CROSS-REFERENCE
to related German patent applications, the disclosures of which are incorporated by reference: DE 20 2004 005 241, filed 26 MAR. 2004; DE 20 2004 019 084, filed 27 NOV. 2004; and DE 20 2004 019 852.5, filed 15 DEC. 2004.
FIELD OF THE INVENTION
The invention relates to a heat sink for absorbing and removing heat from a power component, e.g. a microprocessor, a microcontroller, an ASIC (Application-Specific Integrated Circuit), a laser, or the like. The invention is similarly suitable for the cooling of power components having a high heat flux density.
BACKGROUND
Power electronics components that require cooling must not heat up above specific limit temperatures. Because reduced conductor widths also mean that the surface area of processors or other components is becoming ever smaller, the result is a sharp increase in the density of the heat flux to be discharged, i.e. the heat flux density. This makes it more difficult to apply the principle of so-called heat spreading, since regions inside a heat sink that are remote from the component that is to be cooled can contribute effectively to heat transfer only if their temperature is significantly higher than the temperature of the cooling fluid flowing past them. The flow velocities must also be sufficiently high, and the thermal boundary layers consequently sufficiently thin, to allow heat to be discharged effectively.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a novel heat sink.
The invention provides a heat sink through which a cooling fluid flows during operation, and which offers a large surface area. In the heat sink, cold cooling fluid initially encounters regions having high absolute temperatures. Advantageously, it is possible largely to prevent mixing between cooling fluid that has already heated up and fresh, cold cooling fluid, before the latter encounters regions having the highest absolute temperatures. The regions having the highest absolute temperatures can also experience high incident flow velocities of the cooling fluid, so that turbulent flow, and consequently optimum heat transfer, can be obtained there.
For that purpose, there are arranged, in a depression or cavity, a plurality of thermally conductive elements that project from the bottom of that depression into the path of the cooling fluid, so that the latter thereby flows around them during operation. This yields a corresponding enlargement of the heat-transferring surfaces. The thermally conductive elements and their interstices preferably have dimensions that can be produced using economical methods, e.g. by milling, electrodischarge machining methods, casting methods, forming, stamping, pressing, etc. It has also been found that heat transfer to the cooling fluid can be favorably influenced by a suitable surface treatment, preferably by sandblasting.
The contour of the depression, which can also be referred to as a concave configuration, basin, or cavity, can be adapted to the requirements of a substrate that is to be cooled. A cavity in the shape of a part of a sphere (calotte), for example, can easily be manufactured by milling. Rotational conic sections can alternatively be used, for example a cavity in the shape of a rotational paraboloid or the like.
The housing of a heat sink of this kind can comprise attachment capabilities, with which the housing can be mounted onto existing attachment points.
The inlet element preferably has at its outlet a nozzle field that can comprise, for example, round nozzles or slit nozzles. The cooling fluid is accelerated as it flows through such a nozzle field. As a result, the cold cooling fluid has a high velocity when it encounters regions having high absolute temperatures.
The nozzle field can also cause a portion of the cooling fluid to be deflected as it flows through the nozzle field, which can be considered a result of the so-called Coanda effect (named for Henry Coanda, 1886-1972). By exploiting this effect, portions of the cold cooling fluid can be directed in controlled fashion onto regions of the thermally conductive elements located farther out, thus improving heat transfer.
