IoT and high-end systems are driving the need for increased sophistication in analysis tools.
Ed.: This is the second of an occasional series by the authors of the 2019 iNEMI Roadmap. This information is excerpted from the roadmap, available from iNEMI (inemi.org/2019-roadmap-overview).
Previous iNEMI roadmaps indicated that increasing cost and time pressures were driving the electronics industry to rely more heavily on modeling, simulation and design tools (MS&DT) over experimental prototyping during product and technology development. That theme continues to be relevant and central.
The emerging challenges in the MS&DT area are bound on the one end by the emergence of the Internet of Things (IoT) and growth of many connected and smart devices, and on the other end by the complex, high-speed, high-bandwidth devices that represent the growing high-end (HE) systems market. This latter market supports and comprises the connected infrastructure for communicating and processing large amounts of data – the “big data” of the connected world comprised of smart and connected devices.
Thus, challenges for MS&DT now include analyzing at very high speeds/frequencies and being able to incorporate and holistically design and analyze complex integrated devices and systems. The revolutionary pace of innovation in the HE space now provides a compute environment and platform that can revolutionize the level of detail and the size of problem that can be analyzed. Having tools that can take advantage of this computing power in an effective manner is another critical area for development for the MS&DT area.
Being able to model and analyze accurately in the face of increasing complexity across many applications and markets is a key challenge. The complexity and diversity of devices and systems under the umbrella of these two growing areas drives a need for increased sophistication in the area of MS&DT, which also drives the need for improved modeling inputs – most critically, material parameters. This, in turn, drives a dependency on improved characterization techniques. In addition, the emergence of improved computing infrastructure to host MS&DT, along with the rise of algorithms that take advantage of these systems (e.g., artificial intelligence and machine learning), provides a rich environment and opportunity for improvements in thearea of MS&DT that take advantage of these capabilities.
In terms of design tools, the MS&DT chapter of the roadmap identified five broad technical areas in need of research and development or further implementation over the next two to five years (listed below in bold).
There is a lack of standardized EDA tools (electrical and thermal) for 2.5-D and 3-D systems. 2.5-D and 3-D-specific EDA toolsets must be developed and validated.
No full-blown EMI (electromagnetic interaction) models address all PCB-level components. Instead, they are limited to single components. Modeling and simulation of the full PCB, including components, must advance up to wireless frequencies (GHz range).
Current silicon-package and package-board co-design tools are not optimized to give correct design/performance tradeoffs and have long iteration times. Tools that can perform co-design optimization, along with electrical modeling capabilities, are needed. Predictability of solution cost vs. performance will also be beneficial. Integration of optical components needs to be addressed in the toolset.
Actual prototype build results are different from simulation results. Modeling and simulation tools with built-in correlation capability with collected data are needed.
Often, designs work well during prototype build and test, but suffer systematic volume production yield loss. Next-generation tools should be able to predict yield and process capability index for volume production. Such simulation should integrate materials, process and line setup variations into an early production margin analysis.
Longer term development and implementation are expected of algorithmic modeling interface (AMI)-based SerDes modeling and simulation tools, and chip-to-chip system channel optimization tools for cost/performance. (The channel includes chip, package, PCB, connector and/or backplane.). Tools will evolve to incorporate machine learning and AI to make optimization faster and more efficient. •