Broadband Error Vector Magnitude Characterization of a GaN Power Amplifier using a Vector Network Analyzer

This work investigates the impact of nonlinear dynamic effects due to charge trapping and other long-memory phenomena on the wideband linearity of power amplifiers (PAs) for 5G communications. The proposed method uses the well-known best linear approximation framework (BLA) to estimate the error vector magnitude (EVM) of the amplifier for the class of modulated signals sharing the same probability density function (pdf) and power spectral density (PSD) as the 5G waveform standards. The dependency of the EVM and the BLA on the large-signal operating point of the PA is studied using random phase multisine signals. In particular, we evaluate the impact of different signal repetition periods in order to excite low-frequency dynamic phenomena across a wide range of time scales. Results, using just standard vector network analyzer relative measurements, are reported for a Gallium Nitride power amplifier for two different 5G-FR1-compliant bandwidths of 20 and 100 MHz at 5.5GHz.