Title :
Statistical design techniques for high-speed circuit boards with correlated structure distributions
Author :
Mikazuki, Tetsuo ; Matsui, Norio
Author_Institution :
Phys. Design Res. Group, NTT Commun. Switching Labs., Musashino, Japan
fDate :
3/1/1994 12:00:00 AM
Abstract :
New statistical techniques have been developed for designing the line structures of high-speed circuit boards. These techniques take into account the correlations between dimensional deviations caused by the fabrication process. Monte Carlo simulation is introduced into the finite-element method in order to calculate distributions of line parameters, and distributed circuit models are used to calculate the corresponding eye-patterns at the receiving end of transmission lines with parallel and stub parts. The simulator has an algorithm to save computer time. This paper compares the calculated and experimental results and discusses the accuracy of the simulation. Calculation results imply that the evaluation of the dimensional distributions due to fabrication processes is very important. The timing and amplitude margins, for example, were found to strongly depend on fabrication deviations. These techniques may be easily applied to the control of fabrication deviations to ensure that high-speed circuit boards meet electrical requirements
Keywords :
Monte Carlo methods; circuit CAD; circuit analysis computing; finite element analysis; printed circuit design; statistical analysis; Monte Carlo simulation; PCB design; correlated structure distributions; distributed circuit models; fabrication deviations; finite-element method; high-speed circuit boards; line structures; statistical design techniques; Acoustic reflection; Circuit noise; Circuit simulation; Communication switching; Computational modeling; Crosstalk; Distributed parameter circuits; Fabrication; Finite element methods; Printed circuits;
Journal_Title :
Components, Packaging, and Manufacturing Technology, Part A, IEEE Transactions on