THERMAL DESING AND ANALYSIS OF HIGH ELECTRONIC BOARDS J. Collado Spring, 2000 CHT Page 2 of 18 TABLE OF CONTENT 1. INTRODUCTION .....................................................................................................................................3 2. THERMAL REQUIREMENTS .................................................................................................................3 3. TEMPERATURE IMPACT ON ELECTRONIC BOARDS .......................................................................4 3.1 4. TEMPERATURE EFFECT ON COMPONENT FAILURE ............................................................................................. 4 THERMAL DESIGN CONSIDERATIONS ...............................................................................................4 4.1 FUNDAMENTAL CONCEPTS ................................................................................................................................. 4 4.2 THERMAL RESISTANCE TO CONDUCTION ........................................................................................................... 5 4.2.1 Component Internal Thermal Resistance ................................................................................................. 5 4.2.2 Component to Mounting Surface Heat Transfer ...................................................................................... 5 4.2.3 PWB to Heat Sink Heat Transfer ............................................................................................................. 6 5. CASE STUDY “THERMAL DESIGN AND ANALYSIS OF A TYPICAL HIGH POWER ELECTRONIC BOARD” ..........................................................................................................................................................8 5.1 5.2 ANALYSIS INPUTS............................................................................................................................................... 8 ANALYSIS METHOD ................................................................................................................................... 10 6. RESULTS ..............................................................................................................................................11 7. CONCLUSIONS ....................................................................................................................................11 8. REFERENCE.........................................................................................................................................11 APPENDIX .....................................................................................................................................................17 Page 3 of 18 1. Introduction The reliability of an electronic device depends primarily on its operating temperature. For maximum reliability, devices must operate at temperature below their maximum rated value. Heat affects the electrical characteristics of semiconductors, the chemical properties of metals used in semiconductors devices and integrated circuits production and the mechanical properties of printed wiring board (PWB) materials. Heat affects these materials in a form detrimental to circuit operation. For example, gold and aluminum can form inter-metallic compounds at high temperatures, thus weakening lead bonds within the component. High temperatures also increase the activity rate of corrosive elements in solder pastes. The majority of electronic components, passive and active, as well as the traces on the printed wiring boards generate heat when they are operated. This heat raises the temperature of the component in which it originates, the surface upon which the component is mounted, and other devices close to it. The thermal management of high power PWBs is of increasing importance as the power density of components and circuits continues to rise. The situation is complicated by the severe environmental requirements, limited weight tolerance, and the use of boards with multiple sheet of copper embedded in the electrically insulating boards. Given the high thermal conductivity of copper, the thermal properties are drastically altered by the embedded layers, possibly introducing overall anisotropy. As a result of these conditions, thermal analysis must be an integrated step with component placement, routing, mechanical analysis, and reliability prediction. The use of thermal analysis software helps to identify thermal problems during the early stages of design. It also provides various options to resolve possible thermal problems during the design process. This process reduces costly corrections and reduces design cycle time by predicting the thermal performance before the first prototype is developed. A study case selected to illustrate the modeling techniques in the thermal design of a high power PWBs is a power supply board of an electronic engine control. The engine controller is subjected to severe environmental constraints. External unit cooling is provided mainly by natural convection and, to a smaller extent, to thermal radiation to the engine bay. The ambient air temperature can range from -55C to 75C (118C during unit qualification testing). Inside the engine controller are 6 boards, 2 power supply boards, 2 Processor boards, and 2 Input boards. Each board is bonded to an aluminum heat sink, and the heat sink are screwed to the chassis to maximize conduction cooling of the boards. The total power dissipation of a single power supply board is 25.8 Watts. Thermal math modeling of the design was accomplished using SINDA. SINDA (Systems Improved Numerical Differencing Analyzer) is a finite difference thermal analysis program. 