Issue link: https://iconnect007.uberflip.com/i/304995
18 SMT Magazine • May 2014 destructive techniques to identify BGA joint qualities and prevent escape. We did a study for BGA crack several years ago with AXI, time domain reflectometry (TDR), 2DX, and cross- section/SEM comparison; however, we didn't have good detected results from AXI [1] . During the study, we realized that 2D X-ray with tilting angle detector has the capabilities to identify BGA crack at 5 mi- crons or higher. Recently more AXI machines using both lami- nography and tomosynthesis technologies have better capa- bilities to detect BGA defects; therefore we would like to de- velop the optimization algo- rithm and threshold settings to identify HIP on BGA for high volume products. Based on 2DX images, we have identified two boards con- taining HIP: #495 with 45 HIPs, and #266 with 2 HIPs. With the HIP defective boards, we have tested them at four sites, with four different AXI machines. AXI1 is a laminography machine; AXI2, AXI3, and AXI4 are tomosynthesis machines from differ- ent vendors. We worked together with site en- gineers and vendors' support engineers for this project. We focused on HIP defects escaped %, false call PPM and gage repeatability & repro- ducibility (gage R&R) with that BGA. The old axiom "A picture is worth a thou- sand words" proves accurate. Indeed, a high- quality microstructure of a solder joint provides "sights' and "insights" into the state of solder joint integrity and its anticipated behavior 2 . Similar benefits can be achieved using high- quality 2D and 3D offline X-ray inspection im- ages and data. Through this study, we try to understand more about AXI machines capabili- ties, especially for improving AXI programming optimization used for our best practices. Al- though machine's testing conditions were not exactly the same, but close, the philosophy for achieving optimization is similar. Look at the measurement data first, finding the difference between good solder joint and defective solder joint; and then balance the defects escaped % and false call numbers to find the right thresh- old settings. For the high volume products, monitoring AXI programs is an ongoing process based on further fine-tuning with feedback from ICT, functional test and 2D/3D offline X-ray inspection. Experiments and analysis Two boards with HIP de- fects were from production SMT line of a different site. The BGA has 1,017 balls with ball diameter of 21 mils for outer balls and 19 mils for in- ner balls of the pad. The HIP defective pin (ball) is identi- fied at 2DX machine with tilting angle detector. Tilt- ing (oblique) angle is 55–68 degrees; and rotation of the X-ray detector 0–360 degrees around the examined joint. This is not trivial, but it is very easily accomplished using the 2DX equipment. The oblique and rotation angles of the X-ray detector are key factors for identify- ing BGA defects, such as HIP, open, and small crack 1 . There are a total of 47 HIP pins on the two boards based on 2DX images (resolution 0.1μm). Some 2DX HIP images are shown in Figures 1, where pins with arrow are HIP, except for pin number G32 which is crack (9μm). For this project, the "defects escaped %" and "false call PPM" are calculated with these equa- tions: Defects escaped % = # of HIP escaped / total # of HIPs, where total # of HIPs is 47 (equation 1) False call PPM = (total # of false call / total pins # tested) X10, where total tested pins number is 2034 for two boards (equation 2) The gage R&R results are calculated with SPC tool Minitab with nine sets of variable data by three operators from each machine. Data of BGA diameter for pad and middle slice, and some main AXI algorithm (open outlier, neigh- HEaD-IN-PILLoW X-Ray INSPECTIoN continues feATuRe The old axiom "A picture is worth a thousand words" proves accurate. Indeed, a high- quality microstructure of a solder joint provides "sights" and "insights" into the state of solder joint integrity and its anticipated behavior. " "