The Rationale and Requirements for Thermal Management and Analysis

Objectives

Background

Work Accomplished

The first step wasthe  determination of thermally influenced performance metrics. This allowed the identification of the criteria for maximum uncertainty in temperature and spatial and temporal temperature gradients, for acceptable variations in life or performance predictions.

The next step was to identify all significant sources of uncertainty, including transient variations in environmental thermal conditions, uncertainties in thermophysical properties and contact resistances.

Through system level computational fluid dynamics type thermal analysis, the propagation of these uncertainties to those at the component level was evaluated. It was found that when the predicted changes in temperatures or gradients at the components are comparable to or larger than their respective allowable uncertainties, a better characterization of the responsible parameters is warranted. For cases where these parametric uncertainties have only a small impact on the component level temperatures and gradients, the overall uncertainty will be determined by the inaccuracies in the thermal simulations.

An initial background survey was performed on the effects of thermal conditions on the performance of microelectronic components. The study includes effects of both high and low temperature on the electrical parameters of bipolar junction transistors and metal semiconductor field effect transistors. Some of the important axide are listed below.

• There exist semiconductor device technologies which can perform satisfactorily beyond the conventional temperature limits prescribed by different civilian and military standards.

 
• Parameter shifts in electronics devices occur for both high and low temperatures. Some parameter shifts and degradation mechanisms are significant only at low temperatures.

 
• The mechanisms for electrical parameter shifts vary significantly with temperature ranges.

 
• Most of the parameter shifts in microelectronic devices are temporary and reversible. With reduction (or increase) in temperature, electrical parameters like current gain, leakage current etc. return to normal level.

 
• Along with progressive changes, temperature/performance relationships also exhibit threshold behavior.

Reports on temperature dependent performance parameter variation are available. A background study of the available standards and guidelines for avionics thermal conditions was performed, along with comparison of the standards, with available thermal data from commercial aircrafts. These standards include IPC, ARINC 600 and DOD-STD-1788. It is observed that the actual thermal profiles of avionics equipments often vary significantly from these standards.

Possible sources of uncertainties in the thermal analysis of avionics equipments have been identified. These sources include the following:

• Environmental profile: This includes uncertainties in ambient conditions e.g. temperature, pressure, and ambient fluid.

 
• Operating conditions: These uncertainties arise from the variability in the actual functional profile of any electronic device. The actual and average power output of devices, clock rate of device and other factors which determine the amount of heat generated vary with time.

 
• Thermophysical properties: These uncertainties arise from the lack of detailed information about the materials used, the variations of thermophysical properties with ambient conditions, chance variations in properties.
These different types of uncertainties were divided into two groups. The uncertainties which are integral parts of the operation of a system are called intrinsic. The other types of uncertainties which are possible to characterize and control are called extrinsic.

Based on the review of a prior thermal study of a forced convection cooled aircraft Transformer/Rectifier module, a new set of range of parameters is determined.

These parameter sets were chosen for a complete analysis of the effects of various types of uncertainties. However, the complete analysis at the level was found to be computationally expensive.

Two models of avionics Transformer/Rectifier units of ELDEC corporation was obtained from Flomerics, Inc.

Initial parametric studies were performed on these models to test the effects of different variations. In particular, the effects of the following are tested.

• Material properties : thermal conductivity, contact resistances
• Thermal analysis assumptions : values of external heat transfer coefficients, radiation view factors
• Numerical modeling : grid sizes, convergence criteria.
Based on the results of these preliminary studies, the previous study parameters were modified. A new set of parameters are developed for 2^k-n fractional factorial experimentation. ^*These sets were continually revised based on the results obtained from the initial blocks of study.

Parametric computer simulations of the TR unit using FLOTHERM software were performed in blocks. Results were analyzed and statistical effects of different factors were determined after each block of simulations. The relative effects of individual and combined effects of parametric variations were studied

FLOTHERM model of a natural convection cooled laptop computer was developed using published data on Tape Carrier Packaged Pentium computer. A less detailed experimental design matrix was used for thermal simulation of the laptop computer model.

For both the models, the individual and cumulative effects of uncertainties were mapped into representative performance parameter variations for bipolar and MOS devices.

The results were used to develop sets of guidelines for the thermal design process and the accuracy needs for thermal analysis.
 

Presentation for Members Only