Efficient direct and iterative electrodynamic analysis of geometrically complex MIC and MMIC structures
Efficient direct and iterative electrodynamic analysis of geometrically complex MIC and MMIC structures
复制标题
DOI：
10.1002/jnm.1660020306
复制
发表时间：
1989-09
影响因子：
1.6
通讯作者：
W. Wertgen;R. Jansen
中科院分区：
文献类型：
--
作者：
W. Wertgen;R. Jansen
由于电路封装密度和结构复杂性的增加，严格的全波分析技术目前在MIC，特别是MMIC的设计中受到了很大的关注。在本文中，与这些技术和以前的相关工作的问题进行了简要概述的导言。为了获得自洽性，电动绿色函数和相关的条款，然后制定屏蔽（M）MIC问题。最后的算子方程的数值解导出和相关的函数空间，并进行了讨论。本文的中心部分介绍了一种新的数值解使用离散化绿色的功能数据库技术。由此产生的线性方程组的几何复杂性，涉及多达约1000个未知数的直接反演求解;对于更高数量的未知数，迭代解生成。作为替代数据库技术的开发，各种谱域迭代解决方案已被写入和测试。这包括应用共轭梯度法的正常运营商方程（CGN算法），隐式迭代Galerkin近似称为修改的平面共轭梯度技术（MPCG）和单调收敛的迭代程序是一个版本的共轭剩余算法（CR）。补充这一点，（M）MIC设计数据的提取从获得的数值三维解和误差考虑。本文最后与各种分析的例子中，高的几何复杂性和验证的一些结果与测量和其他来源的数值数据进行比较。典型工作站（Micro VAX、HP 9000等）所需的CPU时间是温和的，从而呈现的技术作为有用的MIC和MMIC设计问题的解决方案。
Rigorous full -wave analysis techniques are presently receiving much attention in the design of MICs and, in particular, of MMICs due to increasing circuit packing densities and structural complexity. In this paper, the problems associated with such techniques and previous related work are briefly outlined in the introduction. To obtain self -consistency, the electrodynamic Green's functions and related terms are then formulated for the shielded (M)MIC problem. The final operator equation for the numerical solutions derived and the associated functions space are presented and discussed. The central portion of the paper describes a new numerical solution using a discretized Green's function database technique. The resulting linear system of equations is solved by direct inversion for geometrical complexities involving up to about 1000 unknowns; for a higher number of unknowns, an iterative solution is generated. As an alternative to the database technique developed, a variety of spectral domain iterative solutions has been written and tested as well. This includes application of the conjugate gradient method to the normal operator equation (CGN algorithm), an implicit iterative Galerkin approximation called the modified planar conjugate gradient technique (MPCG) and monotonically convergent iteration procedure being a version of the conjugate residual algorithm (CR). Supplementary to this, the extraction of (M)MIC design data from the numerical 3D solutions obtained and error considerations are presented. The paper concludes with a variety of analysis examples of medium to high geometrical complexity and with verification of some results by comparison with measurements and with numerical data from other sources. CPU times required on typical workstations (Micro VAX, HP 9000, etc.) are moderate, thus rendering the techniques presented as useful in the solution of MIC and MMIC design problems.