technical note august 1997 thermal management for high-power board-mounted power modules introduction board-mounted power modules (bmpms) enhance the capabilities of advanced computer and communi- cations systems by providing ?xible power architec- tures; however, proper cooling of the power modules is required for reliable and consistent operation. main- taining the operating case temperature (t c ) within the speci?d range keeps internal-component tempera- tures within their speci?ations, which, in turn, helps keep the expected mean time between failures (mtbf) from falling below the speci?d rating. tyco's fc-, fe-, and fw-series 50 w to 200 w bmpms are designed with high ef?iency as a pri- mary goal. the 5? output units have typical full-load ef?iencies greater than 80%, which result in less heat dissipation and lower module case tempera- tures. furthermore, these modules use temperature- resistant components, such as ceramic capacitors, that do not degrade during prolonged exposure to high temperatures, as do aluminum electrolytic capacitors. this application note provides the information to ver- ify that adequate cooling is present in a given operat- ing environment. a thermal model is included that can be used to design heat sinks and other means for meeting the cooling and reliability requirements for a given application. the information can be applied to all tyco high-power bmpms in the 121.9 mm x 63.5 mm x 12.7 mm (4.8?n. x 2.5?n. x 0.5?n.) package. basic thermal management proper cooling can be veri?d by measuring the t c of the module at the location indicated in figure 1. t c must not exceed 95?c (85 ? for 200 w modules) while operating in the ?al system con?uration. after the module has reached thermal equilibrium, the measurement can be made with a thermocouple or surface probe. if a heat sink is mounted to the case, make the measurement on the base of the heat sink as close as possible to the indicated position, taking into account the contact resistance between the mounting surface and the heat sink (see detailed thermal model section). 8-582(c).c note: top view, pin locations are for reference. dimensions shown in millimeters and (inches). figure 1. case temperature measurement while this is a valid method to check for proper ther- mal management, it makes the assumption that the ?al system con?uration exists and can be used for a test environment or can be modeled. the following graphs provide guidelines to predict the thermal performance of the module for typical con?urations that include heat sinks in natural or forced-air?w environments. the goal of thermal management is to transfer the heat dissipated by the module to the surrounding environment. the amount of power dissipated by the module as heat (p d ) is the difference between the input power (p i ) and output power (p o ) as shown by the equation below: p d = p i ?p o also, module ef?iency ( h ) is de?ed as the ratio of output power to input power, shown by the following equation: the input power term can be eliminated by combin- ing these two equations. they yield the equation below, which can be used to calculate module power dissipation: parallel + sense + out case on/off + in measure case temperature here 18 (0.7) 76 (3.0) fe150a dc-dc power module in:dc 48v, 3.7a out:dc 5v, 30a 150w made in usa lucent protected by u.s. patents: 5,036,452 5,179,365 tuv rheinland 6238 h p o p i ------- = p d p o 1 h () h ------------------------- =
2 2 tyco electronics corp. technical note august 1997 high-power board-mounted power modules thermal mangement for basic thermal management (continued) however, ef?iency is a nonlinear function of the mod- ule input voltage (v i ) and output current (i o ). typically, a plot of power dissipation versus output current over three different line voltages is given in each module- speci? data sheet. this is because each power mod- ule output voltage has a different power dissipation curve. the typical curves of p d vs. i o for three input voltages for the fe150a power module are shown in figure 2. 8-583(c) figure 2. fe150a power dissipation as heat vs. output load module derating 200 watt power modules please see module speci? data sheets for the fe200 and fw200 series bmpms. these modules have an 85 ? maximum case temperature. forced convection without heat sinks increasing the air?w over the module improves cool- ing. figure 3 shows power derating vs. local ambient temperature (t a ) at air?ws from natural convection to 4.0 m/s (800 ft./min.) 8-587(c) figure 3. forced convection derating without heat sinks the curves in figure 3 were obtained from measure- ments made in a free stream of air approaching a verti- cally oriented module mounted on a printed-wiring board (pwb) in a rectangular passage, as shown in figure 4. 8-690(c) figure 4. location for measurements figure 3 can be used to determine the appropriate air?w for a given set of operating conditions. for example, at p d = 20? and t a = 40 ?, an air?w of 1.0 m/s (200 ft./min.) is suf?ient to keep the module within its ratings. heat sinks several standard heat sinks are available for high- power bmpms, as shown in figures 5 and 6 with their respective thermal resistances for natural convection. the heat sinks mount to the top surface of the module using #4-40 hardware torqued to 0.56 n-m (5 in.-lb.) placing a thermally conductive dry pad or thermal grease between the case and the heat sink minimizes contact resistance and temperature drop. 40 30 20 10 0 0 5 10 15 20 25 30 output current, i o (a) power dissipation (w) v i = 60 v v i = 50 v v i = 40 v 30 power dissipation, p d (w) local ambient temperature, t a ( c) 20 10 020406080 40 100 0 0.1 m/s (20 ft./min.) natural convection 0.5 m/s (100 ft./min.) 1.0 m/s (200 ft./min.) 1.5 m/s (300 ft./min.) 2.0 m/s (400 ft./min.) 2.5 m/s (500 ft./min.) 3.0 m/s (600 ft./min.) 3.5 m/s (700 ft./min.) 4.0 m/s (800 ft./min.) facing pwb air velocity and ambient temperature measured below the module airflow pwb
tyco electronics corp. 3 technical note august 1997 high-power board-mounted power modules thermal mangement for module derating (continued) heat sinks (continued) 8-691(c) figure 5. heat sink with fins oriented along length 8-692(c) figure 6. heat sink with fins oriented along width natural convection with heat sink figures 7 and 8 represent power derating for a module in natural convection with the heat sinks shown in figures 5 and 6. natural convection is the air?w pro- duced when air in contact with a hot surface is heated, causing it to rise. an open environment is required with no external forces moving the air. figures 7 and 8 apply when the module is the only source of heat present in the system. 8-693(c) figure 7. heat-sink derating curves; natural convection; fins oriented along length 8-694(c) figure 8. heat-sink derating curves; natural convection; fins oriented along width with a known p d and t a , the appropriate heat sink can be chosen from the derating curves. for example, if p d ??0? and t a ??0?c with the heat sink oriented along the width, the 0.5?n. heat sink would keep the module within its temperature rating. 0.5 in.
