Thermal System Design

Transcription

1 Thermal System Design Assignment - 1 Aravind Baskar 3/6/15 ME 5201 Matriculation No: A

2 INTRODUCTION/BACKGROUND Excessive on-chip temperature has become a top concern for high-performance microprocessor design as more devices are integrated on a chip. Thermal constraints are the major driving force for wide adoption of multi/many core architectures. Efficient thermal management is critical for many electronic applications, and is a standard part of the design for components such as power electronics modules, multichip modules (MCM) and systems-on-packages (SOP). Heat sinks serve to transfer the generated heat of an electronic system away from the active and passive electronic components and toward the ambient environment. LITERATURE REVIEW With the increase in heat dissipation from microelectronic devices and the reduction in overall form factors, it became an essential practice to optimize heat-sink designs with least trade-offs in material and manufacturing costs. Thermal Modelling is being done to analyse the scope for improving heat transfer and also for dissipating heat more effectively from the source. Chip-Level Thermal Analysis, Modelling, and Optimization Using Multilayer Green's Function by Baohua Wang describes the thermal analysis of chips. Design and Modeling for 3D ICs and Interposers by Madhavan Swaminathan and Ki Jin Han provides insights on cooling of electronic equipment. PROPOSED SOLUTION The proposed solution involves the design of heat sinks to ensure optimal performance of the microprocessor. The solution is detailed below Heat Sink Heat Source Thermal Interface Material Fig. (a) Overall View of Microprocessor Components Fig. (b) Thermal Components The basic methodology behind the solution involves the following assumptions. Heat losses to the surroundings is negligible. Heat transfer coefficient of the fins is uniform. Material costs is only considered for cost analysis.

3 The cooling system of the microprocessor consists of the following components Heat Source (Microprocessor) Heat Spreader Thermal Interface Material Heat Sink To decide on the lid / heat spreader material, the basic ideology was to select a material that has good thermal characteristics and cost effective. There are basically two options either copper / aluminium. Copper has better thermal characteristics than Aluminium but has certain limitations with its cost and manufacturability. So both options have been considered and results are as follows: Material Thermal Conductivity Cost/Kg Density (Wm -1 K -1 ) ($) (kg/m) Copper Aluminium To decide on the thermal interface material, the basic ideology was to select a material that has good thermal characteristics and cost effective. There are many commercially available TIM, selecting the effective TIM is based on the application. The TIM for the above application is thermal gels (GEL 8010/GEL30 from Parker Chromerics) as the thickness needs to be minimum for this application. Fig. (c) Thermal Interface Materials For the heat sink, the size, material and cost play a pivotal role in selection of the best design. The basic purpose of heat sink is to maximise the heat transfer and also maintain the optimal performance of the electronic equipment. For air cooling minimal no. of parts are used whereas in liquid cooling it involves a significant increase in the no. of parts.

4 Initially the air cooling option has been chosen to test whether the current system performance can be achieved. Natural convection cannot maintain the required temperature and has been rejected from the scope. Forced air convection has been taken as one of the methods. This method involves the selection of proper design of the heat sink and also the optimal fan performance. tf Hf b Wf Fig. (d) Heat sink Model Parameter Case Case Case Length of Fin, Lf 50mm 50mm 50mm Height of Fin, Hf 50mm 65mm 80mm Width of Heat Sink, Wf 80mm 80mm 80mm Base thickness, b 5mm 5mm 5mm Thickness of fin, tf 1.5mm 1.5mm 1.5mm Pitch, s 0.51mm 0.51mm 0.51mm No. of Fins, N Thermal Resistance, Rf C/W C/W C/W METHODOLOGY: Thermal Resistance Model: Thermal resistance is a heat property and a measurement of a temperature difference by which an object or material resists a heat flow. Equivalent thermal resistance model of the problem is as follows: Fig. (e) Thermal Resistance Model

5 DISCUSSIONS: CASE : For qf = 0.93m 3 min -1, Aflow = 80*50*10-6 = 0.004m 2 V = qf/aflow = 3.875ms -1 Re = VL/ν= 3.875*0.05/16*10-6 = For flat plate, Nu = 0.664*(Re) 0.5 (Pr) 0.33 = Nu=hL/k, h=34.65w/m 2 K Af=40*(2*(50*45) +1.5*45+2*(1.5*45))*10-6 =0.188m 2 Ab=0.5128*50*39=0.01m 2 Atot= Af + Ab =0.198m 2 Rf = 1/hAtot=0.152 C/W Rtot=Rf+Rfb+Rb+Rtim Rfb=tfb/KfbAfb = 5*10-3 /(400*80*50*10-6 )=0.003 C/W for Cu and C/W for Al Rb =tb/kbab =5*10-3 /(400*80*50*10-6 )= =0.003 C/W for Cu and C/W for Al Rtim=ttim/ktimAtim =0.05*10-3 /(3*80*50*10-6 )= C/W Rtot=0.162 C/W for Cu & C/W for Al Q=T/Rtot = 500 = Ts-Tamb /Rtot TS = 111 C for Cu < 125 C & C for Al < 125 C Within Limit and acceptable Fin efficiency = tanh(ml)/ml=93% for Cu & 87% for Al Fin Effectiveness = 70% for Cu & 90% for Al Total Volume = 0.175*10-3 m 3 Total Raw Material Cost = Cost per kg * Volume * Density =5.9*0.175*10-3 *8900 =$ 9.18 for Cu & $ 2.7 for Al

