Single-tone optimisation of an adaptive-bias Doherty structure

The Doherty PA configuration has become well established as a means of enhancing PA efficiency. The improvement, however, is usually accompanied by deterioration in overall linearity. This paper addresses two important features of the Doherty PA configuration, which offer a means of improving the linearity/efficiency trade-off obtained using a classical adaptive bias implementation. One issue is the need to generate a suitable bias control voltage for the auxiliary PA device. Such control enables the use of two devices with equal periphery for main and auxiliary functions. The second issue is the use of phase compensation to ensure the combined signals sum in the correct manner prescribed by classical analysis. Both of these refinements are demonstrated using an experimental circuit operating at 1.8 GHz. Extensive experimental data are presented on the beneficial effects of phase control, which show that a better compromise between optimum efficiency and linearity can be obtained using different phase offsets at different drive levels. This raises the interesting possibility of adjusting relative phase, bias and relative input magnitude dynamically in order to obtain improved linearity from the Doherty configuration, especially for signals having high peak to average ratios.