4 4 tyco electronics corp. technical note august 1997 high-power board-mounted power modules thermal mangement for basic thermal model another approach for analyzing thermal performance is to model the overall thermal resistance of the module. total thermal resistance ( q ) is de?ed as the maximum case temperature rise ( d t c , max ) divided by the p d of the module: where: q = total thermal resistance d t c , max = maximum case temperature rise p d = power dissipated as heat this can be represented by the simpli?d model shown in figure 9. in this model, p d , d t c , max , and q are analogous to current ?w, voltage drop, and electrical resistance, respectively, in ohm s law. 8-695(c) figure 9. basic thermal-resistance model for fe-, fc-, and fw-series 50 w to 200 w bmpms, the thermal-resistance value vs. air velocity has been determined experimentally and is plotted in figures 10 and 11 for a unit without a heat sink and with six of the heat sinks mentioned in the previous section. note that the highest values on the curves represent natural convection. in a system with free-?wing air and other heat sources, there may be additional air?w. 8-696(c) figure 10. heat-sink resistance curves; fins oriented along width 8-697(c) figure 11. heat-sink resistance curves; fins oriented along length use an air?w measurement just upstream from the module when determining q from these curves (see figure 4). the following examples illustrate how the curves can be used to determine thermal performance under various air?w and heat-sink con?urations. example a. air?w required to maintain t c an fc150a operates at i o = 20 a and t a = 50?c with no heat sink. determine the air?w required to maintain t c , max = 95?c. p d ? 22.5 w for the fc150a; therefore, the necessary thermal resistance ( q ) is: from figure 10, the required air?w is 2.2 m/s (440 ft./min.) q d t c, max p d --------------------- = bmpm p d
tyco electronics corp. 5 technical note august 1997 high-power board-mounted power modules thermal mangement for basic thermal model (continued) example b. determining t c if an air?w of only 0.75 m/s (150 ft./min.) is available for the same arrangement as in example a, determine t c when using the 0.5 in. heat sink oriented along the length. from figure 11, q ? 1.3?c/w; therefore, d t c is: d t c = p d ( q ) d t c = 22.5 (1.3) d t c = 29.3 ? and t c is: t c = t a + d t c t c = 50 + 29.3 t c = 79.3 ? detailed thermal model thermal resistance in the previous section includes heat transfer by conduction, convection, and radiation from the entire module to the surrounding environment. typically, the power module is soldered to a vertically oriented pwb as shown in figure 4. although most of the heat is transferred by convection and radiation from the top mounting surface of the module, signi?ant amounts of heat are also removed by convection from the sides of the module, and by conduction to the pwb and then convection off of the opposite side of the pwb. heat ?w paths in this con?uration are shown in figure 12. 8-698(c) figure 12. boundaries contributing to thermal resistance q t = convection resistance from module top to ambient q b = convection and conduction resistance from module bottom to ambient q e = convection resistance from edge to ambient q h = convection resistance from heat sink to ambient q r = radiation resistance for entire module q i = contact (interface) resistance when a heat sink is mounted to the top surface of a module, q t is equivalent to the sum of q h and q i : q t = q h + q i several types of heat transfer cool the module. in a fan-cooled environment, convection is the predominant mode of heat transfer. radiation is also important, especially in natural convection. conduction is important when using unusual heat sinks or cold-plate methods of cooling. convection heat transfer convection heat transfer is a function of the surface area, d t c , and the heat transfer coef?ient. a simpli?d approach to convection cooling over a ?t plate leads to the following relationship: where: q = thermal resistance c = surface-dependent constant v = air?w in m/s radiation heat transfer radiation is not dependent upon the air?w over the module, but on the temperature difference between the module and the surrounding environment. for a partic- ular module type, q due to radiation can be determined experimentally. for example, tyco s high-power mod- ules operating at t c = 95 ? and t a = 25 ?, result in q r ? ?5?c/w. bmpm with heat sink
