On real-time optimal control of high-precision switching amplifiers

Currently, in the field of precision power amplifiers rather conservative power densities and bandwidths are used. These limitations are mainly due to the use of linear feedback architectures that do not consider constraints on the power amplifier dynamics during synthesis. Hence, to avoid any instability or inaccuracy due to nonlinearities of the power stage, the amplifier is generally operated with a highly conservative power budget. This paper investigates the applicability of reference governors and model predictive control in the field of precision power conversion. It is shown that the large signal behavior can be significantly improved, in terms of safety and accuracy. We also illustrate that dual-mode model predictive control (MPC), which combines a linear state-feedback with the sub-optimal solution of the MPC problem found by a real-time optimization solver, efficiently eliminates the small signal distortions inherently present in the standard real-time MPC implementation. The paper is organized as a case-study on controller synthesis for the opposed current converter, and as such, is well suited for practicing engineers in the field of precision power amplification.