Search for: All records

A growing interest in AI and deep learning applications necessitates the development of efficient power delivery networks (PDNs) to manage broadband impedance in high-speed switching environments. This paper evaluates the electromagnetic performance of horizontal, vertical, and embedded PDNs in fanout packages, with particular emphasis on vertical and embedded configurations, both less explored for noise mitigation. Simulations indicate that embedded PDNs significantly reduce commonmode (CM) and differential-mode (DM) radiation, achieving levels between 20-78 dBµV. Additionally, measurements show that embedded PDNs maintain a lower electric field strength of 1.9 V/m for DM radiation as compared to 2.7 V/m for horizontal PDNs and 11.4 V/m for vertical PDNs. Advanced shielding with 5-µm thick Cu-CoNiFe layers further reduces magnetic field emissions by 30-45 dB. These findings demonstrate that embedded PDNs are a more effective solution for electromagnetic interference (EMI) mitigation in future electronic systems, offering superior noise control and reduced crosstalk.