Estimated reading time: 4 minutes

The Future of PCB Designs

I designed my first PCB using tape and etch-resist pens from RadioShack. I penciled the schematic on graph paper, drew the layout directly onto the single-sided copper-plated board, and then etched it in my parent’s garage. At the time, commercial PCB design wasn’t too different. PCBs were hand drawn on clear film and transferred to the copper board with a photoresist process.

Today, advanced CAD software is readily available for a few thousand dollars, and even for free. This rules-based design allows for quick and efficient schematic capture and PCB layout. The designs are exported in one of a few somewhat-standard file formats, shipped to a fabrication or assembly house, and built within a few days to a few weeks.

File Formats

The largest limitation today is the somewhat-standard data files. The Gerber format—the most common PCB data format, more or less—gets the job done, but it’s a non-intelligent bitmap format that’s past its prime. Anything that’s not obvious or non-machine-readable in a simple bitmap requires an additional file. The bill of materials (BOM), component location and rotation, and any supplementary datasets each need a different file. Keeping all of those files up to date and in sync is an error-prone and time-consuming process. Gerber must go. It’s not practical with today’s complex designs.

There are a few in-process attempts to replace the Gerber file. ODB++ and IPC-2581 are two emerging standards that include all of the data needed to fabricate and assemble a PCB. The sooner the industry universally adopts one of these file formats, the better.

Size and Complexity

One of the best ways to look at the future of PCB design is to start with the applications those boards will go into, and the requirements posed by those applications.

The big chips—ball grid array (BGA) and land grid array (LGA) chips with 1,000 or more pin connections—don’t seem to be getting much bigger. Even today, these are generally rare and expensive field programmable gate arrays (FPGAs) or central processing units (CPUs) from computers. I don’t see a lot of change on this end of the size spectrum. However, the midsize and small chips continue to get smaller and more highly integrated. The high volumes of mobile devices largely drive this. Now, with the popularity of wearable devices, prototype manufacturers will need to get used to designs that will fit onto a dime, including connectors, I/O, and power.

Vehicle automation will also take a significant toll on PCB designers. Automotive PCBs need to be kept down to two or three dozen square inches to allow for reliability in extreme environments. Large boards flex too much and can’t be tucked as easily into every nook and cranny in a car. While these boards are size limited, the processing capability needs to increase tenfold over the next few years.

PCB layout specialists need to be prepared for two size clusters: less than one square inch, and 25–36 square inches. These designs will need to be completed, verified, and tested in less time than ever.

Interchip Routing

One of the handier innovations in the microcontroller world is the inclusion of internal pin to peripheral routing. In some of the more advanced implementations, peripherals, such as pulse width modulation (PWM), can be routed to nearly any pin on the chip. This allows for great flexibility in PCB routing. The number of vias and long routes to the far side of a chip can be significantly reduced and has the potential to make those tiny and super-dense boards we’ll be working on much easier to lay out. However, the internal routing is generally done within the software development tools.

For example, microchip, has a visual tool that allows the software developer to map peripherals to pins. The tool then auto-generates the necessary configuration code. Next, the software developer needs to document the chip pin routing, and someone needs to make a PCB CAD library component matching the part. With this capability, a CAD designer now has to keep track of library parts for multiple microcontrollers and variations of a single microcontroller. If a chip configuration makes for awkward routing, the PCB designer has to go back to the software developer and make a new library component.

CAD schematic and layout tools will need to take on that peripheral/pin configuration job. In the future, it won’t be practical to have a manual path back and forth between the PCB routing and chip configuration. Both autorouters and human-guided routing need to have control over internal chip-pin routing.

Modularity and Supply Chain

Modularity won’t play as big of a role in hardware as it does in software until the component supply chain challenges can be easily accommodated. New chips are coming out—more integrated and specialized—at a pace that’s difficult to maintain. Last year’s module contained a three-chip, six-passive position sensor, and this year’s module might very well be a single-chip, three-passive set. The original chipset may suddenly become unavailable.

Same Look, Different Insides

From a distance, the PCB design process won’t look very different a decade from now. However, the details will be almost unrecognizable. Chip placement will be rules-based and largely automatic. Autorouting will be AI driven and actually work. The autorouter will extend down inside the chip and take peripheral routing out of the software toolset and into the PCB CAD toolset. Lastly, the supply chain variabilities will be automatically accommodated within the CAD software.

Duane Benson is chief technology champion at Screaming Circuits.