Issue link: https://iconnect007.uberflip.com/i/713934
60 The PCB Design Magazine • August 2016 surements using a small thermocouple. I no- ticed that if I moved the thermocouple slight- ly, I would get a different temperature reading. Not by much, maybe 1.0° or 1.5°C. This is more than the resolution of the thermocou- ple, but not enough for me to be satisfied that the differences were real. So I began to won- der if these variations in temperature appeared at lower trace temperatures, say in the 40°C. range. I had a variety of test boards available with different trace widths and thicknesses [3] , but I needed a thermal imager to look at them. A lo- cal firefighter loaned me a Fluke model TI32 im- ager which was invaluable for this investigation (see acknowledgements). Figure 3 is one of the images I obtained from those sample boards. It is of a 100 mil wide, 0.5 oz. trace heated to just under 50°C. The temper- ature range of the image has been compressed from about 40°C to 49.6°C. in order to focus in on the minor temperature variations. It is evi- dent that even at this low a temperature, there are temperature variations along the trace. And the hottest point along the trace is not at the midpoint of the trace. In this instance, it is well to the left of that point. But perhaps even more surprising is the temperature variations across the width of the trace. This was apparent in every observed trace [4] . The hottest part of the trace is along the centerline, but the trace cools as we move closer to the edge. These two effects are what I saw when I moved the thermocouple. One would expect that the thermal con- ductivity of the copper is so good that there would be no thermal gradients possible along a trace (except, perhaps, as one approaches the pads at the ends). We would especially expect that to be the case for such a short distance as the width. But there is a theoretical explana- tion for the trace being cooler along the edge (Figure 4). The trace cools by heat conducting away from the trace through the dielectric [5] . The di- electric conducts heat better in the horizontal (in-plane) direction than it does in the vertical (through-plane) direction. This cooling path is YOUR TRACES HAVE HOT SPOTS! Figure 2: The trace temperature varies considerably along the trace length. Figure 3: Temperature variation along a 100 mil wide trace.