Roughly Speaking

Geoffrey Hazelett

The AI Revolution is Coming to PCB Design

System-driven part selection and routing are closer than ever.

it’s rare for a motivational keynote speaker to have a lasting impact on attendees. At Zuken Innovation World’s US iteration in June, however, Dan Thurmon certainly set the pace for the conference. The author of Off Balance on Purpose and a popular Ted Talk on embracing uncertainty, Thurmon’s address had attendees referencing and even reenacting portions of his opening presentation in their subsequent presentations. The motivational speaker helped set an energetic and educational tone for the conference that followed.

And thanks to the coming AI revolution, uncertainty is what’s ahead.

Zuken Innovation World is a series of conferences around the world dedicated to bringing customers, partners and Zuken experts together. A good portion of the eight concurrent training tracks at the June event in Scottsdale, AZ, was dedicated to sessions led by customer-users sharing design best practices and success stories. Surprisingly for a conference dedicated to a specific supplier, there was never even a hint of sales; this was a full-on conference dedicated to networking and educating, where one could learn about upcoming technologies, best practices, and novel implementations of toolsets that only a user would discover. At times, it was overwhelming to decide which of the concurrent sessions to attend, as multiple competing interesting presentations were occurring!

An experience unique to ZIW is its “360” session, a town hall-style meeting in which anyone is free to ask questions of upper management. With some well-publicized tech CEOs heading to prison for fraud, it is refreshing to experience a leadership team being so open and transparent with customers about opportunities and issues. A hot topic during the town hall was the announcement and release of the new artificial intelligence-enhanced PCB design software package and the laying out of a very aggressive roadmap for customer-trained AI “brains.” Kent McLeroth, CEO and president of Zuken USA, reassured attendees who expressed concerns over the ethics of bringing such capabilities to life and the potential for displacing humans. And despite the stated focus on training the AI to design like a human, he didn’t downplay the difficulty involved in capturing sufficient design data for proper training.

Asked during Zuken 360 how the project could be expedited to get into the hands of designers faster, McLeroth and the team were clear and forthright: cooperation and support from designers who provide good design will help with the training process. The AI brains are in place, they said, but more good design data are needed to train the AI. The promise of secure-site customer-trained AI was also in the aggressive near-term roadmap.

The timing couldn’t be more perfect for Zuken’s launch of its new AI-driven technology for PCB designs. Indeed, Zuken isn’t just hopping on the AI bandwagon; rather, its UK-based R&D team has been pioneering and building this engine for years. The concept that a designer could design functional blocks and the system would make part selections and routing to complete the design is closer than ever. While Zuken’s toolsets are used predominantly in Japan, it has made steady improvements and significant strides that should be of interest to the international community.

Multiple sessions presented by Zuken UK’s Kyle Miller covered the capabilities and some of the details of the methodology used by the teams bringing this project to life. It should also be noted there was significant training for high-speed digital designers on DDR4 layout and design.

I interpreted this embrace of AI by Zuken to be an ongoing shift in PCB design technology, similar to the shift programming languages made to object-oriented languages and away from assembly languages. Assembly, like manual place-and-route, will still have a place in niche environments, but object-oriented languages dominate modern programming environments for their ease of use, handling, and capabilities in allowing a programmer to do more without diving into nitty-gritty details every use. And to address the ethics of such AI-assisted designs, I think we will see more designers, not fewer, as result of AI systems assisting the management of esoteric rules and guidelines. While this might frighten some of the established design community, it should be reassuring to know new designers won’t have to burn their fingers on bad designs, that AI would assist in making the designs a success and in turn, teach the designers good design practices.

Yes, there is uncertainty ahead. But given the aging design and layout population, the ability to pass design practices and training through AI is something to embrace and encourage. I look forward to how AI brains revolutionize the PCB design community, and at the very least, help push the ball forward making PCB design more accessible to new designers. Article ending bug

Geoffrey Hazelett is a contributing editor to PCD&F/CIRCUITS ASSEMBLY. He is a technical sales specialist with more than 10 years’ experience in software quality engineering and sales of signal integrity software. He has a bachelor’s degree in electrical engineering;