2. Thermal Requirements Thermal management of electronic boards, especially high power electronic boards, begins at the procuring activity with the derivation of the thermal environments and performance criteria. This effort culminates in the baseline requirements and other essential thermal design information to be included with the equipment specifications that will be used in defining the thermal management needs of the unit and boards. The requirements should be based on availability and life cycle cost objectives. The following items should typically be included in the baseline thermal requirements of each equipment specification: 1. Maximum junction temperatures and dereating criteria 2. Thermal environments in which the equipment is required to operate: a. Ambient temperature b. Thermal radiation environment c. Solar heat loads 3. Design constraints: a. Noise and vibration limits b. Size limit c. Weight limits Page 4 of 18 3. Temperature Impact on Electronic Boards The temperature at which an electronic device operates is one of the primary determinants of its reliability. Heat dissipated in the device during operation escapes through a path consisting of one or more series of thermal impedances terminating in the surrounding air. The presence of these nonzero thermal impedances cause the temperature of the device to rise above that of the heat sink. 3.1 Temperature Effect on Component Failure High temperature is an enemy of most electronic PWBs and components. It causes slow progressive deterioration that eventually results in catastrophic failure. The time to failure of each part is a statistical function of its stress level and the complex interrelationship of thermal history and chemical structure. The failure of individual parts leads to unit failure. Many failure modes involve changes in the chemical composition or physical structure of the board and the electronic devices. The rate of these chemical reactions or physical changes increases with increasing temperature. Thus, operation at high temperature increases the failure rate, or decreases the mean time between failures, by accelerating the rate at which these undesired changes occur. Conversely, failure is also accelerated at very low temperatures. Thermal stresses resulting from differential thermal contraction of dissimilar material, embrittlement of materials, and shorting and corrosion caused by condensed water trapped within the encapsulation are typical failures induced by very low temperatures. The table below summarizes some of the effects and the possible failure mechanism induced by very high and very low operating temperatures. Condition Hot Temperature Cold Temperature Thermal Cycling Thermal Shock Effects Insulation Deterioration Oxidation Over Expansion Softening/Hardening Chemical Changes Contraction Viscosity Increase Embrittlement Ice Formation Temperature Gradients Expansion/Contraction High Temperature Gradient Possible Failure Modes Shorting Rust Physical Damage Physical Breakdown Dielectric Loss Wear, Structural Failure Loss of Lubricity Structural Failure, Cracked Parts Structural Failure Repeated stress variation causes mechanical failure of part, solder joints, and board delamination Mechanical failure, Cracks, Rupture 4. THERMAL DESIGN CONSIDERATIONS 4.1 Fundamental Concepts Thermal Resistance Energy, in the form of heat, flows naturally from a hotter region to a cooler one. The greater the temperature difference (T), the greater the heat-flow rate (Q) will be. This relation can be written as: T = Q Where the proportionality factor () is called the thermal resistance. If T is expressed in C and Q is in Watts (W), then the units of are C/W. Thermal Dissipation Electronic equipment requires electrical power to operate. Because the parts are not perfectly efficient, much of the power is converted to heat, which causes the temperature of the equipment to rise. The temperature will continue to rise unless the heat can find a path by which to leave the equipment. The purpose of the cooling system is to provide an effective heat path. Therefore, the most important heat-flow rate is usually the thermal dissipation. Sinks In the cooling of equipment, heat must be transferred from the part in which the heat is generated to an ultimate sink. This ultimate sink is a heat reservoir, such as a body of land, ambient air, the sky, or the ocean. It is the final destination of the heat. Such a heat reservoir can be considered to be infinite. That is, it is so large that the heat transferred from the equipment does not produce a significant increase in the reservoir’s temperature. Page 5 of 18 The Cooling Equation The essence of the problem of cooling electronic equipment can now be described. For a component having a dissipation (Q), a sink temperature (T sink) (which could be the ultimate sink, local sink, or heat sink temperature), and package such that the thermal resistance between the part and the sink is , the operating temperature (Tcomp) is given by: Tcomp = Tsink + (Q) 4.2 Thermal Resistance to Conduction 