New 3D thermal quantities help designers address thermal problems as they arise.
Electronics thermal management is the discipline of designing electronics systems to facilitate the effective removal of heat from the active surface of integrated circuits to a colder ambient environment. In doing so, heat passes from the package both directly to the surrounding air and via the PCB on which the IC is mounted. The PCB and, to a lesser extent, the surrounding air thermally couple the various heat sources.
Heat coupling increases as components and PCBs become smaller and more powerful. Designers must take remedial action to bring all components within their respective thermal specifications, but this step is becoming more challenging and constrained, even when preventative measures are taken early in the design process.
For the past 20 years, computational fluid dynamics (CFD) techniques have provided 3D thermal simulations that include views of the air-side heat transfer that predict component junction and case temperatures under actual operating conditions. Designers routinely use these predicted temperatures to judge thermal compliance simply by comparing the simulated temperatures to maximum rated operating temperatures. If the operating temperature exceeds the maximum rated value, there will be at least a potential degradation in the performance of the packaged IC, and at worst, an unacceptable risk of thermo-mechanical failure.
Simulated 3D temperature and flow fields provide detailed and useful information, but give little physical insight into why the temperature field is the way it is. Examining heat flux vectors can yield some insight into the heat removal paths. But the heat flux vector direction and magnitude data do not provide a measure of the ease with which heat leaves the system. Nor do they provide insight as to where and how the heat flux distribution might be better balanced or reconfigured to improve performance.
How easily heat passes from the various sources to the ambient will determine the temperature rise at the sources and all points in between. Heat flow paths are complex and three-dimensional, carrying portions of the heat with varying degrees of ease. Paths that carry a lot of heat but offer large resistances to that heat flow represent bottlenecks. A redesign can relieve these bottlenecks, permitting heat to pass to the ambient more easily and reducing temperature rises along the heat flow path all the way back to the heat source. In addition, there may be unrealized opportunities to introduce new heat flow paths that would permit heat to pass to colder areas and out to the ambient. So a redesign informed by the right information can do more than alleviate bottlenecks; it can also introduce thermal “shortcuts” to bypass them.
Sans a way to identify such thermal bottlenecks and shortcut opportunities within a design, PCB design teams have faced a stark choice. Either bring in thermal experts to resolve thermal problems, or rely on being able to add heat sinks later. Lacking direction from the simulation results as to appropriate remedial action, thermal engineers have traditionally relied on experience and engineering judgment to guide their search for design improvements. Today their work is often supplemented by design of experiments and automatic design optimization capabilities within the thermal simulator. Such approaches take time.
An innovative way to view the thermal behavior of a populated PCB uses two new 3D thermal quantities aptly known as the BottleNeck (Bn) Number and the ShortCut (Sc) Number that, taken together, guide designers to take appropriate, targeted remedial action to address thermal problems as they are encountered.
Vectors, Bottlenecks and Shortcuts
Heat flow can be defined in terms of a heat flow through a given cross-sectional area. This measure is known as a heat flux. The presence of a heat flux vector will always be accompanied by a temperature gradient vector. The temperature gradient field is taken to be an indicator of conductive thermal resistance as, for a given heat flux, the greater the temperature gradient is, the larger the thermal resistance will be.
The dimensionalized Bn number is the dot product of these two vector quantities. At each point in space (Figure 1) where there exists a heat flux vector and temperature gradient vector, the Bn scalar at that point is calculated as:
Bn = Heat flux magnitude × Temperature gradient magnitude × sting bottlenecks.
This is not always true, of course, especially for multiple heat sources (as found on almost every PCB) where heat flow topologies for widely separated components can be quite independent of each other.
To illustrate how this process works in practice, consider a typical air-cooled PCB (Figure 2). The central BGA has highest temperature rise above specification, followed by the two TO-263s above and to the right.
Though it depicts the same PCB, Figure 3 is not the same kind of thermal view as Figure 2. Instead, it shows the Sc number distribution mapped at a point just above the tops of the packages on the board. Although in Figure 3 the largest Sc numbers are associated with the hottest component, this is not always the case. A component might be hot due to the temperature of the surrounding air, rather than its own internal power dissipation. In the case of this centrally located BGA, the large Sc values on its top surface indicate relatively efficient convective heat transfer locally. Therefore, the obvious remedial design action is to add a heat sink. The heat sink acts as an area extender, making it even easier for heat to leave the top of the component and to be carried away by the air. Introducing this design change reduces the BGA’s junction temperature rise by 70%, taking it well below its maximum safe operating limit. With the BGA running thermally compliant, let’s turn our attention to the TO-263 components.
Figure 4 is a Bn plot depicting the Bn distribution in the top signal layer of the PCB. We can see that, after adding the BGA heat sink, the largest thermal bottlenecks exist near the tabs of the two TO-263 devices. Recall that large Bn values do not mean this is the hottest area. Instead, the Bn figures and the plot reveal areas in which a lot of heat flows “downstream” from the heat source, and is highly restricted. Knowing exactly where the bottleneck is, a large copper pad can be added to cover that high bottleneck area, providing a targeted solution to a specifically identified problem. That is effective, efficient engineering.
Having made this modification, best practice methods call for an updated thermal simulation and inspection of the new Bn and Sc distributions. Figure 5 shows Sc on a cross-section through one of the two TO-263s after the addition of the copper pad. The expanded inset view shows large Sc values on the signal layer and the power and ground planes below the new copper pad and TO-263 tabs, indicating a shortcut opportunity between these layers. This agrees with a designer’s intuition, since heat spreads readily in the metallic layers of the PCB, while the dielectric’s low thermal conductivity acts as an effective barrier to heat transfer. Adding thermal vias to create a new heat transfer path down to the buried ground plane is an excellent, practical way to take advantage of this shortcut opportunity. Note that the Sc field pinpoints exactly where the thermal vias should be added for maximum effect.
By examining the Bn and Sc variations in and around the TO-263, the exact shape of the copper pad and the location of an array of thermal vias (shown schematically in Figure 6) can be determined quickly, without resorting to numerous “what if” studies. In this case, adding the pad and vias yielded a 30% drop in the temperature rise of the TO-263 devices, again taking them below their maximum rated temperatures.
The Bn and Sc fields together provide invaluable insight and comprehension about temperature distribution and behavior in an electronics system. By detecting and mapping both thermal constrictions (bottlenecks) and potential shortcuts for more efficient heat transfer, these parameters enable engineers to quickly determine the most promising thermal design changes – those most likely to provide the most efficient, effective results – without years of thermal experience and intuition.
In the example, three Bn- and Sc-inspired thermal design changes were identified quickly, and the resulting “fixes” dramatically reduced the temperature of the three overheating components discovered in the initial simulation. It is an approach that delivers important gains in simulation productivity. Rather than simulating all possible remedial actions for thermal problems and choosing the best one, engineers can see immediately where they need to focus their thermal design effort.