financial backing of Design Solutions came not from
Innotech, but from "friends."
Gerry Hsu, during his years in charge of Cadence's IC CAD operations, visited
Japan often and formed close relations with many people at Innotech. He also
befriended a number of Japanese IC CAD managers, who admired his "can-do"
approach to technical problems. Hsu would personally make sure that the
complaints and suggestions from Japan received top priority within the Cadence
bureaucracy.
Hsu decided he needed "a personal emissary" to his Japanese business associates.
He put an advertisement in a newspaper and hired Noriko Ando, a former airline
stewardess. Hsu decreed that Ando would be the person to deal with Toshiba
Corp.
After she went to Toshiba to announce herself as Cadence's representative,
an astonished Toshiba engineering manager called Cadence Japan to ask what
was going on.
The manager in charge of Cadence's business at Toshiba phoned Joseph Costello
and threatened to quit. Costello managed to smooth--temporarily--the ruffled
feathers, both at Toshiba and at Cadence Japan. But significantly, Ando stayed
on the Cadence payroll, and Hsu continued to deal directly with Cadence's
key Japan customers, using Ando as his information conduit.
When Hsu quit Cadence to form Avant!, Ando was put in charge of Avant! Japan,
and supervised the two main distributors: Design Solutions, which had grown
to about 20 people, and Marubeni Hytech, a much larger, Tokyo-based electronics
trading company affiliated with giant Marubeni Trading Corp.
Having two distributors in the Japan market is not unusual, but dividing
up the sales territory is tricky nevertheless. Hsu, working through Ando,
managed to play one distributor off against another. Design Solutions handled
sales to Sharp, NEC, Rohm and Matsushita (believed to be Avant!'s biggest
customer, signing a multiyear contract last year that was cited in Avant!'s
stock offering information).
Marubeni Hytech handled most of the other major accounts, but on those occasions
when Hsu was unhappy with sales at one account or another, he would give
the order to Ando to yank the account from Marubeni Hytech and give it to
Design Solutions, or vice versa.
The distributors were instructed to talk only to Ando, and not to Hsu directly.
Avant!'s sales grew rapidly, to about $30 million in 1996, and revenues are
expected to increase sharply this year as well. Engineering managers who
shake their heads at the unusual business practices also are quick to say
that Avant! has excellent place-and-route software.
Last September, Avant! went on its own acquisition binge, taking over MetaSoft,
Anagram and Frontline, each of which had its own network of distributors
in Japan. Shortly thereafter, Hsu came to Japan and announced that all of
the distributors would be rolled into one. The new company recently started
operations in Yokohama as Maingate Electronics Corp., with Ando in charge.
Hsu took the unusual step of asking each of the existing distributors to
invest in Maingate, and to transfer the technical support people who had
been handling the Avant! products to Maingate as well.
One person involved with the changeover said that "back in October, people
doubted that Hsu would actually do it. They didn't believe him. But he has
a very strong management style. And in a way, the relationship he has with
Ando is working well for him. She doesn't make her own decisions, and she
is an excellent translator of information back to Gerry himself. If she was
supposed to make decisions on her own, it wouldn't work, but as it stands
now he has good control over everything through her."
Not that everyone is happy. Marubeni Hytech lost 10 sales engineers to Maingate,
and five people left Design Solutions, which derived most of its revenues
from Avant! sales.
But in the modern-day business world, where survival takes precedence over
loyalty, the entire affair is viewed with a surprising amount of equanimity.
Yoshida, who amassed great personal wealth as he built Innotech into a
diversified trading company with several hundred million dollars in annual
revenues, is on the verge of signing a multiyear contract with Cadence that
will tie the two companies even closer together.
One source said "you have to understand Larry's position. He had a large
number of people on his payroll, Cadence sales were getting big enough in
Japan to go direct, and so Design Solutions was set up partly to cover his
rear.
"Japan is not the kind of place where you can lay off 70 people with a few
weeks' severance pay. And if you look at the history of how foreign companies
have treated their distributors here, it is completely understandable."