4.2.1 Component Internal Thermal Resistance The temperature at which a semiconductor device operates is often referred to as the “Junction Temperature”. Heat dissipated in the device during operation escapes through a path consisting of one or more series thermal resistances terminating in the surrounding ambient, refer to Figure below. Junction-Ambient Case-Ambient DIE CASE Junction Junction-Case The presence of these nonzero thermal resistances causes the temperature of the device to rise above that of the ambient. Each of the components of the overall thermal resistance causes a rise in temperature, which is linearly dependent on the power dissipated in the device. The value for each thermal resistance represents the amount of temperature rise across the resistance as a function of the power dissipation. Usually, is given a subscript indicating the two point between which the resistance is measured. Thus the junction temperature of an operating device is given by: TJunction T T Amb ConductionSink Pd case Sink junctioncase Pd case Amb 1 1 case Amb case Sink Where: Tjunction = Component Junction Temperature Tamb = Ambient Temperature Tconduction Sink = Temperature of the Mounting Surface Pd = Power Dissipation junction-case = Junction to Case Thermal Resistance case-Amb = Case to Ambient Thermal Resistance case-sink = Case to Conduction Sink Thermal Resistance The internal thermal resistance of a given device (junction-case and case-Amb) are dependent on several factors, the package type, the package material, the way the die is attached to the package, etc. Normally these thermal resistances are provided by the component manufacturer in the component data sheets. The case-sink is dependent on the way the component is mounted to its mounting surface. This thermal resistance is normally calculated by the thermal design engineer based on the actual mounting configuration used. 4.2.2 Component to Mounting Surface Heat Transfer Heat is generated within the active components in the unit. This power must somehow be transferred to the component mounting surface with minimal temperature rise. The problem can be compounded by extremely high power densities, 0.4 W on a part with a 0.05” X 0.2” case footprint for 40 W/in 2 (typical of diodes). This is an order of magnitude higher than a 208 Pins Quad Flat Pack with a power dissipation equal to 4 W and a case footprint equal to 1” X 1” for 4 W/in2 power density. Page 6 of 18 The heat path from the component to its mounting surface is commonly accounted for by hand calculations for each active component. Many components have similar mounting techniques, so the same hand calculation may apply to numerous parts thus does not need to be redone very often. Due in part to the extremely tight reliability and weight requirements of aerospace electronic equipment, there are three basic design features available to minimize the thermal resistance from the component to its mounting surface. All three involve mainly conduction as the surface area available for radiation directly off the component is negligible, and natural convection cooling is only a secondary cooling path. 4.2.2.1 Component Leads Conduction out of the component leads is always the first path considered. Typical lead resistances vary from 1000C/W per lead for Kovar leads on microcircuits to 12C/W per lead for copper leads on diodes. The lead thermal resistance is given by: Lead = L / (KA) Where, L = Lead length (measured from the edge of the component case to the soldered pad on the PWB) K = Thermal conductivity of lead material A = Cross sectional area of the lead 4.2.2.2 Bonded Joints For any component with significant thermal dissipation, a conductive bonding material could be used to fillet between the component case and its mounting surface. Either electrically conductive (soft solder or metal filled cements) or non-conductive (epoxy or ceramic filled cements) bonding materials may be used, depending on whether electrical isolation is required. The selection of the bonding agent is normally based on component shape, bonding technique, structural (bond strength) and removal requirements, as well as thermal conductivity. The mounting surface is generally a conductive copper plane or aluminum heat sink bonded to the PWB to then spread and carry the heat away from the component. The thermal resistance across rectangular or circular bond interfaces is given by: Lead = L / (KA) Where, L = Bond pad on the PWB) K = Thermal conductivity of bonding material A = Cross sectional area of the bonded interface The thermal resistance for heat flow by conduction between isothermal cylindrical component bonded to the mounting surface is given by: 1 4 KW ( a 2 1)1 / 2 2 2 2 1/ 2 ( a 1)1 / 2 tan( 1 ) 1 ( a 1) tan( 2 ) 1 2 tan tan 2 KW (2 1 ) a 1 a 1 Where, K = Thermal conductivity of bonding material W = Length of the bonded section (or lead) = Fillet angle a = (2x/D + 1) = (x/r + 1) x = Gap between bonded body and mounting surface D = Diameter of cylindrical body r = Radius of cylindrical body 4.2.3 PWB to Heat Sink Heat Transfer Once in the component mounting surface, the heat must somehow be transferred to the heat sink or metal chassis. The PWB is a very poor lateral conductor, thus extreme care must be used in insuring proper Page 7 of 18 paths are available. This heat transfer is generally accounted for by