Dual graphics card cooperative heat dissipation device

Abstract

The application discloses a double-display-card cooperative heat dissipation device, which comprises fin groups one and two, a flow guide cover and a gas conveying mechanism, the fin groups one and two are internally provided with a plurality of heat transfer gaps, the ends of the fin groups one and two are respectively provided with a set of laminar flow sheets, the two sets of laminar flow sheets are arranged in a staggered mode in the flow guide cover, and the staggered positions of the two sets of laminar flow sheets form laminar flow gaps; the staggered positions of the two sets of laminar flow sheets and the inner wall of the flow guide cover form a flow distribution area; the edge of the flow distribution area guiding airflow and the edge of the laminar flow sheet are provided with a gas flow adjusting gap with variable spacing, and the change of the gas flow adjusting gap causes the change of the gas flow proportion of two branch air curtains. The application can simultaneously dissipate heat for two display cards, reduces heat dissipation energy consumption, and reduces the size of the display cards, so that the space of the display cards is further reduced; in addition, the gas flow of the heat dissipation of the two display cards can be adjusted, and the flexibility of practical application is improved.

Description

Technical Field

[0001] This invention relates to the field of heat sink technology, specifically to a dual-graphics card collaborative heat dissipation device. Background Technology

[0002] With the rapid development of the digital economy, the demand for hardware performance in high-end computing scenarios is growing exponentially. From real-time ray tracing in AAA games and professional rendering of 4K / 8K videos to AI large-scale model training and scientific computing, the computing power of a single graphics card is no longer sufficient to meet the demand for efficient processing.

[0003] Against this backdrop, multi-GPU parallel technology, commonly known as "GPU CrossFire," has become a key solution. By coordinating the scheduling of the graphics processing units (GPUs) and memory resources of two or more GPUs, a near-linear increase in computing power can be achieved. In dual-GPU scenarios, the computing power increase can reach 80%-95%, providing core support for high-load tasks.

[0004] However, increased computing power inevitably leads to a surge in power consumption and heat generation. When a graphics card operates at full load, the heat output of its core heat-generating components becomes a major bottleneck for performance. First, as the computing core, the GPU's transistor density has increased 3-5 times with the evolution of manufacturing processes from 14nm to 3nm, resulting in a corresponding increase in power consumption. Mainstream high-performance graphics cards, such as the RTX 4090 and RX 7900 XTX, have already exceeded 450W under full load, and their core temperatures can easily soar above 90℃. Second, memory modules, such as GDDR6X and HBM3, generate 15%-20% of the total power consumption of the graphics card when operating at high bandwidth, and because they are close to the GPU core, this heat can easily accumulate. Third, components in the power supply module, such as MOSFETs (metal-oxide-semiconductor field-effect transistors) and inductors, generate approximately 10% of the heat themselves while providing stable current to the GPU and memory; if heat dissipation is inadequate, this can easily lead to a decrease in power supply efficiency. Therefore, the performance of the heat dissipation module must keep pace with the computing power of the graphics card; otherwise, high temperatures will trigger the GPU to reduce its frequency. For every 10°C increase in temperature, the performance will be reduced by about 5%-8%, or even lead to hardware failure.

[0005] Looking at the current technological upgrade path of heat dissipation modules, active cooling is still the mainstream solution, which relies on a combination of fins, heat pipes, and fans. However, in dual-graphics card scenarios, the following technical bottlenecks have been exposed:

[0006] Limitations of increasing heat exchange area: Increasing fin width or distribution density is a direct way to improve heat dissipation efficiency, but for every 5mm increase in fin width, the overall thickness of the heat dissipation module will increase by 8-10mm. The thickness of a conventional single card heat dissipation module is about 25-35mm, and it is easy to exceed 50mm after widening. On the other hand, if the fin density is too high, such as the spacing is less than 1.5mm, it will lead to increased airflow resistance, which will reduce heat dissipation efficiency.

