Take Advantage of 3D EM Simulation for Densely Packed Designs
by Chris DeMartino, Modelithics
Small form factors are often crucial for today’s applications, leading to scenarios in which surface-mount components are densely packed onto printed circuit boards (PCBs). When components are packed into a small space, the components can couple to one another because of their close proximity. Unfortunately, these coupling effects cannot be captured in simulations when equivalent-circuit models are used for the components. However, it is possible to capture these effects by performing a full-wave 3D electromagnetic (EM) simulation that includes models defined by 3D geometry and detailed material composition. This article will demonstrate how 3D models can be used to accurately predict the real-life performance of designs in which components are located very close to one another.
This analysis is conducted by performing planar EM/circuit co-simulations with Ansys Electronics Desktop and 3D EM simulations with Ansys HFSS. All component models used in these examples are included in the Modelithics® COMPLETE+3D Library, which is a collection of models for components from many popular vendors.
The COMPLETE+3D Library
Included in the Modelithics COMPLETE+3D Library are Microwave Global Models™ for capacitors, inductors, and resistors. In addition to Microwave Global Models, the Modelithics COMPLETE+3D Library adds a collection of 3D EM geometry models for components like inductors, capacitors, filters, packages, etc.1 These models, which are based on physical dimensions and material properties, are intended for full-wave EM simulations that can predict coupling effects. The Modelithics 3D EM geometry models are also encrypted to protect manufacturer IP.
The library also includes what are known as 3D brick models for multi-layer ceramic capacitors (MLCCs).2 As several challenges are associated with creating 3D EM geometry models for MLCCs, 3D brick models have been developed for these components as an alternative hybrid solution. A 3D brick model is a simplified approximation of the capacitor’s physical geometry.
However, 3D brick models alone do not account for internal device parasitics. So, to execute a 3D EM simulation with 3D brick models, the designer must perform a hybrid 3D co-simulation. This hybrid 3D co-simulation method combines the 3D brick model with the Microwave Global Model for the given capacitor to allow a complete 3D EM simulation.
Designers can perform 3D EM simulations that include both 3D EM geometry models and 3D brick models. In these scenarios, the simulation procedure first involves performing a 3D EM simulation that captures any interaction between the components. Next, a final 3D co-simulation is executed that includes the data from the 3D EM simulation along with Microwave Global Model(s) for the capacitor(s) represented by the 3D brick model(s).
Design Overview
For this analysis, a lumped-element bandpass filter design will be presented. The design includes inductors and capacitors from the TDK MHQ1005P and Passive Plus 0402N part families, respectively. The filter is designed using a 4 mil-thick Rogers RO4350B substrate. For comparison, the results of two simulations are presented for the design: a planar EM/circuit co-simulation using Ansys Electronics Desktop and a 3D EM simulation using Ansys HFSS.
Figure 1 shows a simplified schematic of this filter in which TDK MHQ1005P and Passive Plus 0402N devices are used for the inductors and capacitors. Microstrip interconnections must be added to turn this filter into a complete design. Figure 2 shows a layout of the bandpass filter in Ansys Electronics Desktop. The values of the components are shown in Table 1. Notice how the components are located close to one another to allow for a more compact design at the expense of potential EM interactions between components.
A planar EM/circuit co-simulation will be performed using Ansys’ 2.5-D Method of Moments (MoM) EM solver to simulate the layout shown. A schematic must then be created that includes the EM analysis data properly connected to the MHQ1005P inductors and 0402N capacitors. Simulating this schematic produces the final planar EM/circuit co-simulation results.
Figure 3 shows the same bandpass filter in Ansys HFSS. This project includes 3D EM geometry models for the TDK inductors and 3D brick models for the Passive Plus capacitors. A complete 3D EM simulation will be performed using the procedure explained earlier. That is, a 3D EM simulation is first performed, enabling any interactions between components to be captured. The next step is to perform a 3D co-simulation that includes the data from the 3D EM simulation properly connected to the capacitors. Simulating this schematic produces the final 3D EM simulation results.
The bandpass filter shown was also built and measured. Figure 4 illustrates one of the assembled filters that includes the same inductors and capacitors from the simulations. Figures 5 and 6 show the S-parameter results obtained from the planar EM/circuit co-simulation and the 3D EM simulation. They also show measured data of two filters. The 3D EM simulation results agree very well with the measured data.
However, the planar EM/circuit co-simulation results are shifted with respect to the 3D EM simulation results and the measured data. Specifically, the planar EM/circuit co-simulation results exhibit a lower 3 dB bandwidth of 100 MHz less than the measured data (Figure 7). As a result, the planar EM/circuit co-simulation with equivalent-circuit models did not reasonably predict real-life performance with complete accuracy. This discrepancy can be attributed to the close proximity between components, which was significant enough to impact the performance. In contrast, the 3D EM simulation did accurately predict real-life performance. Therefore, this example demonstrates how 3D models can capture coupling effects between components, allowing for accurate simulations of compact designs such as this one.
Conclusion
The analysis presented here should help designers better understand when 3D EM simulations with 3D models may be necessary and the effectiveness of these simulations when needed. For many designs, coupling effects between components are not a significant factor. For these scenarios, a planar EM/circuit co-simulation with Microwave Global Models is an acceptable simulation approach.
However, the design example presented here included components located very close to one another. For a compact design such as this, a designer may be unpleasantly surprised upon observing real-life performance after only performing a planar EM/circuit co-simulation with equivalent-circuit models. Due to the proximity of the components, a planar EM/circuit co-simulation with equivalent-circuit models failed to predict the real-life performance with complete accuracy.
That said, the example shown demonstrated how 3D models can accurately predict real-life performance when coupling effects between components are present. These effects can be captured by Modelithics 3D EM geometry models and 3D brick models. Designers should consider performing a 3D EM simulation for compact designs like the one shown here. In these instances, neglecting to perform a 3D EM simulation may result in a discrepancy when observing real-life performance that can lead to further action like bench tuning or in the worst case, PCB respins.
References
1. I. Bedford, E. Valentino, and L. Dunleavy, “Application Note 63: Introduction to Modelithics 3D Models in HFSS.”
https://www.modelithics.com/Literature/AppNote
2. I. Bedford, “Application Note 78: Using 3D Brick Models™ for Full-wave EM/Circuit Model Co-simulation of MLCC Capacitors in Ansys HFSS.”
https://www.modelithics.com/Literature/AppNote.
(171)