developing a detailed thermal math model of all possible paths within the PWB. The thermal math model must correctly account for the component power density by outlining all significant dissipaters. The most common error resulting in optimistic temperature predictions results from nodes that are too large. This result in much lower power densities being applied to the PWB surface thus missing potential local hot spots. Nodes that are too large also miss the constriction resistance that is present at local hot spots. As a general rule, all significantly dissipating components (over 0.5 Watts) must also be outlined with node boundaries to insure proper power densities, and all mounting points must also be outlined with node boundaries. Models of this type require complex model generating software, which are readily available from commercial software vendors. There are three basic paths to transfer heat from the PWB to the heat sink or chassis. 4.2.3.1 Lateral Conduction to Mounting Points (X-Y Direction) The PWB is a very poor lateral conductor being mostly epoxy (K = 0.01 W/inC) and typically 0.06” to 0.09” thick. Therefore continuos copper planes embedded in the PWB or aluminum heat sinks bonded to the PWB surface are used to enhance this lateral conduction. Copper planes within the PWB are generally 0.0014” to 0.0028” thick per layer and can be used to conduct small quantities of heat to nearby mounting points only if they are continuous. Any discontinuity in the copper adds an extremely large series resistance to the heat flow path. Even a 0.05” gap in the copper plane (sometimes required for electrical isolation) between the component and the PWB mounting point will result in a significant temperature increase. If the discontinuity occurs around the component case, it is doubly bad as it not only prevents conduction to the mounting point, but also prevents any spreading of the heat for radiation and convection heat transfer. These discontinuities must be properly accounted for in the thermal model. Assuming full copper planes around all components and mounting points can have a catastrophic effect as significant series thermal resistances will be missed resulting in optimistic temperature predictions. 4.2.3.2 Conduction Through the PWB (Z-Direction) Even if the heat gets to a local mounting point, the PWB can act like an insulator in preventing heat from being conducted from a top copper layer through the board into the chassis boss. Adding thermal vias around the screw hole is commonly used to reduce that Z-direction resistance at the mounting boss. Thermal via holes are drilled holes, usually 0.016” in diameter of bigger, with 0.001” thick copper plating on the walls. These thermal vias greatly reduce the Z-direction conduction resistance in the PWB epoxy by adding approximately 150C/W per via parallel resistance. For very high power boards or for large, boards in general an aluminum heat sink is usually bonded to the back plane of the board (side opposite to the components) in order to improve the board structural stiffness or to greatly improve the lateral heat transfer capacity of the board. In this case, for components with extremely high thermal dissipations (over 1 Watt), or component with very high power densities, the power can not be efficiently conducted laterally in the copper planes even if they are continuous and must somehow be given a Z-direction path directly to the metal heat sink. The Figure below shows a typical via hole pattern for a high power semiconductor. Copper Pad Component Could be Soldered or Bonded to this Pad Via Hole 0.016” Drilled Hole, With 0.001” Copper (Cu) Plating, Connecting Cu Pad on Top of The PWB With Similar Cu Pad on the Back of The PWB (Closest to Heat Sink) Section A-A Component Footprint Typically no Thermal Vias Are Placed Directly Under The Component Section A-A The approximate thermal resistance across a single thermal via is given by: First we need to consider the lateral thermal resistance within the Copper pad from the heat source to the thermal via for the top copper pad as well as the lateral thermal resistance to heat spreading from the thermal via for the bottom copper pad. This is given by: 1 Where: r2 = Via thermal radius (*) ln( r2 r1 ) 2LK Page 8 of 18 r1 = Thermal Via Outer Radius L = Thickness of Copper Layer K = Thermal Conductivity of Copper (*) The value of r2 can be approximated by considering the total copper pad area (A) and dividing this area by the total number of thermal vias within the copper pad, then if we consider this area equal to the area of a circle we can determine the radius. This approximation assumed that heat will travel an equal distance to each thermal vias. 4 (r2 ) 2 A Total # ofthermalvias Now we need to consider the thermal resistance across the PWB. This is given by: L 2 K 4 2 (rout rin2 ) Where: L = Thickness of the PWB K = Thermal Conductivity of Copper Rout = Outer Radius of the Thermal Via Rin = Inner Radius of the Thermal Via Finally the total thermal resistance across a single thermal via is given by: Total = (21) + 2 But normally a number of thermal vias are used to greatly reduce the thermal resistance across the board, so the effective thermal resistance for a given number of thermal vias is given by: Effective 1 # vias Total 1 The final path that need to be identify when considering the thermal design of the PWB are the thermal path between the heat sink and the chassis and the thermal path between the chassis and the ultimate heat sink (ambient air, cooling liquid, heat exchanger, etc.). For the purpose of this paper these two elements will not be discussed. 