This paper describes a method for process selection of aluminum components in the early stages of design. Aluminum has many advantages in a variety of applications in its manufacturability and recyclability. Yet, engineers who are trained to design steel components do not take full advantage of this material. The main reason is that engineers tend to be unaware of the many economical processing methods for aluminum. We have developed a program that combines preliminary screening of processes with normalized cost analysis. Design compatibility analysis (DCA) ranks each process based on its feasibility with the basic geometry, material, and production requirements. For top candidates, the program employs external cost routines for detailed comparisons. The primary processes considered are extrusion, sheet forming, forging, die casting, and sand casting. The program extends its compatibility and cost analysis to secondary operations such as bending and machining. The program should be useful for engineer training and as a preliminary design tool. The program uses HyperCard as a front-end, Prolog for logic-based analysis, and Excel for cost calculations.

Gallium Nitride high-electron mobility transistors (HEMT) devices show great promise in their ability to tolerate the high temperature environments of advanced radar systems. This paper examines how GaN HEMT junction temperature determination can vary, owing to factors such as packaging variability, measurement error, and uncertainty in material property data. To demonstrate the impact of these variables, this paper uses practical examples of infrared thermography, micro-Raman thermography, device transient electro-thermal response analysis on GaN HEMT devices, and finite element analysis (FEA). These variations in temperature are combined into a probability model to estimate how life prediction will change as a function of these various factors. Keywords IntroductionAs a nascent technology compared to GaAs, Si, or nonsolid state technology, GaN-on-SiC transistors have not established a history of reliability from which end-users of the technology can establish its long term replacement and refurbishment costs.[1] Nonetheless, GaN provides a number of distinct advantages over older technologies, including improved heat transfer properties, wider bandgap energy, higher operational temperatures, and higher frequency performance. [2] In lieu of historical reliability information, the consumers of this technology must depend on accelerated lifetime testing (ALT) of parts where a predicted operational lifetime, on the order of millions of hours, is extrapolated from faster failures (hundreds of hours) achieved at highly elevated temperatures. The validity of this extrapolation is dependent on three assumptions: 1) that the physics of failure for the GaN device is analogous to previous technologies, allowing for a loglinear extrapolation (the Arrhenius model) through timetemperature space, 2) that the ALT is exciting the same predominant failure as occurs in fielded devices under standard operating conditions, and 3) that the operational temperature of the device is known. [3] This paper focuses on this third assumption, using empirical (micro-Raman thermography, transient thermal testing using the T3ster from Mentor Graphics, and midwave infrared thermography) and finite-difference modeling (ANSYS-Fluent) techniques to assess the measure, spatialuniformity, and statistical variability in temperature measurements on GaN transistor devices.

No abstract

No abstract