[0007] The dilemma of optimizing airflow: Increasing fan speed, such as from 1500 rpm to 2500 rpm, can increase airflow, but higher speeds require larger motors, with the diameter increasing from 20mm to 30mm. This not only increases the module weight, with the weight of a single card heatsink increasing from 500g to over 800g, but also causes noise levels to exceed 45 decibels, far exceeding the noise tolerance threshold of 35 decibels for consumer-grade scenarios. If the airflow is increased by optimizing the fan blade angle, such as from 15° to 25°, the blade rotation resistance will increase by more than 40%, requiring a motor with higher torque and higher power, which will also exacerbate the issues of thickness and power consumption.

[0008] Airflow conflicts in multi-GPU configurations: Increased thickness of GPU cooling modules can obstruct airflow between PCIe slots (Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard). A standard PCIe case has approximately 43mm (2 slots) of PCIe slot spacing. If a single GPU's cooling module is 3 slots thick (64.5mm), the distance between adjacent GPUs in parallel will decrease to less than 20mm, resulting in a 30%-50% reduction in airflow and a sharp drop in cooling efficiency, creating a vicious cycle of "the more cooling, the higher the temperature."

[0009] This problem is even more severe in the server and dedicated workstation sectors. These devices typically employ a 1U / 2U rackmount design, with a 1U chassis only 44.45mm high. Their internal space is far more compact than consumer-grade chassis, and they often house 4, 8, or even 16 graphics cards running in parallel. If the cooling module is thicker than 20mm, the airflow channels between adjacent graphics cards will be completely blocked, causing the average graphics card temperature to exceed 95°C. This not only triggers frequent throttling and a 15%-20% performance loss, but also shortens the lifespan of the graphics card by more than 50%. According to the formula for electronic component lifespan, for every 10°C increase in temperature, the lifespan is halved, while simultaneously increasing the frequency and cost of server maintenance.

[0010] In addition, existing graphics cards generally adopt an independent module design of "one card, one heatsink". As a core power-consuming component, the power consumption of the fan is directly linked to its number: the heatsink module of a single high-performance graphics card usually contains 2-4 fans, with a power consumption of about 15-25W. When two graphics cards are running in parallel, the heatsink power consumption reaches 30-50W, which significantly increases the power consumption required for heat dissipation.

[0011] In summary, current dual-GPU or multi-GPU parallel technologies face a core dilemma: how to ensure adequate airflow and cooling efficiency through enhanced heat dissipation modules while simultaneously controlling module thickness and power consumption to fit within compact spaces and reduce costs. Overcoming this dilemma and achieving a balance between efficient heat dissipation, compact size, and low power consumption has become crucial for the large-scale application of dual-GPU or multi-GPU technology in high-end fields such as AI, supercomputing, and professional rendering. This has significant practical implications and industry value for the technological development of this field. Summary of the Invention

[0012] The purpose of this invention is to provide a dual-graphics card collaborative heat dissipation device to solve the problems mentioned in the background art.

[0013] To achieve the above objectives, the present invention provides the following technical solution: a dual graphics card collaborative heat dissipation device, comprising a fin group one, a fin group two, a flow guide shroud and an air delivery mechanism, wherein the fin group one and the fin group two each have multiple heat transfer gaps, and the ends of the fin group one and the fin group two are respectively provided with a set of laminar flow plates, the two sets of laminar flow plates are arranged in an alternating manner inside the flow guide shroud, and the intersection of the two sets of laminar flow plates forms a laminar flow gap;

[0014] The intersection of the two sets of laminar flow plates and the inner wall of the flow guide cover form a flow splitting zone;

[0015] After the gas curtain output by the gas conveying mechanism enters the laminar flow gap, it is divided into branch gas curtains that enter the heat transfer gaps in fin group one and fin group two respectively.

[0016] The diversion zone is equipped with a movable diversion plate. The edge of the diversion plate guides the airflow and the edge of the laminar flow plate to form a variable airflow adjustment gap. The movement of the diversion plate causes the airflow adjustment gap to change, and the change in the airflow adjustment gap causes the airflow ratio of the two branch air curtains to change.

[0017] Preferably, the laminar flow gap and the heat transfer gap are distributed in parallel.

