Electromagnetic Simulation of High-Speed and RF Multi-Chip Modules — An Overview

The coplanar waveguide (CPW) gap discontinuity is an important component of many applications in the monolithic microwave integrated circuits (MMICs). Therefore, accurate modeling of the CPW gap discontinuity is very important and fundamental for the design of MMICs. In this work, a new lumped equiva1ent-circuit model which can accurately describe the characterization of the CPW gap discontinuity from 0 to 150 GHz is proposed. The parameters are extracted from electromagnetic (EM) simulations simply and conveniently without any complicated optimization algorithms. And we make the model scalable in gap length, width and space. Satisfactory fitting accuracy is obtained between the model data and EM simulations, which verifies that our model can be implemented in computer-aided design software to reduce design time and save computer resources.

Embedded wafer level ball grid array (eWLB) or FO-WLP (Fan-out wafer-level packaging) is investigated as a package for MMICs (Monolithic Microwave Integrated Circuit) for automotive radar applications in the 77GHz range. Special focus is put on the thermo-mechanical performance to achieve automotive quality targets. The typical fatigue modes “solder ball fatigue” and “copper fatigue”, evolving during thermo-mechanical stress like cycling on board will be discussed. Simulation as well as experimental preparation results for typical fatigue levels are given. In addition, several influencing parameters are listed and rated regarding their effectiveness. The theoretical framework why solder ball fatigue is the only failure mode causing electrical failure is provided. The impact of different thermo-mechanically driven fatigue modes is discussed. The two important parameters to be considered for the functionality of the Radar system are RF (Radio Frequency) and thermal performance. For elaborating the RF performance with present fatigue modes, the phase shift between different channels and pads is analyzed by full-wave EM (Electromagnetic) simulation. It is found that for fatigue levels up to 90% the phase shift stays below specification for single fatigue modes and may approach specification only for an unlikely combination of all 90% fatigue modes. For assessing the thermal performance with present fatigue modes, thermal simulation as well as thermal measurements are used. Assuming 50% degradation in average for all thermal balls, an increase in RTH of up to about 30% is seen. On average for all thermal measurements, the deviation between measurement and simulation is within ±1°C.

A bond-wire-free interconnection between monolithic microwave integrated circuit (MMIC) and antenna using a stacked patch configuration is investigated. An edge-fed patch on a gallium arsenide (GaAs) MMIC chip drives a patch antenna integrated in the lid of the MMIC package. The lid is formed using a liquid crystal polymer (LCP) substrate. The first implementation using a laminated multichip module (MCM-L) process is presented that covers the designated E-band spectrum for long-range wireless communications (71–86 GHz). Electromagnetic simulations and measurements of antenna radiation patterns agree well over the whole frequency range of interest. Important design aspects and manufacturing tolerances specific for the implementation of the interface in MCM-L millimeter-wave front-ends are presented. An alternative design is proposed for improved radiation patterns across the band.

A method for first-pass MMIC (monolithic microwave integrated circuit) design success is presented. The power of statistical design and modeling is combined with that of circuit-level electromagnetic simulation in a CAD (computer-aided-design) system, providing the ultimate means for predicting production success. The performance of the proposed CAD system is verified by comparing simulated and actual statistical response characteristics of a three-stage, 7-11-GHz low-noise-amplifier GaAs MMIC.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Conexant Systems Inc. builds single and multi-die IC components for industrial and consumer data communications, telecom, and telephony applications. The IC package technology base allows mixing of analog, RF, and digital functions of noncompatible device families on a low cost interconnect substrate. The multichip module (MCM) technologies are currently being used to design radio frequency (RF) MCM products for wireless and broadband network communication applications. Experience in the use of MCM technology has been developed for nearly a decade by product designers, which has in turn resulted in system-on-a-package implementation long before the fad for system-on-a-chip started. In this paper, the primary features of RF MCM design methodology, electrical simulation tools, and design for high volume manufacturing trade-offs are discussed. Each RF MCM may have one to three CMOS, bipolar, and/or GaAs die, and 10 to 100 discretes depending on its size and functionality. In this paper, we discuss a laminate MCM design that meets performance requirements and can be manufactured on a high volume assembly line.

