For the PDF version of this article, click here.
In previous issues of Power Electronics Technology, a series of articles chronicled the evolution of passive components, magnetics, power ICs and discrete power semiconductors over the past three decades. With those past developments providing some perspective, contributing editors Gene Heftman, Steve Grossman and John Day now examine how each of these component areas are expected to advance in the next five years.
This time frame is short enough to allow for some realistic predictions based on the ongoing development trends, yet just far enough out to allow us to ponder some very promising, emerging technologies. For power electronics designers, the road ahead contains both exciting possibilities such as the opportunity to exploit ever-more sophisticated power ICs, and challenges such as the loss of design experience in critical areas such as magnetics.
— David Morrison, Editor, Power Electronics Technology
Semi Processes Push Power ICs Forward
Power management ICs face mounting pressures from all sides of the power supply application spectrum: they must provide higher efficiency, reduced cost, smaller circuit-board area, greater functionality, smaller packages, greater integration and new functions. In addition, power architectures are changing to meet the demands of advanced digital devices such as ASICs, DSPs and microprocessors. The distributed power architectures used to drive such electronics will require power ICs having both isolated and nonisolated topologies. These ICs will be low-profile, small-footprint, high-efficiency power modules with postregulator controllers and multiple PWM controllers on a single chip.
The way to manufacture such power ICs is to bring together analog, digital and power semiconductor technologies on a single chip to carry out the functions of a power control IC. This is sometimes called smart power, where BiCMOS is used for analog functions such as oscillators, drivers, amplifiers and voltage references, CMOS for control logic and DMOS for the power switches.
This type of technology is currently being developed by companies such as National Semiconductor, which calls it voltage-scaled ABCD for analog BiCMOS-DMOS. It will be available in three different “recipes” or processes to meet various power supply requirements. For example, depending on the system voltages the IC will handle, the breakdown voltage of the DMOS transistors will vary, as will the isolation breakdown voltages of the substrate (Fig. 1).
If the application is low-side gate drivers or buck regulators with an internal 45-V MOSFET, then a minimum 45-V breakdown process will suffice. For a 100-V half-bridge gate driver, a recipe with 100-V minimum breakdown and isolation will be required. The idea is to develop a designer's tool kit that can handle any combination of power supply system requirements to produce a suitable device for the application.
The broad challenge of smart power is to integrate intelligence, analog and power functions in such a way as to optimize the power control IC. “The winning solution,” says Mike Briere, executive vice president of research and development for National Semiconductor, “is a chip with the highest density, highest efficiency and lowest cost.” To reach this goal, designers will have to weigh putting more intelligence on the chip while settling for lower power capability, or getting more power through simpler processing to achieve lower cost but sacrificing functionality. These are the kinds of tradeoffs designers will face in the coming years, but semiconductor manufacturers are developing the tools to handle them.
— By Gene Heftman, Contributing Editor
Discrete Power Devices
“The silicon we use today is running out of gas.” That's the view of Alex Lidow, CEO of International Rectifier, who believes that to make sensible predictions about what's ahead requires sizing things up from two different viewpoints — first from the point of view of new substrate materials and then emerging circuit topologies (Fig. 2).
As Lidow points out, new materials, such as silicon carbide (SiC), are already expanding the performance capabilities of discrete power devices, and SiC has become a player in both diodes and MOSFETs. As to the other semiconductor materials — gallium arsenide, gallium nitride and the like — we will have to wait and see.
In support of the second viewpoint, emerging circuit topologies, Lidow says that the performance of discrete devices is becoming more heavily intertwined with the way they are driven. Here we can cite a number of emerging innovations — single-cycle control, matrix converters, bidirectional switches — some waiting in the wings, perhaps, but others will clearly move to center stage — and quite soon.
Lidow then expands his discussion to bring in applications: “This intertwining will become a focus of extreme importance as we move towards the much higher power densities that are crucial in very low-voltage, high-current densities applications such as IT [information technology]. Also, we can be sure that new topologies will emerge, dramatically reducing the cost equation in lighting and in motor drives for ground-based applications such as hybrid cars, white goods and the like.”
With regard to motor drives, there will undoubtedly be major advances, according to Michael O'Neill, an applications engineer at Cree. As he points out, it has already been demonstrated that SiC, 3-hp motor drives employing 15-A IGBTs, with SiC schottkys as antiparallel diodes switching at 16 kHz, can slash overall inverter losses by 33%.
“Right now, the IGBT may still be the king in the motor drive realm with regard to efficiency from one-half to 1000 hp, but eventually it will be supplanted by the SiC MOSFET,” predicts O'Neill. “This is likely to occur sometime in 2006.”
In the past, the race has been to get the best on-resistance (RDS(ON)). But whatever doubts there are about the future, one thing seems certain: The best switching performance will not be measured by RDS(ON), but instead by RQSWITCH, the resistance of the component multiplied by the charge it takes to switch it. Because from the standpoint of switching, the gate must be viewed as a capacitor that must apply sufficient charge to change the state of the capacitor, thereby crossing a voltage threshold that will turn the MOSFET on or off.
