Issue link: https://iconnect007.uberflip.com/i/1536435
40 PCB007 MAGAZINE I JUNE 2025 To keep up with the fast-changing electron- ics market, AOI in the PCB industry needs to evolve continuously toward full autonomy in applying the requested acceptance criteria when checking the quality of the products. ere has been an institutional shi, on top of AOI, at the end of the production process: a valuable in-line additional tool at earlier and potentially multiple stages during man- ufacturing. We can get to that defect identi- fication, prediction, and reduction before so much cost is built into the panel. e concept of flexible AOI in-line is a game changer, and with machine learning and increased pro- cessing power, AOI can process faster, with much more data and, ultimately, be more reliable, and detect more defects. How do you assess the ROI for these new AOI machines? Traditional AOI in PCB has a clear ROI and a well-established rationale for existing in the manufacturing process. Entering the market with a new AOI solution means competing against major established machine builders with several years of experience. It becomes more interesting when we view the ROI of an in-line AOI. e overall ROI of an inline AOI can be substantial because of the value of early detection and prediction, potentially reducing the quantity of damage, scrap, or loss of productivity, which all translates, of course, to dollars. What does "in-line" look like compared to how we've traditionally used AOI in the process? e PCB manufacturing process has many steps. In-line is the ability to use a visual inspection tool during multiple steps before the end of the process. In every step of PCB fabrication, there might be different use cases for the in-line AOI, defects it can look for that inherently represent a different cost/ value. e presence of a tool that takes pic- tures in-line means that there is a quality checkpoint between virtually every one of these steps. e earlier the defect is detected or predicted, the greater the final impact and overall value. Every step (drilling, plating, etc.) has a very different potential for defects. e vision of this technology is to detect not only the defect at the end of a particular stage, but also to predict it. If it decides there might still be room for improvement, then maybe closed-loop and correction is possi- ble. e vision is for the in-line AOI machine to provide this level of intelligence. Can you give me an example of a defect that could be found now that may not have been the case several years ago? In surface treatment, for example, pictures taken can be processed within the device itself (on the edge), then make a decision autonomously in real time, without calling the operator. If it's a surface defect, what is the category, priority, or gravity level? Is it a fatal defect that requires the line to stop, or can the production keep going? Ten to 15 years ago, this level of sophisti- cation was impossible because these tech- nologies and the processing power were not where they are today. Processing the amount of data that is available at that time could not technically bring real-time value. Giovanni Obino