BRIEF FIGURE DESCRIPTION
Further details and advantageous refinements of the invention are evident from the exemplifying embodiments, not to be understood as a limitation of the invention, that are described below and depicted in the drawings, in which:
FIG. 1 is a section through a first embodiment of a heat sink according to the present invention, viewed along line I-I of FIG. 4;
FIG. 2 is a plan view looking in the direction of line II-II of FIG. 1;
FIG. 3 is a three-dimensional depiction analogous to FIG. 1, the section plane corresponding to FIG. 1;
FIG. 4 is a plan view looking in the direction of arrow IV of FIG. 1;
FIG. 5 is a depiction of the heat sink in the closed state;
FIG. 6 is a depiction analogous to FIG. 5, but in which the four pressure springs and the associated bolts are depicted;
FIG. 7 is a depiction analogous to FIG. 1 in which the route of the cooling fluid flow is schematically indicated;
FIG. 8 is a depiction analogous to FIGS. 1, 3, and 7 in which a substrate, e.g. the so-called “die” of a microprocessor, and the heat flux proceeding from it are schematically indicated;
FIG. 9 is a three-dimensional depiction viewed from the lower side of the heat sink, cooling plate 68 having been removed;
FIG. 10 is a greatly enlarged depiction of a nozzle plate having nine round nozzle openings 58;
FIG. 11 is a depiction analogous to FIG. 10 in which, in contrast to FIG. 10, five slit-like nozzle openings are used;
FIG. 12 is a three-dimensional depiction of a variant of FIG. 9 in which cooling plate 68 has likewise been removed;
FIG. 13 is a depiction analogous to FIG. 6, but viewed in longitudinal section through the two front spikes 24 and 30;
FIG. 14 is a section through a variant in which the section plane extends analogously to FIG. 1;
FIG. 15 is an enlarged sectional depiction of the central portion of a heat sink analogous to the depiction of FIG. 14, nozzle plate 59″ and cooling element 84″ not being depicted to scale because of their small dimensions;
FIG. 16 is a very greatly enlarged section viewed in the direction of arrow XVI of FIG. 17 in which the size relationships, among the dimensions of the parts depicted, correspond to those of an embodiment optimized by comparative tests;
FIG. 17 is a section viewed along line XVII-XVII of FIG. 16;
FIG. 18 is an enlarged depiction of detail XVIII of FIG. 17;
FIG. 19 is a three-dimensional depiction of the heat sink of FIGS. 16 through 18, viewed from the side of the cooling element and its retaining plate 114; and
FIG. 20 is a plan view of the heat sink, approximately analogous to FIG. 19.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
As depicted e.g. in FIG. 1 and FIG. 3, a preferred embodiment of a heat sink 20 according to the present invention has an upper part 22 that is manufactured as a shaped part from a suitable material, preferably as an injection-molded part from a plastic or a suitable metal. Upper part 22 here has four tube-like projections (“spikes”) 24, 26, 28, 30, whose shape is best understood from FIGS. 3 through 6 and 13 and which serve to press upper part 22 toward a substrate 32 to be cooled, which is indicated in FIG. 8 with dashed lines. Substrate 32 can be, for example, the so-called “die” of a microcomputer (CPU) which usually has a square or rectangular shape, in which it generates over a small area a high thermal output that results in a corresponding heat flux, which is indicated symbolically by arrows 34 in FIG. 8.
This heat flux 34 must be absorbed by heat sink 20 and transferred via a cooling fluid 36 (FIGS. 1 and 7) to a cooling unit 38 (FIG. 7) of arbitrary design, where it is discharged to the ambient air e.g. by means of a fan 40 that is equipped with a motor 42, as indicated in FIG. 7. Cooling unit 38 can advantageously contain a circulation pump 39 for cooling liquid 36.
Cooling unit 38, the nature of cooling fluid 36, and the further manner in which heat is removed (whether via air or a different fluid) are not part of the present invention and are therefore indicated here only to the extent that appears useful for an understanding of the invention and its context.
Upper part 22 (FIGS. 3-6) has at the bottom a cylindrical recess 46 (FIG. 9) in a downwardly projecting extension 48. Recess 46 is open toward the bottom and is delimited toward the top by lower side 50 of upper part 22. Located on this lower side 50 is a downwardly projecting tube-like protrusion 52 that, in plan view from below, can have the shape of a ring or tube (cf. FIG. 9), and is arranged approximately concentrically with nozzle field 59 at outlet 54 of a supply fitting 56 for cooling fluid 36. Nozzle field 59, e.g. in the form of nine holes 58 as depicted in FIGS. 4, 9, and 10, is arranged in the region of outlet opening 54.
