Our industry seems unwilling to let the ancient data format perish. It should.
Are you still using Gerber? Really?
I’m writing this not so much as a piece of technical information, but more as a plea to my peers and others in the PCB industry. It confounds me that in 2019, with all the smart technology surrounding us, and alternatives available, anyone still uses Gerber. It confounds me, but it doesn’t surprise me. I’ve seen it many times, over many years, and I have my theories as to why. I’ll get to those in a minute.
The alternatives to Gerber, IPC-2581 and ODB++, offer a single source, intelligent dataset used for processes throughout the PCB cycle. Let’s compare and contrast.
Data intelligence. As you probably know, Gerber 274X data are made up of nothing more than primitive shapes, paths, and polygons. Basically, the file is a graphical representation of the design. It includes no intelligence at all. We affectionately refer to it as “dumb Gerber.” Alternatively, IPC-2581 and ODB++ contain intelligent data that include, among other things, stackup, nets, components, pins, traces and vias. In fact, the high level of intelligence has caused some to worry too much information is potentially carried outside the company. Generators of these formats offer options to dumb down the output, however, by turning off files on command. And IPC-2581 allows users to exchange stackup-only data at the beginning of the design process. Thus, all the data necessary for downstream processes are provided, but not enough to reverse-engineer the design.
Data package. A Gerber data package for an eight-layer PCB includes around 17 separate files just for fabrication. To further help assembly and test, designers might include a pick-and-place report, a Fabmaster (CIM) file, and possibly some drawings. A design with blind/buried vias adds another series of drill files. You’ll probably want a Readme file to explain just what it is you’ve sent and the order in which to read it. And don’t forget that milling file, if your design has slots. The plethora of files is ripe for error by way of missing files, misinterpretation of order, or format discrepancies. In contrast, the data package for fab, assembly and test using a smart format includes just two files and eliminates misinterpretation or format errors. (I still send an IPC-D-356 file for electrical test verification.) There is only one file for IPC-2581 and only one directory for ODB++.
Data viewing. While I consider view-only of any data format a mere convenience, it’s most helpful to view it through an intelligent lens. The available Gerber viewers are great, but in the absence of design intelligence you’re basically reduced to an electronic view of Mylar films. Not much help there. I can understand if this were your only option, but since many Gerber viewers also support smart formats (most for free), then I ask: Why?
Those who don’t have a seat of your design software on-hand may rely on fab data for circuit troubleshooting, rework or hand assembly. In any of these tasks, the ability to query comps, comp pins and nets is a huge advantage. A smart format provides that advantage, and because the smart dataset includes design objects, the GUI often includes a data navigator, allowing the selection of design objects by key-in or list. Several companies provide free IPC-2581 viewers; these are accessible at ipc2581.com.
Data DfM. When it comes to artwork verification (DfM analysis), DfM rules applicable to Gerber are extremely limited. In the absence of component body information, component spacing checks, including board-edge clearance, are non-existent. And because Gerber cannot differentiate between a pin, via, or trace, no object-to-object rules can be applied. Thus, Gerber DfM consists mainly of generic clearance and coverage checks. Gerber analysis tools can add enough intelligence to do an electrical check, but this falls far short of any useful verification.
Smart formats on the other hand recognize design objects and thus permit robust verification, including object-specific clearance and coverage, component proximity, manufacturing, and electrical checks. Smart data also support test analysis and fixture generation.
All the major ECAD design vendors provide smart format outputs, and most, if not all, CAM tools read it. There’s no cost to users, and almost no learning curve, to support the smart formats on either side of the import/export equation.
Why, with this obvious set of advantages, are folks still using Gerber? I believe there are a couple reasons. Part of it can be found in the psychology of human nature. For some, resistance to change stems from a fear of the unknown, or lack of understanding the benefits of the proposed change. After all, “we’ve always done it this way.” In the case of Gerber, that’s something like 35 years, and adds fodder to why we need to evolve and innovate. Thirty-five years? Really. Can you think of any productive technology you still use that hasn’t evolved in 35 years?
Advocates must be willing to carry the mantle. They must put themselves in a lead role and advocate for something others are resistant to. To spend some political capital to do what’s right for their company, and for those who resist. You’ll often find most resistance comes from those who would benefit most. You’ll hear many excuses, but I have yet to hear one that has come close to pushing the scales in favor of retaining Gerber. In lieu of technical reasoning, resistance usually boils down to those human characteristics, making the transition more difficult than it should be.
You’ll probably also find that most pushback comes from inside one’s own organization. Persuading contract manufacturers is much easier and usually includes a statement that goes something like, “Our company no longer supports Gerber.” In other words, if you want our business, you’ll adapt and accept the intelligent format we provide.
By sharing the advantages of smart formats, you’ll find most practical people adapt and learn to appreciate the advantages. It’s not always quick, and it’s not often easy. I still have to inform and persuade, but the ability to understand and communicate the advantages allows one to be more effective. Gerber has been the workhorse of our industry and served us well for many, many (did I say “many?”) years, but it’s clear the limitations of a nonintelligent dataset is a deterrent to an evolving industry. It is well past time to put the old horse out to pasture. •