Comparison of S-Parameter Low Frequency Interpolation Methods

When running time-domain simulations, not having a DC point can cause issues. It is thus necessary to employ methods to extract the low-frequency information and the DC point. In previous literature, methods such as linear, polynomial, and discrete Hilbert integrals have been utilized for interpolation, yet there is a lack of comparative analyses involving diverse low-frequency interpolation methods. In this study, six methods of interpolation are compared, including linear, quadratic, cubic spline, PCHIP, Barycentric, and Akima. Through testing, it is found that (i) when the lowest data point is at 300kHz, there is only a subtle difference among these methods, (ii) in the case of long channels, employing Akima and Barycentric interpolations exhibits more significant discrepancies compared to other methods, and (iii) choosing linear or PCHIP for interpolation in the frequency domain would be a more stable option.