As FIG. 9 shows, protrusion 52 can advantageously have, on its inner side and on its side facing away from an outlet opening 61, a bevel 53 that preferably continuously decreases toward that outlet opening 61 in the manner depicted and that extends, in this example, over approximately three-quarters of the circumference of protrusion 52. The result is that the flow resistance constituted by protrusion 52 is not the same in all directions, but rather increases toward opening 61.
At its left (in FIG. 1) edge, recess 46 transitions via this outlet opening 61 into the outlet conduit 60 of an outlet fitting 62. The heated cooling fluid 64 flows through fitting 62 back to cooling unit 38 (cf. FIG. 7).
FIG. 6 schematically shows the manner of attachment of a heat sink 20 according to the present invention. The latter has four spikes 24, 26, 28, 30, and into each of these a respective compression spring 25, 27, 29, 31 is inserted in such a way that it rests on an inwardly projecting flange 24a, 26a, 28a, 30a, respectively, of the relevant spike (cf. FIG. 13).
According to FIG. 6 and FIG. 13, a respective bolt 25a, 27a, 29a, 31a is inserted into each spring, and these bolts are screwed into a countermember (not depicted), thereby loading springs 25 through 31 with a predetermined force. Heat sink 20 is pressed with this force against die 32 (FIG. 8) in order to achieve good thermal transfer and thereby good cooling of die 32.
Cylindrical opening 46 (FIG. 1) has an annular groove 66 or at least an annular shoulder. A heat absorption part 68 (FIG. 2) has a cylindrical periphery 70, and located in this is an annular groove 72 (FIG. 3) complementary to annular groove 66. Located between annular grooves 66 and 72, as depicted, is a sealing ring 74 that seals heat absorption part 68 against upper part 22. Any desired sealing means can be used for this seal, as is self-evident to one skilled in the art. Also possible, where applicable, is a permanent join between upper part 22 and heat absorption part 68, e.g. by adhesive bonding or welding.
Part 68 is manufactured from a material having good thermal conductivity, e.g. copper, aluminum, or silver. The nature of the material used depends, inter alia, on the application and on requirements in terms of service life and operating reliability. As is evident from the various figures, in this example part 68 generally has the shape of a round plate having a flat lower side 76 and an upper side 78, parallel thereto, into which is recessed a depression whose general three-dimensional contour corresponds to the shape of a trough or basin. Lower side 76 is precision-machined to produce optimum thermal contact with substrate 32, a thermoconductive paste 80 usually being arranged between lower side 76 and substrate 32 in order to optimize heat transfer. Experiments have shown that, for high heat flux densities, the thickness of the bottom of part 68 should be as thin as is compatible with mechanical stability.
The trough-shaped depression 82 is arranged so that, in use, its center is located substantially above the center of substrate 32. Depression 82 here has approximately the shape of a part of a sphere or “calotte” but other shapes for this concave structure are also possible, e.g. a rotational paraboloid, or the shape of a flat bowl having a substantially flat bottom.
As is best shown by FIGS. 2 and 3, columnar cooling elements or thermal conduction elements 84 project upward from the bottom of depression 82 as far as upper side 78 of heat absorption part 68. Because of their needle-like appearance, these cooling elements 84 can also be referred to as “pins.” Conduits 86 are located between them, i.e. cooling elements 84 are preferably produced by the fact that corresponding conduits or channels 86 are recessed in crisscross fashion into upper side 78 of part 68, as is clearly evident, from the depictions, to one skilled in the art. Cooling elements 84 are preferably roughened by sandblasting after they are produced, in order further to facilitate heat transfer at their surface.
There are of course different ways to produce columnar cooling elements 84 of this kind, e.g. also by means of suitable electrodischarge machining tools, with which it is possible to achieve an irregular profile for conduits 86 and thereby to influence the flow conditions in a controlled manner so that heat discharge occurs in largely symmetrical and therefore optimized fashion.