[0018] Preferably, the spacing of the laminar flow gaps is greater than the spacing of the heat transfer gaps.

[0019] Preferably, the branched air curtain forms a heat transfer air curtain parallel to the heat transfer gap when it passes through the connection between the laminar flow plate and fin group one and fin group two.

[0020] Preferably, the diffuser plate has a drive mechanism inside to drive its movement. The drive mechanism includes a drive motor, and the drive motor has synchronously rotating screws at both ends. The drive motor is fixed to the inner wall of the diffuser.

[0021] The distributor plate has a threaded hole that engages with the screw thread. When the drive motor drives the screw to rotate in the threaded hole, the distributor plate moves along the screw axis.

[0022] Preferably, the flow divider plate also has a limiting mechanism, the limiting mechanism including a limiting rod, the limiting rod being fixed to the inner wall of the flow guide cover;

[0023] The diverter plate also has a limiting hole, and the limiting rod is inserted into the limiting hole.

[0024] Preferably, the gas delivery mechanism has a turbofan and a second drive motor inside, and the air curtain output by the gas delivery mechanism is generated by the turbofan being driven to rotate by the second drive motor.

[0025] The inner diameter of the output end of the gas delivery mechanism is the same as the inner diameter of the input end of the flow guide.

[0026] Preferably, the inner wall of the flow guide is fitted to the outer edge of the laminar flow plate.

[0027] Preferably, the inner wall of the flow guide at the connection between the laminar flow plate and fin group one and fin group two is arc-shaped.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] 1. This invention places the air delivery mechanism externally, allowing airflow to be input from the side of the graphics card. This achieves the desired effect of increasing the thickness of the heat dissipation module and ensuring sufficient airflow for heat dissipation. Furthermore, placing the air delivery mechanism on the side of the graphics card provides a structural basis for simultaneously cooling two graphics cards.

[0030] 2. This invention, through the diversion zone, disperses the air curtain output by a single air supply mechanism and guides it into fin group one and fin group two, achieving the effect of a single air source simultaneously cooling two graphics cards, reducing the number of fans and lowering the energy consumption of the cooling module.

[0031] 3. This invention achieves the effect of adjusting the airflow ratio to the air curtain inside the two graphics cards by using a variable airflow adjustment gap. The airflow adjustment gap can be moved according to the heat generation relationship of the two graphics cards to ensure that more airflow flows into the graphics card with higher heat generation, thereby coordinating the airflow introduction relationship between the two graphics cards and balancing the overall temperature of the two graphics cards. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the exploded structure of the components of the present invention;

[0033] Figure 2 This is a schematic diagram of the external structure of the present invention;

[0034] Figure 3 This is a schematic diagram of the staggered distribution structure of laminar flow plates in this invention;

[0035] Figure 4 This is a schematic diagram of the location and structure of the diversion zone in this invention;

[0036] Figure 5 This is a schematic diagram of the internal structure of the air deflector of the present invention;

[0037] Figure 6 This is a schematic diagram of the inner wall structure of the flow guide shield of the present invention;

[0038] Figure 7 This is a schematic diagram of the branched air curtain flow direction of the present invention;

[0039] Figure 8 This is a schematic diagram of the connection structure between the flow divider and the drive mechanism of the present invention;

[0040] Figure 9 This is a schematic diagram of the air curtain flow direction structure inside the air guide cover of the present invention;

[0041] Figure 10 This is a schematic diagram of air curtain transmission inside the flow guide when the flow divider is centrally distributed according to the present invention;

[0042] Figure 11 This is a schematic diagram of air curtain transmission inside the flow guide when the flow divider is offset towards the second fin assembly according to the present invention;

[0043] Figure 12 This is a schematic diagram of air curtain transmission inside the shroud when the flow divider is offset towards the fin assembly of the present invention.