According to B. Jayant Baliga, inventor of the IGBT and currently a professor at the North Carolina State University, continuing innovations in silicon 20-V to 30-V power MOSFET cell design for both planar and trench-gate structures are expected to result in improvements in the RQSWITCH from about 20 for the current generation of products to less than five, over the next five years.
— By Steve Grossman, Contributing Editor
Magnetic Technology Needs a Face Lift
Despite tremendous advances in magnetic technology over the years, and the need to push ahead with improvements, there is an uneasiness underlying the future of the technology among industry insiders. One reason is that power magnetics — the design of transformers and inductors — is likened to a black art, knowable only to a select few who choose to master its details. According to Ed Bloom, president of magnetics consultant EJ Bloom Associates, “We're losing our capability in magnetics design across the world. Of all the aspects of analog design, magnetics lags farthest behind because power supplies are relegated to the end in most system designs.” Such thinking could have an adverse impact on the future of magnetics technology in this country.
It begins in the universities, where magnetics is neglected from an educational standpoint, with few opportunities for young engineers to learn power magnetics design. Another problem is that the basic design and specification of power transformers has remained the same over many years; the construction of transformers has changed, but much of the trial-and-error aspects of finalizing a design have not. The number of different designs that come out of a specification will be as varied as the number of manufacturers who build a part to that specification.
Paul Leibman, vice president of marketing at transformer and inductor manufacturer Coilcraft, says, “The main transformer of a switcher is not easily modeled, so a lot of traditional procurement still goes on. Make me a sample and let me try it out just like years ago.”
Even in this day and age of computer software as the ultimate design tool, magnetics design programs have had less impact on advancing the state of the art than had been hoped for. In Bloom's opinion, “Modern software is not a big help, besides which it can be expensive and is not very user friendly. At $20,000 and up, it's not a worthwhile purchase.” Adds Leibman, “Power transformers are very difficult to design from a program. Engineers have to interface heavily with the manufacturer on transformer designs.”
So, what is the future of power magnetics design in the coming years? In the view of Victor Quinn, CTO at Tabtronics, transformer manufacturers and power supply makers must raise the stakes in producing components. “Optimization methods have been a consistent theme in the literature for magnetic components, but high degrees of optimization are not prevalent. Domestic manufacturers must produce more innovative optimized designs that provide better functionality, not simply lower prices,” says Quinn. Cost-reduction efforts will only turn the domestic base into contract manufacturers, and the growth of offshore sourcing will further weaken the magnetics industry.
— By Gene Heftman, Contributing Editor
Passive Components: More than Miniaturization
Miniaturization is major for manufacturers of passive components, but the trend toward tiny is far from their only concern. Other factors include RoHS compliance, the need for new materials and packaging, and emerging market opportunities, including the integration of components into modules.
“Miniaturization in cell phones has driven engineers to use smaller and more sophisticated passive components for power supply designs, as well as for RF,” says Gerry Hubers, product manager for microwave and optical components at Murata Electronics North America. Passive component manufacturers have also had to change the way they do business. “Ten years ago, a company would take full responsibility to design, test and produce a product,” Hubers recalls. “Today, design is in one location, procurement in another and production in still another.”
Hubers adds that the integration of various wireless technologies is creating requirements for new core materials, packaging technologies and manufacturing processes.
AVX Applications Manager Chris Reynolds cites the “green” trend, dictating the use of lead-free and RoHS-compliant products, as “perhaps the biggest noise in the market today.” He said manufacturers of aluminum electrolytics and some low ESR tantalum polymer capacitors are struggling to meet specific targets, such as 3x reflow cycles at 260°C. “As a result,” Reynolds says, “the move to new components, many with new part numbers, or to new technical solutions, will continue to be a major and costly headache for equipment manufacturers.”
Another growing trend, in Reynolds view, is the need for higher-performance passives that offer enhanced parametric performance to achieve higher system frequencies in power supply and other designs. “Low ESR or low ESL capacitors are in high demand for power products,” he says. “Increasingly, high Q inductors and precision resistors or filters are made on wafers and offered in SMD packages, replacing older thick film or wire-based designs and leaded packages.”
Manufacturers are consuming large volumes of 0402 MLCCs and even some 0201 size devices, according to Reyolds. AVX recently introduced an 0402 tantalum capacitor that offers high capacitance in a miniature case size (Fig. 3).
Emerging markets are also providing opportunities for passive component manufacturers. “With the unprecedented increase in fuel costs, and the recently experienced limitations on the power grid in the United States, the need for power-factor correction and power quality improvement is getting considerable attention,” notes Achim Buecklers, president and CEO of EPCOS. “Installation of power-factor correction capacitors is no longer a choice in the U.S.,” he says. “More and more utility companies are either penalizing industrial consumers for power factors below 0.9 or offering incentives to those who maintain it at the desired level.”
— By John Day, Contributing Editor