Weekend Wisdom - Is the age of Gerber over?
I have in my bookshelf, an autographed copy of "The Inventor's Dilemma, The Remarkable Life of H. Joseph Gerber." I purchased the book, written by the subject's son, David Gerber, in early April 2016 with the knowledge that I would be meeting David later that month at the Boston Embedded Systems Conference.
The photo here shows David Gerber and me, both wearing wireless mesh networking show badges, designed for the conference by a different David (Ewing) and built by Screaming Circuits. I listened intently to Gerber's session and then had some time to speak with him one on one after. He signed my copy of his book: "To Duane, With best wishes and admiration for innovators such as you! David"
As of this writing, the book seems to be out of print on Amazon (and I'm not parting with my copy), but if you can find a used issue someplace, I'd highly recommend it. It's a fascinating story of how innovation in one industry (garment manufacturing) can have far-reaching impact on seeming unrelated industries (electronics design and manufacturing).
The file format we know generically as the "Gerber file" goes back decades to the line of Gerber photo plotters first produced in the 1960s. The initial version of the standard we still rely on today was published in 1980. When I joined Screaming Circuits in 2005, an updated version of that file format was virtually the only means of getting PCBs fabricated and parts assembled onto those boards. Six years ago, when I met David Gerber, newer, more capable standards were in the works and even being used to some degree, but the Gerber format was still the most used file format.
Today, the Gerber file is ubiquitous. Everyone knows its limitations, strengths, and how to create electronics with it. That's the good. The bad is that a lot of what is needed for automated assembly processing is not in the most common version of the Gerber file format. Newer standards, like IPC-2581, ODB++, ASCII CAD formats (collectively referred to as: "intelligent CAD data") and even the newest Gerber X3, promise to meet the needs of today's automated factories. But do they?
One of the biggest limitations of the Gerber file is that it doesn't have a means of differentiating a component pad from any other bit of copper. The metal layers are just metal or no metal so there is no direct way to know the position of a component, the rotation or the side of the board. An assembly shop, like us, has to ask for a centroid file, which gives component location, board side and rotation, and a bill of materials, which has all of the other component attributes. We then have to match data from all three files. Some of it can be done with software, but some parts of the process require a human. The new formats promise to have most or all of that in one file suitable for automatic processing.
We do see more and more engineers that can and do deliver intelligent CAD data. That's a good thing. What we don't really know at the moment is how many engineers cannot deliver intelligent CAD data. We'd love to hear your thoughts on the matter.
PCBs, anyone need denim prepreg?
If you're about to place an order on Screaming Circuits www.screamingcircuits.com, are you planning on using an intelligent CAD data format? (We also consider Eagle CAD .brd files as intelligent CAD data). If you aren't, can you do so? If not, why not?