Issue link: https://iconnect007.uberflip.com/i/1479191
60 PCB007 MAGAZINE I SEPTEMBER 2022 he explained how the microvias, as they are densely packed, will support each other. He simulated and looked for that. As he explained it to me, I realized that every time I had a tri- ple stack or four stacked microvia failure, they were widely pitched or in the middle of nowhere. It was this realization: at's what happened all the time. It wasn't the BGA packages. It was always like some SMA that somebody hand- soldered and they didn't re-stack or in the middle of nowhere there was a resistance that kept on changing the component, taking it off and on. at material was expanding so much around that single location and kept on break- ing the microvia. I said, "at's really what's causing this thing. We've got to let the design- ers know that you can't do that in discrete com- ponents. Just do it in your tiny pitch device." And even one millimeter is too widely spread out. My friend refers to "mechanical crosstalk," where all the microvias in close proximity help to support each other. I couldn't find mechani- cal crosstalk as any term online, but it was bril- liant. I like the way he said it. Johnson: Gerry, you really have my interest piqued with that little story. You're talking to the engineers who created the simulation tool, and they've got this figured out—the fact that they have simulated this, and they know these effects. ey also understand at this point how they interrelate. ey have the most reliable microvia solutions figured out thanks to what they know in the simulator, because you can't simulate what you don't know. Partida: at's what our industry has done. We've stacked microvias without knowing what the reliability would be or the cause and effect by the selection of prepreg material. We were blind and did things because we could. Honestly, it does all the impedance calculation. I don't bother at all. I let my guys do the stack- ups. I just plug it in to check the reliability. In my APEX EXPO presentation last year, I said this that is the ability to simulate what happens before we start fabricating. We should be doing it at the design stage. is is as revolutionary in what we do to fabricate circuit boards as it was to do net compare to the Gerber data back in the '90s. It's that much of a game-changer. Shaughnessy: So, how does the simulation tool work? Partida: You just type in whatever pitch you want. A lot of times I don't know from the customers what it is. I'll do a 1 millimeter, 0.8 or 0.65, and you just change the pitch. You change that number. Re-simulate, then okay, it's 10 reflow cycles to failure at 0.5; change it to 0.65. It's still 10. Good. en switch it to 1 millimeter and it's 8, it's still safe. I'm okay with it. Or you run it at 1 millimeter and it's four reflow cycles. You run it at 0.5. It's 8. You say, "I've got to call them back. What is the pitch in your design that you have eight stacked microvias?" We must ask more ques- tions. You can't just look at a print and know what it is because you don't know if they use a rule everywhere in the design or just use it in the tight devices. ey must do the simulation at the begin- ning of the design when they make the rules for the stack and the type of pitch devices and elsewhere. ey really need to do it up front. at's a big paradigm shi for them. You're tell- ing designers who many times don't hear about or deal with the reliability fallout at assembly, especially if they outsource the design. ey're not intimately aware of what is going on with failures, low yields, or D coupon testing. My friend refers to "mechanical crosstalk," where all the microvias in closely proximity help to support each other.