[0044] In the picture:

[0045] 100, Fin Group 1; 200, Fin Group 2;

[0046] 300, laminar flow plate; 310, laminar flow gap; 320, flow splitting zone;

[0047] 400. Fairing;

[0048] 500, Diverter plate; 501, Screw hole; 502, Limiting hole; 510, Drive motor one; 511, Screw; 520, Limiting rod;

[0049] 600. Gas delivery mechanism; 610. Turbofan; 620. Drive motor II. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figures 1 to 12 The present invention provides the following embodiment:

[0052] Please see Figures 1 to 4 A dual-graphics card collaborative heat dissipation device includes a fin group 100, a fin group 200, a shroud 400, and an air delivery mechanism 600.

[0053] The air delivery mechanism 600 is used to input airflow into the shroud 400. The air delivery mechanism 600 contains a turbo fan 610 and a second drive motor 620. The air curtain output by the air delivery mechanism 600 is generated by the turbo fan 610 driven by the second drive motor 620 to rotate. As the air source in the heat dissipation component, the inner diameter of the output end of the air delivery mechanism 600 is the same as the inner diameter of the input end of the shroud 400. That is, the airflow output by the air delivery mechanism 600 continues to be transmitted into the shroud 400 according to the original transmission path, avoiding the formation of vortices at the connection between the two air curtains, which would generate wind noise and increase wind resistance.

[0054] The air delivery mechanism 600, independently located outside of fin group 100 and fin group 200, eliminates the need for the traditional active cooling structure design where the fan is placed on the fin surface. This effectively reduces the thickness of the cooling mechanism, thereby reducing the overall thickness of the graphics card and allowing for a more compact layout. Furthermore, adjacent graphics cards share a single air delivery mechanism 600, eliminating the problem of air intake obstruction caused by adjacent graphics cards being placed too close together. This provides sufficient heat exchange air for the cooling mechanisms of both graphics cards, ensuring good heat dissipation when the two graphics cards are placed in parallel.

[0055] Both fin assembly 100 and fin assembly 200 are made of 6063 aluminum alloy and are precision stamped. The fin thickness is 0.2mm. Both fin assembly 100 and fin assembly 200 have multiple heat transfer gaps with a thickness of 1.2mm. This avoids increasing airflow resistance due to excessively small gaps while ensuring an effective heat exchange area per unit volume. The overall dimensions of fin assembly 100 and fin assembly 200 are: length 300mm, height 120mm, and thickness 22mm. Compared to traditional graphics card cooling modules, the thickness is reduced by 10-35%. Fin group 100 and fin group 200 each correspond to one graphics card. Fin group 100 and fin group 200 respectively contact and adhere to the heat-generating components such as the GPU, memory and power supply module on the graphics card through heat transfer bases. The heat transfer base can be a heat spreader or a copper plate. The heat transfer base conducts the heat from the heat-generating components such as the GPU, memory and power supply module to fin group 100 or fin group 200, and increases the heat transfer area through fin group 100 and fin group 200.

[0056] It is worth noting that the heat transfer gaps in fin group 100 and fin group 200 are perpendicular to the graphics card interface panel, meaning that the hot air transferred in the heat transfer gaps is eventually exhausted through the heat dissipation vents on the interface panel.

[0057] Please see Figures 3 to 5 The ends of fin group 100 and fin group 200 are respectively provided with a set of laminar flow plates 300. The two sets of laminar flow plates 300 are arranged in an alternating manner inside the flow guide shroud 400, and the intersection of the two sets of laminar flow plates 300 forms a laminar flow gap 310.

[0058] The laminar flow plate 300 is used to divert and guide the air curtain inside the input guide shroud 400. It can disperse the airflow output by the air delivery mechanism 600, ensuring that airflow passes through each laminar flow gap 310. The parallel arrangement of the laminar flow plates 300 guides the airflow, thereby forming an air curtain that is transmitted in parallel in the laminar flow gap 310. This avoids the airflow output by the air delivery mechanism 600 from carrying eddies and generating turbulent noise, and reduces the overall wind noise of the heat dissipation module.

[0059] It is worth noting that the laminar flow gap 310 and the heat transfer gap are distributed in parallel, so that the air curtain flowing in the laminar flow gap 310 can still be transmitted in parallel to the heat transfer gap, avoiding the air curtain from hitting the fins and generating eddies, and also preventing the generation of air turbulence noise.

