Design007 Magazine

PCBD-Dec2017

Issue link: http://iconnect007.uberflip.com/i/916068

Contents of this Issue

Navigation

Page 23 of 87

24 The PCB Design Magazine • December 2017 That opened some ad- ditional insights regard- ing whether traces heat uniformly or not (they don't). And that opened even more avenues for study. Soon, we had enough new information for a book [4] . Shaughnessy: After all that research, what ther- mal design issue really stands out in your mind? Brooks: We discovered several very interest- ing insights, but by far the most dramatic was how wrong we had all been regarding via tem- peratures. Our industry- wide rule of thumb has been that a via's cross- sectional area should be the same as the trace cross-sectional area. If it is not, then multiple vias should be used [5] . It turns out that if the cross-sectional areas are equal, then the via is cooler than the trace. And if not, the via can take a lot more current than we had imagined. In most cases, if the trace has been sized cor- rectly, only a single, small via is needed, almost regardless of current level. Shaughnessy: How can you get by with a single, small via, and regardless of current? That seems counterintuitive. Brooks: The reason is that the via length is very small compared to the trace. The trace acts as a heat sink for the via and conducts heat away from the via. We can easily push two to three times the expected current (and more) through a via, and the heat-sinking properties of the trace will keep the via temperature under con- trol. These results are explained in detail in Chapter 7 of the book. Shaughnessy: That result was based on simulations. Is that when the experi- mental work started? Brooks: Yes. We knew that no one would accept those results without ex- perimental verification. I went to Prototron Cir- cuits in Redmond, Wash- ington, and asked if they would provide some test boards for us. They were very generous in provid- ing via test boards, and then several other boards for subsequent testing. The via experimental re- sults confirmed the sim- ulations, as described in Chapter 9 of the book. I could not have done ev- erything that I did in the book without Prototron's contribution. I was extremely fortu- nate that everyone asked for support was willing to help. Eight persons or companies are mentioned in the "Acknowledge - ment" section in the book, each of which pro- vided invaluable services or advice. For example, C-Therm Technologies (Fredericton, New Bruns- wick) took board material samples and measured the thermal conductivity coefficients for us. The Jesse Garant Metrology Center (Windsor, Ontar- io) took X-rays of the via board for us. I am very grateful for it and humbled by it. Shaughnessy: After all that, is there any general conclusion regarding trace currents and tem- peratures that you'd like to share? Brooks: I'd like to highlight two. First, the IPC- 2152 data are worst-case. By that, I mean that a single trace in isolation is a worst case. The data are correct, but we almost never have a trace in isolation. Anything we do in a real-world design sense lowers the temperature. THERMAL MANAGEMENT UPDATE WITH DOUG BROOKS Figure 1: Doug Brooks recently collaborated with Johannes Adam on this book, and their research yielded some interesting results.

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - PCBD-Dec2017