Issue link: https://iconnect007.uberflip.com/i/1436094
30 DESIGN007 MAGAZINE I DECEMBER 2021 Here's where the "art vs. science" discussion comes in, because knowing how much auto- mation to use requires an artist's touch. Many designers still do things manually instead of using the tool's automation to save time and costs, while reducing the number of errors. Don't get me wrong; many designers have great success with, for example, routing boards by hand. While doing things manually may be acceptable within one company's ecosystem, there are many engineering teams and compa- nies who are taking advantage of today's auto- mation and horsepower to be better, faster, and ultimately more successful. I've read a lot recently about the develop- ment of AI and IoT, and I believe AI offers true potential for PCB designers. But we must be willing to think and act outside the box. Indus- try research suggests that there are crucial electronic systems design areas where AI tech- nologies, including machine learning, can be applied to minimize or eliminate the complex- ity of electronic system design work. is topic had my internal gears turning. I reflected on a recent project design in which we implemented hardware IP design re-use with Xpedition Enterprise, and I realized that today's EDA tools already utilize some forms of AI. Implementing a complex schematic- driven constraint approach while creating the schematic, and automating component clus- ter placement while implementing IP design re-use blocks, is possible because of AI. Auto- routing and sketch routing are also the results of limited AI functionality. ese capabilities lead us to significantly reduce our overall design cycle time for three designs vs. the typical manual approach. e potential for long-term cost and labor savings from creating and unitizing the hardware IP re-use blocks alone is huge. But despite the high level of horsepower and automation in our EDA tools, they are still limited because of the human limitation factor. What if we removed the human limitations factors by applying AI to the actual design and layout of these PCBs? If we were to replace the human limitations factor with an AI function- ality that could design boards based on rules and constraints from algorithms of best prac- tices and lessons learned from human design- ers, can you see the potential for success? AI has been implemented for many years in the IC and soware worlds with great success. I don't see why PCB design would not fall in line with this. We already have high-level auto- mation built into today's tools, and AI might not be such a big step aer all. But allowing AI to design PCBs and electronic systems is a scary thought because it will remove much of the human touch from PCB design. I see the art and science of PCB design inter- twined with AI. Will we lose the art form, artistic side, or crasmanship of PCB design? I don't think so. We will just continue to evolve. Whether a design is completed manu- ally or with automation, they both look great. If the mechanical and electrical constraints are implemented and adhered to, and then the final design meets the final requirements, looks are irrelevant of the final design. All we care about is that it works and that it can be built with higher yields, on time, and under budget. In the end, utilizing AI in PCB design will require us to trust the tool and utilize its horse- power to our advantage. And even the most "intelligent" AI tool will always require a human touch. Are we ready to take that step? Many senior designers with many decades of experience would probably say No. But the PCB designers of tomorrow are a different story. ese "digi- tal natives" have grown up with cellphones and mp3s, and they might wind up embracing AI in PCB design. We'll know soon enough. DESIGN007 Stephen V. Chavez, MIT, CID+, is the chairman of PCEA.