Power electronics designers look to the future
Moving parts on military and aerospace equipment are relying more than ever before on electric power, rather than on hydraulics and on other motion technologies.
By John Keller
Moving parts on military and aerospace equipment are relying more than ever before on electric power, rather than on hydraulics and on other motion technologies. In addition, military, aviation, and space equipment yearly increase in complexity and functionality, which intensifies pressure on engineers who design military power systems.
It is this increasing reliance on electrically driven motion, as well as this technology's corresponding creeping complexity, that is giving rise to broad new technology initiatives in the electronic components that condition power and control motion.
Many different electronic components are involved in motion and power conditioning. There are metal oxide silicon field effect transistors (MOSFETs), for example, which are semiconductors used for switching applications in which an input voltage controls the output current. Rectifiers allow electric current to flow in only one direction, and function as one-way valves. A bipolar transistor amplifies by producing an output signal that is greater than its input signal. Insulated gate bipolar transistors (IGBTs) combine the low-saturation and high-breakdown voltages of bipolar transistors with the high impedance of MOSFETs. IGBTs normally do not switch power as quickly as MOSFETs, but are generally less expensive.
Other components also are involved in power management and motion control. Thyristors are solid-state switches that control current in high-voltage applications. Transient voltage suppressors protect against electronic overstress and can reduce equipment failures. Zener diodes control, limit, or regulate circuit voltages. Many other analog components also are involved, and are following a path similar to the increasing complexity of their electronic cousins in digital applications.
At a higher level of integration than power semiconductors are power integrated circuits (ICs) and power modules. Power ICs, such as pulse width modulators, quad comparators, regulators, voltage references, FET drivers, and bridge drivers, can interface the semiconductors that condition power with digital-control circuitry to automate power-control tasks. Power ICs, for example, can help provide automatic surge protection, can automatically control torque, or provide automatic shutdown when temperatures exceed set parameters.
Engineers who design power-related integrated circuits and discrete semiconductors are concentrating on reducing device size and weight and increasing electric switching speed, as well as on developing new packaging techniques to increase efficiency and reduce heat.
At the same time, pressure to increase the use of commercial off-the-shelf (COTS) components in military and aerospace electronic systems is encouraging systems designers to consider where it might make sense to substitute advanced relatively inexpensive commercial-grade components where in years past they might have used military-grade components for the same functions.
Electronics for commercial, military, and aerospace systems are becoming increasingly complex as designers ask these systems to perform ever-more tasks. The most complex electronics involve commercial systems such as cellular telephones, personal digital assistants, and laptop computers, yet military systems designers rely more today on commercial components than they ever have before.
"Complexity is driven by demands for high power and more functionality," comments John Catrambone, vice president of the International Rectifier (IR) Hi-Rel Products division in Leominster, Mass. The IR Hi-Rel division specializes in ruggedized MOSFETs, rectifiers, IGBTs, multichip modules, DC-DC converters, regulators, drivers, operational amplifiers, and related products. About 70 percent of the division's revenue comes from value-added multichip modules, and 30 percent from discrete devices, Catrambone says.
Experts at power discrete semiconductor manufacturer Vishay Siliconix Inc. of Santa Clara, Calif., continue to shrink the size of their junction field effect transistors (JFETs) and MOSFETs for cell phones and portable appliances, says James Crozier, military product line manager at Siliconix.
Packages for the company's MOSFETs and JFETs have moved from the LittleFoot to the MicroFoot, which reduced the device footprint by half, and now have advanced to chip-scale packaging of the devices, which is smaller still, Crozier says. Military systems designers do demand ever-smaller JFETs and MOSFETs, although at a slower rate than do commercial systems designers, he says.
"The military is pushing for more commercialization, away from the full-hermetic packaging to commercial parts with a good pedigree with long mean time between failures," he explains.
One way to shrink the size of these devices — and hence to help power supply designers reduce the overall size of their components — is to lower the resistance of current flow through the MOSFETs and JFETs, he says. Siliconix engineers do this by simultaneously shrinking the number of device cells, yet substantially shrinking the size the cells themselves. These cells essentially are points at which the device controls the current flow. "It is a little junction," Crozier says. "We put them in parallel and they work in concert for lower resistance."
The trend is a quickly growing number of cells per device, which parallels trends in shrinking device sizes and ever-tightening integration in silicon technology across the board, he says. "We used to have 8 million cells, and then we went to 32 million cells, and now we are moving beyond that to 177 million cells. We are getting better at creating small cells with good efficiency."
