Issue link: https://iconnect007.uberflip.com/i/679419
May 2016 • The PCB Design Magazine 27 made to open routing channels and land sizes adjusted automatically to ease routing based on design rules. The AI router could decide which layer and direction to route a bus in order to alleviate bottlenecks that generally occur in the center of the board. The PCB project deliverables such as Gerber, IPC-2581B or ODB++ and pick-and-place files could be automatically generated based on the established design constraints. Fabrication doc- umentation could also be auto-completed based on established standards. On the assembly side, Internet of Things (IoT) manufacturing is currently being de- ployed. This system supports live bidirectional data flow between all electronics manufactur- ing shop floor machines and processes creat- ing a "smart" factory with "plug and play" de- ployment. This allows an organization to over- come the bottleneck of establishing efficient machine-to-machine and machine-to-human communications. I'm sure both you and I could come up with many more ideas to intelligently automate the PCB design process if time permitted. And al- though such a tool is many years off, the con- cept behind the EDA AI tool looks promising. PCB designers can certainly use all the help they can get. And, having the design tool intelligent- ly automate their work flow could substantially speed time-to-market. However, there is already considerable resis- tance to AI with some suggesting it many cre- ate too many erroneous results to be useful, and cannot be trusted. There is also the potential downside that AI might work too well and re- duce the skill level needed for the role of the designer to simply indicating the goal and ex- amining the results—but that is a long way off! Points to Remember: • Dale Parker, CAD manager with Google X, said in his Design Forum keynote that EDA tool vendors need to trash all their old 1990s code and start over, this time with artificial intelli- gence. • Current EDA tools, with all their bells and whistles, are still very limited in automation processes and mostly rely on the skills and fore- sight of the engineer and PCB designer. • Automating many of the tedious steps in set- ting up the initial database would be a good start. • Predictive text could provide self-evident naming conventions for supplier part numbers and database fields greatly speeding-up the de- sign definition. • A selection of predefined library compo- nents could be offered, based on an initial bill of materials, and pre-placed on the schematic predicting the designer's requirements. • The IBIS model's source and load imped- ances could be extracted to assign the required impedance and terminations to each individual transmission line. • The board stackup could be created based on previous designs with similar technology. • Functional blocks that the designer is working on, could be scrutinized in order to an- ticipate which blocks might be useful in future designs. • Intelligent forward and back annotation would be a definite godsend. • Placement changes could be made to open routing channels and land sizes adjusted to ease routing based on design rules. • The project deliverables such as Gerber, IPC-2581B or ODB++ and pick and place files could be automatically generated based on the established design constraints. • IoT manufacturing is currently being de- ployed creating a "smart" factory with plug and play. PCBDESIGN References 1. Richard Quinnell, AI for software devel- opment, www.edn.com. 2. Mentor, Valor software documentation, www.mentor.com. 3. The ICD Stackup and PDN Planner soft- ware can be downloaded from www.icd.com.au. Barry Olney is managing director of In-Circuit Design Pty Ltd (ICD), Australia. This PCB design service bureau specializes in board-level simulation, and has developed the ICD Stackup Planner and ICD PDN Planner software. To read past col- umns, or to contact Olney, click here. THE CASE FOR ARTIFICIAL INTELLIGENCE IN EDA TOOLS