Issue link: https://iconnect007.uberflip.com/i/271756
30 The PCB Magazine • March 2014 nent was attached using SAC305 solder. The assembly lots were then spherical bend tested at a target primary strain rate of 3000 mi- crostrains per second. Microstructures were examined and failure distribution parameters were established for each lot. Board and com- ponent design and construction were identical to those tested in DOE1. Experimental Design: DOE3 The premise for this experiment was that by holding all other factors including board design constant the inherent susceptibility of the resin/glass system would be comparable. Test vehicle assemblies built from a single set of design da ta and specifica- tions were built with OSP solder surface by a variety of qualified suppliers in a variety of laminate resin systems. The PWB suppliers pro- vided sample lots that complied to the design criteria which were then all assembled with the same 32 mm monolithic ceramic device using a single thermal recipe for SMT reflow. This pro- cess did allow for some variation in the mate- rial beyond the resin system. Thickness specifi- cations are normally +/10% which allowed the PWB facilities to utilize their qualified pressing There were no remarkable results from primary attach. Solder conformation was normal, very little voiding was evident and the all of the standard phase compositions were measured. Packages are pre-plated with SAC305 directly over Cu so OSP boards produce Cu6Sn5 inter- faces at both top and bottom interfaces. ENIG boards produce Cu6Sn5 at the top interface and a much more complicated Ni3Sn / Ni4Sn3 / (Cu-Ni)4Sn3 interface at the board side pad. Experimental Design: DOE2 The intent of this experiment was to iden- tify and quantify differences in pad crater resis- tance and therefore survivable strain on a typical filled, phenolic cured laminate after forced BGA rework. Boards supplied in a single laminate from a single source were as- sembled with the 40 mm BGA devices using the standard high thermal mass conditioning and attach reflow excursions using the profile displayed in Figure 5. The lot for comparison was then subjected to two further thermal excursions on standard semiautomated hot gas rework equipment. The initially attached com- ponent was removed the site was dressed using standard methods and then another compo- SPHERICAL BEND TESTING continues Figure 5: smt reflow profile: primary attach.