Issue link: https://iconnect007.uberflip.com/i/713934
August 2016 • The PCB Design Magazine 83 pertaining to one single PCB. Putting more than one PCB in an archive is confusing; even if you use a clever file naming scheme or comprehen- sive explanatory note, you are potentially crea- ting a lot of confusion by putting different PCBs in the same archive. Remember: one PCB, one archive. The question is what must and must not be included in the fabrication data set. Obviou- sly, the archive must contain all data needed to fabricate the board, in a standard and une- quivocal format and manner. Less obviously, the archive must not contain other files, super- fluous data or duplicates. The reason is that the manufacturer must check each file to see if it contains relevant instructions. Superfluous fi- les waste his time and increase the risk that he misses something essential. Aperture list files are superfluous, for example, as all the required aperture information is in the Gerber files. CAD data is also useless as it requires CAD software to handle it. Duplicate information is even wor- se as the fabricator must compare the different files to check whether they contain conforming instructions, and conflicts raise questions about what is now valid. Duplicate image information (e.g., in Gerber and ODB++ format) is especially aggravating as images are complex and hard to compare—which tolerances apply? It is mandatory that one single Gerber file should be provided for each patterned layer (copper, solder mask, legend, etc.) and for each drill sequence present. This rule is violated every time that only one solder mask is provided "because top and bot- tom are identical." This obnoxious habit may save a few bytes but then questions arise about whether a mask was forgotten and which masks must now effectively must be on the board. Confusion. The space saving is illusory anyway as the first thing the CAM operator must do is to create two masks from the single file. A big- ger archive is better than a confusing archive, and when both masks are present, even if they are identical, everything is clear. Also mandatory are all fabrication drawin- gs—in Gerber format—and fabrication in- structions such as finishes, ROHS etc. If it is not mandatory, it must not be there. Finally, and even if this seems like a su- perfluous duplication of data, the CAD netlist should always be included (Chapter 8 in this series). This genuinely original data provides a powerful and essential checksum on the data, and is far better than a netlist generated from the image data, which is only a reverse-engine- ered approximation. Make the fabrication data set as simple as possible, but no simpler. PCBDESIGN Karel Tavernier is managing director of Ucamco. THE GERBER GUIDE, CHAPTERS 17 & 18 Scientists at the Energy Depart- ment's National Renewable Energy Laboratory (NREL), in collaboration with researchers at Shanghai Jiao Tong University (SJTU), devised a method to improve perovskite solar cells, making them more efficient and reliable with higher repro- ducibility. The research, funded by the U.S. Department of Energy SunShot Initiative, involved hybrid halide perovskite solar cells and revealed treating them with a specific solution of methyl ammonium bro- mide (MABr) would repair defects, im- proving efficiency. The scientists came up with a better method, using what's called the Ost- wald ripening process. The process in- volves small crystals dissolving and then redepositing onto larger crystals. The researchers were able to induce the Ostwald ripening process by treating the perovskite with a MABr solution. The perovskite cells treated with MABr were shown to be more efficient than those without the treatment. NREL Technique Leads to Improved Perovskite Solar Cells