Power semiconductor designers at Advanced Power Technology (APT) in Bend, Ore., also are tackling issues related to device complexity. APT manufactures high-current power semiconductors for power conversion, control, and amplification in avionics and other high-reliability applications. Their products include MOSFETs, IGBTs, diodes, power modules, and RF transistors.
"We are taking the basic MOSFET we have qualified since 1984, and are improving it with lower electrical resistance," says Gary Matthai, senior product marketing engineer at APT. "The lower the resistance, the more power you can get out of it. Our latest technology has improved in every area — DC losses, switching losses, and thermal resistance."
APT experts are improving functionality and efficiency by boosting the complexity of their devices in a proprietary process that yields the PowerMOS7 MOSFET. "Our product is linear cell technology," Matthai says. "The gate sources are 'interdigitated.' Our typical competitor has cells; the more cells they can get into a space, is how they reduce their resistance. We have interdigitated fingers. Everything we do is to get more power out of the discrete devices."
Designers at the Fairchild Semiconductor Discrete Power division In North Branch, N.J., are improving the functionality of their devices also by adding complexity. "One activity to add functionality we call the sync FET — an integration of a Schottky diode with a MOSFET," says Steve Ahrens, director of business development for discrete power at Fairchild Semiconductor. "It improves performance in synchronous DC-DC converters. Today, Schottkys are used, but externally or co-packaged with MOSFETs. We use tighter integration of the two devices so the resistance and inductance is minimized."
Complexity also is increasing for power ICs and discrete semiconductors that designers use for space applications. Chip engineers at Intersil Corp. in Palm Bay, Fla., are designing power ICs for satellites that not only must be small, dense, and lightweight, but that also are able to withstand the rigors of space-based radiation.
In satellite electronic systems, "the trend is distributed power, where the power bus voltages are in the 100-to-120-volt range," explains Tom Marshall, Intersil's manager of space products. "That for us embraces distributed power deployment. Everything we design is to work distributed power architectures, to provide power at point of use."
Intersil provides a family of about 18 radiation-hardened power-management ICs called Star*Power. Radiation-hardening analog circuitry such as the Star*Power ICs calls for innovative approaches such as isolating PN junctions from one another such that they cannot accidentally conduct current, Marshall explains. Without this kind of transistor isolation, a charged particle from space could propagate current throughout the device and burn it out.
Other power device suppliers are approaching the need for increased functionality by automating certain tasks within the chip. At Advanced Linear Devices in Sunnyvale, Calif., chip designers are using what they call electrically programmable analog devices (EPADs). These devices are floating-gate devices that enable designers to electrically program to a voltage level down to millivolts, explains Robert Chao, product manager at Advanced Linear.
"The device is part of the chip technology in the integrated circuit," Chao says. "It is a small transistor device that can be made inexpensively that sends a series of voltages from a PC program to the device, and the device is left behind with that setting."
Advanced Linear manufactures low-power linear integrated circuits for industrial control, instruments, automotive, and telecommunications. In addition to EPADs, the company offers MOSFETs, rail-to-rail operational amplifiers, voltage comparators, analog switches, timers, and matched transistor pairs.
Not only must power component manufacturers package their devices for increased complexity and functionality, but they also must package devices to remove as much heat as possible.
"Packaging trends are in two arenas — one is the trend toward surface mount, and the second is in power dissipation — or power management," says Jeff Urish, director of sales and marketing at Semicoa Semiconductors in Costa Mesa, Calif. Semicoa provides power transistors and photodiodes for high-reliability military and space applications. Power semiconductor packaging, Urish says, enables designers to build small form factors and improve thermal management. Thermal management often depends on the materials used to make the semiconductor.
Managing heat also is a primary factor of devices that manage high-voltage systems. At Solid State Devices Inc. of La Mirada, Calif., engineers are using exotic materials such as silicon carbide for the hermetically sealed rectifiers they build for military and aerospace applications. Solid State Devices also manufactures Schottky rectifiers, MOSFETs, IGBTs, power modules, hybrids, and Darlington thyristors.
The company's hermetic rectifiers range from 2 to 40 amps, and run voltages as high as 600 volts, says Arnold Applebaum, president and founder of Solid State Devices. "They also can run at very high temperatures," Applebaum points out. These devices in high-temperature packages could run at 300 to 400 degrees Celsius, yet more typically 250 C is their highest operating temperature in modified surface-mount leadless chip carrier packages made from ceramic materials, he says.