[0060] The spacing of the laminar flow gap 310 is larger than the spacing of the heat transfer gap. The air curtain output from the laminar flow gap 310 can be input into multiple heat transfer gaps. On the one hand, this reduces the density of the laminar flow gap 310 and prevents excessive wind resistance. On the other hand, it ensures the uniformity of airflow in each heat transfer gap and guarantees comprehensive heat dissipation.

[0061] Please see Figure 5 The inner wall of the flow guide 400 is attached to the outer edge of the laminar flow plate 300. The inner wall of the flow guide 400 at the connection between the laminar flow plate 300 and the fin group 100 and the fin group 200 is arc-shaped to avoid the air curtain from hitting the fins and generating eddies, thus preventing the generation of air turbulence noise.

[0062] Please see Figures 4 to 7 The intersection of the two sets of laminar flow plates 300 and the inner wall of the flow guide shroud 400 form a flow diversion zone 320. The flow diversion zone 320 is used to divert the air curtain transmitted in the laminar flow gap 310. After the air curtain output by the gas delivery mechanism 600 enters the laminar flow gap 310, it is diverted through the flow diversion zone 320 to form branch air curtains that enter the heat transfer gaps in fin group one 100 and fin group two 200 respectively.

[0063] Please see Figure 7When the branch air curtain passes through the connection between the laminar flow plate 300 and the fin group 100 and fin group 200, it forms a heat transfer air curtain parallel to the heat transfer gap. The heat transfer air curtain is transmitted in the heat transfer gap and comes into contact with the surfaces of the fin group 100 and fin group 200 to exchange heat. It carries away the heat conducted by the graphics card heating element on the fin group 100 and fin group 200 and finally exhausts it through the graphics card interface panel with heat dissipation window, thus achieving the effect of cooling the graphics card.

[0064] It is worth noting that, please refer to Figure 9 and Figure 10 A movable flow divider 500 is provided in the flow divider zone 320. The edge of the flow divider 500 guides the airflow and the edge of the laminar flow plate 300 form a variable airflow adjustment gap. The movement of the flow divider 500 causes the airflow adjustment gap to change, and the change in the airflow adjustment gap causes the airflow ratio of the two branch air curtains to change.

[0065] The splitter plate 500 has a drive mechanism inside that drives its movement. The drive mechanism includes a drive motor 510. The drive motor 510 has synchronously rotating screws 511 at both ends. The drive motor 510 is fixed to the inner wall of the shroud 400. The drive motor 510 adopts a micro DC stepper motor such as SUMTOR-28HS2806A4 and is equipped with a photoelectric encoder. It can achieve rotation at a specific angle, thereby controlling a specific amount of rotation and realizing precise movement of the splitter plate 500.

[0066] Secondly, the drive motor 510 is electrically connected to the two graphics card boards respectively, and is used to make feedback adjustments based on the temperature difference between the two graphics card boards.

[0067] The manifold 500 has a threaded hole 501 that engages with the screw 511. When the drive motor 510 drives the screw 511 to rotate within the threaded hole 501, the manifold 500 moves along the axis of the screw 511. The movement of the manifold 500 causes a change in the airflow regulation gap.

[0068] The flow divider 500 also has a limiting mechanism, which includes a limiting rod 520, which is fixed to the inner wall of the flow guide 400; the flow divider 500 also has a limiting hole 502, into which the limiting rod 520 is inserted.

[0069] The limiting mechanism is used to limit the movement path of the splitter plate 500, keep the stroke at both ends of the splitter plate 500 the same, and prevent the problem of airflow difference caused by displacement deviation at both ends during the movement of the splitter plate 500.

[0070] It is worth noting that this heat dissipation device also includes a temperature detection and automatic adjustment system. Specifically, thermistor temperature sensors are embedded in fin group 100 and fin group 200 near the GPU core, respectively. The sensors have an accuracy of ±0.5℃ and can collect the core temperature data of the two graphics cards in real time. The temperature signal is transmitted to the microcontroller controller via an AD converter. The microcontroller controller can be an STM32 series microcontroller, such as the STM32F103. When the temperature difference between the two graphics cards exceeds 5℃, the drive motor 510 is automatically triggered to adjust the position of the air distribution plate 500 to change the airflow distribution ratio until the temperature of the two graphics cards tends to balance.

