Issue link: https://iconnect007.uberflip.com/i/1101604
12 DESIGN007 MAGAZINE I APRIL 2019 digital product model from concept to manu- facturing hand-off with a lot of multidiscipline collaboration required along the way (such as schematic/layout, ECAD/MCAD, harness/ board, system/boards). Each step requires smart data to collaborate, and many checks along the way to ensure that nothing is cor- rupted during the collaboration process. At the PCB design stage, the major elements include: • Component information (e.g., footprint, pin contact areas, pin numbers, 3D geometry, package type, part number, simulation models, cost, status, compliance, etc.) • Schematics • Multiboard structure • 3D board (e.g., metal traces and shapes; net properties; HDI, blind, buried, and/or through vias; materials stackup; rigid-flex regions; etc.) • System connectivity • Constraint sets for performance, mechanical structure, and manufacturability Back in the origins of CAD, designs were much less intelligent (designers were just hap- py it wasn't tape). But over time, attributes have been added so that a design isn't just 2D geometries; it's filled with "smarts," such as net names and types, rule sets, impedances, materials, manufacturers, region types, and package constructs. Unfortunately, often, all of this intelligence is stripped away when the data is sent to manufacturing, burdening them to recreate the intelligence to effectively tool the job. The PCB manufacturing industry speaks of the goal to build a lot size of one. The trend in electronics is certainly in the direction of smaller lot sizes but think about the manu- facturing front-end engineering responsibili- ties today. They must take in a multitude of design data files, each of various formats and frequently with incomplete or conflicting data. Because the majority does not contain intelli- gent attributes or properties, a manufacturing engineer is tasked with reverse engineering the data so that the proper work instructions, tool- ing, and programming can be done. Whenever the data package is incomplete or contains contradicting data, the job is put on hold while it is resolved with the customer. Optimizing for a lot size of one is impossible if your front-end engineering consumes days of manual effort. Contrast that with Amazon. Do you think they could be proficient fulfilling your order of a lot size of one without automa- tion? Optimization equates to efficiency—it's a low-cost producer without sacrificing quality. An excellent paper by Roland Berger, "In- dustry 4.0 and Its Impact on Electronics As- sembly," defines 20 building blocks of Industry 4.0, and one of them is DFM. At a basic level, this is the process of determining if the PCB design can be manufactured without issue, or ideally, whether the PCB design has been op- timized for manufacturing—that is, it can be manufactured and tested at the lowest possible cost with the highest yield and reliability. If the PCB design data input to the DFM software has the full set of attributes and properties as ex- ists in the electronic design automation (EDA) data, then the DFM software can automatically identify which DFM checks are applicable for that design and what is the correct technology classification to use in selecting the right DFM rules values. For instance, if the design data is represented in ODB++ format, the receiving DFM system will know the construction is sequential lami- nation and which layers each of the drill files span. It can then perform DFM checks based on microvia and HDI technology without any human intervention. Without that level of in- telligence, the DFM user would need to re-es- tablish the layer construction and associated drill files to the appropriate layers before they then have to determine the DFM rule sets to use. You might say it's not rocket science, but what happens when the drill drawing contra- dicts the layer names in the data files? Smart data eliminates that potential risk and allows for a streamlined DFM process (Figure 2). On the assembly side of DFM, we need to know component and lead types and IPC clas-