Issue link: https://iconnect007.uberflip.com/i/1040234
22 DESIGN007 MAGAZINE I OCTOBER 2018 As you know from previous columns, a gen- eral rule of thumb is that the thinner the dielec- tric of prepreg, the higher the resin content, and the higher the resin content, the lower the Dk. If a Dk is listed for a 0.004" dielectric sub- section as 4.5, we must use thin pregs as low as 3.3 for FR-4/406 derivatives and 3.87 for the thinnest pregs on Nelco N4000-29 (a com- monly allowed material for RoHS compliance and the elevated temperatures at assembly due to alternative surface finishes) to achieve this dielectric. Ultimately, this can result in the need for either altering the specified dielec- trics or resizing impedance lines, which is less desirable. The bottom line is this—avoid specific Dk or material information unless it is critical to the design. It is better to call out the material type via IPC-4101 or the number used on the mate- rial specification. Specifying Dk or dielectric numbers in your stackup may precipitate a call or email from your manufacturer and require a negotiation, which takes valuable time. Pre-Quote Software Another example is IPC-2581, which is an all- inclusive consortium package that includes the bill of materials (BOM) and assembly informa- tion as well as design and image output data in three formats. As long as there is no conflict- ing information between the image data—and sometimes there is—this dataset can be used. I am writing this from the standpoint of quick but accurate PCB quotes, not the validity of a given dataset, which means I don't have any issues with an IPC-2581 output for manufacturing. However, for fast and accurate pre-quote analyses, the additional files in datasets, such as IPC-2581 and others that include multiple image outputs, can slow down the quote pro - cess. We have seen as many as six separate ODB++ outputs—some as an ODB folder, some as TGZ files, and some as ODB inside a ZIP file—all in the same dataset. Many fabricators use pre-quote software packages, such as INSIGHT or Integr8tor. Both are out - standing for what they are intended. A pre- quote design check to ensure obvious things like trace and space meet the defined copper weight and that a Drill file exists and matches the drawing. Even though the datasets should all be the same in theory, input pre-quote packages must be edited so that only one dataset is checked. Multiple sets will stop the operation and ask for more user information, which again, takes time out of the quote process. Additionally, file naming conventions are another way to streamline the quote process. Many design systems have a default for file names as do drawing stackup templates. Most do not match exactly, and if there are no Z-axes layer designators on the file itself, sometimes it is difficult to tell what each layer should be. For example, the image data file template may call inner plane layers "GP1" for Gerber plane 1, but your stackup may show the real names of the layer functions. Thus, GP1 of your image data may be called "GND" or "PWR" on the stackup describing the layer function more than the layer name. This is not an issue most of the time if you provide an EXTREP file or extension repre- sentation file that describes the layer versus stackup layer names. Much like with multiple image datasets, incoming pre-quote software packages will have trouble with identifying each layer and will pause for the operator to give the system more information to continue with an analysis. As far as the pre-quote analy- sis is concerned, if the layer names and stackup layers can match and multiple datasets do not need to be culled before running the pre-quote software, the faster the pre-quote software works, the quicker you get your quote. Again, the purpose of this column is not to suggest you do not use IPC-2581, but to suggest you The bottom line is this— avoid specific Dk or material information unless it is critical to the design.