6 CASE : For qf = 0.93m 3 min -1, Aflow = 80*50*10-6 = 0.004m 2 V = qf/aflow = 3.875ms -1 Re = VL/ν= 3.875*0.065/16*10-6 = For flat plate, Nu = 0.664*(Re) 0.5 (Pr) 0.33 = Nu=hL/k, h=30.39w/m 2 K Af=40*(2*(50*60) +1.5*50+2*(1.5*60))*10-6 =0.250m 2 Ab=0.5128*50*39=0.01m 2 Atot= Af + Ab =0.252m 2 Rf = 1/hAtot=0.130 C/W Rtot=Rf+Rfb+Rb+Rtim Rfb=tfb/KfbAfb = 5*10-3 /(400*80*50*10-6 )=0.003 C/W for Cu and C/W for Al Rb =tb/kbab =5*10-3 /(400*80*50*10-6 )= =0.003 C/W for Cu and C/W for Al Rtim=ttim/ktimAtim =0.05*10-3 /(3*80*50*10-6 )= C/W Rtot=0.14 C/W for Cu & C/W for Al Q=T/Rtot = 500 = Ts-Tamb /Rtot TS = 100 C for Cu < 125 C & C for Al < 125 C Fin Efficiency = tanh(ml)/(ml) = 89% for Cu & 81% for Al Fin Effectiveness = 83% for Cu & 105% for Al Within Limit and acceptable Total Volume = 0.22*10-3 m 3 Total Raw Material Cost = Cost per kg * Volume * Density =5.9*0.19*10-3 *8900 =$ for Cu and $ 3 for Al CASE : For qf = 0.93m 3 min -1, Aflow = 80*50*10-6 = 0.004m 2 V = qf/aflow = 3.875ms -1 Re = VL/ν= 3.875*0.080/16*10-6 = 19375

7 For flat plate, Nu = 0.664*(Re) 0.5 (Pr) 0.33 = Nu=hL/k, h=27.4w/m 2 K Af=40*(2*(50*75) +1.5*50+2*(1.5*75))*10-6 =0.313m 2 Ab=0.5128*50*39=0.01m 2 Atot= Af + Ab =0.32m 2 Rf = 1/hAtot=0.11 C/W Rtot=Rf+Rfb+Rb+Rtim Rfb=tfb/KfbAfb = 5*10-3 /(400*80*50*10-6 )=0.003 C/W for Cu and C/W for Al Rb =tb/kbab =5*10-3 /(400*80*50*10-6 )= =0.003 C/W for Cu and C/W for Al Rtim=ttim/ktimAtim =0.05*10-3 /(3*80*50*10-6 )= C/W Rtot=0.12 C/W for Cu & C/W for Al Q=T/Rtot = 500 = Ts-Tamb /Rtot TS = 90 C for Cu < 125 C & C for Al < 125 C Within Limit and acceptable Fin Efficiency = tanh(ml)/(ml) = 86% for Cu & 76% for Al Fin Effectiveness = 94% for Cu & 116% for Al Total Volume = 0.25*10-3 m 3 Total Raw Material Cost = Cost per kg * Volume * Density =5.9*0.25*10-3 *8900 =$ for Cu and $ 4 for Al FAN SELECTION: 3115FS (80 X 38L) is chosen as the fan for providing the required air flow and the power consumption is around 10W. CONCLUSION: Based on the of design cases it is found that the design with 65 mm fin height is the best option. Aluminium is chosen as the material as it has good thermal properties and better cost effectiveness. The use of heat pipe will further enhance the heat transfer and is proposed as the alternative in case of higher heat dissipation requirements.

8 REFERENCES: 1. Advanced Materials for Thermal Management of Electronic Packaging by Xingcun Colin Tong ISBN Fundamentals of Microsystem Packaging by Prof. Rao R. Tummala Georgia Institute of Technology Tata McGraw-Hill E-resource Manufacturer of fans Manufacturer of thermal interface materials Chip-Level Thermal Analysis, Modelling, and Optimization Using Multilayer Green's Function by Baohua Wang. 7. Design and Modeling for 3D ICs and Interposers by Madhavan Swaminathan and Ki Jin Han 8. Energy reduction and performance maximization through improved cooling by David Copeland Oracle 9. Analytical Forced Convection Modeling of Plate Fin Heat Sinks by P. Teertstra, M.M. Yovanovich and J.R. Culham 10. OPTIMUM DESIGN AND SELECTION OF HEAT SINKS by Seri Lee of Aavid Engineering Inc. 11. CPU Thermal Management from AMD 12.