Joseph Costello, chief executive officer of Cadence, was kept informed of
the situation as it developed but took a pragmatic approach as well. Getting
confrontational with Yoshida could have resulted in losing a valuable group
of technical sales people in Japan, where non-Japanese companies have difficulty
hiring engineers.
The pending agreement between Innotech and Cadence, according to reliable
sources, includes provisions that will put an end to the practice of referring
Cadence rivals to Design Solutions. And revenues at Cadence are growing rapidly
as Costello's gamble on the consulting business has paid off handsomely,
with Japanese customers gradually warming to the idea.
Suzuki, the president of Design Solutions, said the loss of the Avant! line
to Maingate Electronics is "a headache, but I am not particularly upset about
it. When companies get to a certain size they want to set up their own sales
subsidiary. I did the same thing when I worked at Mentor, and trading companies
can't avoid these kinds of changes. Now we at Design Solutions just have
to go out and look for new business, and in the fast-changing world of
electronics, we will find it."
Another source with direct knowledge of the situation said the people at
Design Solutions need not worry too much about their paychecks. Innotech,
the company which helped set up Design Solutions two and half years ago,
will continue to support them until new lines of business are developed,
this source said.
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
OnTrak On the Lam
Lam buys OnTrak to secure foothold in wafer polishers
By Margaret Ryan, EE Times
Fremont, Calif. -- Lam Research Corp. disclosed plans to acquire OnTrak
Systems Inc., a supplier of wafer-polishing equipment, in an exchange of
stock valued at $225 million.
For Lam, the acquisition creates a foothold in the chemical mechanical
planarization (CMP) cleaning systems market, the most rapidly growing segment
of the $21 billion semiconductor-equipment industry. The CMP market is expected
to grow at a rate of 31 percent through 2001, according to Dataquest Inc.
(San Jose, Calif.).
The deal also calls for OnTrak Systems chairman and chief executive James
W. Bagley to become Lam's chief executive officer and to join Lam chairman
Roger Emerick in the newly created Office of the Chairman. Bagley had been
chief operating officer and vice chairman of Applied Materials Inc. before
joining OnTrak in June 1996; with Lam, he will take the helm of one of Applied's
top rivals. Way Tu, president of Lam, will report to Bagley.
In a separate announcement last week, Lam said it would report a loss of
about $1.40 per share, or about $42.8 million, for the quarter ending March
31. The loss is related to customers' pushing back orders for equipment and
to a faster-than-expected move to the next generation of etching tools; the
shift left Lam with an excess inventory of older products.
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
Battery Talk
Portable debate turns to batteries
By Ron Wilson, EE Times
Santa Clara, Calif. -- Batteries and battery management pushed the
envelope at Portable by Design, as vendors described several new chemistries,
each of which could make important contributions to portable equipment.
Additional papers discussed the latest in battery management technology,
often based on the Intel/Duracell Smart Battery Data Specification.
The shift from NiCd and Ni-metal-hydride batteries to lithium-based cells
was virtually a given for the conference. One presenter claimed that nearly
all the new notebook computers introduced in 1997 would use some sort of
lithium battery system.
But the new lithium cells, while offering advantages in operating life for
the space and weight, have brought their own problems, some papers warned.
The cells require careful supervision by a microcontroller or dedicated battery
management IC to achieve their potential. In the extreme, intelligent battery
management is necessary in lithium-ion batteries to prevent fire or explosion.
In answer to this challenge, papers discussed a host of battery-management
strategies based around the Intel System Management bus (SMbus) and the SBDS.
Dedicated ICs, such as those described by Benchmarq Microelectronics Inc.,
can be integrated into a battery pack to provide a full range of management
features, disconnecting the battery from an overcharge, protecting it from
exhaustion, controlling the charging process in conjunction with a smart
charger and providing an accurate capacity gauge.
Other papers, notably one from Upal Sengupta, principal engineer at Rayovac
Corp., argued in favor of a microcontroller-based rather than dedicated approach.
