Issue link: https://iconnect007.uberflip.com/i/652863
46 The PCB Design Magazine • March 2016 the trace measured with thermocouples. The hottest point on the 27 mil trace was 114 o C, measured to the left of the via. The via temper- ature was 109 o C, about 4% COOLER than the trace! The hottest point on the 200 mil trace was 44.5 o C, measured at the via. The temperature to the left of the via was about 40.5 o C (found by experiment to be the maximum temperature of the trace under similar conditions without a via). It is important to note that: A 6.6 Amp current through the 27 mil-wide trace results in a via temperature of 109 o C, while a higher current of 8.6 Amps in a larger (200 mil) trace results in a much lower via temperature of only 44.5 o C. This confirms that it is the trace that is controlling the via temperature. We wanted to run a simulation of the ac- tual test board, since it had different dimen- sions than those hypothesized in our first pa- per. The simulation technique is described in previous papers on the UltraCAD website. The simulation results are provided, and compared to the empirical test results, in Table 1. As can be seen, the simulation results compare quite favorably to the measured results. The thermal profiles for the two simulations are provided in Figure 2. Two questions come to mind: Why doesn't the via heat up more? The answer is that the thermal conductivity between the via and the copper trace is so good that the via can't heat up (much) more than the trace. Any excess heat conducts away from the via into the trace and is then conducted into the board material—the same material that is cooling the trace. So as long as the trace is sized correctly to carry the current, the via can't get much hotter (maybe less than 10% at most) than the board. Why is the via COOLER than the trace in the 27 mil case? Since the via and trace conducting areas are the same size, one might expect the via and the trace to be the same temperature. But the via looks something like an internal conduc- tor, completely surrounded by board material. The trace has conducting material underneath it, but air above it. And the board is more ef- ficient at cooling the trace than is the air. This is the same reason IPC 2152 found that internal traces run cooler than external traces, all other things equal. The conclusion? It would appear that a sin- gle, 10 mil diameter via is adequate for conduct- Figure 2: Thermal profiles of a 27 mil trace (top) carrying 6.65 A and a 200 mil trace (bottom) carrying 8.55 A. hoW many Vias does it take to…?