Issue link: https://iconnect007.uberflip.com/i/1534953
replaced on the boards, but when they cooled down, they made a mechanical connection again. Even if you cycled at lower temperatures, they would always pass testing, get into the field, and have early-life failures. During failure analysis, we would find out that the micro- via was separated. Good micro- vias don't do that, so what hap- pened? We explored how we might detect this, and concluded that the easiest way was to run coupons through multiple sim- ulated reflows and monitor the resistance of a daisy-chain of via structures. Do they break dur- ing the simulated reflow cycles? In coming up with this, we observed that assemblers would attach components to side one first, then to side two. They might put some connectors on it. They might have a bad component or two and remove the compo- nent and then replace the com- ponent back onto the board. We could have as many as six sol- der processes prior to the PCBA being completed. So, how do we simulate that entire component attachment process cycle and be believable? We created IPC test method 2.6.27 and said, "We can do six simulated reflow cycles, which will somewhat represent what you do when you attach a com- ponent to a board. That should allow us to find these types of failures." In the Method, we created profiles at 230°C for the leaded solders, at 245°C for the in- between solders, and 260°C for lead-free solders. The response was, "That's great, but I would also like to know what's going on after I put the compo- nents on. Do I have a little bit of life left after the component attachment process?" We then decided to perform this multiple reflow assembly simulation first, followed by 100 thermal shock cycles, ensuring they are con- ducted below the glass transi- tion temperature Tg of the base material. This approach pre- vents overstressing the material and avoids introducing acceler- ation factors that do not occur in real life. That way, we gain a lit- tle bit of life information as well. That's how the current require- ment came forward: six cycles of reflow simulation followed by 100 cycles of thermal shock. I set out to make a single piece