Microwave Journal European Editor Richard Mumford recently had the opportunity to follow-up this month’s product feature article on CST STUDIO SUITE™ version 2008 with a discussion with Dr. Bernhard Wagner, Managing Director of Sales, Marketing and Finance at CST - Computer Simulation Technology.
Dr. Wagner gained his Masters (Diplom Ingenieur) in electrical engineering from the University of Technology, Darmstadt, Germany, a “Diplome de Specialisation” in Telecommunications, from the Ecole Supérieure d'Electricité, France, and a PhD in the area of numerical simulation of electromagnetic fields. Prior to joining CST, Dr. Wagner worked as a consultant, providing simulation and design services with the company CSS, which he co-founded in 1996. In his current position at CST, Dr. Wagner is committed to developing CST's presence world-wide, and was instrumental in the founding of CST of America® Inc. and CST of Korea.
MWJ: Thank you for taking time from your busy schedule to meet with Microwave Journal. CST is a pioneer of 3D EM software and your portfolio now includes numerous solvers including Transient, Frequency Domain, Eigenmode, fast S-Parameter and Integral Equation. What is the reasoning behind providing these various solver types?
BW: Most vendors strive to improve their “favourite” method. This is quite understandable as almost every method has room for improvement and it is easier to continue along a well known path then to develop expertise in a neighbouring area. In addition, vendors tend to believe that their method is superior to others. The truth, however, is that no one method is perfect for each application. In order to ensure optimum performance, we have developed the “Complete Technology for 3D EM” approach, integrating various methods seamlessly in one interface. This means that once the structure model has been set up, the most appropriate technology, in terms of speed and accuracy, can be selected to solve it without changing the setup or interface.
MWJ: Do the various solvers address diverse specialized high-frequency engineering problems?
BW: Some solvers are indeed designed to tackle smaller segments of the overall EM market, whereas both the Time domain and Frequency domain solvers address a very broad range of applications. There are standard recommendations indicating when to use which of the two general purpose solvers, but at the risk of sounding complicated, it is not possible to say that one solver is always the best choice for antennas and the other better for all PCB structures.
MWJ: Does a single engineer often use multiple solvers for a single project? Do engineers often mix and match these different solvers for a single engineering effort and if so could you give an interesting example?
BW: Our customers use the time domain for the majority of applications, but more than one third of our users have already added other solvers from CST STUDIO SUITE™, notably the frequency domain solver. There are numerous reasons to use multiple solvers for the same project. It can simply be the need for cross verification for critical designs. But it can also be that various parts of the same structure are simulated with various solvers and the results are then assembled. To illustrate, let’s look at the analysis of a complex antenna placed on a large object. First the antenna is calculated alone using a general purpose solver such as the Time Domain solver. The fields from this analysis can then be used as a source in the Integral Equation solver which in turn calculates the effects of placing the antenna on a large object.
MWJ: How do these multiple solvers fit into the aim of aiding workflow?
BW: Having the right tool for the job is basic to efficient electromagnetic analysis, but if they cannot communicate with each other, unnecessary overhead in model preparation and structure imports will occur. CST’s approach ensures a minimum of time is spent, as once the model has been set-up the engineer can decide which numerical method to choose, without having to change interface.
MWJ: Specific solvers are likely to get their imported data from very specific sources such as EDA or 3D CAD, so did your development team have to tweak the automation for a given workflow or to optimize the solvers?
BW: As the model setup is the same for all solvers, all imported data can be used in every solver. The difference is that the various solvers react differently in case of geometric inaccuracies of the imported model. We have recently developed a new technique that makes our time domain solver capable of dealing with CAD models tolerances. This is particularly important for the analysis of complete handsets or complex models from the EDA world. Solvers based on tetrahedral meshes are typically less robust. To solve this problem, CST offers an intelligent and automatic healing process, which can overcome many, but not all, difficulties inherent in flawed data.
MWJ:Who is driving the call for improved workflow — the software vendors, leading technologists, rank and file device manufacturers or others?
BW: The most immediate pressure comes from real world applications that our customers want to simulate today. Among others, mobile phone companies have steadily increased their demands for workflow automation. The trend is no longer to simplify the model, which can be a tedious process, but to include all parts even those that can be neglected in the EM simulation. This saves valuable working time and reduces the necessity for engineers to decide which parts are critical to keep in the simulation model.
MWJ: Are organizations adopting workflow automation as fast as it can be developed or is there a time lag between software innovations and application in the field?
BW: This varies alot and cannot be answered in general way. It is clear that some companies see a stronger potential in streamlining the workflow than others.
MWJ: Are you working with anyone specific in defining and testing new workflow automation?
BW: Not really. We work with a large number of customers and try to resolve all workflow issues as this, sooner or later, will be beneficial to other users as well. Requests detailing what users would like to do in 3 or 5 years from now are essential to our roadmap planning. We are, of course, cooperating with some of the large EDA players to develop concepts for the future.
MWJ: Workflow diversity is dependent on the practicality of the engineering problems and organizational infrastructure, so is the customization of workflow critical?
