Here is the abstract you requested from the IMAPS_2011 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|Studying the Impact of Return Current Path on the EM Simulation of High-Speed Package Designs|
|Keywords: return path, package, EM simulation|
|In high-end flip-chip and ball-grid array (BGA) packages, multiple power and ground planes are used to keep power supply noise low; vias are also used to connect power planes at multiple locations. Electrical performances of power supply systems have been characterized by using effective inductors, however the model is valid at low frequencies only. For high-end packages, the frequency range of interest is from DC to tens of GHz, therefore there are several resonant frequencies and the effective inductor model may become totally invalid. Accurate characterization of packages necessitates three dimensional (3D) electromagnetic field simulations to take into account all the various electromagnetic interactions. Typical multi-layer package designs contain many fine details which dramatically increase the simulation time and memory requirements for commercial 3D electromagnetic (EM) field solvers. It is therefore common practice to truncate the three-dimensional package model leaving only a few important features surrounding the nets of interest in order to reduce the simulation time without sacrificing the accuracy of the results. This approach is based on the assumption that the return currents and electromagnetic fields at high frequencies primarily exist in the proximity of the signal trace. In this paper we demonstrate that this simplification can have a significant impact of the simulation results if it is not performed carefully and it can introduce spurious/non physical resonances. In order to study the effects of truncation we first examine the impact of the simulation conditions (both physical and numerical) on a simplified test structure. Through these results we try to develop an understanding of boundary conditions and the role of cavity resonances. Next we apply this knowledge to the simulation of a realistic complex multilayer package model and demonstrate that through properly truncating the geometry good correlation with measurement results can be achieved over a wide frequency range. Finally, we present general simulation guidelines to improve model accuracy and will also discuss trade-offs related to accuracy vs. simulation time.|
|Antonio Ciccomancini Scogna, Principal Engineer
CST of America