The world’s electronics industry is in a period of extensive expansion. Everyday objects are getting smarter as embedded electronics appear in a broader range of devices than ever before. Established electromagnetic areas, including antenna design and placement, actuators, sensors and electromagnetic compatibility, are also on the rise thanks to this widespread global electrification.
As a result, the simulation and modelling of both low- and high-frequency electronics has seen explosive growth in the last few years to bring innovative products to the market in shorter timescales and to match the pace of consumer demand.
This rapid growth shows no signs of stopping and applies to both recognised electromagnetic (EM) markets including automotive, marine, aerospace, communications, power engineering, consumer goods, but also to new, upcoming applications. For example, the Internet of Things (IoT), autonomous driving, the conversion of mechanical systems to use electrical power, the introduction of smarter products with sensors and actuators in the connected world, and 3D printing applications all now need to integrate EM simulation and modelling to a greater or lesser extent.
This presents many challenges from both a technical and human perspective. For example, Ulrich Jakobus, senior vice president – electromagnetic solutions at Altair, said: ‘One challenge is to map the reality into an EM simulation model, both in terms of geometry and material characteristics. For the geometry, Altair has great expertise in terms of CAD import / de-featuring and meshing through FE modelling tools in the HyperWorks platform, complementing the native capabilities of the EM products. But often the decision remains with the EM engineer what level of geometrical detail is required for a particular EM simulation.’
For instance, in an urban propagation modelling scenario, is it necessary to include all the trees and the leaves on the branches? And how do you adapt this model for different seasons when the leaves will and will not be present? It’s a thorny issue, as Jakobus added: ‘Typically, some level of abstraction or model simplification is necessary when mapping the reality into an EM simulation model based on experience.’
Then, there are the technical challenges of covering the whole frequency spectrum. These are two-fold. Firstly, the applications are very different so, for example, the optimisation of an induction motor versus the bio-electromagnetic modelling of an MRI system represent two completely different problems. And, secondly, in order to provide efficient EM solutions, different algorithms must be provided, including full wave, high frequency asymptotic, 2D solvers, empirical models, and so on.
Jakobus said: ‘Altair manages this very well through many different solvers available in its EM tools, partially also hybridised with full mutual coupling. For example, using an integral equation-based full wave solver to model an antenna while using a high-frequency asymptotic solver to describe the interaction of the platform where the antenna is mounted on.’
What’s more, EM simulation and modelling tools not only need to link more directly with common electronics design environments but also introduce greater automation and simplified assembly modelling into the process. Lawrence Williams, director of technology at ANSYS, said: ‘Many electrical engineers avoided studying 3D electromagnetics because their interest was digital design, but high speeds and higher electronics density has thrust EM effects upon many more engineers. ANSYS has responded by making EM analysis link more directly with common electronics design environments.‘
When it comes to simulating EM at high frequencies, existing tools will also need substantially more computational power. Jiyoun Munn, technical product manager of RF (radio frequency) at COMSOL, explained: ‘This could be resolved through the use of cluster computing. However, in one of our recent releases, the RF Module introduced an example simulation app, which simulates a single slot-coupled microstrip patch antenna that is fabricated on a multilayered low temperature co-fired ceramic (LTCC) substrate.’
The results include the far-field radiation pattern of the antenna array and its directivity. The far-field radiation pattern is approximated by multiplying the array factor and the single antenna radiation pattern to perform an efficient far-field analysis without simulating a complicated full array model. Munn said: ‘We wanted to show how to efficiently evaluate many antenna components by running a 90-second analysis that gives you reliable results, rather than running the analysis for about two days, which is what would happen if a different approach was used.’
Over the past several years, COMSOL has focused on bringing an intuitive and easy-to-use yet powerful user interface for multiphysics modelling. Munn added: ‘Simulation speed is, of course, important to everyone as well. When it comes to EM simulation, everything is relative to the wavelength and that defines the size of the problem. So, for subwavelength devices such as mobile phone antennas, simulation specialists can easily model the entire object since it does not require intensive computational resources. To address large-scale problems, COMSOL allows this process to be expedited through not only full 3D simulations but also very efficient 2D axisymmetric formulation for symmetric structures.’
With electromagnetic simulations penetrating new domains, EM simulation engineers often face the challenge of missing electromagnetic material parameters, which suppliers are often not in a position to provide. Jakobus said: ‘So, simplified or generic material parameters must be used, or they must be obtained through measurements or from other separate EM simulations to be then used at a macroscopic level.’
Altair’s vision is to transform design and decision making by applying simulation, machine learning and optimisation available throughout the lifecycle of its increasingly-electronic products. ‘This is achieved by a combination of offering simulation software within the HyperWorks platform of innovation, complemented by our software-related services and consultancy offerings,’ according to Jakobus.
3D printing company PROTIQ uses Altair’s suite of electromagnetic simulation and modelling tools to develop innovative geometries for induction heating components. These new geometries are then additively manufactured in high conductive copper. Johannes Lohn, head of research and development at PROTIQ, explained: ‘Simulating the magnetic flux and the inductive heating helps us to avoid iterative testing with expensive prototype geometries.’
Conventionally, manufactured inductor coils are limited to rectangular or round cross sections as their effect and form of the magnetic field is well known. Lohn said: ‘These limitations in geometry can be overcome by additive manufacturing, which enables us to increase efficiency or reduce cycle times. The challenge that we face with the new freedom in inductor design, is that the magnetic flux also becomes way more complex. Simulating the heating process in 3D helps us to understand the effect of new designs and to develop complex new geometries.’