Sheet- and film-type soft magnetic materials have been widely used to suppress unexpected noise in mobile and wearable devices using ferromagnetic resonance (FMR) losses [1][2]. The designs of these magnetic noise suppressors are mainly based on large-scale electromagnetic field simulations and "trial and error" experiences. In previous studies, the effect of the complex permeability of the magnetic sheet and film on circuit parameters of transmission lines was investigated by a simple magnetic circuit analysis using only the cross-sectional size and material parameters [3]. However, the effect on the reflection and the transmission coefficients have not been estimated quantitatively and the design guideline of the noise suppressor has not been clarified.In this paper, the estimation method of the reflection and the transmission coefficients of a transmission line with a noise suppressor is investigated including the effect of the complex permeability. A 95 μm-width microstrip line (MSL) with a Co-Zr-Nb film is discussed as a test bench. The electrical equivalent circuit parameters of the MSL with Co-Zr-Nb film were calculated by the magnetic circuit analysis and the characteristic impedance was estimated. The thickness of the Co-Zr-Nb film was determined considering impedance matching below 1 GHz range, where it was assumed that a signal frequency range, and the conductive noise suppression by the Co-Zr-Nb film at 3 GHz was estimated.Figure 1 illustrates the magnetic circuit in the cross-sectional view of the MSL with the Co-Zr-Nb film. The MSL was fabricated on low-temperature co-fired ceramic (LTCC) substrate with the thickness h=100 µm and the relative permittivity of 9.8. The width, thickness and length of the signal line are ws=95 μm, ts=3 μm, and 10 mm, respectively. The Co-Zr-Nb film was fabricated using radio frequency (RF) sputtering and deposited on a SiO2 substrate. The film thickness t was obtained by the magnetic circuit analysis and the electromagnetic simulation so that the characteristic impedance of the MSL with and without the film have a same value. The electrical equivalent circuit of the MSL with the Co-Zr-Nb film was obtained to calculate the characteristic impedance. The circuit consists of series impedance Z = R + jωL = (Rm+Re+Rs) + jωLm and parallel conductance Y = jωC, where Rm and Lm were calculated by the magnetic circuit analysis considering the effect of the complex permeability of the Co-Zr-Nb film. Rs and Re were the resistance of the signal line including the skin effect and the increase of resistance due to the eddy current loss generated in the film [3], respectively. C = 2 pF calculated by experimental value at 0.1 GHz [4].Figure 2(a) shows the estimated characteristic impedance Z0 = (Z/Y)0.5 with t = 0.1-1 μm. Below 1 GHz, the impedance is roughly constant and the estimated impedance of the MSL with 1-μm-thick film is almost the same as the measured value of MSL without the film. Therefore, t = 1 μm was used to calculate the reflection and the transmission coefficient.Figure 2(b) and (c) shows the reflection coefficient |S11|, transmission coefficient |S21|, and the noise suppression Ploss/Pin = 1-(|s21|2 +|s11|2) of the MSL with 1-μm-thick film obtained by the electrical equivalent circuit analysis. In the case with the film, |S11| and |S21| were almost the same as the measured value within about 8 dB and 0.1 dB below 5 GHz, respectively. |S11| increased up to 8 dB below 1 GHz but the value kept less than -25 dB. On the other hand, |S21| decreased as the frequency increased because of the combination of FMR and Joule losses in each film at 3 GHz. Ploss/Pin is maximized at 3 GHz.In summary, the reflection and the transmission coefficients of a transmission line with a noise suppressor were quantitatively estimated including the effect of the complex permeability using the magnetic circuit analysis using the cross-sectional size and material parameters. The thickness of the Co-Zr-Nb film was determined considering characteristic impedance matching. As a result, the reflection coefficient was lower than -25 dB below 1 GHz, and the conductive noise suppression was obtained at 3 GHz. Those results give a way to clarify the design guideline of the noise suppressors. This work was supported in part by JSPS KAKENHI Grant Number 20K04497. Part of this work was carried out under the Cooperative Research Project Program of the Research Institute of Electrical Communication, Tohoku University. **

