Issue link: https://iconnect007.uberflip.com/i/1011746
26 PCB007 MAGAZINE I AUGUST 2018 Does it exceed the capabilities of the material? Now we go to UL. We're not allowed to do that under UL. Holden: It has nothing to do with the materi- al. It has everything to do with soldering and the number of reflow cycles a board is going to see, and the consensus was that you can throw away all the MIL-spec thermal cycling and the IST and the other thermal cycling—they don't find a defect on these microvias. But at least six passes at this 260°C peak, unless you're us- ing bismuth solder and then the spec is 230°C peak instead of 260°C for lead-free, but you have to be monitored by a four-wire system be- cause a crack and a split can occur and then it self-heals itself and then will only fail in the field a couple years later. We put out a white paper called IPC-WP-023. Have you seen that or read that white paper? It's about 30 pages. It's about the hidden reli- ability threats in microvias. Osborn: To be honest with you, no. Unless it's required by our customers, we really wouldn't explore that on our own. Holden: It's funny, even one level deep micro- vias are failing the test. The data is bad be- cause as much as 40% of milaero boards are now being thrown away because they are fail- ing the reliability testing. The reason we called the conference was because the fabricators can't afford to have an overage of 40% to meet the quality today, but we still don't have a root cause as to why these microvia boards fail so often. Osborn: It doesn't play in our arena. It's not something our customers have brought to our attention. Krick: We have done a very small handful of microvias and stack microvias here, none of which has been military. I believe it's all been commercial level. Osborn: Almost all the military, though, are running under the IPC specs. Knapp: And, we've heard no complaints with internal or field parts that we've sent. We had one that was a suspect IST coupon or a D coupon, but other than that we haven't heard of any. Holden: Appendix A from IPC-2221 about the D coupon has a whole new generation of D coupons to better test the finer geometries as well as drilled or laser-drilled blind vias. A couple of the talks at that conference were on that. Krick: I still personally believe that until the laminate suppliers can control the Z-axis ex- pansion, or you can come up with a more ro- bust dynamic of hole creation, I don't think it's going to hold up. I don't think that you can subject laminate to that number of cycles nor plating integrity for that matter, particu- larly when you're looking at sub five-mil mi- crovias and sub five-mil dielectric. I certain- ly don't know technology-wise of where it is to that or where you could go with it today. I mean we certainly see the axis here and we toggle cycle all the stuff that we do but you certainly can't expect the Z-axis expansion to be less than a few mils before it's going to end up pulling a four-mil stacked via apart. That's mechanical. Osborn: But when it cools down, it works. Krick: Then it heals right back. That is correct. Knapp: Again, specs don't necessarily reflect reality. Osborn: It goes back to communication. Feinberg: Well, I want to thank everybody. This has been a very interesting call. Osborn: This was fun. You're welcome. PCB007