Issue link: https://iconnect007.uberflip.com/i/1129312
44 PCB007 MAGAZINE I JUNE 2019 be able to go further and classify more types of defects, such as small nicks, protrusions, or false calls. The biggest challenge is to classify false calls because they can appear in many different ways, which is why we need to em- ploy AI. We'll have to see how far it can take us because it's still a developing field. Will it be able to completely replace physical verifi- cation? That's still an open question. I don't believe it's possible in the near term. There's still going to be some physical verification, but that's where we're going and what customers are asking from us. Everybody wants to reduce physical verification, so that's the idea. Another product is CIMS Quality Control, or sometimes we call it the CIMS Quality Cen- ter (CQC). The idea is to display aggregated live quality information from AOI—for exam- ple, how many opens and shorts we detected, where the defects are located, and what's the yield. The live data as it comes out of AOI is displayed in one configurable dashboard. In this example, you can see only one type of in- formation but the user can set as many dash- boards as they want. For instance, you can have utilization in real time and see how ma- ny panels you've already scanned, how many panels have been verified, what the yield is, and if it goes up or down. If your shorts go up, it can trigger an alarm, so you know in real time that you have to fix something within the process and so on. We want to make it very visual and feed this information in real time so that you can have it on your laptop or big monitor in the AOI room, for example, so that anyone can look at it. You can set up different types of graphs or charts that are most relevant for you. If you're a production manager, you may be more in - terested in utilization, and quality control per- sonnel will be more interested in types of de- fects, their location, and where they're coming from. Matties: What do you mean when you said it's in "real time?" Because this isn't an inline pro- cess. Kaplan: It's AOI in real time, so as information comes out of AOI or verification, that's typi- cally where you capture it. That's when you know for sure what type of defect it is, so as soon as you make that decision, information is displayed immediately. It's accumulating and you can have it for a day, for the last hour, for even for the last 30 days, but you get the most value if it's live. That's why we call it the Quality Control Center; it provides the quality data as it comes in rather than waiting until electrical test, for example, when it would be too late the fix the process. And it's a pretty simple program that can run on any computer; you don't need spe- cific hardware to run it. All of the data comes out of our server. We centralize the data col- lection from our AOI and VVS systems. Every- thing goes into one server, and this program pulls out the data and extracts all of the infor- mation that you're interested in. Matties: Is there a server at the customer's lo- cation or a server at your location? Kaplan: There's a server at the customer's loca- tion where all of the data is consolidated. Matties: So, you're an equipment builder, but you're also developing a lot of software. Kaplan: Absolutely. All of the equipment today is becoming smarter and more software heavy than hardware. At the end of the day, hard- ware is quite simple. You eventually build ev- erything with some off-the-shelf component, so the expertise comes from software. Eventu- ally, everything will become software. Matties: And now, with the move to Industry 4.0 smart factories, when this data becomes Everybody wants to reduce physical verification, so that's the idea.