‘In the future, we will use EM modelling and simulation to build an even deeper understanding of the induction heating process and the generated magnetic fields,’ Lohn added.
The increasing proportion of electronics in our vehicles means the automotive industry, in particular, is increasingly using EM simulation and modelling. Jean-Claude Kedzia, product manager of computational electromagnetics at ESI Group, explained: ‘The major trend that could be identified in the past few years (and is still alive and well) is the desire/willingness of industrial customers to manage fully equipped models featuring all relevant contributors to the overall electromagnetic environment and/or to the product behaviour (including the 3D structure with materials, internal wiring and cabling, electronic equipment, antennas, sensors, etc.). This is mainly observed in the automotive sector, but also in aeronautics or defence for instance.’
The major emerging application automotive application areas include advanced driver assistance systems (ADAS) and connected cars with intelligent traffic systems (ITS). These newer systems need to be modelled in conjunction with the traditional electronic components in a vehicle.
The resulting complexity means that more than one software tool is required for this myriad of components and systems. Kedzia said: ‘From a practical standpoint (also because of the multi-scale modelling that is needed), several solvers should be used to handle all those contributors.’
This requires the introduction of chaining, coupling and/or hybrid techniques, as well as co-simulation, which is being used in many places, according to Kedzia, who added: ‘When the customer’s’ request is to handle a fully equipped model within its real operating environment, the key challenge for the software provider is to combine various simulation tools in an easy to use but accurate manner. The key criterion is not to make it ‘perfect’ but to make it ‘smart’ with relevant assumptions duly accepted by the customer (and well communicated to them) and with clear limitations.’
A key feature of ESI Group’s CEM One solution is its ability to manage integrated sensors within their operating environment. ‘Those devices can be first characterised through various options and then embedded within their full 3D operating environment with limited effort. The sensor location and orientation can be easily changed, no need for tedious and time-consuming remeshing work, emitting devices can be combined within the same model,’ Kedzia added.
The next logical progression for the automotive sector is self-driving cars, which will be heavily instrumented with sensors, sensor processing and fusion, embedded controls, and links to the network.
Kedzia said: ‘When it comes to driverless cars, one extra feature should be added related to connected drive in the city with its huge diversity, namely an urban environment with all those scenarios that may occur in real life.’
Kedzia added: ‘Standard 3D simulation does not fit (in real time) and dedicated platforms are thus needed. Within ESI Group, the answer relies on the Pro-SiVIC platform featuring RADAR sensors (short and long range) but also cameras, LIDAR devices and other ultrasonic equipment.’
Similar to 4G and the 3G network before it, 5G is the next-generation wireless network, which is widely predicted to unleash a deluge of new innovations thanks its huge network capacity and ultra-low latency.
The 5G network aims to be 100 times faster than the 4G LTE communications standard and increase broadband connection speeds by up to 10 times.
However, the challenges of 5G are ‘huge’, according to Williams, who added: ‘Today, companies like Skyworks use multiple radios simultaneously to achieve some 5G functionality on existing 4G networks using carrier aggregation. Multiple radios operating simultaneously generate more heat. As a result, thermal analysis coupled with the electromagnetics becomes very important. Of course, higher frequencies (millimetre-wave) are also a huge new challenge [with 5G networks].’
In the case of 5G, a COMSOL customer has used the company’s simulation and modelling tools to design specialised connectors for high speed RF applications. Munn said: ‘The opportunity here is to use simulation to optimise the RF connectors that will transmit the data to appear electrically invisible. Because COMSOL is inherently multiphysics in this situation, you are able to model the complexity that occurs in developing products for the next generation of communication.’
Munn added: ‘Our customers working on EM simulation are exploring next generation RF, microwave, and millimetre wave applications such as 5G, IoT, and high-speed interconnects. When researchers want to closely examine products that will bring 5G to life beyond the electromagnetic spectrum of microwaves and optics, they often need to incorporate for example, heat transfer and structural mechanics.’
The resulting expansion of electromagnetic devices into a broader range of products and devices means, in turn, that a broader range of individuals now need to understand and work with electromagnetic simulation and modelling techniques to effectively design these components and systems.
Consequently, simulation and modelling vendors must put an increased emphasis on the accessibility and user-friendliness of their EM tools. This theme is prevalent across the board. Williams said: ‘ANSYS has also built much greater automation and simplified assembly modelling into the process. Design organisations now can build out customised design flows that leverage the physics without the engineer needing to become an EM or thermal expert.’
Williams added: ‘We have built out a platform that integrates semiconductors, IC packaging, printed circuit boards into the analysis to predict full product behaviour comprehensively.’
Additionally, COMSOL’s Application Builder provides simulation experts with the tools needed to turn their detailed physics and mathematical models into easy-to-use simulation apps for use by everyone in their organisation and beyond.
However, building these complex models for systems with an increasingly larger EM footprint can also cause issues for simulation experts. Kedzia said: ‘An obvious extra challenge is the human effort (workload) required to access the results. No matter what the CPU time is (provided this value remains acceptable), the key point is the time spent by the operator to prepare the model, to access input data, to gather mandatory information, to clean the CAD model, etc.’
In some cases, the required information is so difficult and uneasy to access that experimental measurements are preferred: a typical example is the equivalent scheme of onboard electronic equipment, but the same comment also applies to basic parameters characterising plastic coating of wires. Kedzia added: ‘This is not a big deal when dealing with a simple twisted pair, but quite relevant when hundreds of wires are gathered within the same bundle.’
‘For a successful deployment, numerical simulation should allow easy to get measured data to be inputted instead of very complicated (and quite uneasy to determine) models,’ Kedzia concluded.