Jim Wilson
My experience in electronics cooling has taught me that it is often a fairy tale that thermal engineers fully understand requirements related to their designs and analyses, or that they fully understand the implications of subjecting thermal requirements on suppliers or coworkers. A rush to get a quick answer or to use the latest feature in an analysis software code often means jumping right into the analysis modeling tool and neglects the critical first steps related to refining the problem statement. The fact of this column is that taking the effort to comprehend thermal design and analysis requirements will minimize wasted effort and reduce the time required to optimize a design.
Near the start of thermal design or analysis task, it is important to understand why you are doing the work (often directly related to the phase of the product development cycle you are supporting) and what criteria will be used to assess when the task is complete. A few typical reasons for performing thermal analysis are:
-Assure that components will be within manufactures rated temperature limits
-Contractually obligated to perform the analysis
-Provide guidance to a design team for optimization considering thermal effects
-Develop performance metrics for electronics that are temperature sensitive
-Prevent thermally related failures in the products use
-Support reliability estimates for the product
-Develop thermal requirements for a supplier
-Provide thermal characterization information to a customer
-Develop the thermal architecture of a system [1]
In the case of design optimization, the completion criteria are often directly related to the available budget and schedule. For this type of work, the successful thermal engineers learn to tailor their effort to the available time and money. Experience is very beneficial, but even in its absence thermal engineers can perform analysis that is commensurate with the resolution of the design data. Hand calculations may be all that is needed to properly refine requirements, assess sensitivities, and impact the product design in the initial stages. Making an informed decision on the appropriate level of detail for an analysis can be difficult. Determining the resolution of boundary conditions and other thermal parameters requires effort and some knowledge of the operating environment.
Once you have established why you are doing the work and when you will be done, it is appropriate to assess the design information available. A few common areas where uncertainty is often present are:
– Dissipated power estimates – The values at an early stage may only represent a rough estimate and may contain excessive margin. A common thought process is to see if there are any thermal issues with a worst case estimate. If the subsequent analysis indicates acceptable performance then there is not a pressing need to revisit the thermal design, but the excessive margin is still present. In fairness to complex systems, actual thermal dissipation is sometimes not known with certainty until late in the design or testing phase.
– Material properties – Just because a product data sheet lists a value does not mean this value is accurate. Items like thermal conductivity are difficult to measure [2], and thermal interface material performance often depends on how it is used. Your design probably does not replicate the thermal characterization environment used to develop the advertised data.
– Accurate understanding of thermal margin in supplier delivered parts and requirements. It is often difficult to find supporting information that provides the basis for the cooling requirements (flowrates, mounting temperatures, etc) for some parts. For example, if a design goal is minimizing margin to enable a low mass design, using overly conservative vendor supplied parameters is not optimum.
– An understanding of how the parts are built and the associated tolerances. Parts are rarely built as perfectly as the design shows up in a solid model and the subsequent thermal model. Attachment layers may have voiding and the geometry is constrained by the manufacturing processes.
– Accuracy of the boundary conditions. In some cases the external boundary is reasonably well understood (a specified operating environment) but in some cases the thermal interface to another part of the system is described. Understanding and developing thermal requirements at this level usually takes time, especially if an understanding thermal margin is required. The reader is also cautioned against indiscriminate use of a fixed temperature boundary condition. [3]
After developing an understanding of what is needed to perform the thermal tasks and assessing the design information, the requirements for selecting appropriate software tools are usually evident. For example, the resolution of the modeling tool selected should be proportional to the accuracy of the design parameters and assumptions.
If the tasks include developing thermal requirements, giving consideration to the lists above at least allows a documentation of how they were developed and provides an initial set of information for sensitivity information.
While most thermal engineers are capable of solving a well posed problem, the fact that the problem is well posed should motivate one to make sure that the underlying assumptions required to define the problem in this manner are correct and applicable. Effective electronics cooling design work requires understanding how the system responds to its thermal environment and how the system changes as its thermal characteristics change. Additionally, relevant thermal requirements imposed on the system or derived from the system require attention at all phases of the design process. This helps avoid having a detailed design and analysis against requirements that are no longer completely relevant. A continuous collaborative exchange with other engineers is necessary for consistency between thermal requirements and the resulting design.
References
[1] Johnson, S. and Holt, B., “Thermal Facts and Fairy Tales: Evolving the Role of the Thermal Engineer from Analyst to Architect”, ElectronicsCooling, December 2013.
[2] Wilson, J., “Technical Data Summary”, ElectronicsCooling, August 2009.
[3] Wilson, J., “Thermal Facts and Fairy Tales: Fixed Temperature and Infinite Heatsinking”, ElectronicsCooling, December 2010.
