PCB007 Magazine


Issue link: http://iconnect007.uberflip.com/i/1050827

Contents of this Issue


Page 63 of 111

64 PCB007 MAGAZINE I NOVEMBER 2018 have very different properties, such as stress and strain properties, fracture mechanics, etc. The Sherlock software can model all that. We combine the modeling with the failure analysis work to do what the people in the automotive industry, for example, are calling "shift left." This is instead of waiting until the last por- tion of the product development cycle where you're doing low-rate initial production, and then verification validation qualification test- ing—if you discover problems there, it's very expensive. What the shift left means is that they try to do pre-failure analysis and reliabili- ty physics modeling right at the schematic cap- ture and board layout portion of the design de- velopment sequence. Matties: Predictive engineering, basically. Brown: Exactly. We can model all those stress- es, strains, and ruptured things you're looking for while it's still in software. If you're going to have problems, you want to find those early. In the early part of my career, I worked at Rock- et Research Corporation, which made rocket engines for satellites and things like that. The watchword there was if you had a problem, where and when do you want to find it? Do you want to find it when it shows up or when it blows up? In the rocket business, sometimes blowing up is literal. That's an analogy for the solution set that we're trying to offer people nowadays. You can do that shift left kind of thing and find the problem when it shows up, as opposed to when it blows up in testing. Matties: That's a new shift in thinking though. Do you see more companies looking for this? Brown: More and more people are looking at it because it's a heck of a lot less expensive. You find issues earlier, and time is money. That's particularly true in the case of automotive. The automotive industry is working on new stan- dards, not only for the circuit board applica- tions, but also within the integrated circuits themselves. As geometries are shrinking in the integrated circuits, as well as at the board lev- el, the CMOS properties inside memory ele- ments and processors are not as long-lived as they used to be. In the old days, people would say, "It's not a tube anymore, it's a transistor—it's all solid state; nothing moves, and nothing will break." What we're finding is those geometries are shrinking in integrated circuits, then at the atomic level, things really do move. We can now model that movement mathematically— for example, electromigration in the metalliza- tion of the integrated circuit. If semiconduc- tor manufacturers are going to have a problem with electromigration, we can help them figure that out before they go with the production. In the case of end users in automotive, which is such a harsh environment, if you're going to have those kinds of issues, you want to be able to figure that out. Matties: Obviously, it's driven by the cost of failure, right? The higher the cost, the more likely they are to look at it on the front end? Brown: It's also the consequence of failure. Matties: That's part of the cost, for sure. As technology changes, I would think that you're seeing new types of failures. What trends and failures are you seeing? Brown: One of the big surprises for all of us was a presentation at the SMTA Pan Pacific Micro- electronics Symposium about three years ago. What the shift left means is that they try to do pre-failure analysis and reliability physics modeling right at the schematic capture and board layout portion of the design development sequence.

Articles in this issue

Archives of this issue

view archives of PCB007 Magazine - PCB007-Nov2018