A Wideband and Scalable Model of Spiral Inductors Using Space-Mapping Neural Network

Abstract

A wideband and scalable model of RF CMOS spiral inductors by virtue of a novel space-mapping neural network (SMNN) is presented. A new modified 2-equivalent circuit is used for constructing the SMNN model. This new modeling approach also exploits merits of space-mapping technology. This SMNN model has much enhanced learning and generalization capabilities. In comparison with the conventional neural network and the original 2model, this new SMNN model can map the input-output relationships with fewer hidden neurons and have higher reliability for generalization. As a consequence, this SMNN model can run as fast as an approximate equivalent circuit, yet preserve the accuracy of detailed electromagnetic simulations. Experiments are included to demonstrate merits and efficiency of this new approach. Index Terms-Modeling, neural networks, space mapping, spiral inductor. 0018-9480/$25.00 © 2007 IEEE Gaofeng Wang (S'93-M'95-SM'01) received the Ph.D. degree in electrical engineering from the University of Wisconsin-Milwaukee, in 1993, and the Ph.D. degree in scientific computing from Stanford University, Stanford, CA, in 2001.From 1988 to 1990, he was with the has been the Chief Technology Officer (CTO) with Siargo Inc., San Jose, CA. He is also currently a Professor and the Head of the CJ Huang Information Technical Research Institute, Wuhan University. He has authored or coauthored over 120 publications. He holds six U.S. patents. His research and development interests include integrated circuit (IC) and microelectromechanical systems (MEMS) design and simulation, computational electromagnetics, electronic design automation, and wavelet applications in engineering.

Analysis of scalable two-π equivalent-circuit model for on-chip spiral inductors

“…According to the Eqs. (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14), the number of independent elements in Figure 1 is reduced form 28 to only seven (L s1 , L p11 , R p11 , C ox1 , R sub1 , C sub1 , C s ). Then, these seven independent model elements will be calculated by the scalable model comprises of the geometry and process parameters.…”

Broadband scalable compact circuit model for on‐chip spiral inductors by neural network

“…For example, so as to study the dependency of the self-resonance frequency on design parameters during the design phase, the model needs to be accurate beyond the self-resonance frequency so that a reliable sensitivity analysis of the self-resonance frequency can be performed. 14 It is obvious that the proposed parameter-extraction technique for the physics-based T topology model is reliable and can be used for developing a scalable model for on-chip spiral inductors. The space mapping by neural network can be effectively used to construct the broadband and scalable model for on-chip spiral inductors.…”

“…7 The study of applying artificial neural network (ANN) techniques for modeling of microwave passive devices has been reported. [14][15][16][17][18] The ANN can easily map complex nonlinear input-output relationships through an automated training process.…”

A new model of on-chip inductors on ferrite film using KB-FDSMN neural network

“…It is a common practice to use the S-parameters for training the neural network models of RF and microwave components [24]. The S-parameters are herein calculated by using the full-wave EM simulation HFSS, from Ansoft Corporation [25].…”