Introduction
Solid state power transmitters (for a wide range of radar and communication applications) are usually formed by a chain of MMIC power amplifiers, [1]. In order to guarantee the correct performance of the complete systems in terms of their reliability it is essential to determine the thermal behavior of those devices and their mounting structure, [2]. The purpose of this work is to provide a simple methodology for system engineers to measure and predict the peak temperature on the power amplifier MMICs used in their systems under typical bias conditions. A large number of different methods for assessing the peak temperature of power devices have been previously proposed. The majority of these methods are unable to handle complex structures like those shared by commercial MMIC power amplifiers, which are usually formed by the combination of different stages of parallel single devices. Some methods need the amplifier to operate under untypical working conditions, [3], or may actually alter its electrical and thermal properties, [4], [5]. Others require unusual device topologies, [6], or sophisticated measurement equipment and techniques, [7], [8], [9]. Hot spots and peak temperatures can be also determined by the resolution of the heat flux 3-D problem in the amplifier's multiple-layer configuration by numerical techniques, [10], [11]. Although this last procedure is more precise and capable of reproducing the real problem than other methods [12], it usually requires the thermal properties and physical dimensions of the different layers to be precisely known. For instance it is always hard to determine the physical and thermal properties of the attachment layer used to solder or glue the amplifier to its metallic carrier, [13]. In this paper a new indirect electrical method [14] to determine the peak temperature of complex MMIC power amplifiers is presented. The method requires that the amplifier be formed at least by two different stages, and that their drain supplies be independent as in fig. 1. The proposed method makes use of the drain current of one of the amplifying stages as a temperature sensitive parameter (TSP). Other stages are used to heat up the previous one through thermal coupling. Numerical techniques are also employed to obtain the temperature profile on the MMIC surface. The suggested method does not need any physical or thermal information regarding the attachment layer to be known and, unlike other procedures, requires the use of basic and inexpensive measurement equipment. The suggested technique avoids possible electrical damage on the amplifiers under test because untypical operating conditions are not needed. This technique was successfully applied to a commercial X-band MMIC power amplifier and its thermal behavior was characterized under different bias, temperature and mounting conditions. A simple DC electro-thermal model was then derived for the single FET device which, properly combined, forms the complete MMIC amplifying structure. Experimental DC measurements over a wide range of ambient temperatures for the complete MMIC structure are precisely predicted by the derived electro-thermal model, validating the single FET extracted model and the suggested new indirect thermal measurement technique.