Search for: All records

This paper proposes a novel multi-objective inverse optimization method for high-speed interconnects based on a cascaded deep neural network (DNN) structure, which can efficiently optimize characteristic impedance, insertion loss, and far-end crosstalk (FEXT) simultaneously. Parameter optimization for high-speed interconnects is essential to the signal integrity and electrical performance of complex designs such as multilayer printed circuit boards (PCBs) and chiplets. Conventional optimization approaches often rely on numerous optimization iterations, which is highly time-consuming, especially in highdimensional parameter spaces. This paper proposes a novel DNNbased method by cascading an inverse-prediction network and a forward-prediction network to achieve multi-objective optimization for characteristic impedance, insertion loss, and FEXT by optimizing the trace width, trace spacing, and pair-topair distance. Further, by incorporating an integer programming technique, parameter optimization of multilayer PCBs, including the PCB stackup and design parameters of each signal layer, can be accomplished in seconds, much more efficiently than the conventional optimization approaches. 

This study shows that placing four NFPs in selective PCB layers significantly improves signal integrity, reducing minimum jitter from 18.28 ps to 12.81 ps and maximum jitter from 27.66 ps to 22.03 ps. 

The accuracy of mixed-mode S-parameter conversion is important for crosstalk mitigation in high-speed digital systems. However, the conventional mixed-mode S-parameter formulation assumes equal even-mode and odd-mode impedances ( Zoo=Zoe ), which limits its applicability, particularly in tightly coupled differential structures. In this article, we propose a novel mixed-mode S-parameter generalization (generalized M1/M2 approach) using an N-differential port network, which allows for multipair (i.e., pair-to-pair) crosstalk analysis on coupled differential systems, given by: [Smm]i×i= ([M1]i×i×[Ss]i×i+[M2]i×i)×([M1]i×i+ [M2]i×i×[Ss])−1. The proposed M1/M2 formulation eliminates the need for renormalization by integrating mode-dependent coupling factors koo and koe, ensuring a more physically meaningful representation of mixed-mode S-parameters, thereby improving the accuracy of both intrapair and interpair crosstalk analysis in high-speed digital systems. The effectiveness of the proposed M1/M2 approach is demonstrated through intrapair and interpair analysis on tightly coupled striplines, revealing peak-to-peak variations in differential return loss, interpair near-end crosstalk, and far-end crosstalk. Validation using a differential setup with commercial tools (Balun approach) confirmed the formulation's accuracy, with errors below 1%. In addition, measurement validation on a microstrip differential pair highlighted the model's scalability and precision, emphasizing the importance of incorporating mode-dependent impedance variations.