CFD Thermal Analysis for Improved Reflow

Estimated reading time: 8 minutes

There is an urgent need for PCB design solutions that merge operator expertise on oven behavior with deep knowledge of the PCB’s layout, particularly the copper content and component placement, and their effect on thermal properties. One CFD-based analysis environment tests out possible scenarios based on PCB design data, without taking the reflow oven off-line. In CFD-based reflow profiling, the reflow machine only has to verify the CFD tool’s findings once the PCB design is released for manufacturing.

By John Wilson, Mechanical Analysis Division, Mentor Graphics Corporation

Every process engineer knows that reflow soldering must balance heat, duration, and materials. Critical to successful reflow soldering is the knowledge about those materials, which helps achieve consistency and quality in the finished product. In most PCB manufacturing environments, reflow oven settings historically have been determined by engineers who apply both their own expertise and an empirical trial-and-error approach. No two PCB designs behave exactly the same inside the oven, so engineers perform tests using samples of newly designed PCBs. Thermocouples are attached, reflow cycles performed, results observed. Eventually the engineer reaches a conclusion in the form of nominal oven settings that repeatably produce good wetting, flow, and other crucial soldering characteristics.

These methods are workable but they take time, diverting a valuable reflow oven from its normal profitable duties. Moreover, that time usually comes out of a development schedule that seems too short as it is. Accuracy with this method can depend too much on individual profiling skills.

Innovative new solutions are now emerging: toolsets based on computational fluid dynamics (CFD) analysis software specifically targeted at reflow applications. Such products are designed to help process engineers determine reflow oven parameters quickly and reliably. Underlying this breakthrough is a new generation of easy-to-use desktop CFD tools usable by every engineer, not just analysis specialists. Using information interfaced from the PCB design database, the CFD tools can produce accurate PCB thermal profiles in minutes.

Changes and Challenges in Reflow

The challenge of reflow profiling is akin to that of other thermal issues in electronic design: technologies grow more complex and margins get smaller. The same factors that affect thermal management during the design phase have similarly affected manufacturing. Process engineers attempting to develop reflow oven profiles have seen their challenges intensify for several reasons:

First, today’s PCBs incorporate more metallic layers (more copper) than in the past, leading to a more complicated thermal response as boards proceed through the reflow oven.

Secondly, environmental concerns are changing the business. The world is moving toward lead-free manufacturing, a trend accelerated by Removal of Hazardous Substances (RoHS) directives in the European Union and elsewhere. Lead-free media have a narrower range of workable melt and flow temperatures than their lead-based predecessors, which of course impacts the reflow profile. In addition, lead-free soldering requires higher temperatures, which engineers must tightly control or risk damaging the very components that are being soldered.

Lastly, automated design tools to assists the process engineer have not kept pace with the challenges.

Clearly, there is an urgent need for solutions that merge operator expertise on oven behavior with deep knowledge of the PCB’s layout, particularly the copper content and component placement, and their effect on thermal properties.

A Reflow Profile

For years, most process engineers have begun the development of a reflow oven profile by looking back into archives for profiles of PCBs similar to the current design. Some manufacturers have even codified this kind of information into detailed databases in an attempt to democratize the profiling art.

The next step is to apply some proven rules of thumb about the board’s component population. Variables include:

Thermal mass: smaller components heat up more quickly.

Component location: grouped components heat up more slowly. For evenly spaced group of components, those near the edges and especially in the corners heat up more quickly.

Package type: ceramic packages have a lower specific heat than plastic packages. Cermanic packages tend to heat up more slowly.

Once these elements of the design are accounted for, profilers attach thermocouples on a test board in areas estimated to have the highest and lowest temperatures. The reflow oven must then be taken out of service (typically with grudging approval from the manufacturing manager) for as long as it takes to develop a successful profile. It may require multiple iterations and the thermal response of an actual populated board is nearly impossible to predict using the rules of thumb alone.

Oven Profiling Advances

One methodology lets the process engineer put their experience and knowledge to work in a suitable CFD-based analysis environment. Even before the PCB design is complete, design data can be used to inform countless cycles of “what-if” testing in the CFD software environment, all without any demands on the reflow oven’s time. Once the PCB design is ready for manufacturing, a quick run through the reflow machine will verify the CFD tool’s findings and production can commence. This is the inherent promise of CFD-based reflow profiling.

Figure 1. Maximum body temperature results for two scenarios.

