The following workflow illustrates one of the ways in which simulation can assist with the design process of a flexible, stripline, multichannel I/O interface.
The main motivation of the simulation-driven approach is to reduce the number of design iterations and enhance the performance and scalability of high-density quantum computing interconnect systems.
One of the key elements of the I/O architecture can be referred to as “flex-to-coax segment”. It can integrate components such as low-pass filters, attenuators, flex-PCB and flex-to-coax or flex-to-flex RF transitions to enable the modularity of the system.
To optimize performance of that segment, the electromagnetic (EM) simulation models can be used to evaluate and compare different designs. The workflow enables analyses of critical RF metrics including return loss, insertion loss, and crosstalk between channels, referred to as near- , far- end crosstalk (NEXT, FEXT).
Fig. 1. View of a flexible, stripline, multichannel I/O interface.
Components of a flex-to-coax segment
Let us focus on the simulation of the flex-to-coax segment.
All 3D components, which are shown in Fig. 2, can either be imported from external E-, M- CAD tools, or be built and parametrized in CST Studio Suite.
The evaluated structure contains four RF channels, which pass through two modules and a flex-PCB. Two of these channels incorporate low-pass filters that are housed within a module. Both modules are equipped with SMA connectors to facilitate seamless integration with external coaxial cables.
The modules and flex-PCB assembly are secured in place using clamps to ensure stable alignment and electrical connection within the segment.
Fig. 2. Detailed view of the simulation model of a flex-to-coax segment assembly.
Flex-PCB
A dedicated tool called EDA Import Wizard is available inside CST Studio Suite. This tool is used to import PCB layouts into the simulation environment, while preserving their stack-up and routing information. A 3D simulation model of the PCB can be bent automatically, based on bending information stored in the EDA data (available for Cadence layout files), or bent interactively using the CST bend tool (e.g. for ODB++ files).
Fig. 3. Import of the flex-PCB into CST Studio Suite software.
Fig. 4. PCB stack-up view.
The PCB stack-up configuration, including its material properties, is obtained from the EDA file during the import procedure.
It can be further customized by the user inside CST at any time.
The flex-PCB used in the workflow contains a total of four RF transmission channels.
Two stripline channels are routed on “layer 1” surrounded by GND “guard” areas along the sides of the PCB, and two others are routed on “layer 3”.
A 3D simulation can quickly provide the user with multiple results (shown in Fig.5) such as characteristic impedance (Zo), TDR and S-Parameters: return loss (RL), insertion loss (IL), allowing for a quick evaluation of a design. Based on how well the current model fulfils the design criteria, the user can perform parametric study and/or optimization to make further improvements.
Fig. 5. EM simulation results of the RF channel between port 3 and 7.
Low pass filter
CST Studio Suite offers a Filter Designer 3D tool that speeds up and optimizes the filter design process.
A low pass filter (LPF) properties, selected to be used in this workflow, are shown in Fig. 6. The filter has been chosen to operate as a stripline transmission line, and it will be placed inside the module.
The filter designer tool allows the user to quickly browse through different filter technologies available for the current design, and select the one that best matches the design requirements.
Fig. 6. View of the filter design in CST software.
The Filter Designer 3D enables rapid transformation of filter specifications such as bandwidth, ripple, and transmission zeros, into a 3D filter model.
Key capabilities include:
Automatic generation of coupling matrices and practical filter topologies without requiring manual synthesis.
Automatic conversion of synthesized filter into a parameterized 3D EM model within CST Studio Suite for seamless electromagnetic validation.
An integrated workflow that connects circuit-level filter synthesis with full-wave EM simulation, reducing design iterations and accelerating development.
Automated tuning and optimization of the filter assembly to achieve target performance efficiently.
Additional information is either available through e-webinar, or in CST Knowledge Database.
3D E-field view in the passband | 3D E-field view in the stopband |
Fig. 7. LPF: Results view of the EM simulation.
Figure 7 shows the following results: RL, IL and group delay of a filter prototype, which was selected for the workflow.
