On the use of RF bursts for identification of amplifier kernels

In this paper we show a simple approach for the extraction of third-order kernels for Volterra-based models of wideband amplifiers. RF bursts are applied to the input of the amplifier under test, and the kernels are identified through the acquired samples of the complex envelope. This procedure has been applied to obtain the third-order behavioral model of a commercial amplifier using a training RF burst. The extracted kernels have been employed to predict the response of the amplifier under diverse excitations levels and signals, including RF bursts with different levels compared to those used for parameter extraction and two-tone signals, showing a good correspondence with the measured waveforms. The obtained results demonstrate a promising approach to capture the amplifier third-order kernels with independence of the excitation signal, bandwidth or level.