A Bias Elimination Strategy for Multi-Active-Bridge Converters Based on Switching-Frequency Ripple

The performance of a multi-active-bridge (MAB) converter can be degraded by dc biases in the currents flowing through its multiwinding transformer (MWT). Existing strategies estimate this bias either by measuring quantities in the reactive network, which increases circuit complexity and cost due to the sensing arrangement, or by directly sampling a noisy voltage ripple, compromising the signal-to-noise ratio (SNR) of the bias measurement. This paper proposes a method that utilizes the switching-frequency component of the terminal voltage ripple to estimate the bias current in a MAB converter. The amplitude of this ripple component is shown to be proportional to the bias, enabling a simple and noise-resilient implementation using low-cost conditioning circuits. Based on the proposed estimation approach, a control strategy is developed to actively eliminate dc biases in the currents flowing through the MWT. The proposed technique is validated through both simulations and experiments on a quadruple-active-bridge (QAB) converter, demonstrating improved efficiency and the overall effectiveness of the method.