We're so used to thinking of technology migrating from military applications to the civilian world that it's surprising to find an opposite migration. Satellite communication, spread-spectrum modulation, and fly-by-wire come immediately to mind as examples of the traditional military-to-commercial pathway (see "The Latest Mil-Aero Developments," p. 52).
Hence, it's a little startling to hear about chip sets expressly designed to control washing-machine motors mutating into technology for solar-powered unmanned tactical airships hovering high over battlefields. A little less surprising is the story of how a state-of-the-art foundry led a company back into the mil-aero market after abandoning it more than a decade ago. It could now do things for far less money than in the old days.
First, let's delve into that eyebrow-raising, washing-machine-motor to solar-powered-airship transition. International Rectifier maintains a high-reliability division in Leominster, Mass. Rick Furtney, vice president of the business unit, says that hi-rel comes down to radiation-hardening MOSFET devices, extending temperature ranges, ruggedizing packaging, and providing screening and traceability (see "Radiation-Hardening Power Devices," p. 50). On the traditional military side, IR's radiation-hardened products go into satellites, launch vehicles, and weapons.
For space exploration, the company works with Caltech's Jet Propulsion Laboratory on advanced platforms for the next-generation Martian space laboratory, a follow-up to the Mars Rover missions. IR also has hi-rel devices in aircraft ranging from the F-22 to the Airbus A380. It's interesting to note that the F-22 has more than 100 different dc-dc power modules. Other hi-rel products go into medical products, where applications range from motor control in power saws and high-speed drills to voltage converters in defibrillators.
So where do the washing machines and blimps come in? Lockheed-Martin's High Altitude Airship (HAA) is an unmanned lighter-than-air vehicle intended to operate above the jet stream, where it can use its electric engines to hold position against moderate winds (Fig. 1). Surveying an area 700 miles in diameter and a volume of millions of cubic miles of airspace, the HAA will serve as a telecommunications relay, a weather observer, or a surveillance platform.
While the artist's-conception HAA illustrations on these pages suggest deployment in the distant future, proof-of-concept efforts are under way (Fig. 2). New technologies under evaluation include high-strength fabrics to minimize hull weight, thin-film solar arrays for power, and lightweight propulsion units—where the washing-machine connection comes in.
FPGA-BASED MOTOR CONTROL
Motor control is a key technology area in IR's commercial sector. This is particularly true outside the U.S., where the efficient and silent running of motors in white goods gets more attention than it does here.
The company's iMotion design platform targets variable-speed motion-control systems in the 200-W to 4-kW range (Fig. 3). It integrates digital, analog, and power technologies in a mixed-signal chip set and a targeted development system based on an FPGA-configured signal architecture. The system executes complete field-oriented motor control and a pulse-width-modulation algorithm loop.
It isn't based on general-purpose DSPs and microcontrollers. As a result, iMotion doesn't require coding. Instead, the system designer applies the iMotion development software and intellectual-property (IP) library when configuring an FPGA with one of the supplied control algorithms. IR's iMotion IP library can be used to design variable-speed, motion-control systems with or without encoders. Feedback control-loop bandwidth extends up to 5 kHz.
The analog section of iMotion consists of three high-voltage ICs: a three-phase gate driver, a current-sensing IC, and a power-supply controller. These components interface with IR's 600-V non-punch-through (NPT) insulated-gate bipolar transistors (IGBTs) for the power stage. The analog and power components are offered as SIP-packaged integrated power modules that constitute a complete, isolated three-phase inverter power stage, including gate drivers and auxiliary circuitry.
Furtney says the applicability of iMotion to a project that's as different from its original purpose as the HAA illustrates the changing face of military procurement. "The challenge in military today arises from a combination of small product volumes and tightly squeezed research and development budgets," he says.
"In the case of iMotion, migrating the concept from commercial is literally saving the military hundreds of thousands of dollars in R&D." He adds that there is a further advantage in doing it IR's way, rather than with a DSP architecture, because it prevents code storage and integrity problems.
Nor is the HAA some kind of fluke. Beyond the HAA, Furtney says that IR is getting iMotion design-ins on advanced fuel pumps for aviation, including motor control for aircraft control-surface actuators. He notes that the U.S. Navy's next-generation aircraft carrier will be all-electric, and IR will contribute to its systems as well.
The next-generation carrier will be a significant departure in terms of electric power usage. Writing in the U.S. Naval Institute's Proceedings magazine, Captain J. Talbot Manvel Jr. says the next generation of carriers will be Nimitz-class hulls run by nuclear power that will be home to as many as 75 aircraft. They will include a new electrical power generation and distribution system with substantially more generating capacity and the ability to support new, cost-saving electric technologies.
