Applying memory polynomial model to linearize wideband multi-bit delta-sigma based transmitters

This paper presents a new predistortion linearization technique based on the memory polynomial model of the power amplifier (PA) and a multi-bit delta-sigma modulator (DSM), in order to avoid reverse modeling and the associated instability problems. In this topology, a multi-bit DSM is utilized as a digital predistorter by locating the forward model of the PA in its feedback path. The technique has two main advantages. First, it applies the forward model instead of the reverse model of the PA, which leads to less implementation complexity and, therefore, greater robustness against instability. Second, using a dynamic model instead of a static model makes it more applicable for the linearization of wideband signals, due to the compensation for memory effects. To justify its advantages, linearizers based on both static and dynamic models have been designed and tested. For a 4 MHz WiMAX signal, the measurement results show an improvement of 17.2 dB in the signal-to-noise and distortion ratio (SNDR) by using a dynamic polynomial model, compared to an improvement of 13 dB for a static model.