Sengupta pointed out that microcontroller firmware could be altered to take
into account the needs and behaviors of multiple battery chemistries. Even
within a particular group, such as lithium batteries, the charging and discharge
characteristics can vary widely.
Whatever approach was taken, there was general agreement that safety and
accurate capacity measurement were vital. The safety issue spoke for itself,
particularly in light of Li-ion batteries' ability to grow tiny slivers of
metallic lithium, which could then short out a cell, resulting in fire.
But capacity measurement emerged as an equally important issue. Speakers
from SystemSoft Corp. pointed out that with typical battery measurement systems
today, capacity is only measured within about 10 percent. By having a more
accurate knowledge of the remaining energy in the battery, and by carefully
choosing a power-management scheme in the computer to best use the remaining
power, the paper estimated that operating life could be extended by as much
as 20 percent with the same hardware.
But the greatest changes in portable equipment life may come from the batteries
themselves. Despite the current wave of enthusiasm over Li-ion batteries,
several competing technologies--some recent and some not yet ready for
production--spoke their piece. They focused not only on energy density but
also on manageability and safe disposal.
Perhaps the most familiar of the competing technologies was RAM--a rechargeable
alkaline battery developed by Battery Technologies Inc. and marketed by several
major companies. RAM cells, according to a paper, offer discharge rates in
the hundreds of milliamps, hundreds of useful recharge cycles under modest
operating conditions and extremely low self-discharge. They are used primarily
to replace conventional alkaline cells in consumer applications, but they
also have OEM uses.
A more unusual approach came from AER Energy Resources Inc., which described
a unique Zinc-air battery design: the cell essentially respires, giving off
oxygen during charging and absorbing it during discharge. A Zinc-air battery
uses a combination of air tubes and a fan to provide or remove air when the
battery is in use and to isolate the battery from air when it is not in use.
Zinc-air technology, according to AER, offers much lighter weight per Watt-hour
than even Li-ion batteries, at about the same volume. The batteries are also
potentially much less expensive than lithium solutions, and can be disposed
of as household waste. This latter point is important because, unlike Li-ion
batteries that provide hundreds of recharge cycles in even heavy operation,
Zinc-air batteries are reduced to half capacity by about the 50th deep discharge.
AER marketing manager Tim Cutler estimated that in practice, the Zinc-air
battery in a notebook computer would have to be replaced every couple of
months.
Other new and promising technologies used lithium ions, but in different
ways. Li-ion polymer batteries, discussed by Powerdex division of Gould
Electronics Inc., offer essentially the same performance as liquid-electrolyte
Li-ion batteries. But because the electrolyte is a flexible polymer instead
of a liquid, cells can be made in virtually any shape, from millimeter-thick
sheets the size of a notebook-computer LCD panel to complex curved surfaces.
Another emerging technology came from Battery Engineering Inc. The company
has devised a Li-ion cell in which both anode and cathode are carbon, and
in which the electrolyte is an inorganic material containing sulfur dioxide.
A sacrificial electrode of metallic Lithium is placed in the cell, and over
the first few operating cycles, the metal is transported into the carbon
electrodes.
This gives the cell the remarkable characteristic of actually increasing
its energy capacity during the first 100 or so cycles. The cell shows an
initial capacity of about 90 mA-hours, which degrades only slightly over
250 cycles.
More important, the cell exhibits one very valuable characteristic of previous
sulfur-dioxide/lithium cells--it is chemically resistant to overcharging.
Cells overcharged to many times their capacity would show no abnormal behavior
on discharge. But previous sulfur-dioxide cells had the unfortunate habit
of exploding violently during operation. Replacing the lithium anodes in
these cells with the carbon anode, and inserting the sacrificial lithium
plate, eliminated the growth of Lithium dendrites and the explosions.
The new battery technologies promise greater energy density, greater safety,
more flexible packaging and less need for battery management. But none offers
an enormous increase in operating time for portable equipment. As one speaker
commented, "We have made big advances in energy density. But the systems
vendors will find ways to use more current, so the operating hours have stayed
about the same."
