Search for: All records

Semiconductor thermal management is becoming a bottleneck challenge that restricts further development in the electronics industry. Compromising processor thermal requirements will impact the processor performance and reliability. Heat sinks are designed to increase the available surface area of an electronic component and allow for more heat to be easily dissipated. As a result, the thermal characterization of the heat sinks plays a critical role in electronics thermal management. In this study, a flexible experimental apparatus is designed, built, and assembled to characterize and test various electronics components in different aerodynamics and thermal conditions. This novel experimental apparatus allows for controlled characterization of the various heat sinks with different heights as well as realistic scenarios with air bypass at server level. Moreover, a general guideline on precise experimental procedure to characterize air cooled heat sinks is developed. The results show that introduced method reduces the experimental error by 26%. 

Abstract The practice of commissioning data centers (DCs) is necessary to confirm the compliance of the cooling system to the information technology equipment (ITE) load (design capacity). In a typical DC, there are different types of ITE, each having its physical characteristics. Considering these geometrical and internal differences among ITE, it is infeasible to use the actual ITE as a self-simulator. Hence, a separate device called load bank is employed for that purpose. Load banks create a dummy thermal load to analyze, test, and stress the cooling infrastructure. Available commercial load banks do not accurately replicate a server's airflow patterns and transient heat signatures which are governed by thermal inertia, energy dissipation, flow resistance, and fan system behavior. In this study, a novel prototype of the server called server simulator was designed and built with different components to be used as a server mockup. The server simulator accurately captured air resistance, heat dissipation, and the functionality of actual server behavior. Experimental data showed up to 93% improvement in ITE passive and active flow curves using the designed server simulator compared to the commercial load bank. Furthermore, the experimental results demonstrated a below 5% discrepancy on the critical back pressure and free delivery point between the actual ITE and the designed server simulator. In addition, experimental data indicated that the developed server simulator improved the actual ITE thermal mass by 27% compared to the commercial load bank.