Current radio frequency (RF) surface coils are made of rigid or semi-flexible materials with limited bending properties. Flexible RF coils are not only more ergonomic, but they also offer another opportunity to increase the signal-to-noise ratio, which is a fundamental limiting factor in magnetic resonance imaging (MRI) quality. Electromagnetic (EM) simulation is important for the design and performance evaluation of MR RF coils. However, EM simulation is often time-consuming due to the large number of mesh cells and complex calculations required for dimension modeling of the flexible distributed capacitance coils with lesser dielectric substrate thickness. To improve the simulation efficiency, a novel method was proposed whereby the dielectric thickness and dielectric constant were simultaneously increased, which allows to reduce the number of mesh cells while maintain the accuracy of the simulation. The performance was validated through a comparison with the actual dimension simulation approach. The results demonstrate that the proposed method saves approximately 78% on both one simulation time and the total simulation time, whereas the variations of simulation results of both EM fields and specific absorption rate (SAR) are less than 7%. The feasibility of the proposed novel method was validated through the performance of the distributed capacitance coil prototype in phantom studies.

This article tackles the generalized synthesis of antenna arrays using two-order deep learning. Existing deep learning-assisted antenna synthesis approaches mainly rely on model training using electromagnetic (EM) simulation data, and hence feature limited generalization ability and the need for a huge amount of EM simulations. The proposed two-order deep learning method uses the first-order model to learn the generic features of radiation patterns from the data efficiently generated by applying conventional array factors. After that, the second-order model learns from EM simulations to capture the detailed pattern variations due to concrete coupling effects in the case of a specific array arrangement with different operating frequencies, radiation structures, and feeding schemes. Therefore, the two-order DL model can predict the radiation patterns of a series of antenna arrays while reducing the needed amount of EM simulation data. Implementation is carried out on a series of patch antenna arrays to verify the feasibility and robustness of the proposed approach. The validation includes conditions for the array operating at arbitrary new frequencies, with modified radiation structures or new feeding schemes. The results show that the proposed two-order model provides a prediction accuracy of about 84 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\%$</tex-math></inline-formula> for a series of 1 × 4 antenna arrays, clearly outperforming the existing regular one-order DL model, which obtains around 65 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\%$</tex-math></inline-formula> with the same EM simulation data. The proposed method reveals a promising direction for applying deep learning to assist the design and analysis of antenna arrays.

We inferred the effective sheath width using the cutoff probe and incorporating a full-wave three-dimensional electromagnetic (EM) simulation. The EM simulation reproduced the experimentally obtained plasma-sheath resonance (PSR) on the microwave transmission (S21) spectrum well. The PSR frequency has a one-to-one correspondence with the width of the vacuum layer assumed to be the effective sheath in the EM simulation model. The sheath width was estimated by matching the S21 spectra of the experiment and the EM simulation for different widths of the sheath. We found that the inferred sheath widths quantitatively and qualitatively agree with the sheath width measured by incorporating an equivalent circuit model. These results demonstrate the excellent potential of the cutoff probe for inferring the effective sheath width from its experimental spectrum data.

This paper studies the accuracy and limitations of different electromagnetic (EM) simulation approaches for dielectric radar benchmark targets. In comparison to perfect electric conductors (PEC), dielectric targets are much more complex from an EM point of view. However, dielectric materials are widely used for modern aircraft and unmanned aerial vehicles (UAVs). Apart from that, EM simulation tools nowadays play an important role in the design of aerial vehicles and radar systems. In order to evaluate the accuracy and performance of different EM simulation approaches, the results obtained using Physical Optics in combination with a ray tracing approach are compared to full-wave reference data.

An important step of the microwave design process is the adjustment of geometry and material parameters of the structure under consideration to make it meet given performance requirements. Nowadays, it is typically conducted using full-wave electromagnetic (EM) simulations. Because accurate high-fidelity simulations are computationally expensive, automation of this process is quite challenging. In particular, the use of conventional numerical optimization algorithms may be prohibitive as these methods normally require a large number of objective function evaluations (and, consequently, EM simulations) to converge. The adjoint sensitivity technique that recently become available in commercial EM simulation software packages can be utilized to speed up the EM-driven design optimization process either by utilizing the sensitivity information in conventional gradient-based algorithms or by combining it with surrogate-based approaches. Here, several recent methods and algorithms for microwave design optimization using adjoint sensitivity are reviewed. We discuss advantages and disadvantages of these techniques and illustrate them through numerical examples.