5. CASE STUDY “Thermal Design and Analysis of a Typical High Power Electronic Board” 5.1 Analysis Inputs Physical Design The EEC150 (Electronic Engine Control) consists of the chassis, 2 Power Supply boards, 2 Input boards, 2 Processor boards, 7 interconnect boards, 6 pressure sensors, and the PB sensor heater (high pressure). Internally the unit is broken down into two channels, A and B, with each channel consisting of a power supply board, an input board, and a processor board. The two rectangular sections of the chassis are geometrically identical to each other. Each section is machined from a large block of alloy A356 (AMS4218) aluminum. The base and side walls are 0.110” thick with overall unit dimensions being 18.38” X 14.35” X 5.05” for the fully assembled. Each section of the housing also provides an internal network of ribs 0.1” thick and 0.2” high. This rib network provides additional support for the board mounting posts. There are two different types of post, those only supporting the Power Supply board and the power supply heat sink, 0.317” high, and those supporting the Power Supply board, power supply heat sink and the Processor/Input module (Processor board, Input board, and common heat sink), 0.417” high with 0.1” high aluminum spacers (.517” total height). The outer diameter for all the posts is 0.5” and the inner diameter is 0.17”. The two rectangular sections of the housing are held together by means of 18 #10-32 screws. All external and internal surfaces are hard Page 9 of 18 anodized. The unit mounts to the outer section of the engine case across four vibration isolators. These vibration isolators also conductively isolate the unit from the engine case. Each of the 6 PWB’s integrating the unit, 2 Power Supply boards, 2 Input boards, and the 2 Processor boards, consist of multilayers of copper and Type GFG epoxy sheets held together with layers of Pre-Preg material (adhesive). There are 12 copper planes integrating each of the Power Supply boards for a nominal board thickness of 0.074”, and 10 copper planes for each of the Input and Processor boards with a nominal board thickness of 0.064”. In all cases, the copper layers in the PWB’s consist of 1 oz/ft 2 of copper per plane, this yields an effective thickness of 0.0014” per layer. Figures 2 shows the component layout for the Power Supply, which is the only board that we will consider for detailed analysis on this report. Each Power supply PWB is bonded to an alloy AMS4027 aluminum heat sink 0.08” thick. The bonding material used in the bonding process is 3M-Y9469 with a nominal bond thickness of 0.005”. The Power Supply heat sinks are attached one to each half of the housing by mean of 26 #6-32 screws. Component Mounting The method used to mount the different electronic components on the boards is of paramount importance in determining their operating temperatures. In this thermal design all the electronic components are mounted directly to the PWB’s. As previously mentioned, the critical factors to determine the mounting method of an electronic component are its physical size, power dissipation, and lead material (where applicable). The following is a description of the mounting methods used for all the components in the Power Supply boards. Appendix shows a component specific description of the mounting methods. 1. All surface mounted components are soldered directly to copper pads on the top surface of the PWB’s. 2. High power surface mounted transitors, diodes, and resistors (in excess of .5 Watts) have been soldered to enlarged copper pads with a large number of thermal vias connecting top copper pads to identical copper pads on the opposite side of the PWB near the heat sink. Thermal vias reduce the thermal resistance across the PWB while the enlarged copper pads help to minimize the thermal resistance across the PWB by increasing the effective heat transfer area. Refer to Figure 3 for a typical mounting configuration of a high power surface mounted component. 3. Medium power surface mounted resistors (less than 0.4 Watts) are soldered to enlarge copper pads on the top surface of the PWB’s. 4. None of the Flat Pack, PQFP, CQFP, SOIC, and DIP type components in the unit will be bonded. The thermal dissipation within most of these types of components is low to moderate. The components in this group with moderate power dissipation levels are large in size and their lead material is copper. These two factors combine provide adequate heat transfer capability for these types of components. Thermal Configuration During the steady state operation of the unit, the heat dissipated in operating electronic components is conducted out through the case and leads (where applicable), across any solder or bonding (where applicable) material present to the mounting surface. Once the heat enters the PWB’s, it is spread laterally and conducted downward to the aluminum heat sink. From the heat sink, heat is conducted laterally, across the screw interfaces, and to the mounting points. In this design the mounting points are the metal posts coming off the housing. The