. . Extraction of a Nonlinear Amplifier Model from Measured Characteristics Aladdin H. Assisi 1 , Ahmad Sileem Mahmoud 2 and Mustafa M. Ahmad 3 Egyptian Armed Forces sisi@msa.edu.eg Abstract In this paper a mathematical analysis to extract a nonlinear amplifier model from measured characteristics was developed and verified on a case study. Two methods were proposed. The first can be used to extract a fifth order polynomial model by inserting a sine wave at the amplifier input and measuring the output harmonics at different input power levels. It can be used to estimate other nonlinear behaviors of the amplifier; such as intermodulation and desensitization. The other model can be extracted by measuring the two-tone intermodulation spectrum and deduce other amplifier characteristics. Both methods were mathematically analyzed and experimentally verified. 1. Introduction There are many ways to characterize amplifier nonlinearity. Two-tone intermodulation distortion is one of the most common. Collantes et al developed a modeling technique based on the use of cubic B-splines that are not forced to pass through the measurement points. Such approximation splines avoid the interpolation of noisy measurement [5]. Kamal and Eassa used genetic programming to solve curve fitting problems without need to assume the equation shape [1]. Maas introduced six valuable ‘laws’ to be followed for reliable intermodulation modeling and analysis of practical amplifiers [3]. In a recent paper, the authors proposed a new measure for amplifier nonlinearity, namely, the small signal gain degradation due to large signal capture of the amplifier power. This criterion for linearity evaluation is especially important in communication repeaters where received signals are not a priory known [6]. The “gain capture” phenomenon degrades also the performance of wide-band saturated power amplifiers in jamming systems. Moreover, it affects wide-open receiving amplifiers in EW systems, where neither signal frequencies nor their levels are a-priori known. In such receivers, gain capture phenomenon can be used to define an upper limit for the dynamic range. Amplifier gains for small and large signals were analyzed and a mathematical formula for each was driven [6]. Those gain prediction formulae are functions of the large signal amplitude and the odd coefficients of a fifth degree polynomial expansion of the amplifier input-output characteristic. Therefore, it is necessary to get those coefficients for an amplifier to estimate its large and small signal gains. Since the input-output characteristics of an amplifier and the intermodulation spectrum of the same amplifier are inter-dependent, any of them can be predicted from the other. In this study, we are going to analyze the measured input-output characteristics of an amplifier to extract the coefficients of a fifth degree polynomial representing its nonlinear behavior. In this way, we can predict different nonlinear performance phenomena of this specific amplifier, such as intermodulation and small signal gain degradation. On the other hand, the intermodulation output spectrum of an amplifier is analyzed to extract a set of 5 th degree polynomial coefficients that can be used to predict the input-output characteristics of the same amplifier. It is worth noting that we have not used known curve fitting techniques with a single valued set of coefficients covering the entire dynamic range. Instead, we developed a set of coefficients representing the measured characteristics at each input power independently. This gives a model that follows the amplifier characteristic point by point. This model does not describe phase properties of the nonlinear amplifier; since it does not consider its time delay. 26th NATIONAL RADIO SCIENCE CONFERENCE, NRSC’2009 Future University, 5th Compound, New Cairo, Egypt, March 17 – 19, 2009 1 D02