RoMPE User's Manual Version 1.2

Abstract

OSA's first product RoMPE is a very powerful program for extraction of FET model parameters from measurement data. Unveiled in New York at the 1988 IEEE MTT-S International Microwave Symposium, RoMPE is the world's first commercially available software system dedicated to nonlinear FET model parameter extraction suitable for microwave circuit design. RoMPE (an acronym for robust model Parameter Extractor) is a program dedicated to the extraction of FET model parameters from measurement data. It benefits from several years of theoretical research and incorporates the latest results published in the IEEE MTT -S International Microwave Symposium. RoMPE accepts measurement data in the form of small-signal S-parameter and/or DC bias measurements. Depending on what type of data is available, RoMPE can be used to extract DC model parameters, small-signal model parameters, or, as a unique feature, both DC and small-signal parameters simultaneously. RoMPE offers the combined power of novel techniques in device modeling and the state-of-the-art gradient-based L1 and L2 optimizers through a friendly and uncomplicated user-interface. The input file adopts a format similar to that of Super-Compact and Microwave Harmonica. The program is also enhanced by high-quality graphics display. Currently, RoMPE supports two popular FET models, namely the Materka and Kacprzak model [and the Microwave Harmonica model. The novel approach pioneered by Bandler, Chen and Daijavad, recently extended by Bandler, Chen, Ye and Zhang and implemented in RoMPE provides two distinct advantages. One is to process simultaneously measurements at multiple bias points, and another one to optimize simultaneously small-signal and DC parameters. The motivation of multi-circuit modeling, i.e., modeling through simultaneous processing of multiple sets of measurements, is to strengthen the model identifiability and to enforce a unique solution. This principle has been amply demonstrated and is applicable to all engineering modeling problems. In our case, measurements at multiple bias points can be utilized to achieve a robust solution, and this can be applied to small-signal modeling, DC modeling and large-signal modeling.