Issue link: https://iconnect007.uberflip.com/i/304995
May 2014 • SMT Magazine 21 HEaD-IN-PILLoW X-Ray INSPECTIoN continues feATuRe Figure 3: axi 1 "open outlier" measurement data for middle ball and pad slices (board #495). Table 2: axi 1 detected hip and false call vary with threshold settings (board #495). Table 3: Testing summary with of AXI 1. for open outlier < -3 for pad slice, and > 3 for middle ball slice which are the main algorithms that can detect HIP. The rest of the parameters for open outlier can be used with default num- bers if the test results are fine. Table 3 listed AXI 1 testing results summary by using equations 1 and 2 from site 1; site 2; and site 3. The site 1 used default threshold set- ting; site 2 and site 3 use the same threshold settings which are listed in the first two lows of Table 2 for testing two boards. The data tells us that AXI 1 has the capabilities to detect HIP with the right algorithm thresholds settings. B. aXI 2 (Tomosynthesis) There are a few AXI 2 machines at Flex- tronics. For the AXI 2 machine, the threshold settings for neighbor outlier is < -3.5 for pad, and > 4.5 for middle ball. Table 4 listed AXI 2 test results summary from site 1. Two detected HIP pins (in green), and two escaped HIP (in red), are listed in Figure 4. Similarly, there is not a huge difference on pin images between good and HIP pins. In comparisons to AXI1 versus AXI2, the AXI 2 (tomosynthesis) seems to have better HIP detection than AXI 1 (lami - nography).