This study proposes an antenna simulation framework powered by graphics processing units (GPUs) based on an open-source electromagnetic (EM) simulation software (gprMax) for machine learning applications of antenna design and optimization. Furthermore, it compares the simulation results with those obtained through commercial EM software. The proposed software framework for machine learning and surrogate model applications will produce antenna data sets consisting of a large number of antenna simulation results using GPUs. Although machine learning methods can attain the optimum solutions for many problems, they are known to be data-hungry and require a great deal of samples for the training stage of the algorithms. However, producing a sufficient number of training samples in EM applications within a limited time is challenging due to the high computational complexity of EM simulations. Therefore, GPUs are utilized in this study to simulate a large number of antennas with predefined or random antenna shape parameters to produce data sets. Moreover, this study also compares various machine learning and deep learning models in terms of antenna parameter estimation performance. This study demonstrates that an entry-level GPU substantially outperforms a high-end CPU in terms of computational performance, while a high-end gaming GPU can achieve around 18 times more computational performance compared to a high-end CPU. Moreover, it is shown that the open-source EM simulation software can deliver similar results to those obtained via commercial software in the simulation of microstrip antennas when the spatial resolution of the simulations is sufficiently fine.

A millimeter-wave (mmWave) gallium nitride (GaN) high-power amplifier (HPA) monolithic microwave-integrated circuit (MMIC) was implemented, considering a source via effect. In this paper, we introduce guidelines for designing GaN HPA MMICs, from device sizing to meeting high-power specifications, power matching considering source via effects, schematic design of three-stage amplifier structures, and electromagnetic (EM) simulation. Based on the results of load pull simulation and small-signal maximum stable gain (MSG) simulation, the GaN high-electron-mobility transistor (HEMT) size was selected to be 8 × 70 μm. However, since the source via model provided by the foundry was significantly different from the EM results, it was necessary to readjust the power matching considering this. Additionally, when selecting the source via size, the larger the size, the easier the matching, but since the layout of the peripheral bias circuit is not possible, a compromise was required considering the actual layout. To prevent in-band oscillation, an RC parallel circuit was added to the input matching circuit, and low-frequency oscillation was solved by adding a gate resistor on the PCB module. The proposed PA was fabricated with a commercial 0.1 μm GaN HEMT MMIC process. It exhibited 38.56 to 39.71 dBm output power (Pout), 14.2 to 16.7 dB linear gain, and 14.1% to 18.2% power-added efficiency (PAE) in the upper Ka band. The fabricated GaN power amplifier MMIC shows competitive Pout in the upper Ka band above 33 GHz.

In this manuscript, a new lumped equivalent-circuit model is proposed for the simulation of the coplanar waveguide (CPW) step discontinuity within the frequency range of 0 to 150 GHz. With the computer-aided modeling and appropriate parameter extraction method, the parameters can be extracted from electromagnetic (EM) simulations without any optimization. Furthermore, we also make the model scalable. Excellent agreement is obtained between the model data and electromagnetic simulations over a considerable range of frequencies and device geometries, which also verifies the validity of our model.

Driven by the continued advance of integrated circuit (IC) technology, large-scale electromagnetic (EM) modeling and simulation problems have been encountered across the design of on-chip circuits, package, and die-package interfaces. To give a few examples, full-chip post-layout performance verification that includes transmission line and full-wave effects, system-level signal and noise integrity analysis, and the characterization of broadband die-package interaction for clean power delivery. Two major difficulties exist for the analysis of the aforementioned large-scale problems. One is the large problem size. The modeling of a die, a package, and a combined die-package system results in numerical problems of ultra-large scale, requiring billions of parameters to describe them accurately. The other is the multiscale nature of the problem. An electromagnetic simulator is required to span scale ranges of at least 10000:1 to analyze a combined die-package system. Recently, a time domain Orthogonal Finite-Element Reduction-Recovery method (OrFE-RR) was developed to handle the large problem size associated with the design of very large-scale integrated circuits [1]. In this method, a set of orthogonal prism vector bases are developed. With this set of bases, the original ultralarge scale 3-D system of order TV is rigorously reduced to a 2-D single-layered system of order Mwith negligible computational cost, where Mis much less than N. The reduced 2D system is diagonal, and hence can be solved readily. After obtaining the solution of the reduced system, the rest of the solutions can be recovered in linear complexity. This method entails no theoretical approximation. It applies to any arbitrarily-shaped multilayer structure involving inhomogeneous materials. Numerical experiments have demonstrated its accuracy and capacity in simulating very large-scale on-chip, package, and die-package interaction problems.