PCB AOI(Auto Optical Inspection) device extracts too many possible defect points. Also, since the inspection accuracy of the inspection software used in the PCB inspection is very low, which is 50 ~ 80%, it can not be used to judge whether the defect point extracted from AOI is a genuine or not. As a result, it requires the complete visual inspection for all extracted defect points from AOI at the VRS (Verification and Repair Station) verification stage. The visual inspection workload is not only a burden on the PCB manufacturer, but also it can not maintain the uniform PCB quality because the inspection result is different for each visual inspection worker. In case of long working time, the worker fatigue increases and the human error is often triggered.
The existing PCB process has separated production and inspection processes. If they are separated, there is a risk of a foreign substance, stabbing, and external damage caused by floating matters during the PCB transfer process, which will seriously undermine PCB quality. It also requires additional facilities such as clean-room and AGV (Automatic Guided Vehicle System) and it is hard to manage the record.
In addition, since the inspection accuracy is low, additional visual inspection is required at the VRS verification stage, making it difficult to construct an in-line system.
Because of the excessive inspection of AOI, it takes much time to take the visual inspection (1 original board, 1 Minute of AOI/10 Minutes of VRS Re-inspection), and it is nearly impossible to make in-line in the current visual inspection method.
Artificial intelligence inspection using AIPi Software extracts the accepted products with an accuracy of 99.9% on the basis of exterior inspection. The images judged to be accepted in AIPi Software are judged by the reliable accuracy at the level of visual inspection by the inspection expert, so no further visual inspection is required. For example, assuming AOI has extracted 2,000 possible defective points, AIPi Software determines approximately 90% of 1,800 defective points as accepted products and about 10% of defective points as the object requiring the additional visual inspection. In this case, the amount of work required for the worker to undergo a visual inspection is reduced to one-tenth.
AIPi In-Line Package is the world's only innovative solution that combines the high inspection accuracy of AIPi Software with AIPi Scanning System (High-Sensitivity Color TDI Camera) to integrate PCB production-inspection process.
AIPi Scanning System can simultaneously scan the number of images equivalent to 6 AOIs, and scan the entire area of the original PCB board in 10 seconds by diving it as 6 parts.
In addition, AIPI Software performs PCB inspection within 14 seconds with GPU-based high-speed parallel processing. AIPi In-Line Package combines 10 seconds of scanning time of Scanning system and 14 seconds of Software inspection in real time, reduces 24 seconds to 15 seconds, and enables the in-line with 40 times faster Tact Time than existing visual inspection.
AIPi acquired the database for location and type of defect verified from tens of PCB inspection expert through prior learning with deep learning by PCB inspection using artificial intelligence. It also can learn tens of thousands of PCB patterns and PCB exterior images through the self-image generation algorithm as well as accepted/defective images from PCB inspection expert and can inspect various types of PCB defects thoroughly.
In addition, AIPi is easy to learn after a new inspection condition occurs. As AIPi learns new types of accepted/defective image data that will be generated in the field and accepted/defective image data that will be created from new clients, it will show the more advanced type of inspection performance as time passes.
AIPi precisely detects and visualizes defective positions in pixels, which could not be achieved by the previous inspection. It not only can display large defective locations but also detect the fine defects.
AIPi distinguishes between types of genuine/false defects that were only visible to the naked eye. You can determine the type of defects in the pattern inspection step as well as the defects in the visual inspection step. This type of bad information is used as predictive maintenance data to improve yield.
AIPi Software applies the AOI equipment and VRS equipment used by the customer site while maintaining it. AIPi Software accepts the potential defective points extracted from AOI and confirms 90% of the potential defective points as accepted ones, and extracts the remaining 10% potential defective point to the VRS verification stage. This allows up to 90% reduction in the amount of visual inspection that is required, resulting in economic benefits.
With the packaged AIPi Software, you can operate normally within 3 weeks. (Definition of inspection criteria for one week, AI learning and inspection for two weeks)
Laon People recommends 'AIPi' In-Line Package, a premium solution for clients who need to build new lines or build advanced In-Line systems. AIPi In-line Package is a scanning system that can replace 6 AOI devices and it scans the large-sized PCB. It is also an innovative solution that can cover the verification by 12 to 18 VRS devices by applying AIPi Software.
If you apply AIPi In-Line Package, you can solve the deviation by workers with AI algorithm, and you do not have to worry about the quality deviation of each different device. It prevents external damage during PCB transfer, eliminates the need for a separate clean room, and facilitates systematic lot history management in a structured In-Line structure. You can also implement next-generation smart factories that can monitor real-time processes with high-capacity/high-precision/high-speed image inspection.
Implementation of AIPi In-Line Package, which can drastically reduce the investment cost required to set-up the inspection process, can be completed in about 7 weeks. (Definition of facilities and inspection criteria for five weeks, AI learning and inspection for two weeks)