6 6 tyco electronics corp. technical note august 1997 high-power board-mounted power modules thermal mangement for detailed thermal model (continued) detailed thermal-resistance model thermal resistances can be represented in an electri- cal analogy as resistances in parallel (see figure 13). this model is valid for forced convection. natural convection can be estimated using v = 0.25 m/s (50 ft./min.) for open environments with no additional heat sources. 8-699(c) note: v is measured in m/s. figure 13. detailed thermal-resistance model the following examples illustrate how this detailed model can be used to solve speci? thermal problems and to analyze unusual thermal applications. example c. custom heat-sink design an FE150H is operated at p o =150 w with natural convection and t a = 50 ?. q r ? 30 ?/w represents the radiation thermal resistance from the sides and back, and q i = 0.15 ?/w is a conservative value for the contact resistance between the heat sink and the mounting surface. determine the thermal resistance of the smallest heat sink required for this application. using the data sheet for the FE150H, p d ? 24 w is calculated. in order to maintain t c , max = 95?c, the overall module resistance must be: from figure 13, the equivalent resistance to heat ?w from other than the top surface of the module is: q other = q r || q e || q b q other = 1/(1/ q r + 1/ q e + 1/ q b) q other = 1/(1/30 + 1/20 + 1/30) q other = 8.6 ?/w the thermal resistance for heat ?w off of the top surface of the module should be: q t = 1/(1/ q total ?1/ q other ) q t = 1/(1/1.9 ?1/8.6) q t = 2.4 ?/w and the heat sink requires a thermal resistance of: q h = q t ? q i q h = 2.4 ?0.15 q h = 2.25 ?/w example d. contact resistance using the module and heat sink selected in example c, determine the temperature drop from the surface of the module to the heat sink. first determine the heat ?w through the top surface of the module. this is given by: p d , top = (t c , max ?t a )/ q t p d , top = (95 ?50)/2.4 p d , top = 18.8 w the temperature drop is heat ?w multiplied by the resistance: d t = p d , top ( q i ) d t = 18.8(0.15) d t = 2.8 ? therefore, to keep the module from overheating, do not allow the temperature on the top surface of the heat sink to exceed: t h = t c , max ? d t t h = 95 ?2.8 t h = 92.2 ? the contact resistance between the top surface of the module and the heat sink should not be allowed to exceed 0.2 ?/w. typically, with an appropriate dry pad, q i < 0.15 ?/w. radiation edge bottom top bmpm p d 15
tyco electronics corp. 7 technical note august 1997 high-power board-mounted power modules thermal mangement for detailed thermal model (continued) conduction through a cold plate in some instances, heat can be removed from the top surface of the module by conduction through a solid object. one-dimensional conduction heat transfer can be expressed as: q = l/ka where: q = thermal resistance l = heat travel distance hot to cold (m) k = material conductivity (w/m ??) a = material cross-section area (m 2 ) for example, an fe150a operating at i o = 30 a and t a ??0?c in natural convection is mounted to a copper base plate that conducts heat to a cooling surface held at 40 ? (t c s). the base plate cross-section is 0.000806 m 2 , its length is 0.127 m and k = 380w/m ??. from the ef?iency curves, p d = 33 w. determine the case temperature (t c ). the thermal resistance of the base plate is: q bp = l/ka q bp = 0.127/(380 ?0.000806) q bp = 0.4 ?/w so the overall module resistance is: q total = q other || q bp q total = 1/(1/ q other + 1/ q bp ) q total = 1/(1/8.6 + 1/0.4) q total = 0.4 ?/w and t c is: t c = t c s + p d ( q total ) t c = 40 + 33(0.4) t c = 53.2 ? in this example, thermal resistance from the top sur- face is low enough that other resistances are negligible when computing overall resistance. operation in narrow spaces figure 4 shows the typical mounting con?uration with the module oriented vertically in a rectangular pas- sage. when clearance between the facing board and the top surface of the module is reduced, air?w between these two surfaces drops, thereby lowering the heat transfer. this increases the value of q t . flow interference begins when the clearance approaches the boundary layer thickness on the top surface of the module. when operating the module in tight spacing, make a calculation of the boundary layer thickness and use a higher value of q t , if necessary, when calculating q total . for assistance, contact a tyco application engi- neer. horizontal orientation in some applications, the module is operated in natural convection and oriented horizontally as shown in figure 14. in this situation, use q total = 4.8 ?/w for overall module thermal resistance. 8-700(c) figure 14. horizontal orientation
tyco electronics power systems, inc. 3000 skyline drive, mesquite, tx 75149, usa +1-800-526-7819 fax: +1-888-315-5182 (outside u.s.a.: +1-972-284-2626 , fax: +1-972-284-2900) http://power.tycoelectronics.com tyco electronics corporation reserves the right to make changes to the product(s) or information contained herein without notic e. no liability is assumed as a result of their use or application. no rights under any patent accompany the sale of any such product(s) or information. ?2001 tyco electronics corporation, harrisburg, pa. all international rights reserved. printed in u.s.a. august 1997 tn97-009eps (replaces tn92-004eps) printed on recycled paper
|