As already described, the depression 82 represents an enveloping body that is interrupted by pins 84. In this embodiment, depression 82 ends before the cylindrical periphery 70 of cooling plate 68, so that a flat rim segment 90 is created there. Together with lower side 50 of upper part 22, cylindrical recess 46, and protrusion 52, it forms an annular conduit 92 that intersects at the left (in FIG. 1) with outlet conduit 60, since the center axis of outlet conduit 60 coincides approximately with the outer rim of conduit 92, so that the heated cooling fluid 64, which flows in an approximately radial direction out of conduits 86 (FIG. 3) of part 68, is collected in this annular conduit 92 and flows through it and through outlet opening 61 to outlet conduit 60.
As a result, heated cooling fluid 64 is directed around the central portion of cooling plate 68 and consequently cannot mix with cold cooling fluid 36 that is supplied through supply fitting 56 to the central portion (inside ring 52). There consequently exists, inside ring 52, a zone with very cold cooling fluid 36 which serves to provide the most intense cooling of substrate 32 (FIG. 8) at the point where it evolves the most heat. Outside ring 52, mixing-in of heated cooling fluid 64 takes place in annular conduit 92, and the heat removal there is consequently less intense.
FIGS. 1, 3, 9, and 14 through 16 show that a plate 59 having the previously described nozzle openings 58 is provided on the lower side of inlet conduit 54. This plate 59 is also depicted in greatly enlarged fashion in FIG. 10, and constitutes a nozzle field.
Different kinds of nozzles can, of course, be used. FIG. 11, for example, shows a nozzle field 59′ in which parallel slit-shaped nozzles 58′ are present.
During operation, under specific flow conditions that are easy to ascertain empirically, nozzles 58, 58′ cause a constriction of the inflowing cooling fluid 36. The latter flows more quickly as a result, and upon encountering cooling elements (pins) 84 brings about intense turbulence and consequently better heat transfer.
Mode of Operation
During operation, cooled cooling fluid 36 is supplied from cooling unit 38 to supply fitting 56 and sprayed at high velocity through nozzle field 59 (FIGS. 3, 4, 7, 9, 10, 14 through 16) inside ring 52 onto heat absorption part 68, preferably (as indicated in FIG. 7) being spread out by the so-called Coanda effect and thereby causing homogeneous, turbulent cooling of this central region (inside ring 52). The Coanda effect occurs at the nozzles that are located at the edges of the nozzle field, for example the topmost and bottommost nozzles 58″ in FIG. 16.
The cooling fluid flows outward through conduits 86 (FIG. 2) in an approximately radial direction, since the annular stopper 52 rests on cooling elements 84 and therefore forces cooling fluid 36 to flow through conduits 86 (and not past them), also cooling the radially outer region 90 of heat absorption part 68.
From conduits 86, heated cooling fluid 64 travels into annular conduit 92, and through the latter via outlet opening 61 to outflow conduit 60 and back to cooling device 38 where it discharges its heat, for example, to the ambient air, as indicated by fan 40.
It should be noted that it is also possible to use as the cooling fluid, for example, a boiling cooling fluid which evaporates at a temperature that is below the maximum temperature of substrate 32 that is to be cooled.
The trough-shaped depression 82 yields the additional advantage that heat sink 20 is insensitive to the slightly oblique positions that often occur in practical use, for example, in a computer; this is because, as indicated by arrows 34 in FIG. 8, a certain forced convection does occur in heat sink 20 and depends little on its position.
What is obtained, by means of the present invention, is thus a heat sink 20 through which a cooling fluid flows, which offers a large surface area, and in which cold cooling fluid first encounters regions having high absolute temperatures. The invention prevents already-heated cooling fluid 64 from mixing with fresh, cold fluid 36, before the latter encounters the regions having the highest absolute temperatures. This means, conversely, that already-heated fluid is withdrawn as quickly as possible from areas having the highest absolute temperatures. In addition, the regions having the highest absolute temperatures also experience an incident flow of cooling fluid 36 at high flow velocities.
As one skilled in the art may gather from FIGS. 1, 3, and 4, without ring 52, the cooling fluid 36 that is supplied would flow in a direct path from inlet fitting 56 to outlet fitting 62 and intensively cool only the left portion of heat absorption part 68, so that substrate 34 (FIG. 8) would also be cooled more intensely on its left side than on its right.