"What separates us is packaging, technology, and taking the packages and technology to make a functional building block," Applebaum says. The building block he refers to is a lightweight and high-density power module.
Other companies are investigating exotic semiconductor materials, as well. At International Rectifier, engineers are working with aluminum silicon carbide composite substrates to stiffen and stabilize the aluminum device substrate, Catrambone says.
Other materials in use at IR include berilium oxide ceramic — better known as BEO — and aluminum nitride ceramic. These materials facilitate current flow and move heat efficiently away from the devices, Catrambone says. In addition, IR engineers are looking into placing copper foil on top of the ceramic substrate, which acts as an interconnect. "It means more current, and carries away more heat," Catrambone says. This approach would be a replacement to thick-film substrates, he says.
At Powerex, a joint venture of Mitsubishi and General Electric, in Youngwood, Pa., designers also are looking to aluminum nitride, as well as alumina, to help them manage heat. Powerex experts are involved in demanding military applications such as the Lockheed Martin F-35 Joint Strike Fighter, in which systems integrators seek to replace hydraulic pumps and pipes with electric motors and wires.
"The limitation of applications like that is not being able to pull the heat out of this many discretes," says Duane Prusia, director of rectifier/thyristor products and custom modules at Powerex. "Now people want to go to the die level and put it directly on a substrate, and parallel the die together for a higher-current switch."
To do this, Powerex designers are using aluminum nitride or alumina to improve thermal conductivity. "Alumina has conductivity of 35 watts per meter Kelvin, where aluminum nitride has 170 watts per meter Kelvin," Prusia explains. Company experts are using aluminum silicon carbide for base-plate material, which "conducts heat away, but is very light." This approach is an alternative to using traditional copper base plating, he says.
Back at Solid State Devices, company engineers find that silicon carbide is particularly adept at siphoning excess heat away from sensitive transistors on rectifiers and other power electronics, Applebaum says. The more heat that designers can move efficiently away from the device, the faster the device can switch electric current, and the higher the voltage the device can tolerate, he says.
Today silicon carbide technology is in its infancy, he points out, but should improve with time as chip fabricators learn to handle the technology as they have with silicon. Although Solid State Devices today concentrates its silicon carbide efforts on rectifiers, this technology should extend to other power components in the near future. The next stage for Solid State Devices will involve FETs.
For high-voltage applications at or above 300 volts, "silicon carbide technology will probably dominate in the future," Applebaum says. "They will be fabricating FETs and IGBTs, and that will be an important breakthrough. Now we can make rectifiers and get very high yields, but in three to five years we will have more sophisticated devices."
Silicon carbide should provide a substantial improvement in device efficiency over today's silicon materials, Applebaum says. "Today the FETs and IGBTs are out of silicon, so their speeds have a limitation on switching frequencies and the temperatures they can operate at. Also, with silicon carbide, the voltages can be in thousands of volts. Right now FETs are around 1,000 volts, and in the future with silicon carbide they will be around 4,000 volts."
Applebaum calls the challenges of fabricating high-yield silicon carbide wafers simply "a matter of time," as the devices are improving month-by-month. "It's running the same course as silicon," he says. "At first silicon had a lot of defects and as things
developed they made larger and better-quality wafers, which meant larger and better devices."
Power electronics manufacturers also share other similarities with digital device makers — particularly the pressure to reduce component costs and tap into commercial best-manufacturing practices to improve reliability, boost yields, and bring down costs. One approach is to use plastic-encapsulated power semiconductors rather than the more rugged and expensive ceramic and metal packages.
"The military is pushing more to commercialization, and away from full hermetic, to use commercial parts with a good pedigree and long mean-time between failures," says Crozier of Siliconix. "More military guys asking for these commercial plastic parts."
Although he says the military is typically a few years behind commercial systems designers in adopting commercial-grade power components, "we see increased demand for high-reliability plastic parts," Crozier says. "We offer 'ROTS,' or ruggedized off-the-shelf components, with some burn-in on a plastic part to weed out infant mortality and offer high reliability. The military is embracing that."
Although military systems designers first adopted plastic-packaged digital circuits, the need in the military for increasing amounts of COTS usage also is moving into the analog power applications. "We are seeing designers go to the COTS products because as some of the military programs move to plastic, they are moving from the full-up hermetic MOSFETS to the plastic MOSFETS, and to some of the plastic IGBTs," says APT's Matthai.
