Design007 Magazine

Design007-Nov2018

Issue link: https://iconnect007.uberflip.com/i/1053050

Contents of this Issue

Navigation

Page 17 of 105

18 DESIGN007 MAGAZINE I NOVEMBER 2018 Brown: What we see in the modeling and labo- ratory is that when people stack up microvias, they can actually be less reliable than plated through-holes. The kind of design configuration that seems to work better is one where if you have to go a long distance through a board, you go a little way, shoot over on the internal layer, go down a little bit, and shoot over through another spot. As I mentioned earlier, these material configurations in today's boards are extremely complex structures. They have all of those different properties of polymers and met - als and things like that, so yes, you can model the behavior of those microvias and then see what kinds of failures are anticipated. When we initially tried to validate this reli - ability physics package about 10 years ago, we had some struggles with that validation pro- cess. We wanted a high correlation between what the model was showing and what we could demonstrate either in the field or labora- tory with temperature cycling or shock vibra- tion. What we discovered was that the material properties of the polymers, for example, in the prepreg and laminates in the board, weren't precisely per the manufacturer's datasheets. There were changes in the modulus of elastic - ity, in the X, Y, Z, and coefficient of thermal expansion (CTE) that didn't match up exactly with the manufacturer's datasheets. When we were going through the validation phase, part of the software development process was to match up those real-world physical con - stants with the way that they really play out inside board structures, rather than what's on the datasheet for just a poly- mer layer. Holden: Nobody has gone to DfR yet with this because we think we have a containment process, so if the coupons pass, they'll assemble the boards, but we haven't figured out how to find the root cause. As many as 40% of the boards are being returned. They're failing the con - tainment process, which is expensive for the fabricator and for schedules and time- lines. We have dozens of theories about why it is, but no real hard evidence. Matties: Thanks for your time today, Dock. Is there anything that we haven't talked about that you feel we should share with the industry? Brown: One of the things that I would like to see a broader spreading of the knowledge that engineering is as much as art as it is a science, which is one of the things I talk about in my design for excellence classes at both SMTAI and IPC APEX EXPO. Part of engineering being an art means that as engineers, we are called upon to develop good intuitions as well as good analytical programs. Analyses and com- puter simulations are expensive. You can't ana- lyze everything, so the art part of that comes in knowing when to deploy more expensive modeling techniques, and when you can allow your intuition to take place. Back to the example of automotive; for instance, there were design rules that people of various industries applied in the old days. The origin of those design rules for tempera- ture cycling was a relatively smooth diurnal day-night cycle where things gradually heat up in the morning, get hot around 4:00 p.m., and cool off at night. The next morning, you go through that same thing in a gentle once- a-day diurnal cycle. What some of the people in automotive are now discovering is that info- tainment products that are in the cabin of the automobile are seeing temperature cycling pro- files that are dramatically different.

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - Design007-Nov2018