The Earth is no Longer Flat
On my morning walk yesterday – through some rain-forest-like trails near our house in Redmond, Washington – I was thinking about writing this blog, and the PCB industry’s adoption of IPC-2581 … I was reminded of something I read years ago that seemed like a reasonable parallel. Speaking about leadership and innovation, the book said that if you put a procession of pine tree caterpillars around the rim of a flower pot, they’ll follow each other in a continuous circle until they all starve. (Recently added to my bucket list. Comment below, if you’ve tried it!) Honestly, I don’t think this is too far off from what we’re witnessing with respect to the gradual migration from Gerber to the more flexible fab file formats … The best predictor of what you’ll do today is what you did yesterday.
The Gerber format has been the industry’s workhorse for longer than most of us have been around—when relatively flat, double-sided boards ruled the earth. As such, Gerber is tuned to this flat-board world. Separate sources, including fab drawings, spreadsheets, and PDFs are used to introduce the Z-direction for multi-layer PCB design. Valor’s ODB++ is a good bit broader in scope, but it’s proprietary. I think that most everyone in our industry would agree that a comprehensive, vendor-neutral industry standard for fabricating a PCB is a good idea. Having said that, it’s not quite that simple ….
High Speed, Multi-Layer PCB Stackup Design
My own interest in IPC-2581 is for the purpose of managing high-speed signal integrity on high layer-count PCBs. While in the process of launching one of the newer EDA software companies, Z-zero, my team and I had to make decisions on our own data format, as well as the external formats we would support out of the blocks. In doing so, we evaluated the common, standard formats for conveying PCB design intent to fabricators, as it related to stackups (the purpose of our software).
At Z-zero, we put our heads together—researching Gerber, ODB++, and IPC-2581 for their support of stackup features, dielectric materials, and electrical properties, as well as copper roughness. A format that didn’t support these parameters would be of no use in high-speed PCB stackup design. In our research, we concluded that we favored the flexibility of an XML format rather than flat ASCII. All of the above narrowed things down to ODB++ and IPC-2581, and we decided to lean heavily in the 2581 format in designing our Z-planner stackup database in our first (2017) release. We’ve since added an import/export for ODB++, but our base format aligns closely with 2581.
Specific to stackup formats, it seems like the Wild West out there. I typically see Excel stackups, PDFs, and JPEGs … I’m waiting to see a stackup on a lunch napkin. Besides wishing they were in more-accessible formats, there isn’t much that can be done with the image-based formats – and there are as many Excel stackup formats as there are fabricators. Add in the stackups used by the various SI tools on the market, and it’s almost complete anarchy. This segment of the PCB design world is ripe for IPC-2581, in fact. It has every parameter we need, and it’s well documented.
Industry Adoption
I’ve heard it said that as much of 80 percent of boards fabricated today still rely on Gerber data – a format that has no knowledge that there’s a Z-direction in a PCB, dielectrics, or copper roughness. In the multi-layer, high-speed world that we live in, that honestly baffles me. Obviously, you can build a board using Gerber – and you can convey stackup details in the fabrication drawing. I get it. But if you’ve moved beyond the flat, double-sided boards Gerber was designed for, you might want to consider moving to a format that handles everything you can throw into a PCB.
No one’s saying that change is easy, but to my thinking, the heavy lifting has already been done.