Design007 Magazine


Issue link:

Contents of this Issue


Page 21 of 87

22 The PCB Design Magazine • December 2017 by Andy Shaughnessy I-CONNECT007 I had the opportunity to talk with our regular contributor Doug Brooks recently. He has been doing some research on temperature effects on PCB traces over the last few years, and I wanted to check the status of his latest thermal efforts. He discussed his work with Dr. Johannes Adam, why temperature charts based on a trace in isolation are inaccurate, and how the industry remained so wrong about PCB temperatures. Andy Shaughnessy: You have done some work on thermal man- agement lately. How did that project start? Doug Brooks: I wrote an article in the mid- '90s on trace current/ temperature effects, and I used two data sources: the then- current IPC data and some data I found in a 1968 Design News (DN) article. The DN temperatures were about 30–40% higher than the IPC tempera- tures and I wondered why. I began to suspect that it was because of the differences in the way the temperatures were measured or calculated. In looking for a way to confirm that hypoth- esis, I ran across an article about three years ago written by Dr. Johannes Adam, and I contacted him. It turns out that Johannes had written a com- puter simulation program called TRM (Thermal Risk Management) [1] that was well suited for me to use to look at the data I had used in the arti- cle. He offered me a license for the software and we used TRM to simulate the IPC trace data in IPC-2152 [2] and the earlier data from DN. The simulations were very successful. Shaughnessy: What did you find out? Brooks: It turns out the DN data were unreli- able! Shaughnessy: And all of this took place over several years? Brooks: No. That was just the beginning. It was so easy to simulate the IPC trace data that we began to simulate more realistic scenarios. The IPC data apply to a 6-inch trace in isolation. We began to look at what happens when we change things: change the length, change the pad sizes, add additional adjacent traces, add planes below the trace or on the other side of the board, more common layout condi- tions like those. Then we wondered if we could simulate the temperature of a via, something that is difficult to do in practice and to our knowledge had not been done before. When that was successful, we looked at fusing tempera- tures. This is something I had written about earlier and had developed some basic rules for based on Onderdonk's Equation [3] . Thermal Management Update with Doug Brooks FEATURE INTERVIEW Doug Brooks

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - PCBD-Dec2017