The two methods presented in this article estimate first-, third-, and fifth-order Volterra kernels for an RF power amplifier (PA) using common RF laboratory instruments. In the first method, single measurement readings are employed to determine the kernels, while the second approach utilizes least-squares curve adaptation of measured data. Comparisons between the two methods discussed here and others described in existing publications reveal that these methods are easier to perform.
When an amplifier operates in the nonlinear region, it produces unwanted signals, namely harmonics. The second-, third-, and higher-order harmonics are outside of the amplifier’s bandwidth and can thus be filtered out. However, unwanted signals close to the fundamental (intermodulation products) cannot be filtered out and will interfere with amplified input signals and cause nonlinearity (Fig. 1). Furthermore, channel selectivity, detection efficiency, quality, and more importantly, spectral efficiency are impacted, leading to poor wireless-network capacity.1, 2