Ring 52, acting as a stopper, counteracts this and forces cooling fluid 36 to flow in all directions and thereby to cool cooling plate 68 more homogeneously. If ring 52 has the same dimensions everywhere as depicted in FIG. 1, however, asymmetrical cooling can nevertheless occur because cooling fluid 36 always takes the path of least resistance, i.e. in FIG. 1 from inflow 56 for the most part directly to the left to outlet 62.
For this reason, in FIG. 9, ring 52 is provided with bevel or chamfer 53, which is greatest where it is located opposite to outlet opening 61. This bevel 53 could also be arranged on the outer side of ring 52. The bevel creates, for cooling fluid 36 that flows in via nozzles 58, a relatively high flow resistance for a direct flow (labeled 90 in FIG. 9) to outlet 61, whereas the flow resistance for flows 92, 94, which proceed via points having a large bevel 53, is less. The result is that the cooling of part 68 (FIG. 2) is altogether more homogeneous, i.e. that large temperature gradients in cooling plate 68 are avoided.
FIG. 12 shows a variant of this. Here base 52B of ring 52 is of solid configuration and has the same height h everywhere. Protrusions 100 of various lengths project upward in FIG. 12 from this base 52B in the manner depicted. They are separated from one another by valleys or conduits 98. These conduits are labeled 98L on the left in FIG. 12, and 98R on the right in the vicinity of outlet opening 61. It is thus evident that protrusions 100 are low on the left side in FIG. 12, and high on the right side.
Protrusions 100 are so configured that they project, in FIG. 2, into interstices 86 between cooling elements 84 and act there, depending on their length, as orifices or restrictions of varying intensity. As a result, in FIG. 12, direct cooling fluid flow 102 from nozzle field 58 to outlet 61 is greatly restricted, whereas cooling fluid flows 104, 106 from the left side in FIG. 12 to outlet 61 are restricted only slightly or not at all, so that overall, cooling plate 68 is as a whole cooled symmetrically and not one-sidedly.
As described with reference to FIG. 6, housing 22 has attachment capabilities with which heat sink 20 can be mounted onto previously existing attachment points that are located in the system of which component 32 to be cooled is a part. Attachment to the base of substrate 32, to the associated main circuit board, or to another component, using so-called “clip” technology, is alternatively possible.
A heat sink 20 of this kind can be manufactured on the whole very inexpensively, since upper housing part 22 with its complicated shape can be manufactured inexpensively as an injection-molded part, and can be optimized for the requirements of a specific processor type. Pins 84 of cooling plate 68, and their interstices 86, preferably have dimensions that can be manufactured using economical production methods, e.g. a width for pins 84 on the order of less than 2 mm. The same is true analogously for the arrangement according to FIG. 12. With reference to FIG. 16, dimensions will be described below that may be regarded as optimum, based on present knowledge.
The inserted cooling plate 68 has a shoulder or annular groove 72 that makes available some of the sealing edges for fluid-tight sealing between upper part 22 and cooling plate 68. At least one other sealing edge (annular groove 66) is made available by upper part 22. Upper part 22 has at least one inlet fitting 56 and at least one outlet fitting 62. Inlet fitting 56 is located at the center, and outlet fitting 62 intersects with its center axis approximately the outer rim of annular conduit 92. The interpenetration of the elements resulting therefrom yields a rectangular outlet 61 out of annular conduit 92, and this outlet has a large cross section and consequently a low flow resistance. The corner edges of the interpenetration can be rounded off for further reduction of the flow resistance.
Inlet fitting 56 preferably has, at its inner end, a diaphragm 59, in which round nozzles 58 or slit-shaped nozzles 58′ can be provided, so as to define a nozzle field. Cooling fluid 36 is accelerated as it flows through this nozzle field 59 or 59′, and the fluid stream is in fact constricted even further after exiting from the nozzle field (Coanda effect), resulting in a further increase in flow velocity. Cold cooling fluid 36 thus has a high velocity when it encounters regions having high absolute temperatures. In addition, the surfaces are enlarged in a balanced relationship by way of pins 84. Corresponding ribs could also be used instead of pins 84; this is not depicted.
