This paper presents a hybrid microgrid configuration with a flying capacitor dual-output (FCDO) converter. The output ports of the FCDO converter can directly interface ac sources/loads operating at different amplitudes and frequencies without additional ac/dc/ac converter units. Compared to the conventional configuration, the hybrid microgrid with the FCDO converter operates at multilevel voltages, reduced power conversion stages, less power switch count, and fewer control loops. The paper also presents a cascaded model predictive control (CMPC) algorithm for such configuration to control the variables, namely three-phase dual-output currents, ac/dc bus and FC voltages, and active/reactive power. The proposed CMPC sequentially executes multiple single-objective MPC units with adaptive dynamic reference (ADR) models to control the multivariable. The controller first obtains the optimum voltage vector for each output port by minimizing the output current errors, where the ADR model generates the appropriate references. Finally, the controller identifies the optimum state from the determined voltage vector pair by minimizing the FC voltage errors. Unlike conventional MPC, the CMPC algorithm reduces the computational burden of the controller and attains multivariable control without additional control loops and weighting factors. Furthermore, the converter's performance with CMPC algorithm is validated experimentally on a low-power hybrid microgrid.