[0071] The specific process of coordinated heat dissipation is as follows:

[0072] When both graphics cards are running, the drive motor 620 of the air delivery mechanism 600 starts, driving the turbofan 610 to rotate at high speed. This creates negative pressure zones at both ends of the air delivery mechanism 600, drawing air from inside the chassis into the turbofan 610. Outside ambient temperature air enters the chassis due to the negative pressure or the action of the intake fan. The rotating turbofan 610 compresses the air, forming a stable air curtain. This air curtain flows along the output end of the air delivery mechanism 600 into the input end of the shroud 400. Because the inner diameters of the air delivery mechanism 600 and the shroud 400 are matched and their inner walls are smooth, there is no significant pressure loss during airflow, maintaining a stable flow rate.

[0073] After entering the flow guide shroud 400, the air curtain first flows through the laminar flow gap 310 formed by two sets of staggered laminar flow vanes 300. The staggered structure of the laminar flow vanes 300 can cut and organize the turbulent airflow, transforming it into a parallel laminar flow state. At this time, the uniformity of airflow velocity within the laminar flow gap 310 is increased to over 90%. Subsequently, the laminar flow enters the flow splitting zone 320, where, guided by the flow splitting plate 500, it forms two branch air curtains along the airflow adjustment gaps on both sides, guiding the heat transfer gaps between fin group one 100 and fin group two 200.

[0074] Please see Figure 9 and Figure 10 When the two graphics cards generate the same amount of heat, the splitter 500 is located in the middle of the splitter zone 320, and the airflow adjustment gaps on both sides are equal, both being 2.5mm. The flow ratio of the two branch air curtains is approximately 1:1, and they are evenly introduced into fin group one 100 and fin group two 200 respectively.

[0075] Please see Figure 11 and Figure 12When one of the graphics cards experiences a temperature increase due to increased load, such as graphics card one reaching a core temperature of 92℃ and graphics card two reaching 85℃, the temperature sensor transmits a signal to the controller. The controller drives motor one 510 to rotate, which in turn drives screw 511 to rotate in screw hole 501. Through the meshing relationship between screw 511 and screw hole 501, the splitter plate 500 is driven to move towards graphics card two, increasing the airflow adjustment gap corresponding to graphics card one (e.g., increasing it to 4mm) and decreasing the corresponding gap for graphics card two (e.g., decreasing it to 1mm). At this time, the branch air curtain flow to graphics card one increases by more than 30%, ensuring that the high-load graphics card receives more sufficient cooling airflow.

[0076] After the branched air curtain enters the heat transfer gap, due to the connection and guiding design between the laminar flow plate 300 and fin group 100 and fin group 200, the airflow direction remains parallel to the heat transfer gap, forming a heat transfer air curtain that closely adheres to the fin surface. Fin group 100 and fin group 200 absorb the heat conducted by the graphics card core, memory, and power supply module through the heat spreader. When the heat transfer air curtain flows through the heat transfer gap, it undergoes forced convection heat transfer with the high-temperature fins, increasing the airflow temperature. Subsequently, it is discharged along the output end of the fin group to the outside of the chassis or the server cooling duct, completing the heat dissipation cycle.

[0077] During this process, since the fin assembly is only 22mm thick, the airflow channel between the slots will not be squeezed when the two graphics cards are running in parallel, and the air intake volume is increased by 40%-60% compared with the traditional cooling solution. At the same time, a single air delivery mechanism of 600 can meet the cooling needs of the two graphics cards, reducing the number of fans by 2-3 compared with the traditional "one card, two fans" design, and reducing the power consumption from 30-50W to 12W, reducing energy consumption by more than 70%.