Ring 52 forces cooling fluid 36 to flow through at the foot of pins 84, and only small flow resistances are subsequently imposed on the fluid in outer annular conduit 92, so that backflow and mixing with cold cooling fluid 36 is made additionally difficult. Upper part 22 can also possess, for this purpose, an outlet cross section that is larger than the inlet cross section.
FIG. 13 is a depiction analogous to FIG. 6 but in section. The reference characters are the same as in FIG. 6. It is evident that parts 25a, 31a have internal threads 25a′, 31a′, and that springs 25 and 31 are loaded when parts 25a, 31a are pulled downward, for example by screw threads (not shown) that serve for attachment to a circuit board.
FIG. 14 shows a variant that is constructed very similarly to FIG. 1 but, in contrast to FIG. 1, has two outflow fittings 62, 63.
The left half of FIG. 14 corresponds to the left half of FIG. 1, likewise cooling or heat absorption plate 68; the same reference characters are therefore used for these parts. An outflow fitting is located on that side, as in FIG. 1.
Additionally present in the right half of FIG. 14 is a second outflow fitting 63 that is located symmetrically opposite first outflow fitting 62. Cooling fluid 36 thus flows in centrally through inflow fitting 56, flows through nozzle field 59 onto impact plate 68 where it absorbs heat from the latter, and then splits, i.e. one half flows to the left to outflow fitting 62, and the other half to the right to outflow fitting 63. A flow 64′ therefore exists in fitting 62 and a flow 64″ in fitting 63, and these fluid flows are then combined into one outflow 64 by means of connections that are indicated merely schematically.
The advantage in the context of FIG. 14 is that cooling plate 68 is cooled more symmetrically than in the context of FIG. 1 since, in the context of FIG. 1, the left half can be cooled somewhat better than the right because of the geometry of the arrangement.
In FIG. 14, as in FIG. 1, ring 52 is used to prevent mixing of cold fluid 36 with already-heated fluid 64′, 64″, and this ring can once again be formed with bevels or protrusions which are so configured that a slightly elevated flow resistance for the cooling fluid is created in the vicinity of outlet fittings 62, 63.
FIG. 16 shows a portion of a practical exemplifying embodiment of the invention that has proven particularly successful in comparative tests. This depiction corresponds to section XVI-XVI of FIG. 15, but contains details that cannot be graphically depicted in FIG. 15.
The lower (in FIG. 16) portion of nozzle field 59″ is shown cut away in order to improve comprehension of the location and size, in relation to the location of openings 58″ of nozzle field 59″, of cooling elements 84″ that are used there. The location of openings 58″ in the lower part of FIG. 16 is indicated by dot-dash lines.
As FIG. 15 schematically shows, openings 58″ are arranged, in relation to the columnar cooling elements 84″ and conduits 82″ produced between them, so that a stream 36″ of cooling liquid 36 that exits from an opening 58″ during operation is aimed at the intersection of a conduit 82″h extending horizontally in FIG. 16 and a conduit 82″v extending vertically in FIG. 16. As depicted in FIG. 16, a conduit crossing 82″cr of this kind is therefore visible through each nozzle opening 58″, i.e. a stream 36″ is aimed at the deepest point of a conduit 82″, namely the point at which the temperature generated by electronic component 32 (FIG. 15) is highest.
The arrangement of nozzles 58″ is selected so that heat is removed from cooling elements 84″ as homogeneously as possible. In FIG. 16, a central vertical row 96 of four evenly distributed nozzle openings 58″ is provided and, parallel thereto but at a lateral offset, a left row 98 and a right row 100 each having three nozzles 58″. Good results in tests have been obtained with this arrangement. Dimensions a, d, D, h, and L are noted in FIGS. 15 and 16; d denotes the (usually identical) inside widths of conduits 82″h and 82″v. The cross section of cooling elements 84″ is preferably approximately square, and one such square has a side length L. Tests have shown that the cross section of cooling elements 84, 84′, 84″ should be approximately proportional to width d of conduits 82″, and that the best results are obtained when
L=(1.4 . . . 2.0)*d (1),
where L and d are measured in mm. For a dimension d=0.3 mm, for example, a value L of 0.4 to 0.6 mm has proven particularly favorable, i.e. a cross section of approximately 0.1 to approximately 0.4 mm2.