CFD-based profiling did not immediately realize its potential. CFD is a powerful design tool in which a 3D model of the system is created and the governing equations for airflow and heat transfer (convection, conduction, and radiation) are calculated. Conventional CFD software requires a large time investment for actual model creation. Steps range from meshing (a particularly time-consuming procedure) to setting boundary conditions, and post-processing the relevant results. A full-scale CFD set up and simulation might take two days. Even so, the CFD approach is more accurate than other analytical methods because it takes the full geometry of the system into account and relies on fewer simplifying assumptions. Over the last 20 years, CFD has become an integral part of the thermal and airflow design processes, but it has not made much of a mark on reflow profiling.

However, in recent years, major advances have enabled CFD to be applied practically to the reflow process design for the first time. Improvements to automation algorithms for meshing, reflow oven zone representation, and reporting have reduced those two-day simulations to 20-minute exercises. Equally important, the process engineer does not have to become a CFD guru to make this work. Today’s CFD software is tailored to perform reflow analysis, including specific set ups for the familiar reflow machine variables. Analysis result displays are also easy for the non-specialist to understand.

Considering Copper

As implied above, the CFD capability is only part of the reflow profiling automation story. Taking advantage of PCB design data already residing within every enterprise’s EDA tools is the other factor. Some EDA tools now available can export complete, ready-to-simulate CFD thermal analysis models of PCBs. The thermal model should include the PCB layout, board and component dimensions, and PCB stack-up as well as detailed copper distribution information for each layer.

Figure 2. Peak body temperature results.

Timely application of CFD-based reflow profiling processes can benefit a design by allowing the engineer to quickly study the effects of conveyer speeds, zone temperatures, and forced-convection air speeds. The results of the analysis predict time above liquidus (TAL) and maximum body temperature for each component. The latter value can be used to accurately locate thermocouples for physical temperature profiling in the oven – eliminating guesswork.

Proving Analytical Reflow Thermal Profiling

The experiment summarized in Table 1 studies the effect of varying certain inputs on the predicted transient behavior of the reflow process. A virtual four-layer PCB is loaded with 35 components. It passes through a simulated reflow oven with a “ramp-soak-spike” temperature profile, part of the CFD analysis software. The oven in this experiment includes eight heating zones and three cooling zones and is heated from the top and bottom. The ambient temperature used is 24°C and the baseline oven profile has a peak temperature of 268°C.

Table 1 summarizes the results of five design scenarios including Scenario 1, the baseline. Scenarios 2, 3, and 4 vary the conveyer speed, zone temperatures (with a uniform increase), and forced air convection speed individually. Scenario 5 changes all of the settings simultaneously. The two right-most columns in Table 1 present the maximum and minimum body temperatures.

Each change individually renders an increase in body temperatures between 5° and 10°C. When changing all the parameters at once, as in Scenario 5, the body temperatures increased by about 20°C compared to the baseline. Increasing the zone time (decreasing the conveyer speed) was the only change that reduced the temperature gradient between components.

Figure 1.shows the maximum body temperature map for Scenarios 1 and 5. These plots could be used to determine the best points for thermocouple attachment during profile validation testing.

Figure 2.displays the maximum temperatures in terms of “Over Peak” and “Under Melt” temperatures. Components that fall within the process window are displayed in wireframe form. In none of the scenarios did a component exceed the allowable peak body temperature (PBT).

In the final design, the maximum body temperature was within the desired process window on all but three components. The TAL varied from 5 to 30 sec.

It is worth noting here that the solution time for each of the cases was about 10 minutes. All-day CFD computations are a thing of the past. No special hardware was required; the analysis was performed using a laptop PC with a 2-GHz CPU and 2-GB RAM. The computations ran in the background while other applications were active.

This study highlights the variation in PCB thermal response that results from modifying just three of the reflow oven parameters available to the process engineer. Each PCB and associated reflow temperature profile will exhibit unique dynamic thermal behavior. The study confirms that CFD-based analysis is an effective tool for developing a detailed prediction of the PCB’s response in an actual reflow oven.

Conclusion

Today’s optimized CFD tools take the art and science of reflow oven profiling to a new level. Until recently, the interaction between PCBs, components, and the oven itself simply couldn’t be studied effectively as part of the overall design process. All that has changed with the advent of CFD applications designed for use by process engineers rather than thermal specialists. Using CFD models and existing EDA layout data, PCB production personnel can work off-line to predict reflow profiles quickly and accurately.

John Wilson, consulting engineering manager, Mechanical Analysis Division, Mentor Graphics Corporation, may be contacted at johnr_wilson@mentor.com.