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
Oracle NCs
Oracle rolls out StrongArm X86-based network computers
By Rick Boyd-Merrit, EE Times
Maynard, Mass.-- Oracle Corp. will drive its concept of the network
computer (NC) forward at its Open World conference in Japan next month when
it launches two NC reference designs: one based on Digital Equipment Corp.'s
StrongARM implementation and another on the Intel X86.
Though Oracle is keeping the X86 design under wraps until the April 16 event
in Tokyo, it went public with the StrongARM reference platform last week.
It's pitching the Digital Network Appliance a more powerful, more
standards-complaint upgrade of the ARM 7500FE-based reference design, created
by U.K.-based Advanced RISC Machines Ltd. and unveiled by Oracle in the summer.
Oracle claims 11 companies licensed and built systems around that earlier
design but that the reference platform's relatively low performance and
ARM-specific nature limited its appeal.
The 7500FE "had everything from a 28.8 kbit/second modem to 25-Mbit/s ATM
interfaces, but they were not industry-standard interfaces," said Jim Lynch,
vice president of business development for Network Computer Inc. (NCI), an
Oracle spinoff that designs NC software. Systems based on the 7500FE, Lynch
said, were suitable for "low-end browsing and light applications" on consumer
NCs but not for "substantial Java applications " in a business environment."
In that earlier reference design, the 33-MHz 7500FE integrated chip acted
as central processor, video accelerator, memory manager and I/O controller.
It ran ARM's RiscOS, used ARM-specific firmware and employed a unified memory
architecture that relied on systems memory for frame buffer chores.
By contrast, Digital's design uses the 233-MHz StrongARM and supports PCI
and ISA peripherals, including Iomega's Zip drives, which have proved popular
in some NC designs. The reference platform runs a version of public-domain
NetBSD Unix that has been tailored for NCs by NCI and uses firmware compliant
with the Open Firmware Initiative. It contains 8 to 64 Mbytes of system memory
plus a separate Cirrus Logic graphics chip supporting both TV and SVGA displays
and its own frame buffer.
Bill-of-materials costs for the Digital design are said to range from $150
to $250, depending on the configuration.
Digital Equipment will provide the full electric and mechanical specification
for the design free as part of its Strong-ARM sales program. The company
intends to rely on outside manufacturers to build the StrongARM NCs; those
companies, in turn, will source through Digital's systems business.
"We want to have these NCs built by a company whose business is high-volume,
low-cost manufacturing," said Bob Supnik, vice president of research and
advanced design for Digital. Supnik said his group worked with NCI for nine
months on the StrongARM NC.
He said the business model is similar to one employed by IBM Corp. in that
IBM makes and sells a variety of servers, PCs and networking gear but sources
its IBM Network Station from terminal maker NCD.
On the software side, NCI's April launch in Tokyo will include its NC Access
software suite, a bundle of systems and applications-level programs designed
to act as a complete NC environment. Based on NetBSD, the package includes
Oracle's own Web browser, e-mail client and productivity applications as
well as support for Java.
NetAccess is expected to ship in volume later this year, Lynch said.
Lynch positioned NCI's switch from its hardware dependence on ARM to its
relationships with Digital and Intel as a planned upgrade and expansion.
"We selected the ARM environment in part because of the StrongARM," Lynch
said. "When we started talking to ARM, we also started talking to Digital.
The reference design from Digital is part of an evolution.
"We will have an Intel reference design as well [at the April event]," Lynch
said. "And we will support other hardware architectures as our users demand
them."
Two companies have already announced plans to build network computers based
on the Digital StrongARM design. Funai Electric Company Ltd. (Osaka, Japan)
will build consumer-oriented devices; Aranex Inc. (Bedford, Mass.) will build
business-class NCs.
Both companies had licensed the earlier, ARM 7500FE-based reference design.
(c) 1997 CMP Media, Inc
[This article comes from EE Times in a joint cooperative effort
with the Motley Fool. For more articles like it, please look at Fool's Gold
every weekend or simply go to the Fool's Gold Mine and page through our back
issues, which all have clever and cool EE Times articles in
them.] |
|