For example, these carriers will use electromagnetic rather than steam-propelled catapults. According to Manvel, the old steam catapults, while reliable, were manpower-intensive and costly to operate and maintain. Beyond the catapults, he says that the next-generation carriers' enhanced electrical-generating capacity and improved electrical distribution system also will help replace steam with electricity in many other shipboard systems, all of which will provide more potential for technologies such as iMotion.
PLATFORM ASICs & SMIF PODS REKINDLE A BUSINESS
John Bendekovic, director of Military/Aerospace Marketing at LSI Logic, says that the company abandoned the military business a decade ago. However, a convergence of factors made possible by its RapidChip platform-ASIC technology recently led the company to again offer its services to the military .
Qualified Manufacturers List (QML) foundries aren't a new phenomenon. A total of 19 ASIC vendors are on the QML, but LSI is the first with a qualified 0.13-µm process technology (see "QML Sets New Standards," p. 48, and "Structured ASIC Trims NRE Costs For Low-Volume Mil-Aero Chips," p. 50).
"When LSI Logic introduced the platform ASIC about two years ago, it was targeted at companies in the telecommunications and consumer applications arena," Bendekovic says. "But then the mil-aero customers started calling, saying they wanted what we were offering the civilian customers, but 'we just want you to give us X,' where X might be any of a number of things—a QML process, or mil-temperature testing, or radiation testing, or DO-254 compliance for commercial avionics customers." (The latter is the Radio Technical Commission for Aeronautics Document RTCA/DO-254, titled "Design Assurance Guidance for Airborne Electronic
"UNTOUCHED BY HUMAN HANDS"
A modern fab makes it relatively easy to provide options that were very difficult 15 or 20 years ago. That's why LSI is back in the mil-aero business. The company builds its QML'd ASICs in its Gresham, Ore., facility, where wafer movement is based on SMIF (Standard Mechanical InterFace) production.
In that environment, which has come to permeate the industry, blank wafers are loaded into sealed transport "pods" at the beginning of processing and remain sealed in those boxes except when they're inside each successive piece of processing equipment. Total automation and record-keeping leads to die-level traceability, which is as important to the military for security as it is to the foundry and its commercial customers for process control and order tracking.
SMIF-based total automation is ideal for QML, RTCA, and even the National Security Agency. Bendekovic boasts that LSI's initial QML audit took place in November, and when the auditors returned in March for the final audit, what was supposed to take five days wound up being only three.
A lot of the military's acceptance of SMIF handling comes down to the fact that it considers an SMIF pod a "room." "Years ago, we would need a dedicated facility to get into the mil-aero market," says Bendekovic. "We would construct a 'SCIF,' a Secret Compartmentalized Information Facility. This would have been essentially a concrete room in which all classified processing had to take place, and it needed to be staffed by cleared personnel, contain its own dedicated tooling, and run on its own autonomous infrastructures."
"Now 15 years later, when we go for class-5 processing, some of the same rules still apply—you have to have a dedicated room, you have to have cleared people, you have to have autonomous IT systems. But now a lot of the same investment that we've made for other reasons coincidentally makes it much easier to meet the QML requirements," he continues.
In this case, he says, SMIF handling takes the human element out of the picture. Instead of people pushing wafers from machine to machine, there's an automated transport system that cannot be casually breached. If an SMIF pod full of wafers doesn't arrive at a processing station on time, or if a pod arrives at the wrong station for the recipe, an alarm occurs. The NSA and DoD regard that as lowering the security risk.
The SMIF pods that carry the 25 wafers around are considered "rooms" because it is difficult to get into a pod without breaking it. Treating a pod as a room means a non-cleared employee may be present in the same area as the pod, just as in the old days, a non-cleared person could be present out in the hallway next to the SCIF. (Of course, the personnel responsible for running the equipment need clearances.)
OUTSOURCING SEEN HARMFUL
Some foundries on the QML list are located in countries outside the U.S. Yet Bendekovic notes that recently there has been a lot more focus on on-shore processing and mask generation, and in the case of the platform ASIC, the metallization and the back-end manufacturing steps.
LSI often accomplishes all of that at its Gresham plant, or by using local subcontractors for masks, wirebond and flip-chip assembly, and hermetic packaging. In addition, if all-U.S.-citizen handling is required, it's simple to swap shift assignments so that everyone on the appropriate shift is a citizen.
For more on the U.S. government's concerns about dependence on offshore foundries, check out www.acq.osd.mil/dsb/reports.htm and click on "High Performance Micro-chip Supply." This report was produced by the Defense Science Board and released last February by the Office of the Under Secretary of Defense For Acquisition, Technology, and Logistics.
The contents of this 118-page report contain nothing that would surprise any working engineer, but the document is written so that it's accessible to any reader with a good education and offers some compelling reasons for keeping the manufacture of custom chips onshore.
|NEED MORE INFORMATION?|
Lockheed-Martin HAA Program
Space Data Corp.
U.S. Army Space and Missile Defense Command