Issue link: https://iconnect007.uberflip.com/i/1307491
18 DESIGN007 MAGAZINE I NOVEMBER 2020 so closely with our suppliers, reliability teams, manufacturing teams, and producibility teams, we continue to collaborate to optimize our pro- cesses and designs. That way, our end-product is produced using industry best practices and les- sons learned to meet all necessary requirements. And latent failures, which is what you just described, are a problem that you don't ini- tially see; it just happens all of a sudden. The worst case is to have a latent failure out in the field because it potentially has a significant negative impact as it can ruin your company's reputation, costs hundreds of millions of dol- lars, or causes permanent injury or death. Shaughnessy: And they still haven't figured out exactly what's causing that. Holden: Somebody might have, but they won't share it with us. We tried to get the data, and although the scientists and the process engi- neers agreed to it, the legal department and management shut down the sharing of the data. My best hope right now is that we're going to get the results from the European Space Agency, where they've done staggered and stacked vias. They use four different meth- ods of testing, which gives us a whole lot of data about which methods of testing the final board are reliable, and which ones aren't, and how they manage to slip through. IPC paid for the European Space Agency to make a set of these panels for us in the U.S. using a laminate that we're accustomed to using because the European Space Agency is not using laminates that our military aerospace compa- nies in North America are using. But once we get those panels and get them tested, we'll start to have some data. We didn't do a design of experiments, so we're not going to get at the root cause, but we are going to get at what's reliable and what's not, how much can you stack, etc. I'm depending on Marc Carter—founder of Aeromarc and former director of technology transfer for IPC—and his experiments. Marc has process variables that he's changed. Hope- fully, he can provide insight into what the root cause of military via failure might be this year because he's actually going to build boards using direct metallization as well as electroless copper. A lot of experts tell me that the solution is simple, "Go to direct metallization," but the military doesn't allow that. Marc's data, which the Army is paying for, may say, "You can make it work with electroless copper," or, "sometimes it doesn't work, but it always works with direct metallization." That's kind of an answer. Shaughnessy: What would you say are some of the things that designers can do to help achieve more reliable vias? Chavez: The biggest thing that I always say is to communicate with your suppliers. You must have that collaboration and communication. The second thing is that it has to be designed to industry best practices. It's okay to push the envelope, but if you're pushing the envelope, you had better have justification as to why you're doing it, as well as collaboration with your suppliers to back up what you're doing. With aerospace, the volumes are significantly less than what you would see on the com- mercial side, with volumes of 1,000 or even 100,000 units. In aerospace, you could have 250 units, and that's a lifetime run. Holden: I always tell people, "If you have to do something different, for heaven's sake, start with a test vehicle." Chavez: A lot of the times, especially within these large corporations, you have a budget for that. With other projects, you often don't have that luxury. With your first prototype or article, you have one shot at it. If you're going to do that, you need to have all the stakehold- ers at the table. That way, you're doing it right from the beginning, and you have everyone's input to give you your best shot at success. Shaughnessy: Do you use landless vias? Chavez: No. I do know that's out there, but we do not, to my knowledge. Holden: Landless is not Class 3. Because of the way they define it, Class 3 has lower reliability