Abstract
Paul A. RoderFour Pi Systems, San Diego, CA, a subsidiary of Hewlett-Packard Company.ABSTRACTLearning algorithms are introduced for use in the inspection of cross-sectional X-ray images ofsolder joints. These learning algorithms improve measurement accuracy by accounting for localizedshading effects, which can occur when inspecting double-sided printed circuit board assemblies (PCBA).Two specific examples are discussed. The first is an algorithm for detection of solder short defects. Thesecond algorithm utilizes learning to generate more accurate statistical process control (SPC)measurements.1. INTRODUCTION : X-RAY LAMINOGRAPHYX-ray inspection has gained favor as a method of detecting soldering defects on printed circuitboard assembly (PCBA) production lines. Solder joint defects remain a major problem in the electronicsindustry, since degradedjoint quality can cause board failures in the field over time. Traditional in-circuittesting can only detect a more limited set of defects which exist before shipping the product. In order todetect solder joint defects, human visual inspection has commonly been used. Studies have shown poorcorrelation in defect calls between different inspectors viewing the same boards, so such human inspectioncan be somewhat subjective. Also, human inspection is becoming increasingly less useful due to therapidly increasing solderjoint density and the need to use new technologies that utilize solderjoints hiddenunder components, notably the newer packaging technique known as Ball Grid Array (BGA).Traditional X-ray inspection techniques involve transmission X-ray images. These images do nothave a specific plane offocus, so any opaque objects either above or below the solderjoint to be analyzedwill obscure the joint. Typically, transmission X-ray images are only used on single-sided circuit boards,since devices and solderjoints on the opposite side ofthe board render manyjoints uninspectable ondouble-sided boards.In order to inspect double-sided boards, a technique must be used to image the specific layers ofthe board where the solder joints exist, and mask out interfering objects above and below the joints. FourPi Systems' patented technology creates cross-sectional laminographic images of individual layers of thesePCBAs. These laminographic images are created by a rotating X-ray beam synchronized with a rotatingdetector, so that only objects in the focal plane appear clearly, while objects above or below the focal planeare smeared out. A time integration is performed on the image, usually for one cycle of rotation,generating a clear image of only the objects within the focal plane. Objects outside of this slice areblurred and faded, and usually become invisible.Performing the cross-sectional imaging utilizing this scanned beam laminography hardware setupcreates some computational benefits over other methods such as tomography. The integration is done inhardware, and the computational complexity of tomography is avoided. Also, only the important slices(usually the pad level slices on both the top or bottom of the board for surface mount devices) can bequickly imaged and analyzed without calculations over the entire volume, as in tomography.58
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.