Printed circuit board (PCB) design has long been
considered a less-glamorous aspect of engineering than, say, microprocessor design
or application-software design. This, despite the fact that in areas like
communications-board design or consumer, the challenges confronting PCB designers
are bleeding edge.
It wasn't always
this way and PCB design can be really cool again, according
to one engineer and self-described inveterate doodler.
In fact, Saar Drimer argues that thinking
outside the PCB box means not only reframing decades-old form-vs.-function
questions, but laying the groundwork to inspire a new generation of engineers.
"I'm looking for applications where the
form contributes to function, like perhaps RF design, flexible circuits--where
the free hand and freedom of shape contribute to the function of the
board," Drimer said in a
Drimer is a Cambridge, U.K.-based hardware
engineer who has been designing different-shaped PCBs and has developed a tool
to that end, through his company Boldport.
He built a board honoring the late analog engineering guru Bob Pease and has
turned out PCBs in the shapes of butterflies, board games, and business cards.
He argues that boards in the 1960s didn't necessarily
adhere to 45-degree corners (and by extension, path routing), but the rise of
commercial EDA vendors ended that in the name of tool uniformity and
Fair enough. Every industry evolution or
creation is a tradeoff of existing processes and technologies for something
that promises additional productivity. But, Drimer argues, the only thing
preventing PCB designers from getting creative is a slow-to-change EDA world
that won't alter tried-and-true tools.
"EDA companies are like a huge ship sailing
on with little incentive to change," Drimer said. He argued the usefulness
of EDA tools is reaching "a certain limit" and companies "need
to offer people innovative and creative ways to manage those large designs."
He notes that in the software world, the dynamic
nature of open-source software has opened up a world of creative opportunity for
"The more openness that comes from EDA and
manufacturers, the more you would allow creative people to use the information for
better productivity and better project management. There will be more
innovation, more innovation in DRC, more innovation in routing, even though
routing is a difficult problem."
There are existing tools that can handle
"curvy" traces, but they're limited, said Drimer (pictured, left). Indeed, he argues that 45-degree angles aren't
optimal for high-speed board design. He expanded:
do want to go curvy in some cases because then the width of the trace is
uniform as it meanders, as it turns. Whereas, the 45-degree angle has kinks,
turns, and the characteristic impedance changes with those angles."
The tradeoff with
Drimer's approach, he noted, is that integration isn't optimized since extra area is
needed for layout.
As for the next
generation of engineers, there's potential here, Drimer said. The open-source
Rasperry Pi phenomenon is appreciated by engineers because they understand the
effort that went into innovating the platform. But a grade-school student
can't understand that. What the student might understand better is different
shaped boards, different color options, and so forth.
He rails about tools and design convention, but he's
no gadfly: His 2009 research dissertation was on security within volatile FPGAs.
Drimer right now is figuring out what the
business model is and seeking investors. Clearly he can design and produce
lovely and functional PC board designs. He offers a free open-source design
tool via Apache License 2.0 called PCBmodE written in Python. He's
considering a freemium model involving domain-specific add-on modules for
PCBmodE that, he added, address industry-specific challenges that aren't well served with
The existing version supports two-layer
board design but can scale to four or more, Drimer said.
"On top of productising PCBmodE, we're in the
process of designing a few products using PCBmodE as a tool. It's still early
days but we're working on a few concepts that we hope will work out to be good
--Bob Pease PCB; Secret Smart Watch; ESL Lives! (Great Reads
Last year I made a circular PCB with components and traces at 5-degree intervals. The clock-line is nearly 2 feet long, and being able to make it a uniform circular trace definitely improved the signal integrity. I have the 14" x 14" PCB hanging in my office; most people assume it's for a tester-head but they are wrong. It's for a large neon clock. To save costs, a single square PCB is fabbed, then I cut-out the circular main-board. The triangular-shaped corners are the other 4 PCBs needed for power supplies, display-drivers, connectors, etc. Wish I could post some jpeg's......
Thanks Brian for the writeup! I do think that the EDA industry needs a bit of a shake-up ;)
On the Boldport blog, I write about some of the issues we face, particularly these articles:
And for some background for PCBmodE
I'm looking forward to hearing people's thoughts about what I'm trying to achieve.