Since the filter has been solved using a 3D full-wave solver (FEM), the user also has access to the 3D electromagnetic (EM) field distribution.
Flex-to-coax segment - assembly
The user can assemble and test the entire flex-to-coax segment once all of its crucial components have been designed and optimized.
Figure 8 shows RF channel description within the model. Figures 9a and 9b show the simulation results in the form of the S-Parameters for the entire flex-to-coax segment.
Fig. 8. View of the RF channels inside the assembly. Some housing elements are hidden from view.
Fig. 9a. EM simulation results of the assembly model: RF channel without LPF.
Fig. 9b. EM simulation results of the assembly model: RF channel with LPF.
Flex-to-coax segment: addressing potential issues in the design process
Although the results appear satisfactory, the engineer may have encountered several issues during the design process.
This is precisely where electromagnetic simulation comes into its own. It can provide valuable insights to help resolve issues at an early stage of development.
Let us consider the following scenario:
What would happen if the mechanical connection provided by the clamps, which are holding the modules and the flex-PCB together, did not provide adequate electromagnetic shielding?
Fig. 10. EM shielding of the clamps.
3D E-field view at 1100MHz |
Fig. 11. Simulation results of the EM shielding of the clamps.
As can be seen in Fig. 11, insufficient “watertightness” along the clamps seems to cause an EM resonance at 1100 MHz of the entire segment, resulting in a significant increase in EM interference to the surrounding area.
Simulation enables users to test various crosstalk suppression techniques. This step is especially beneficial when the aim is to deliver a well-optimized prototype in a cost-efficient manner.
As shown in Fig. 12, introducing metal dividers inside the module can significantly reduce NEXT between RF channels.
XTALK initialXTALK with isolation |
Fig. 12. RF channel isolation within the module.
Let us consider another scenario:
Is additional EM shielding of the flex-PCB essential?
The overall effectiveness of different shielding techniques can be investigated with simulation.
In this workflow example, the ink encapsulation technique is tested.
The selective openings in the flex-PCB, which are filled with conductive silver ink, will be called "fpc vias“ (Fig. 13).
The best way to test the model is to run an "immunity test“ with a dedicated virtual reverberation chamber workflow, offered by CST Studio Suite.
The virtual reverberation chamber workflow, in a nutshell, is as follows: The plane wave solver (PWS) ensures field uniformity inside the working volume by producing a set of field sources. The PWS delivers these field sources by using analytical calculations with Monte Carlo iterations to solve sets of waves with randomized polarizations and delays. The field sources are later used to excite the 3D model in a full-wave analysis to ensure that the standard deviation of the electric field uniformity does not exceed 3 dB across the entire computational domain and frequency range.
More information can be found in CST Knowledge Database: QA00000435955
Fig. 13. EM shielding of the flex-PCB.
Fig. 14a. EM shielding: PCB with fpc vias vs. PCB without fpc vias.
Fig. 14b. EM shielding: Different view on the results.
The outside noise level (e-field strength) has been set to 1 V/m.
As illustrated in Figure 14a, the immunity is around 10 dB better across the entire simulated frequency range for channels routed on layer 1 of the flex-PCB compared to channels on layer 3.
Moreover, it can be seen that the presence of fpc vias exerts an influence on the electrical length of the assembly, which pushes EM resonances towards higher frequencies.
CST Studio Suite streamlines the post-processing of results, providing users with a range of options for reviewing and analyzing their data. In this example, the ratio between two cases is demonstrated in Fig. 14b.
HF materials: Copper vs. Niobium Titanium
CST Studio Suite provides users with the option to test a variety of conductors. In order to meet the demand for quantum computing applications, it is necessary to simulate a superconductor material such as niobium titanium (NbTi).
In the workflow, a comparison is presented between that superconductor and a typical HF conductor made out of copper. These materials are used in the flex-PCB stack-up.
As shown in Fig. 16, the use of NbTi instead of copper will reduce the overall insertion loss across the RF channel.
Fig. 15. HF material setup in CST software.
Fig. 16. HF material comparison: S-Parameter results: IL, RL.