International Rectifier also is moving into the high-reliability plastic power semiconductor market. The company's Hi-Rel Products division will offer plastic-encapsulated discrete MOSFETS that meet the guidelines of the Qualified Parts List — or QPL — by the end of this year, says IR's Catrambone.
IR's plastic QPL MOSFETs, which are targeted to military and aerospace applications that require COTS parts, also will be screened to the standards of the Joint Army-Navy Technical Exchange, better known as JAN TX, Catrambone says. These parts are for voltage switching applications, he says.
Despite pressure to move to COTS, however, analog power companies still deal more routinely than their digital counterparts with military specifications and custom-designed components. Yet the approach of using COTS wherever possible in military systems is beginning to apply to power devices as well as digital components.
Two of the most applicable military standards in the power semiconductor business are MIL-PRF-19500 for discrete devices, and MIL-PRF-38534 for power ICs, explains International Rectifier's Catrambone.
For discrete devices, MIL-PRF-19500 outlines standards for shock, vibration, space-level testing, and temperature extremes, explains Semicoa's Urish. Although some in the industry may claim that military specifications by their nature run contrary to the spirit of COTS, Urish says mil specs such as 38534 and 19500 are a much better option than custom approaches that use Source-Control Drawings (SCDs).
"Most military and commercial satellite manufacturers are on a trend of using standard mil specs rather than Source-Control Drawings, which are inherently costly, Urish says. "To control costs in the commercial arena, satellite manufacturers are evaluating MIL-PRF-19500 technology."
Despite the company's support for MIL-PRF-19500, Urish points out that Semicoa engineers do provide custom-designed parts when system designers demand such a part. "Semicoa is receptive to custom designs that are classically small-volume specialty applications," he says.
IGBTs vs. MOSFETs
The COTS vs. mil-spec debate is actually about cost vs. capability. Systems designers are trying to get the best capability from their systems at the least possible cost. That has some engineers thinking not only about mil-spec vs. COTS, but also about how different devices relate to one another. Specifically, some designers are starting to consider the use of IGBTs where they once would have considered only MOSFETS.
Experts at Richardson Electronics in LaFox, Ill., specialize in industrial applications rather than military applications, but still deal routinely with engineers who seek the most capability at the lowest cost. "Occasionally we see customers pull back on the specifications," says Ross Waite, business unit manager of power semiconductors at Richardson Electronics.
"IGBTs have become the standard device with respect to new designs," Waite says. "They are getting faster, and can go up to 100 kHz, where that application used to be all MOSFETs."
Waite makes clear, however, that IGBTs cannot substitute for MOSFETs in all applications. "MOSFETs are not becoming obsolete," he says. "In switching speed they are still the preferred devices." Where switching speeds must exceed 100 kHz, MOSFETs often will be the default choice, but where switching speeds are 100 kHz or slower, designers are starting to think IGBTs.
Waite points out that IGBTs are smaller and less-expensive devices than MOSFETs. "It all comes around to cost, and the IGBTs, when they get into the 100 kHz realm, will cost less than a MOSFET," he says. "But when it comes to absolute high-speed switching, the MOSFET is still your preferred device."
Prusia at Powerex says that today the functional switch-speed cutoff between IGBTs and MOSFETs is 20 kHz. "Anything of 20 kHz and below we would push IGBT technology, where a MOSFET will operate very efficiently at the higher frequencies," he says.
Chip designers push the state of the art in motor-control technology
Power component manufacturers such as Data Device Corp. (DDC) in Bohemia, N.Y., and Aeroflex Inc. in Plainview, N.Y., are trying to take power electronics related to motor control to the next level.
DDC offers an integrated device on a circuit board and subassembly that drives three-phase motors based on a feedback element. This component is a central part of new generations of brushless DC motors, says James Zuber, DDC's product manager for motion controllers and drivers.
"In the old days we had two brushes wired to a power supply, and outputs of the brushes contacted a rotating commutator to generate a magnetic field, which made the motor rotate," Zuber explains. "The old DC motor had arcing inside, which would be the brushes moving the current itself."
This arcing created the sparks that are visible inside of old motors, he says. "What's happened is with later technology we revamped the motor to remove the brushes and put them on the outside of the motor sleeves. The improvement was removing the brushes because of arcing and reliability issues, putting magnetic fields on the outside of the motor, and moving the magnets to the rotor section."
