Issue link: https://iconnect007.uberflip.com/i/1189040
16 SMT007 MAGAZINE I DECEMBER 2019 Manor: Yes. With smaller companies, it's all about flexibility because they usually deal with the smaller customers, and what they give them is the ability to do a very quick cycle and prototyping. They're usually very close to their customers. The advantage is the cus- tomers trust them with the data; many times, it's around aerospace and automotive. They're in the U.S. and Germany and they work with the big OEMs to do a very quick turnaround of data. We have a customer in Colorado where the big thing is we get the order on Monday, and we ship it on Friday. It could be 10 boards, and when we get the data Monday, we do the NPI, start programming, do a few boards, see that it's okay, and then do the batch of 20–50 and ship them on Friday. The automation is on the process level, but they have to be very flexible because setting up the line, and all of that stuff is less of an issue. With the bigger companies, it's more about reducing headcount and coming to a situation. We've seen a couple of customers now with assembly lines in Italy with no human interven- tion. We have a couple of other ones as well; one of them is doing a lot of white goods and panels for dishwashers. The whole thing with the plastics is done completely automated, and for them, it's all about reduced headcount and people and improving the quality. That's what they want to automate, and they need the soft- ware to do that type of automation. Matties: You mentioned CFX and how you have tools. How do all these packages play together? Manor: CFX is definitely an enabler to do data acquisition. It's going to reduce the cost of integrating into the factory floor space. It depends on what type of data you want. If you just want to do reporting and dashboards and give the factory the ability to know what's the production going like, how are we doing from a production perspective? CFX can do that data today. It still lacks the capabilities to do more advanced data output and just-in-time material management or Kanban-style material man- agement—all of the analytics that we've been talking about. Here, you need a wider set of data, but it's not sure that all of this is in CFX. As time evolves, I imagine that this will expand. In the meantime, we will have to kind of take some data from CFX and proprietary interfaces. We have our OML, which we have been using as our own data format, and we have a couple of OEMs who have taken OML and have made their suppliers who supply cus- tom machines do an OML format. Many cus- tomers want exact trace; they want to do stuff like automatic program change, where if there is a new product, they automatically send the new program to all the machines. They want to do stuff like bad board marks. If there is a panel with a couple of defective boards, they want to automatically tell the machines "don't place components here." These types of more complicated communication are what some of the customers are requesting to get this level of automation and digitalization; this is what we need to focus on. How do we get the data and the API to do something like this? Matties: With the digital factory, having a lot size of one is probably the best approach. You're loading up your machines for multiple boards with the components, but what's your thought on a lot size of one or reducing lot size in terms of increasing yield, performance and throughput? Manor: It's happening, and it's one of the key challenges we see today in the market. If I look at the challenge, a lot size of one is definitely one of the key challenges today. And we have many customers who say that the average batch size a couple of years ago was 5,000, and today, the average batch size is around 100. We've seen a couple of customers now with assembly lines in Italy with no human intervention.