On-Time Mismatch Based System Identific...

This work introduces a fast and low-complexity natural frequency estimation concept for power converters, enabling their adaptive control. Neither a high resolution analog to digital converter nor complex arithmetic operations, such as multiplications, are required by the proposed approach. Instead, the mismatch between the on-time of the control signal and the effective on-time at the switching node, occurring under light load conditions, is exploited. This mismatch is caused by the parasitic drain-source capacitance and the parasitic body diode oft he power MOSFETs. It is not necessary that the value of the parasitic capacitance is known nor is it required to stay constant, e.g., over temperature. Due to its simplicity, the proposed method is well suited for practical on-chip realizations. The introduced concept is experimentally verified on the example of a buck converter with different output filter sets. A digital controller is adaptively tuned based on the estimated natural frequency, resulting in an improved dynamic performance. For the presented converter, the magnitude of the output voltage transient in response to a load change was reduced by 30%.