Issue link: https://resources.pcb.cadence.com/i/1057145
2 while others perform correctly as expected. For example, a PCB layout containing very thin pieces of copper created in the design tool by rule would be correct per the schematic. And if spaced properly it would pass DRC. However that same slivers, being so thin, could potentially detach on the physical PCB and inadver- tently connect itself to other copper elements during assembly, thus creating shorts on some PCBs but not on others. So the sliver would pass DRC verifiction, but in real-world manufac- turing the sliver could cause some PCBs to fail. Without DFM, this problem would go on undetected and would result in scrap or rework. Until recently, DFM analysis was either left to the PCB fabricator or assembly engineer to manage, or could only be performed by companies that had the financial resources to purchase high-end DFM analysis software and support a dedicated staff to run DFM analysis. Most PCB designers would perform only a DRC analysis and visual inspection of the design before submitting the design to be manufactured. Manufacturers who know that DFM issues like acid traps, slivers and starved thermal pad connections can decrease manufacturing yields and increase costs take it upon themselves to analyze the design, often making modificationsto ensure that the design can be built with maximum yields and lowest costs. So long as the finished PCB functioned properly, the design engineer was content. So why move DFM into the PCB design flow There are several reasons: cost of finished PCBs, maintaining design intent, and the potential for future design failure. Figure 2 Acid traps have the potential of trapping acid during the PCB etching process longer than intended and can eat away a connection, making the circuit defective. It can cost a PCB manufacturer as much as 20% of the cost of the PCBs for CAM engineering, the processing and tooling of design data to prepare it for manufacture. This additional cost is built into the end price that users pay to have physical PCBs fabricated. So, theoretically, designs submitted without DFM defects are less expensive to manufac-turer than ones with DFM defects. One could deduce that it's better to pay a little more to have a manufacturer ensure the design can be built. However, this creates other issues that are not so desirable. Figure 3 Small copper slivers can detach themselves during assembly, float aound during soldering and inadvertently reconnect themselves anywhere on the PCB, potentially tying multiple nets together. To take a design that has DFM issues and make it comply with the manufactur-ing process, a CAM engineer may need to modify the design data. What this means is that the layout provided to manufacturing may not be 100% consistent with the finished PCB. Issues with electromagnetic interference, signal integrity, cross talk, etc., which are commonplace in today's high-tech electronics and are addressed in design engineering, may be unknowingly re-introduced into the design as it's reworked for manufacturing. There is also no guarantee that a CAM engineer will communicate the design changes back to engineering to be incorporated into the original PCB design database. So not only is the design layout different between engineering and manufacturing, but what happens when a second manufacturing build is required or the design is released to a different manufacturer for volume production? Consider this real-life scenario: a design engineer designs a PCB, runs DRC analysis and determines that the design is correct. He creates PCB manufacturing filesand sends the filesoff to a manu- facturer to have prototypes made. The manufacturing engineer runs his analysis on the PCB filesto ensure the design can be fabricated and identifiesdefects in the design that could result in scrap or low yields. Wanting to deliver a good product, the manufacturer fixs the issues, builds the PCBs and ships back the finishedprototypes without communicating what changes were made. Back in the lab, the design engineer tests the prototypes and they work successfully. That's great; however, unbeknownst to the design engineer, his prototypes are different from his PCB manufacturing fies. Now the design engineer releases the manufacturing fils for high volume production from a different manufacturer who specializes in production PCBs. This manufac- turer, for one reason or another, chooses not to run an analysis prior to manufacturing and therefore doesn't detect the same issues as the prototype manufacturer. They build and ship the finished PCBs back to the customer. The boards are assembled and tested and, oddly enough, some, most, or all of the PCBs fail. Why? Because the design data still contained the original DFM errors in the manufacturing filethat were corrected in prototype, but never incorporated for production. The result was thousands of dollars in material being scrapped but – even