contact resistance value used for the #6-32 screws connecting the Power Supply heat sinks to the housing is 1.73 C/W. These values include the resistance across the dry bolted interface as well as the resistance through the post. The later case also includes the thermal resistance across the double interface of the spacers. The determination of this contact resistance is based on empirical charts and it will not be discussed in this paper. During steady state operation there is thermal radiation interchange between the PWB’s that are facing each other and between the PWB’s facing the housing and the housing, as well as natural convection within the unit. Once the heat is in the chassis shelf, it conducts outwardly to the unit walls. Heat is also transferred between the two halves of the housing across the bolted interface. The contact resistance term used for #10-32 screws was 1.6 C/W, again how this value was obtained from empirical charts and it will not be discussed in this paper. The heat is dissipated to the aircraft environment in two ways. First, it is transferred by means of natural convection to the Nacelle environment. The ambient air heat transfer Page 10 of 18 coefficient for the operating unit is 4 BTUHrFT2F. The value of the external air heat transfer coefficient was provided by the engine manufacturer. Secondly, heat is also radiated to the Engine case as well as to the Nacelle wall. A vast majority of the heat within this unit is transferred to the environment by convection rather than radiation. Conduction to the engine case is negligible when compared to convection and radiation off the unit, therefore it was ignored for the purpose of analysis. Operating Conditions The method of analysis considered steady state operation for typical power dissipation (75.4 Watts). The analysis also assumed that both channels, A and B, were operating simultaneously during unit operation. These power levels were used to determine case, and junction (where applicable) temperatures for each individual component within the power supply boards. Environments The only thermal environment analyzed corresponds to the EEC150 qualification test environment. Under this operating condition the Nacelle wall, Engine Wall, and Nacelle air temperature were taken to be 118C. The Nacelle air temperature was used as a convection heat sink. The Nacelle wall temperature and the Engine wall temperature were used as radiation sinks. Material Properties Air Aluminum AMS4218 Aluminum AMS4027 Bond, type 3MY9469 Copper C11000 Copper C19400 GFG Solder Mask Hard Anodized Thermal Conductivity WinC 7.774E-4 39.20E-1 41.20E-1 4.400E-3 97.00E-1 65.94E-1 1.311E-2 4.400E-3 N/A Optical Property Emissivity N/A 0.10 0.10 N/A 0.10 0.10 0.68 N/A 0.60 Assumptions 1) The Nacelle air temperature, the Nacelle wall temperature, and the Engine case temperature are isothermal and infinite heat sinks. 2) The heat transfer coefficient for the internal air (inside the unit) is 0.0037 Win2C. 3) A view factor of 1.0 between the PWBs facing each other and the PWBs facing the housing. 4) Conduction across electrical connectors is negligible. 5) The Nacelle wall and the Engine case surface emissivities are 0.5. 5.2 ANALYSIS METHOD A detailed unit level mathematical thermal model was developed for the EEC, PWBs, and heat sinks to represent all significant thermal exchange within the elements integrating the unit as well as the thermal exchange to the aircraft engine environment. Although individual models were created to represent the different PWBs and heat sinks integrating the unit, all of them were connected to form a single unit thermal model. The model accounts for conduction from the PWBs to the heat sinks, and from the heat sinks to the chassis. The model also accounts for PWB to PWB, PWB to chassis, and chassis to environment thermal radiation. Convection from the chassis to the environment as well as internally was also accounted for in the model. The power distribution within the PWBs was such as to represent the actual power density within the unit. A detailed thermal model of a Power Supply board was also created. The model account for all significant thermal path within the board, including the interlayer connections, all significant copper pads and the thermal vias connecting top and bottom copper pads. The mounting post, internal air, and housing and board average temperatures predicted by the unit level thermal model were used as conduction, convection, and radiation heat sinks respectively. The nodes and connectors for the unit level thermal model as well as the boundaries were created numerically using SINDA/G. This software is a finite difference thermal analysis program. SINDA/G utilized Page 11 of 18 a steady state routine STDSTL to arrive at a solution. The node and connectors for the detailed Power Supply thermal model was created geometrically using SINDA/3D. SINDA/3D is a pre and post processor for SINDA/G. The output from the pre-processor is a SINDA/G input file. SINDA/G is used to arrive at a solution, and the output from SINDA/G is used by the post-processor to create color contour plots of the predicted temperatures among other things. 