Optimisation of a Doherty power amplifier based on dual-input characterisation

The success of the Doherty architecture compared to other efficiency enhancement techniques derives mainly from its simple design and full-RF nature, not requiring complex digital signal processing to achieve high back-off efficiency. In this work we propose a design strategy for the optimisation of a Doherty power amplifier to mitigate the typical practical issues of this architecture related to inaccuracy of the non-linear model and of the manufacturing. The approach is based on the experimental characterisation of a dual-input Doherty prototype without input section. This test structure is obtained from a single-input Doherty amplifier, designed only through non-linear simulations, by removing the input section and allowing for separate control of the two RF inputs. From the collected data, approximated functions for the phase shift and power splitting versus frequency are identified to be realizable in hardware with RF networks. Compared to the reference single-input Doherty stage, a significantly improved behavior is registered in terms of output power (up to 2.7 dB), efficiency at saturation and back-off (30 % and 15 % respectively) and power gain (2 dB).