[0078] Performance testing and effect verification:

[0079] Two RTX 4090 graphics cards were selected to build a dual-graphics card test platform. The device described in this embodiment and a traditional dual-fan independent cooling module were compared and tested. The ambient temperature was 25℃, and the test load was AI model training, running at full load. The test results are shown in the table below:

[0080]

[0081] As can be seen from the above test results, the dual-graphics card collaborative heat dissipation device in this embodiment significantly improves heat dissipation efficiency and temperature balance while controlling the thickness and power consumption of the heat dissipation module. It effectively solves the contradiction of "high-efficiency heat dissipation - compact size - low power consumption" in traditional solutions and can fully meet the heat dissipation requirements of dual graphics cards in parallel in high-end computing scenarios.

[0082] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A dual-graphics card collaborative heat dissipation device, comprising a first fin assembly (100), a second fin assembly (200), a shroud (400), and an air delivery mechanism (600), wherein both the first fin assembly (100) and the second fin assembly (200) have multiple heat transfer gaps, characterized in that: The ends of the first fin group (100) and the second fin group (200) are respectively provided with a set of laminar flow plates (300). The two sets of laminar flow plates (300) are arranged in an alternating manner inside the flow guide (400), and the intersection of the two sets of laminar flow plates (300) forms a laminar flow gap (310). The intersection of the two sets of laminar flow plates (300) and the inner wall of the flow guide (400) form a flow diversion zone (320). After the gas curtain output by the gas conveying mechanism (600) enters the laminar flow gap (310), it is divided by the flow splitting zone (320) to form branched gas curtains that enter the heat transfer gaps in fin group one (100) and fin group two (200) respectively. The diversion zone (320) is provided with a movable diversion plate (500). The edge of the diversion plate (500) guiding the airflow and the edge of the laminar flow plate (300) form a variable airflow adjustment gap. The movement of the diversion plate (500) causes the airflow adjustment gap to change, and the change of the airflow adjustment gap causes the airflow ratio of the two branch air curtains to change.

2. The dual-graphics card collaborative heat dissipation device according to claim 1, characterized in that: The laminar flow gap (310) and the heat transfer gap are distributed in parallel.

3. The dual-graphics card collaborative heat dissipation device according to claim 2, characterized in that: The spacing of the laminar flow gap (310) is greater than the spacing of the heat transfer gap.

4. The dual-graphics card collaborative heat dissipation device according to claim 3, characterized in that: The branched air curtain forms a heat transfer air curtain parallel to the heat transfer gap when it passes through the connection between the laminar flow plate (300) and the first fin group (100) and the second fin group (200).

5. The dual-graphics card collaborative heat dissipation device according to claim 1, characterized in that: The diverter plate (500) has a drive mechanism inside that drives it to move. The drive mechanism includes a drive motor (510), and the two ends of the drive motor (510) have synchronously rotating screws (511). The drive motor (510) is fixed to the inner wall of the shroud (400). The diverter plate (500) has a threaded hole (501) that engages with the screw (511). When the drive motor (510) drives the screw (511) to rotate in the threaded hole (501), the diverter plate (500) moves along the axis of the screw (511).

6. The dual-graphics card collaborative heat dissipation device according to claim 5, characterized in that: The diverter plate (500) also has a limiting mechanism, which includes a limiting rod (520) and the limiting rod (520) is fixed to the inner wall of the flow guide (400); The diverter plate (500) also has a limiting hole (502), and the limiting rod (520) is inserted into the limiting hole (502).

7. The dual-graphics card collaborative heat dissipation device according to claim 1, characterized in that: The gas delivery mechanism (600) has a turbo fan (610) and a second drive motor (620) inside. The air curtain output by the gas delivery mechanism (600) is generated by the second drive motor (620) driving the turbo fan (610) to rotate. The inner diameter of the output end of the gas delivery mechanism (600) is the same as the inner diameter of the input end of the flow guide (400).

8. The dual-graphics card collaborative heat dissipation device according to claim 1, characterized in that: The inner wall of the flow guide (400) is attached to the outer edge of the laminar flow plate (300).

9. A dual-graphics card collaborative heat dissipation device according to claim 8, characterized in that: The inner wall of the flow guide shroud (400) at the connection between the laminar flow plate (300) and fin group one (100) and fin group two (200) is arc-shaped.

Images