The next generation of automobiles — whether hybrid, fuel cell or combustion engine — promises better fuel efficiency and greater options for in-car entertainment systems, including connections for computers, video players, navigation systems and MP3 players. However, these additional electronics systems present additional challenges for auto designers and manufacturers.
The waste heat from the growing number of electronic devices — such as insulated gate bipolar transistor (IGBT)/metal-oxide semiconductor field-effect transistors (MOSFET) modules, power ICs or PICs, passive components, and batteries — strains the capabilities of conventional air/liquid-cooling systems as they are called on to handle multiple high-heat-flux components spread throughout the automobile (Figure 1).
However, innovative technology for cold plates developed at Amulaire can deliver the thermal efficiency and small size required, while also taking advantage of existing cooling loops.
RETHINKING THERMAL DESIGNS
Electronic modules and integrated circuits (ICs) continue to shrink and become more powerful, and all must be cooled in an efficient, reliable manner. These thermal solutions must be somewhat ingenious to meet automobile manufacturers' requirements for decreased size and weight without any sacrifice in performance. Obviously, it wouldn't work to strap on a 50-pound aluminum heat sink with a blower attached.
Automobile designers already have many years of experience working with liquid cooling, and they are not afraid of mixing electronics and liquid in the same application. In addition, they already have a liquid-cooling loop circulating through the target zone for cooling the engine. The challenge becomes finding a way to use existing pumps, hoses and radiators within the engine compartment, tapping off these components to also cool the new electronics. And the only way to do this is with a very efficient cold plate, because you also must consider added pressure drop caused by such heat sinks. (If it is too restrictive, you will need a separate thermal loop for your electronics — which is cost-prohibitive.)
Let's look at a typical 100 kW drive in a hybrid vehicle. The waste heat is typically 2%, or 2000 Watts of heat that must be removed. Generally, an aluminum cold plate will experience a 30-40 °C rise per 1000 Watts of heat (depending on flow and the design of the cold plate). With typical inlet fluid temperatures around 85-100 °C, coupled with a 30-40 °C rise at the cold plate, the electronics you are trying to cool might experience case and junction temperatures well in excess of 130-140 °C, which is above the IGBT manufacturers' specifications.
Engineers have considered using micro channels and spray cooling for thermal improvements, but this approach would greatly increase the pressure drop within the loop. The goal is to use the existing loop within the engine compartment, thus removing the additional expense of redundant hoses, pumps and radiators. Adding a better cold plate would solve many other issues while reducing the cost of a separate thermal loop just for the electronics (Figure 2).
The traditional aluminum-and-copper embedded tube technology is not efficient enough to satisfy the thermal requirements needed to remove 2000 W with at 85-105 °C inlet temperature. Because aluminum thermal conductivity is approximately 200 W/mK (cast aluminum is roughly 160 W/mK) and copper is 400 W/mK, copper is the better choice in higher-heat-flux applications. However, copper is more expensive and difficult to machine, and it weighs more than aluminum. The design of a cold plate must take these issues into consideration and use the copper only where it is needed.
Amulaire has found that efficient copper cold plates allow thermal engineers to continue using the existing thermal loop within the engine compartment, while adding the flexibility to shrink their electric drives or make them more powerful.
Thermal modeling shows that in liquid cooling, heat does not spread more than a few millimeters in any direction. Applying copper fins directly underneath the heat source enables the system to remove the heat more efficiently. (Of course, increasing the surface area and turbulent flow under these hot spots also contributes to the efficiency of the cold plate.)
The thermal performance of copper cold plates is many times greater than that of existing alumi-num technology. In addition, large aluminum-cast cold plates can be reduced to the size of the power electronics, in some cases becoming an integral part of the modules themselves. Because these cold plates are more efficient, you can reduce the size of your drive, reduce the number of IGBT modules used, or use a smaller IGBT and take advantage of the thermal efficiencies of the cold plate.
When designing a cold plate for the automobile market, another consideration is high flow rate, which affects the shape of the pins. If you are going to tap into an existing cooling loop, you don't want to introduce a high-restriction solution. Pressure drop is an important aspect of cold-plate performance. Rounded, staggered copper pin fins in liquid-cooling systems provide very high efficiency and allow for a low-pressure drop. Because the rounded pins have no sharp edges, the water moves unimpeded around and over them, while the staggered geometry assists heat removal by breaking down the stagnant boundary layer of fluid that forms around each pin.
In the automotive electronics systems being discussed here, the heat transfer is from the hot spot — e.g., PICs — to the copper base plate, and the water circulates around the copper pins. As a result, it serves no thermal benefit to have the entire cold plate made from copper, so the cover of the cold plate does not have to be copper. A combination of copper base plate and aluminum can be used. However, you must add inhibitors to the loop to avoid galvanic corrosion issues. You can also incorporate composite materials as the cover design.
Other space-saving designs include a double-sided cold plate, which can cool six IGBT modules at once (Figure 3). This design is ideal for regeneration systems in braking appli-cations. Because the IGBT modules are not typically running at the same time the driver brakes, the efficiency of the cold plates is not compromised, allowing for compact electric motor drives (Figure 4).
All of the cold plates described so far are designed as bolt-on designs (Figure 5), which means you can purchase a catalog IGBT from your favorite vendor, add some thermal interface material (TIM), otherwise known as grease, and you are ready to go. This approach offers plug-and-play advantages.
Further integration of this technology is possible, however, and is being used by early adopters who are mounting the PICs directly on top of the base plate. Integration removes at least one layer of TIM and a layer of copper, which will reduce the weight and cost of the overall solution (Figure 6). It will also improve the thermal efficiency and give the designer the flexibility to shrink the drives even more, thereby designing a robust, powerful solution without sacrificing the full loads of the available IGBT modules.
The technology exists today to integrate cold plate advances into PICs. The market reality, though, is that many PIC vendors might prefer to concentrate on selling more general-purpose modules without a specific cooling technology inte-grated, leaving the thermal solution details to their customers. On the other hand, because the thermal benefit of integrated pin fin base plates is so great (Figure 7) — and the high-performance needs of the automobile segment represent high-volume potential — we predict that integrated modules will eventually become the norm for PIC vendors' offerings.
The need for new cooling technologies will continue to grow, and it only makes sense to remove as much thermal resistance as possible. The next generation of vehicles will have more electric and electronic systems with increased power density for multifunctionality and multiple missions. The drive-by-wire trend will increasingly replace current mechanical and hydraulic actuators with electrical ones.
The increasing waste heat from these electrical devices will present greater challenges for thermal technologies. By using the innovative technology and continuing the creative use of its integration, future thermal improvements will add additional flexibility to auto designers to achieve their desired results. This example of the exotic, ingenious use of common materials can provide a cost-effective, lightweight, high-flow solution the market has been searching for.
David P. Bono is vice president at Amulaire Thermal Technology in San Diego, Calif. He can be reached at [email protected].
Editor's note: NanoPin is a trademark of Amulaire.