Issue link: https://iconnect007.uberflip.com/i/1479191
58 PCB007 MAGAZINE I SEPTEMBER 2022 But for the military guys who have ASICS that cost hundreds of thousands of dollars each, the stakes are much higher. If it is for space, then it can only be assembled once for flight; it cannot be taken off and reused. e military packag- ing then was a much wider pitch than the com- mercial guys who were stacking microvias ini- tially. When we looked at where the failures were happening, they were still happening with the commercial guys who were going three or four deep stacking microvias. ey weren't trying to make short, squatty, wide-diameter micro- vias because they were using thicker dielec- trics to get wider lines for impedance. Conse- quently, we went for a time where there really didn't seem to be a problem. en it became, "We see a fracture at the target pad on the stack of the microvias," and everybody thought there was a weakness in the electroless copper. We all came up with these rules of thumb: Don't stack more than two. A lot of DOEs were done, and they almost always concluded, "Do two stacks and then stagger off." at seemed to work. Even fabricators we would work with had rules like, "Keep your aspect ratio for a sin- gle microvia at 0.75 to one. If you're stacking them, keep them at 0.6 to one." at seemed to work; we got good results. Now, during this time we employed reflow resistance testing to monitor the strength of connections in the finished product. We started learning more about what works, and what doesn't work. Some designs would slip through, where they do a three-stack on tight pitch, and they were passing. We were asking that if our rule of thumb was only two, then why is it working at three? When you look at the design, it's a 0.4 mm pitch. Last year, you interviewed me about the ability to simulate microvia reliability. As more material got into the simulation so- ware, and as we had more experience with it, it became evident that the simulation was show- ing us that the tighter pitch devices could han- dle more stacked microvias and even smaller diameters. e simulation showed that it could survive reflow. IPC accepted my paper for 2023, so my pre- sentation will show that it is possible to stack deeper in certain package densities, but else- where on the board, you would need to stag- ger because of the material expansion of micro- vias stacked in the middle of nowhere. e via is fighting against the expansion of the material itself at 360°F. But in a BGA, it is supported by the other microvias around it, like rivets hold- ing the material together. at's why they sur- vived. at's why the commercial boards in the past had survived—because they used smaller pack- ages. at's what I want to share, in simulation as well as in D coupons. I want to put a tight, dense, 0.4-mm pitch and do a three stack. On the second one, put it on 1 millimeter pitch on the same coupon to demonstrate that there's a population difference between three stack and a tight pitch passing six reflows. en when they're a 100-mil pitch, you will see more fail- ures when they're wider spaced. Now I have proof to give designers that, yes, you have these three levels of connections, these three microvias and connections. In Gerry Partida