Cooling elements 84″ preferably have a square cross section for ease of manufacture. If a different cross section is selected, e.g. a cylindrical cross section, the average cross section is taken as the starting point, i.e. the weighted average of the cross sections of the individual cooling elements, referred to as Q. This is usually in the range from 0.1 to 0.4 mm2. The relationship between this value and the inside width d of conduits 82″ between cooling elements 84″ is preferably constrained as follows:
d=(0.25 . . . 0.5)*exp(0.5*lnQ) (2),
where d is measured in mm and Q in mm2, and lnQ is the natural logarithm of Q.
This therefore yields, based on present knowledge, a preferred value range for the value d when Q is known.
As the size of an electronic power component 32 to be cooled decreases, its heat flux density usually increases, and it is then necessary to adapt the size of cooling elements 84″, i.e. parts 84″ become even smaller and, according to equation (1), the width of conduits 82″ also becomes even smaller. Diameter D of nozzles 58″ also decreases correspondingly in this case.
Round nozzles 58″ have a preferred diameter D of approximately 1 to approximately 1.2 mm. The distance h noted in FIG. 15 between nozzle plate 59 and the bottom of conduits 82″ at their deepest point has, based on present knowledge, an optimum when the following is true:
h=(2 . . . 3)*D (3),
i.e. for optimum results, this distance h is approximately two to three times the diameter D of a nozzle 58″. This distance h is in any case greater than D. (Diameter D is normally approximately the same for all nozzles 58″, in the interest of simple manufacture. The depiction in FIG. 15 is schematic, and therefore cannot show these dimensional relationships, to which reference is made.)
The preferred vertical center-to-center spacing a between two adjacent nozzles is obtained from the dimensions indicated, i.e.
a=n*L+n*d (4)
where n=2, 3, . . .
If n=2 (as depicted), then
a=2*0.6+2*0.3=1.8 mm (5).
In comparative tests, these dimensions resulted in very good heat removal from component 32 indicated schematically in FIG. 15. A structure of this kind is particularly well suited for components 32 having small dimensions and a high heat flux density, e.g. for processors.
Each two adjacent nozzles 58″ of center row 96 form an isosceles triangle with one adjacent nozzle of row 98 or row 100, the length s of the sides being
s=1.12*a (6).
In comparative tests, an arrangement of this kind has proven effective for removing the quantity of heat that is produced in such a way that local temperature peaks do not occur.
FIG. 17 shows heat sink 20″ of FIG. 16 in the assembled state. It has an upper part 22″ (shaped part) made of plastic, on which are provided an inlet fitting 56″ in the center and an outlet fitting 62″ laterally, the latter leading obliquely upward at an angle alpha of preferably approximately 70 degrees.
Located at the lower (in FIG. 17) end of inlet 56″ is nozzle plate 59″, whose shape is most apparent from FIG. 16. Upper part 22″ has at the bottom a cylindrical recess 46″ in which heat absorption part 68″, sealed by means of an O-ring 74″, is arranged.
Heat absorption part 68″ preferably has approximately the shape of a disk, and has on its lower side a cylindrical protrusion 110 on which is located a rectangular protrusion 112 that, during use, rests with a surface 112u against a heat-emitting part, e.g. against an IC or a microprocessor, as depicted in FIG. 8 for heat-emitting component 32.
Heat absorption part 68″ is retained by a supporting part, here in the shape of a retaining plate 114 (FIGS. 17, 19, and 20) made of steel or the like. Retaining plate 114 has, at its center, a hole 115 through which protrusion 110 extends. For clarification, those parts that belong to heat absorption part 68″ are highlighted in gray in FIG. 20. The configuration is very clearly evident from FIG. 19. Retaining plate 114 is retained on upper part 22″ by four bolts 116.
