Issue link: https://iconnect007.uberflip.com/i/1305670
NOVEMBER 2020 I SMT007 MAGAZINE 29 defects. It's really up to each vendor to deter- mine whether adjustments are to be made manually or autonomously. Most vendors will provide a means of selecting between manual and automatic mode. As I mentioned earlier, it is very difficult to determine which adjustments need to occur throughout the line based on the post-reflow analysis of defect data. You can certainly make assumptions, but those assumptions are so far down the line that it becomes increasingly dif- ficult to determine exactly where the defect occurred. From the customer's perspective, I com- pletely understand that we are suggesting a substantial investment in test equipment; how- ever, the absolute best way to get a handle on the entire production line is to have testing at each interval within the line, including post- solder deposition, post-placement, and post- reflow. That is really what industry 4.0 is all about. Johnson: Based on this trend of moving toward inserting inspection functions throughout the, does that cause test and inspection equipment manufacturers like you to specialize a feature set for a particular spot in the line? D'Amico: Absolutely. An SPI system is specifi- cally designed to test for defects related to sol- der paste deposition such as excessive or insuf- ficient solder, X-Y offset with respect to pad location, solder bridging, shape deformity, etc. A post-placement AOI system would be pro- grammed to test for component presence or absence, part markings, proper rotation, offset, etc. And a post-reflow 3D AOI system would be programmed to test for all of the same parame- ters as a post-placement AOI system, as well as co-planarity, for each device, solder joint integ- rity, and solder bridging. Johnson: Talking about getting information from this equipment and feeding the trending data back, and performing the analysis, is that analysis taking place on your system or is that happening in the database? D'Amico: The answer is basically yes. Under some circumstances, data may be available through direct machine-to-machine communi- cation. This machine data, however, is also col- lected within the inspection database for anal- ysis. This is commonly referred to as big data. By applying a subset of artificial intelligence (AI) called deep learning to this big data, we can determine the optimal process parameters for maximum quality and efficiency. When setting up a new PCB, an SMT engi- neer will program the process parameters for each system within the manufacturing line