Abstract: The commutation error of the ironless-stator permanent magnet brushless DC motor (IPMBLDCM) can degrade motor performance. The present BLDCM drivers mostly employ a single-stage full-bridge inverter or a two-stage topology with a full-bridge rear stage, and the circuit state under various commutation error scenarios is investigated on this basis. Meanwhile, present commutation error correction approaches mostly employ PI controllers, but their effectiveness declines when dealing with nonlinear time-varying systems. To address the commutation error problem in a two-stage topology drive consisting of a Buck converter cascaded with a three-phase half-bridge, the motor terminal voltage before and after the commutation moment in electric and dynamic braking states is analyzed in detail in various commutation error cases, and a method of commutation error characterization that can be used in four-quadrant operation state is provided. On this basis, a controller based on the model-free adaptive control method is designed, as well as a one-step estimation method for commutation error using the model studied in this paper, which accelerates the controller's convergence speed and allows for the rapid identification and correction of the commutation error. Finally, the proposed theory and controller are tested experimentally using a digitally based motor control system.