BW: Absolutely, many customers have a very individual workflow and it is therefore requisite that a software package offers the ability to easily connect to other packages e.g. via OLE interface, macro language, parameterized models, and a wide range of import and export capabilities.
MWJ: Can you explain more about customized workflow? For instance, is your support team becoming more focused on helping organizations to develop the appropriate customized workflow automation? And will CST ever have a ‘services’ group dedicated to designing and customizing workflow, design kits, etc.?
BW: We haven’t really considered having a special group dedicated to customization, but our team does include experts in this area. It should also be mentioned that, on a number of occasions, our R+D team has been heavily involved in the implementation of very specific functions or program calls to simplify the customization process, and this type of help will continue to be available. MWJ: Can model import be simple, quick and accurate or is it really a case where design complexity is such that workflow is an activity that should be addressed by skilled professionals?
BW: We’ve consistently focused on developing and updating our import options and we have certainly succeeded in simplifying the procedure in many respects. On the other hand, the ever increasing complexity of structures continues to set new challenges. It would be presumptuous to assume that we can stay on top of these developments - also as a user - without the appropriate knowledge. That said, we can import structures easily today that seemed near to impossible a couple of years ago.
MWJ: What should someone implementing such a workflow be most concerned about with regard to design re-entry and likely issues such as defining ports (sources and loads)? For instance, is there a way to identify objects as likely ports from the model source in the case of a high number of terminals, vias or other multiple objects? Is this operation best handled in CST DESIGN ENVIRONMENT?
BW: Most of a layout’s features can be imported directly: the material and stack-up information, the geometry itself, components, drilling holes, vias and the pads. The error probability is very low, and thanks to the advanced healing capabilities, this process is very robust. Usually no port information is available in the layout, and so they have to be added manually in the CST import layout view. The port location is stored and available for further use.
MWJ: What is the state of the art with regard to a robust bi-directional workflow, wherein designs brought into an EM simulator for analysis and even parametric optimization can be exported back into the original EDA tool with the appropriate mechanical detail?
BW: Currently the workflow is very unidirectional, though the COM/DCOM mechanism, which we have been using for a long time, can be very helpful in passing on information about required parameter changes on and to realise a high degree of automation. Back annotation will be one of the key concerns in future developments.
MWJ: Does co-simulation with tools from Agilent and AWR mean that those environments can also control geometric and material parameters of the 3D model or simulation parameters? Are the simulation results (S-parameter) directly embedded into the circuit or system simulation? Is any field data passed back to these environments?
BW: There are basically two ways to integrate a 3D EM tool into a circuit simulator like Agilent ADS or AWR Microwave Office. The first option is to use the layout created inside this simulator as an import file. In this case all parameters are taken from the circuit simulator (geometry as well as materials), and the ports are created automatically, in fact everything is set-up in the circuit simulator and the 3D EM tool is nearly invisible. The other option is to create a parameterized 3D model and make it available to the circuit simulator. The circuit simulator can control all 3D parameters, geometry or material, which has been made accessible. CST has developed a very efficient result caching and interpolation technology, which makes EM/circuit co-simulation really feasible. CST MWS returns the S-parameters for the circuit and system simulation. Field data is generally more difficult to handle, which means that high-level tools cannot deal with it efficiently. Our own block-schematic tool, CST DESIGN STUDIO, can also create and access the respective fields.
MWJ: The ability to input field data from other tools such as Sigrity’s Speed2000 and SimLab’s PCBmod for use in a CST based simulation is a new and very interesting feature. Is the intended application for an engineer to be able to couple the radiation from a complex high speed PCB to a device chassis for EMI/EMC studies?
BW: Yes, that is the idea.
MWJ: Do you envisage other useful applications for this capability?
BW: EMI/EMC studies are the main focus here. It would of course be interesting to model the feedback as well.
MWJ: CST appears to be having to support and develop two quite different use models for its technology. One is within your own environment, requiring all the development effort to design a user interface and desired automation/functionality. The other is to embed your solvers in someone else’s environment as seamlessly as possible. In the not too distant future, which do you see as having the greater number of engineers using it and would this ever lead to two truly separate products?
BW: It is true that there are 2 or even more possible ways to use our software. The traditional one can be typified by an engineer with a deep understanding of electromagnetic field theory who designs and invents new field-based devices, or tries to investigate secondary effects of the devices. This user performs his virtual experiments inside our design environment and heavily relies on all the work that goes into improving CST STUDIO SUITE as a whole. The second way lies mainly in the usage of our software to add the accuracy of 3D to otherwise high-level models to circuit and PCB simulators. Here the solution engine shall be hidden from the user as well as possible. Although the number of engineers who can use our software in the second way is much larger, the first type of engineers will remain our main clientele for a longer time to come.
MWJ: Bernhard, I’d like to thank for taking the time to answer a few of our questions about various EM solver technologies and the evolving engineering requirements for integration into various workflows. It seems that EM simulation software has entered the mainstream of design and will continue to improve in capacity and automation in order to keep pace with engineering complexity. Cheers.