6. RESULTS The results from the thermal analysis are various and can be presented in various forms. Figure 4 is one example of how the results could be presented. Figure 1 shows the temperature delta increase across a PWB 0.08” thick for a particular type of component (TO-220) as a function of power dissipation and for several mounting configurations. The main purpose of the Figure is to show that similar temperature delta increase can be achieved by different mounting configuration. For example when there is no sufficient board area to have large copper pads on the board so that the heat transfer area is large we can compensate by increasing the number of thermal via holes. In the event that large number of thermal via holes could not be used because of electrical routing problems we can compensate by increasing the copper pad size. Results are also shown in the form of temperature color contour plots. Figures 5 and 6 show the predicted temperature for the same board for two different type of component mounting configurations. Both Figures are based on the same board level power dissipation and environment, but Figure 5 shows the predicted board temperature when no means of thermal management is used and Figure 6 shows the predicted board temperature when enlarged copper pads and thermal via holes are used to mount all the high power components. We can see that there is an 8C temperature delta decrease on the predicted surface temperature for the case where the thermal via holes are used (Figure 6). This shows the effectiveness of the thermal via holes in reducing the thermal resistance across the PWB. Results of the thermal design are also presented in the Appendix where the actual thermal design for each of the high power components is shown. This is very important because the mechanical designer would require this information in order to incorporate the changes into the design. 7. CONCLUSIONS The subject of thermal design of any electronic printing wiring board (PWB) is a very complex process not only from the thermal stand point but from all aspects. In this paper, it was the author intention to present the very basic concepts behind proper thermal design and thermal management of high power PWBs. There are many areas of the thermal design of PWB that were only mentioned or not mentioned at all in this paper. Passive thermal systems that rely basically on improving thermal conduction across a board were discussed in this paper. Active thermal management systems, thermo-electric coolers, phase change materials, force convection, and liquid cooling, were not discussed. These types of systems are often used when the power dissipation is of the order of hundreds of Watts. On the other hand, it was the intention of the author to show that the thermal design of a PWB is a team effort between the electrical, mechanical, structural designers. Proper thermal management often time involves making compromises between what is the optimum design and what is economically feasible. 8. REFERENCE 1) Thomas, L. C., “Heat Transfer, Professional Version,” Prentice-Hall, 1993. 2) Mullen, J., “How to Use Surface Mount Technology,” The Staff of the Texas Instruments Information Publishing Center, 1984. 3) Nakayama, W. and Bergles, A. E., “Cooling Electronic Equipment: Past, Present, and Future,” 1988. 4) Lazzaro, G. and Andrikowich, T., “Thermal Management of Aero and Space Electronic Boards,” 1998. 5) Graebner, J. E., “Thermal Conductivity of Printed Wiring Boards,” AT&T Bell Laboratories, 1995. 6) EPP Staff Report, “Considerations in Device Cooling,” Electronic Packaging and Production, July 1981. 7) Azar, K., “The History of Power Dissipation,” Electronic Cooling, January 2000. 8) Swager, A. W., “Circuit Desing Requires Thermal Expertise,” Electronic Design News, June 1989. Page 12 of 18 Figure 1. EEC150. Isometric view of typical unit Figure 2. EEC150. Power Supply Board Component Layout Page 13 of 18 TJUNCTION THERMAL VIAS 0.019” O. D./ 10 PLCS THETAJ=C= 6.3 C/W 0.165” TCASE COMPONENT FOOTPRINT A 0.380” 0.190” THETACFG= 3.7 C/W A THETACFG= 37.4 C/W THETA10 VIAS=7.7C/W COPPER PAD TOP OF PWB 0.330” SOLDER 0.003”TK. THETAC-A= 65 C/W COMPONENT COPPER PAD 0.0014”TK. 0.33” X 0.38” CFG 0.006”TK THETASM= 4.1 C/W COPPER GROUND PLANE 0.0014” TK. / 2 PLCS CFG 0.052” TK. COPPER PAD 0.38” X 0.48” X 0.0014” 3M Y9469 BOND 0.005” TK. SOLDER MASK 0.003” TK. SECTION A-A THETABOND= 6.2 C/W TSINK TAMB. AL HEAT SINK (AMS4027) 0.07” TK. Figure 3. EEC150. Mounting Configuration for a Typical High Power Surface Mounted Transistor TO-220 TEMPERATURE DELTA ACROSS A BOARD 0.08" THICK ACTUAL PAD SIZE USED IN CALCULATION IS X TIMES THE FOOTPRINT OF THE COMPONENT (O.183in 2) 70 65 60 TEMPERATURE DELTA ACROSS THE BOARD (C) 55 60 V IA S/1.5 PA D 60 V IA S/2.0 PA D 60 V IA S/2.5 PA D 60 V IA S/3.0 PA D 60 V IA S/3.5 PA D 50 V IA S/1.5 PA D 50 V IA S/2.0 PA D 50 V IA S/2.5 PA D 50 V IA S/3.0 PA D 50 V IA S/3.5 PA D 50 V IA S/4.0 PA D 40 V IA S/1.5 PA D 40 V IA S/2.0 PA D 40 V IA S/2.5 PA D 50 45 40 40 V IA S/3.0 PA D 40 V IA S/3.5 PA D 40 V IA S/4.0 PA D 30 V IA S/1.5 PA D 30 V IA S/2.0 PA D 30 V IA S/2.5 PA D 30 V IA S/3.0 PA D 30 V IA S/3.5 PA D 30 V IA S/4.0 PA D 20 V IA S/1.5 PA D 20 V IA S/2.0 PA D 20 V IA S/2.5 PA D 20 V IA S/3.0 PA D 20 V IA S/3.5 PA D 20 V IA S/4.0 PA D 10 V IA S/1.5 PA D 10 V IA S/2.0 PA D 35 30 25 20 15 10 10 10 10 V IA S/2.5 PA D V IA S/3.0 PA D V IA S/3.5 PA D V IA S/4.0 PA D 10 5 0 0 0.25 0.5 SLOPE=THETA C-S 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75 3 3.25 3.5 3.75 4 POWER DISSIPATION (W) Figure 3. TO-220. Temperature Delta Across A 0.08” Thick PWB for Various Mounting Configurations Page 15 of 18 Figure 4. EEC150. Power Supply Board Predicted Temperature (Using Thermal Management Techniques Described in the Appendix) Page 16 of 18 Figure 5. EEC150. Power Supply Board Predicted Temperature (Without Using Any Thermal Management) Page 17 of 18 APPENDIX MOUNTING METHODS USED FOR THE DIFFERENT HIGH POWER COMPONENTS ON THE EEC150 1. The components mounted using this configuration have their respective bases soldered to enlarged copper pads on the top of the board. These copper pads are connected to similar copper pads on the opposite side of the board, closed to the heat sink. The connection between the two copper pads is made via a number of thermal vias. The actual size of the copper pads and the total number of thermal vias connecting the two pads vary with component and power dissipation. The table above shows the actual copper pads sizes and total number of thermal vias used for the different components. Figure 1 and 2 shows tipical configurations for this method. 2. The metal end caps on this type of components are each to be soldered to enlarged copper pads on the top of the board. The enlarged copper pads increase the effective heat transfer area for the given parts. By increasing the heat transfer area, the temperature delta increase across the board is reduced. The actual size of the copper pads varies with component and power dissipation. The table above shows the actual copper pad sizes used for the different components. POWER SUPPLY BOARD SCHEMATIC IDENT. CR201 CR202, 203, 207 CR206, 215 CR208 CR211 CR212 Q201, Q213 Q202, 214 Q203, 215 Q208, 212 Q702 Q704, 705, 718-720, 732-735 Q712, 713 Q722, 724 Q726, 727 Q729, 730 Q738 Q740-742, 746 Q744 R201 R219 R221 R244 R248 R702 R710, 711, 767-769, 834-837 R740 R752, 753 R765-767 R803, 810 R838 R870-872, 894 R915, 924 U204 U205 U206 U207 PART NUMBER MUR3020PT MURB1620 MURB1620 MURB1620 MURS360T3_4A MURS360T3_4A IRF540S IRF540S IRF540S IRF640S IRF9530 MTD6N10 IRF9530 IRF9530 MTD6N10 IRF9530 IRF9530 MTD6N10 IRF9530 M55342K09B1E 93077-R402F 88034-K27R4 93077-R200F 88034-K3091 93076-R133F 93076-301F 93076-2R37F 93076-R162F 93076-2R37F 93076-R301F 93076-2R37F 93076-301F 88034-K2001 MC7815BD2T MC7915BD2T MC7815BD2T MC7915BD2T CASE TYPICAL MAXIMUM MOUNTING TOP COPPER 2 TYPE POWER (W) POWER (W) METHOD PAD SIZE (IN ) TO-252 0.000 3.225 1 0.890 X 0.790 D2PAK 1.003 1.604 1 0.594 X 0.693 D2PAK 1.029 2.069 1 0.594 X 0.693 D2PAK 0.415 1.245 1 0.594 X 0.594 DO-214 0.228 0.416 2 0.198 X 0.298 DO-214 0.204 0.445 2 0.198 X 0.396 D2PAK 0.451 1.016 1 0.594 X 0.693 D2PAK 1.967 1.991 1 0.594 X 0.693 D2PAK 1.967 1.991 1 0.594 X 0.693 TO-236AB 0.599 1.288 1 0.594 X 0.644 SMD-220 0.337 0.862 1 0.594 X 0.693 TO-252 0.108 0.270 1 0.297 X 0.397 SMD-220 0.267 0.742 1 0.594 X 0.693 SMD-220 0.087 0.237 1 0.594 X 0.693 TO-252 0.125 0.388 1 0.297 X 0.397 SMD-220 0.347 0.939 1 0.594 X 0.693 SMD-220 0.469 1.705 1 0.594 X 0.693 TO-252 0.108 0.270 1 0.297 X 0.397 SMD-220 0.076 0.176 1 0.594 X 0.693 CHIP 0.085 0.117 2 0.074 X 0.148 CHIP 0.147 0.283 2 0.149 X 0.228 CHIP 0.326 0.326 2 0.198 X 0.198 CHIP 0.073 0.640 2 0.149 X 0.228 CHIP 0.160 0.170 2 0.074 X 0.148 CHIP 0.128 0.298 2 0.089 X 0.198 CHIP 0.065 0.163 2 0.100 X 0.100 CHIP 0.028 0.371 2 0.148 X 0.297 CHIP 0.124 0.333 1 0.148 X 0.198 CHIP 0.023 0.225 2 0.100 X 0.100 CHIP 0.075 0.204 2 0.100 X 0.100 CHIP 0.024 0.292 2 0.198 X 0.198 CHIP 0.065 0.163 2 0.100 X 0.100 CHIP 0.288 0.410 2 0.148 X 0.198 TO-220 0.490 0.588 1 0.594 X 0.693 TO-220 0.546 1.509 1 0.594 X 0.693 TO-220 0.648 1.635 1 0.594 X 0.693 TO-220 0.476 0.571 1 0.594 X 0.693 TOTAL NUMBER OF THERMAL VIAS REQUIRED BOTTOM COPPER PAD SIZE (IN2) 0.890 X 0.790 0.594 X 0.693 0.594 X 0.693 0.594 X 0.594 N/ A N/ A 0.594 X 0.693 0.594 X 0.693 0.594 X 0.693 0.594 X 0.644 0.594 X 0.693 0.297 X 0.397 0.594 X 0.693 0.594 X 0.693 0.297 X 0.397 0.594 X 0.693 0.594 X 0.693 0.297 X 0.397 0.594 X 0.693 N/ A N/ A N/ A N/ A N/ A N/ A N/ A N/ A 0.148 X 0.198 N/ A N/ A N/ A N/ A N/ A 0.594 X 0.693 0.594 X 0.693 0.594 X 0.693 0.594 X 0.693 THERMAL VIAS / PART 40 21 21 37 0 0 21 21 21 29 33 6 36 28 6 21 23 6 14 0 0 0 0 0 0 0 0 4 0 0 0 0 0 28 28 28 28 TOTAL VIAS PER GROUP 40 63 42 37 0 0 42 42 42 58 33 48 74 56 12 42 23 24 14 0 0 0 0 0 0 0 0 8 0 0 0 0 0 17 17 17 17 768 Page 18 of 18 0.693” COMPONENT FOOTPRINT A THERMAL VIAS 0.016” O. D./ 21 PLCS A 0.594” 0.380” COPPER PAD TOP OF PWB 0.380” SOLDER 0.003”TK. COMPONENT COPPER PAD 0.0021”TK. 0.594” X 0.693” CFG 0.009”TK COPPER GROUND PLANE 0.0014” TK. / 2 PLCS CFG 0.052” TK. COPPER PAD 0.594” X 0.693” X 0.0021” 3M Y9469 BOND 0.005” TK. SOLDER MASK 0.003” TK. SECTION A-A AL HEAT SINK (AMS4027) 0.08” TK. FIGURE 1. EEC150. THERMAL CONFIGURATION FOR IRF540S TRANSISTORS ON THE POWER SUPPLY BOARD, D2PAK TYPE CASE / MAXIMUM POWER DISSIPATION 1.991W A A THERMAL VIAS / 4 PLCS SOLDER PAD 2 PLCS COPPER PAD 2 PLCS COMPONENT METAL END CAP / 2PLCS SOLDER 0.003” THICK COPPER PAD 0.0021”TK./ 2 PLCS THERMAL VIA 4PLCS COPPER GROUND PLANE 0.0014” TK. / 2 PLCS CFG 0.052” TK. SOLDER MASK 0.003” THICK COPPER PAD 0.148” X 0.198” X 0.0021” 3M Y9469 BOND 0.005” TK. SECTION A-A AL HEAT SINK (AMS4027) 0.08” TK. FIGURE 2. EEC150. THERMAL CONFIGURATION FOR R752 AND R753 ON THE POWER SUPPLY BOARD, CHIP TYPE RESISTORS / MAXIMUM POWER DISSIPATION 0.333 W