Issue link: https://iconnect007.uberflip.com/i/960594
46 SMT007 MAGAZINE I APRIL 2018 each cell. For both dimensions, the balls on the SMD pads were essentially the same across all the cells. This was expected since these dimen- sions were dictated by the package substrate SRO and solder ball size, which were nomi- nally. Likewise, the balls on NSMD pads were the same among the cells. Since solder wets the NSMD pad side- walls, it's expected that the height of these pads would be lower than an SMD pad of the same diameter. Surface evolver simulations predicted the NSMD pads would be 22 µm shorter if the same 0.45 mm pad were used. To compensate, the NSMD pads were designed slightly smaller. Surface evolver predicted only a 3 µm difference for the compensated NSMD pads. Figure 6 shows that an actual difference of 6 µm was observed. No difference was apparent in ball diameter. Room temperature solder ball coplanarity is summarized in Figure 7. There is a tendency for the hybrid Cells 2-3 and 5-6 to be slightly higher than the SMD Cells 1 and 4 due to the slight difference in ball heights. Also, the low CTE Cells 4-6 were higher than the standard material Cells 1-3. This was unexpected and is opposite the warpage results shown in the next section. Reasons for this discrepancy are under investigation. In any event, all are well within the case outline (package outline) specification of 200 µm max. Warpage Figure 8 plots the freestanding component warpage across the thermal cycle tempera- Figure 6: Ball height measurement by RVSI scanner. Error bars represent the data range. Figure 7: Solder ball coplanarity at room temperature. Figure 5: Ball diameter measurement by RVSI scanner. Error bars represent the data range. Figure 8: Package warpage by TherMoiré.