I want to simulate a high-frequency inductor coil, where the frequency is so high that the parasitic capacitance of the coil cannot be ignored, and it even exceeds the self-resonant frequency of the coil. However, I found that the simulated self-resonant frequency is higher than expected. Upon analysis, I suspect that this might be due to the mesh in the time-domain solver being too coarse, with mesh sizes larger than the spacing between coil turns, causing the mesh to directly bridge the gaps between turns. At the same time, because the spacing between turns is very small, if I try to refine the mesh to make it smaller than the gap between turns, the number of mesh cells becomes excessively large, reaching 400 to 500 million. Are there any good solutions for this situation? Or how should high-frequency coils be handled in simulations?