Silicon Carbide (SiC) power transistors have fast switching speed but are more inclined to be affected by parasitics. Miller capacitor ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C<sub>gd</sub></i> ) and common source inductor ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L<sub>s</sub><sub>com</sub></i> ) are two major cross coupled parameters across gate driver loop and power loop, which cause overshoots, crosstalk and decreased switching speed. To ensure a reliable and efficient switching, the impacts of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C<sub>gd</sub></i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L<sub>s</sub><sub>com</sub></i> need to be identified and fully compensated in the gate driver. However, state-of-arts failed to provide a straightforward compensation approach with the coupling effects extracted and un-coupled model accurately represented. In this paper, a novel SiC device model with decoupled equivalent miller capacitance and common-mode inductance extracted for both gate driver loop and power loop is presented following the Miller theorem in micro-electronics. The decoupled equivalent circuit is verified to have decent accuracy in predicting the actual SiC switching transients. Furthermore, this paper proposed an improved resonant gate driver (RGD) design using the proposed decoupling model under the effects of common-mode inductance. The pre-charging current and driving pulse width are adjusted to ensure a full resonance driving transient with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L<sub>scom</sub></i> . The experimental results show that impact of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L<sub>scom</sub></i> is fully compensated, where SiC's <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">di<sub>ds</sub>/dt</i> is improved from 1.88A/ns to 2.102A/ns, which is close to the results without <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L<sub>scom</sub></i> .