When passive heat management isn’t enough, fans or blowers are “cool” solutions.

It has been said that good things come in small packages. This is true of electronics up to a point. Chip companies compete for bragging rights over package size. Less is more. Increased density permits faster operation due to reduced lag in shorter traces. This trend extends beyond just chips; shrinking the entire product represents a significant “marketing breakthrough.”

Meanwhile, a compact board will have more wattage to dissipate per unit area. The power supplies are spread around the board with dedicated regulators assigned to each chip even when the voltage requirements are the same from one device to another.

Linear power supplies vs. switch mode power supplies (SMPS). Traditional linear power supplies are not as efficient as switch-mode power supplies. The difference is that linear power supplies create much more heat than SMPS devices. That heat doesn’t just materialize out of nothing. Power is wasted to create the hot spot.

Switch-mode power supplies avoid this problem but create a different issue. The tradeoff is that the SMPS circuits generate considerable noise, which must be managed with a large LC filter. While the linear power supply produces continuous power at a given voltage, the “switch mode” toggles the full input power on and off to produce an average power sufficient for the device under load.

We use the filter to smooth the SMPS output voltage so that the device behaves according to the app notes. The placement and routing of the filter and other components are key factors in the performance of the power supply. Most mobile devices will use this technology (Figure 1).

Figure 1. A graphics accelerator card or GPU uses a heatsink with pin fins for dissipating the thermal load. Note the left edge of the image shows two D-PAK (TO-252) components, which are linear regulators. They get warm on their own but avoid the potential noise concern of an SMPS. Since this GPU is ultimately plugged into the wall socket, efficiency isn’t as important as a lower noise floor. (Image: Author)

Heavy copper: A means to turn the whole PCB into a heat sink. As part of the Chrome OS Hardware team, I was tasked with the layout of two boards intended to test different laptop batteries. They were big boards that held several batteries each. Beyond the natural heat of operation, the boards were placed in a type of oven known as a burn-in chamber. It’s a carefully regulated hot box that includes a pen that charts the temperature over time on a scroll of graph paper. This would serve as proof of the environmental conditions of the test.

A bunch of batteries running near their maximum rated output in a torture chamber that cycles the temperature up and down would require a robust board. I went with 4-oz. copper on the outer layers and 3-oz. copper on the two innerlayers. The vias were huge relative to the normal ones. I also designed an aluminum frame to go under the board, which would stiffen it against the weight of all the batteries while also acting as a heat spreader.

Some batteries would not make it through the test, and those “infant failures” were analyzed. While the cells were off-the-shelf items, the battery packs themselves were bespoke units. We needed to know if the cause of failure was the cells or the construction. That’s why the PCB fixture was overdesigned. I took the total expected dissipation of the cells and did 100% derating. If we needed to deal with 45W, the design goal was to accommodate 90W per battery (Figure 2).

Figure 2. This is a board that the fabricator, Sanmina, used in its trade show display. What’s remarkable is that the bottom (layer 4) is a solid plate of copper. Two dielectric layers of Getek separated the three other layers. High-temperature-rated materials and 45-oz. copper across the entire bottom gave this amplifier the robustness required to drive signals from cellular base stations to phones within the cell. No fewer than five meetings preceded the tape-out. (Image: Author)

Active cooling: When passive efforts fall short. Moving air is a good thing. When it happens naturally, we use a term called windchill to factor in the cooling effects beyond the ambient temperature. Internal combustion vehicles typically feature a radiator with coolant flowing through it, while a fan and the vehicle’s movement augment the flow of air through the radiator’s fins.

The same thing plays out on a printed circuit board. Cooling fins are attached to the high-wattage devices, and a small fan can help evacuate the lingering heat. This prolongs the lifespan of the device and permits it to keep running while crunching the numbers. Really thin blowers are incorporated into the z-stack of laptops to prevent thermal shutdown. It’s typical to cut out the board around it so it can be mid-mounted, reducing the impact on the overall thickness of the machine.

Graphic-heavy programs can cause the fan to come on just by opening a sufficiently large design. Running certain ECAD automation is sure to tax the system. Given a choice, I’d prefer a tower computer to a laptop for this reason. The airflow is better, the heat sources are spread out and there’s room for more active cooling. My gamer laptop works for ECAD, but I can hear it struggling right away (Figure 3).

Figure 3. Active cooling methods include the fan over the heat sink over the SoC. Furthermore, heatpipes can be implemented to pull the heat toward a radiator located away from the hot spot(s). This is from an old “gamer” computer where graphics are a priority. (Image: Author)

Wrapping up the discussion on thermal challenges. Success in thermal management includes paying attention to the ground paddles that underpin the QFN/QFP packages with perimeter pins and a big central pad that is grounded. Space constrains us to put vias in the big pad. Microvias are naturally better, even if they are not rated to dissipate as much wattage as a full-plated through-hole via. The soldering process is less of a risk with microvias. Plated through-hole vias should be plugged with a solder mask or filled with thermally conductive material for best results. An open via hole is an invitation for solder to creep away from where it’s required under the flatpack device.

The specific methods will vary depending on the application and, of course, the chipset itself. The form factor will determine the available solutions. It could involve the use of graphene or other exotic materials to extract thermal energy from the system. As parts and systems shrink, we’re tasked with ever-increasing current density. We need to consider thermal management during floor planning so that there’s space between hot spots.

Although a connector doesn’t create heat on its own, it may be under-provisioned in terms of the number and size of the pins to carry the current. Breather holes in the chassis and other efforts may be considered in the service of thermal management. Derating components can prolong the mean time between failures. A well-planned layout will include considerations for thermal management. Aim for higher reliability for lower ongoing costs related to field service or warranty claims. It all starts with keeping your cool.

John Burkhert, Jr. is a principle PCB designer in retirement. For the past several years, he has been sharing what he has learned for the sake of helping fresh and ambitious PCB designers. The knowledge is passed along through stories and lessons learned from three decades of design, including the most basic one-layer board up to the high-reliability rigid-flex HDI designs for aerospace and military applications. His well-earned free time is spent on a bike, or with a mic doing a karaoke jam.