This approach no longer requires current to flow through the brushes, which caused a notorious reduction in reliability. "Now we have a more reliable motor, but we still need to energize the appropriate fields on the motor at the right time," Zuber says. "Our devices energize those fields by providing the timing on which to energize them. This is called commutation circuitry."
DDC can provide systems designers either with only the switches necessary to time the magnetic fields in the motor, or can provide intelligent circuitry that can automatically control factors such as speed and output torque, Zuber says.
Aeroflex, meanwhile, has developed two radiation-tolerant Class K motor driver modules for spaceborne applications such as gimbal drivers, pointing mechanisms, solar array drivers, actuator assemblies, antenna drives, reaction wheel assemblies, and control motor gyroscopes, says Joseph Castaldo, strategic business manager at Aeroflex.
The motors can withstand 300 kilorads of radiation, using radiation-tolerant components and proprietary circuit designs, Castaldo says. Aeroflex is offering these devices through a licensing agreement with Honeywell International that allows Aeroflex to manufacture Honeywell's motor drivers exclusively as catalog items.
For more information contact DDC by phone at 631-567-5757, by post at 105 Wilbur Place, Bohemia, N.Y., 11716, or on the World Wide Web at http://www.ddc-web.com. Contact Aeroflex by phone at 516-694-6700, by fax at 516-694-4823, by post at 35 South Service Road, Plainview, N.Y., 11803, or on the World Wide Web at http://www.aeroflex.com/.
Navy researchers seek to develop 25-megawatt propulsion motors for future warships
by John Keller
ARLINGTON, Va. Experts in the U.S. Office of Naval Research (ONR) in Arlington, Va., are trying to develop a full-scale, high-power, lightweight electric propulsion system for future Navy warships.
Toward this end, ONR researchers are gathering industry white papers on technology demonstrators that, at minimum, involve 25-megawatt motors, motor controllers, 26-megawatt generators, transformer/switchgear, and related controls and ancillary equipment.
Experts say advanced motors of this kind would be substantially lighter and more efficient than existing motors or those under engineering development. Navy officials make clear that this kind of propulsion motor must be manufactured within the United States.
Navy scientists say they want the demonstrator and subsequent production system to be not only compatible with Navy electric warship concepts, but also viable for commercial and industrial applications such as cruise ship propulsion or large industrial drives.
A key part of this project is to make electric propulsion power available and apportioned to ship service loads such as propulsion, future weapon systems, and high-power sensors.
Navy experts are specifically asking for revolutionary ideas, and say they will not consider evolutionary or incremental improvements in current electric machinery designs. Instead, they are looking for approaches that allow a substantial increase in torque and power density, with weights, dimensions, and aspect ratios compatible with efficient pod propulsion of 30 knots or faster.
Scientists say they would like industry proposals that involve technologies such as superconducting windings and magnetics, efficient cryogenic cooling, direct liquid cooling for heat removal, and intrinsically low-acoustic-noise machine design. On the other hand, Navy officials say they are far less interested in motor drive, transformer, and switchgear technologies.
Industry awards for this program will be in three phases:
- preliminary design of the full-scale demonstrator system, integration of the proposed technology into an electric warship concept, plans for manufacturing and acceptance testing, and assessment of the industrial base and potential market share of the technology;
- detailed design and long lead item procurement specifications for the motor, motor drive. and ancillaries; and
- motor demonstrator manufacture, factory acceptance testing of the motor demonstrator, and delivery of the motor, motor drive, and ancillaries to a yet-to-be-determined U.S. Navy facility.
For technical questions, contact Thomas Calvert, program officer of the ONR Ship Hull, Mechanical, and Electrical Systems S&T Division, by phone at 410-591-1898, by fax at 703-696-0308, by e-mail at Thomas_Calvert@onr.navy.mil, or by post at Code ONR 334, Office of Naval Research, Ballston Centre, Tower One, 800 N. Quincy St., Arlington, Va.
For business questions contact Todd Hanson, senior contracting officer at ONR's contract and grant awards management activity, by phone at 703-696-2009 by fax at 703-696-3365, by e-mail at Todd_Hanson@onr.navy.mil, or by post at Code ONR 254, Office of Naval Research, Ballston Centre, Tower One, 800 N. Quincy St., Arlington, Va.