Has PCB design run out of steam? - Embedded.com

Has PCB design run out of steam?

The multiple conflicting demands and mandates on PC boards may soon result in there being no acceptable design solution.

Every system and circuit design is a balancing act among trade-offs of size, weight, power, reliability, performance, and many other parameters. After all, that’s what engineering design is largely about, working through that combination of hard numbers and analysis leavened by experience and judgment to assess, balance, and execute the trade-offs.

In some cases, one or a few objectives may have the most critical weighting by far, so other objectives may be sacrificed to some extent to satisfy those top ones. In other cases, the weighting and balancing among trade-offs and choices is a more reflective and iterative process: “should we give up 5% of runtime if it improves sensor precision by 15%?,” is the sort of soft-judgment question which is often hard to quantify and harder to answer.  Of course, many design objectives are just that: aspirational goals where some must be met while others are ones for which the design strives but are not absolute “must haves.”

Among the must-haves are the many EMC, efficiency, and safety mandates that various regulatory and standards organizations bodies (both government and industry) have put in place. Unlike performance objectives where there is some give-and take, many of these mandates are absolute: you either meet them or the design will not be approved and certified. They have little “wiggle room” or areas for trade-offs, except for the specific approach and tactics you choose to meet some of them.

I thought about this issue as I finished the final part of a ten-part series “Evaluation of EMC Emissions and Ground Techniques on 1- and 2-layer PCBs with Power Converters” published by In Compliance (this final part has links to all nine previous parts). The series detailed the many performance- and mandate-related issues associated with EMC-acceptable design as well as the somewhat-related ground considerations (Figure 1).

click for full size image

Figure 1: A circuit board must be simultaneously viewed and assessed across its thermal, power-distribution, and EMC domains, as well as other perspectives. (Image sources: Electronic Concepts and Engineering, Inc.; Open Airbus Cockpit; ResearchGate).

Reading this article series was both exhilarating and discouraging. It was the former because it showed how much was understood and explained by the three authors with respect to theory, practice, measurements, and more. Using this insight, designers should have a better chance at devising a layout for power, ground, and DC/DC switching regulators which would meet system-performance needs and pass certification. So far, so good.

But the article also had me concerned: it made me even more aware of the many expectations placed on the design and, by extension, on the design team. There are so many best-practice and often conflicting guidelines points to adhere to, and so many trade-offs where doing something good also has a negative effect. Some of these imperatives are defined by the laws of physics and Maxwell’s equations while others due to well-intentioned regulatory-compliance standards. In many cases, you need a compliance expert to guide you to, through, and past the bewildering thicket of standards.

The above-cited article concerned relatively simple one- and two-sided PC boards, but many designs are now on boards with four, eight, and more layers. In some ways, having additional layers available makes it easier to meet the EMC and ground requirements, as there are more degrees of freedom for ground planes and other good things; in other ways, having them complicates design as there are many more paths for signal routing and current flow, emission sources, and emission-sensitive pickup points.

We’re asking a lot of circuit and board designs these days, of course. We routinely have modest-size boards handling tens and hundreds of amps, which inherently brings IR drop and connection-resistance issues. Plus, nearly of that supply current gets converted to heat, so there are thermal issues in dissipating all that heat to that magical, mystical place called “away.”

At some point, I fear we will run out of running room. The multiple demands—DC electrical, signal integrity, EMC, thermal, isolation, creepage/clearance—placed on PC board design will yield either a null set, or the need for severe compromise on power levels, circuit density, thermal density, EMC performance, size…it’s a long list.

Certainly, thinking that we’ve reached the end of what we can deliver and that we can’t go further is not new in engineering. Somehow, we find a way to get past it with new materials, techniques, components, and other innovations. After all, that’s been the story of Moore’s “law” (a brilliant and prescient supposition, but not a law, sorry) at each major node. Perhaps the engineering imperative is somewhat like Samuel Beckett wrote at the end of his 1953 novel “The Unnamable”: “… you must go on. I can’t go on. I’ll go on.” Or it could be what Samuel C. Florman was referring to by the title of his book “The Existential Pleasures of Engineering.”

Still, at some point, radical new approaches are needed. Right now, I don’t see any such breakthroughs, nor do I see designers stopping their attempt at fitting more components and more functions at higher frequencies and dissipation onto those PC boards. Perhaps there is a three-prong fork in the PC-board road: one path leads to a dead end and we stop; the middle path is one of slow, steady, incremental progress; and the final path is some sort of somewhat revolutionary approach such as a shift to fully integrated optics which are low power and also have no EMC issues.

What’s your sense of the board-design situation? Are we asymptotically approaching a limit on what we can achieve with the technology?  Will the path ahead be one of steady progress made possible by a series of small steps? Or will some breakthrough technology we really don’t see clearly yet change the entire situation and enable significant advances, just as the transition from hand-placed components and hand-wired circuitry to pick-and-place and PC boards did?

>> This article was originally published on our sister site, EDN.

Related Contents:

For more Embedded, subscribe to Embedded’s weekly email newsletter.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.