Upper part 22″ has, at its four corners, four attachment holes 117, each of which has a hollow-cylindrical extension 117A that, as shown in FIG. 17, is elevated above bolts 116. Holes 117 are located on radial enlargements 119 of housing part 22″ that project outward from the latter in the form of lugs, as shown best by FIGS. 19 and 20.
Heat absorption part 68″ is thereby largely relieved of load-bearing functions, and in its central part, i.e. in the region of rectangular protrusion 112, can be very thin, having for example, as depicted, a thickness of less than 1 mm in the central region; this is very advantageous in terms of good cooling, since excellent heat transfer is obtained as a result.
Heat absorption part 68″ has, on its upper (in FIGS. 17-18) side, a basin- or trough-shaped recess 118, in the form of a spherical cavity or “calotte” in the present exemplifying embodiment. Upper part 22″ has an annular recess 120 that is located opposite the outer portion of recess 118 and forms, together with that portion, an annular conduit 122 whose cross section is approximately lens-shaped and which is in direct liquid communication with outflow fitting 62″.
Located at the deepest point of spherical cavity 118 in FIG. 18 are cooling elements 84″, which are shown greatly enlarged in FIGS. 16 and 18. Some of cooling elements 84″ are located, in FIG. 18, directly beneath nozzle plate 59″. As shown in FIG. 18, the distance from nozzle plate 59″ to bottom 82″ of spherical cavity 118 has a value h whose optimum magnitude is obtained from equation (3) and is preferably less than 5 mm. In the exemplifying embodiment, h has an optimized value of approximately 2.5 mm.
Longitudinal axes 126 of nozzles 58″ preferably extend through the center planes of valleys 82″h (cf. FIGS. 16 and 18). This allows particularly good cooling, since the cooling fluid can then flow from above directly into these valleys and, at the bottom of them, can effect impact cooling with a high heat transfer coefficient.
Radially outside those cooling elements 84″ that are located directly beneath nozzle plate 59″, an annular protrusion 128 of upper part 22″ is in contact against cooling elements 84″ there, so that during operation, the cooling fluid cannot flow away over the cooling elements 84″ there but instead must flow between them through valleys 82″. This prevents cold and hot cooling fluid from mixing, which would reduce the cooling efficiency.
Located radially outside annular protrusion 128 are cooling elements 84″x whose height increases toward the outside, additionally improving heat transfer there.
As is particularly apparent from FIG. 17, annular conduit 122 has a large cross section, so that the flow resistance for the cooling fluid that flows in through inflow 56″ in the direction of arrow 36 (FIG. 17) differs little over all the cardinal directions. Cooling at the center, i.e. beneath nozzle plate 59″, is particularly good, since the cooling fluid is coldest there and flows along lines 126 (FIG. 18) directly between cooling elements 84″ and directly cools the bottom of spherical cavity 118, which optimizes cooling at the center of the object (32 in FIG. 15) that is to be cooled.
Reinforcing ribs 130 are provided on the upper side of upper part 22″, partly in order to enhance the mechanical stiffness of upper part 22″ and partly to give it a pleasant appearance that identifies its origin.
As FIG. 17 shows particularly well, spherical cavity 118 extends both into cylindrical protrusion 110 and into rectangular protrusion 112, i.e. spherical cavity 118 extends through opening 115 of plate 114 to a point very close to the lower (in FIGS. 17 and 18) end surface 112u of protrusion 112; this is highly advantageous in terms of heat transfer from this protrusion 112 to the cooling fluid in spherical cavity 118.
Many variants and modifications are possible within the scope of the present invention. For example, the cooling fluid can also flow through the heat sink in the opposite direction, for example when substrate 34 to be cooled requires more intense cooling in its outer regions than in its central regions. These and similar modifications are embraced within the capabilities of one skilled in the art.
Patent Application
20050210906
