Issue link: https://iconnect007.uberflip.com/i/916068
18 The PCB Design Magazine • December 2017 DR. JOHANNES ADAM SOUNDS THE ALARM FOR THERMAL DESIGN From those, and from customers, I learned what data they have and what they would need in order to calculate PCB temperatures before the manufacturing or the lab process. In my eyes, designing printed boards is an art and manufac- turing is amazing. All of this experience, previous and new, then went into my new TRM software. The basic idea is that you have an existing design, whether finished or just in the experimental state. You import the Gerber files from the lay- out software, add components, the dissipated power, and optionally as- sign Ampere values to traces, and then simulate what temperature can be expected under certain ambient conditions. Shaughnessy: So, you think there is kind of a disconnect? Perhaps the EEs don't realize that they don't have all the data? Adam: Yes, and some don't even know the proper data. I guess that universi- ty education is rare as far as this subject is concerned. Typically, thermal resis- tance is all about thermal management. But thermal resistance depends on anything and everything. It depends on the boundary conditions, on the layer count, on the vicinity of a component or the trace you are interested in. Shaughnessy: What do you think PCB design- ers and EEs should take away from this? What would be a couple of things that they should do or should think about, going forward? Adam: They should understand that any com- ponent or trace has neighbors, heated or cold, and they are mounted in or on a very complex geometry of copper on a dielectric. This whole mixture makes it almost impossible to have a single value for a temperature, or a value you can print into some data sheet. And there is the issue of boundary conditions. Electrical engineers get some crude specification, for ex- ample, that it should operate in 85°C ambient. But there's a principle difference: whether the ambient 85°C is the ambient air, or whether it's a solid piece of metal which is 85°C, whether it is air around or some other fluid. Each case will give completely different results. I'm always claiming that no application note can reflect your individual board and ap- plication. If we look at design rules, it's even worse. Typically, they consider a very low layer count where the tempera- ture created by a small amount of heat is ver y high. But, if you deal with multilayers, much more heat will work fine, won't it? In fact, you are mostly on the safe side applying design rules, but you are at risk of overdesigning and losing money. Applying laws of physics to the spe - cific board gives a much more precise answer. The horizon should be wider, because the electri- cal engineer is not the only one who contributes to the design of a PCB or a device. There is the product man- agement, purchase department, marketing and design studio who, for example, want to decide on time and cost. The EE has to be firm and has to have the proper arguments to back up his ideas. Regarding temperature one of the best methods is to show pictures from simulations comparing different options as early as pos- sible. He can do that already in early stages of the design cycle, of course, with some assump- tions on power dissipation, currents, and so on. Finally, he will take a thermal image of the real PCB, but by then has got to be manufactured, to have components and has to be operating, all of which costs at least time and shouldn't require much iterations. Shaughnessy: It seems as if thermal has only be- come a "hot" topic, pardon the pun, in the past Johannes Adam