Thermoelectrically separated printed circuit board power management method, system, circuit board and computer device

Abstract

The application relates to a thermoelectric separation printed circuit board power supply management method, system, circuit board and computer equipment, wherein a temperature sensor array is arranged in the base of a thermoelectric separation printed circuit board, temperature distribution data in the physical base is acquired, real-time power loss and temperature spatial gradient are calculated according to the collected temperature distribution data and output currents of each phase, and a thermal diffusion parameter is introduced to predict the predicted temperature distribution data in a future time window. According to the predicted temperature distribution data, the loads of phases whose predicted temperatures exceed a threshold value are proportionally transferred to phases whose temperatures do not reach the threshold value by changing the driving signals of power devices. In this way, the load redistribution can be completed before the temperature rise exceeds the operation limit, the local hot spot effect caused by high power density is inhibited, the working junction temperature of the semiconductor device is controlled within the rated range, the uneven spatial thermal load distribution is avoided, and the stability and safety of the system are maintained.

Description

Technical Field

[0001] This application relates to the field of power supply management technology for printed circuit boards, and in particular to a method, system, circuit board, and computer device for power supply management of thermoelectrically separated printed circuit boards. Background Technology

[0002] With the increase in power density of power electronic devices, the heat flux per unit area of ​​printed circuit boards (PCBs) has increased significantly. In applications such as inverters for new energy vehicles, high-performance computing modules, and high-power power converters, multi-phase power supply architectures are typically used to handle high-current loads.

[0003] Current printed circuit board (PCB) thermal management technologies are mainly divided into two dimensions: passive heat dissipation and logic control. In terms of physical structure, metal-based insulated printed circuit boards (IMS) or direct thermal path (DTP) technologies are commonly used, transferring heat to the heat dissipation interface by embedding a highly thermally conductive metal substrate (such as a copper block) under the power device. In terms of control logic, the power management system typically distributes current across each phase according to preset electrical parameters. However, when facing high-power dynamic loads, this control method, which relies solely on electrical signals and subsequent single-point temperature feedback, suffers from severe physical hysteresis. It cannot proactively intervene before heat accumulation occurs, ultimately resulting in uneven spatial heat load distribution in multi-phase power supply systems during operation. This easily leads to localized hot spot effects and increases the risk of thermal runaway in semiconductor devices. Summary of the Invention

[0004] In view of this, in order to at least partially improve the above-mentioned problems, this application provides a power supply management method, system, circuit board, and computer device for thermoelectric separation printed circuit boards.

[0005] This application first provides a power supply management method for a thermoelectric separation printed circuit board. The printed circuit board includes multiple power supply phases, each power supply phase is provided with a power device, and the printed circuit board also has a base. A temperature sensor array composed of multiple temperature sensors is embedded in the base. The method includes: The actual temperature inside the base is obtained from each of the temperature sensors, and the real-time output current of the power devices in each power supply phase is obtained simultaneously. The real-time temperature spatial gradient of the base is determined based on the actual temperatures described above. Based on the actual temperature and the real-time output current of each power device, the real-time power loss of each power device is determined. Based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, the predicted temperature distribution data of the base within a preset future time window is predicted, wherein the predicted temperature distribution data includes the predicted temperature value of each power supply phase at the corresponding position on the base. Based on the predicted temperature distribution data, the power allocation weight of each power supply phase is dynamically adjusted by regulating the drive signals of each power device, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold.

[0006] In one embodiment, the base is a copper base embedded in the printed circuit board, and the power device has a metal heat dissipation pad at its bottom; the printed circuit board also includes an insulating dielectric layer surrounding the base, the insulating dielectric layer being formed by laminating a polymer resin containing aluminum nitride particles; The step of obtaining the actual temperature inside the base collected by each of the temperature sensors includes: The actual temperature generated by the operation of each power device and directly conducted to the top of the copper base via the bottom metal heat dissipation pad is obtained.

[0007] In one embodiment, the temperature sensor is a laser-induced graphene temperature sensor; The step of obtaining the actual temperature inside the base collected by each of the temperature sensors includes: The actual temperature inside the base is obtained by the laser-induced graphene temperature sensor disposed in the stepped pressure-reducing blind groove on the side wall of the base; wherein, the pressure-reducing blind groove is filled with a polydimethylsiloxane flexible buffer layer covering the surface of the laser-induced graphene temperature sensor, so as to protect the porous structure of the laser-induced graphene temperature sensor by hydrostatic pressure during the pressing process.

[0008] In one embodiment, determining the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device includes: Based on the actual temperature of each power device, the real-time equivalent impedance corresponding to each power supply phase is dynamically updated. Based on the real-time output current and the real-time equivalent impedance, calculate the real-time power loss of each power device at the current time.

[0009] In one embodiment, predicting the predicted temperature distribution data of the base within a preset future time window based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board includes: The actual temperatures described are integrated into a spatially discrete temperature matrix; A real-time power loss matrix is ​​constructed based on the real-time power loss of each power device. Extract the preset thermal diffusion parameters of the printed circuit board, wherein the thermal diffusion parameters include spatial thermal impedance and thermal diffusion characteristic matrix, a first adjustment coefficient and a second adjustment coefficient; Based on the actual temperature matrix, real-time power loss matrix, real-time temperature spatial gradient, and preset thermal diffusion parameters of the printed circuit board, the predicted temperature distribution data of the base within a preset future time window is predicted.

[0010] In one embodiment, determining the real-time temperature spatial gradient of the base based on each of the actual temperatures includes: Acquire the spatial physical coordinate data of each temperature sensor in the temperature sensor array within the base; The actual temperatures are correlated and mapped with the corresponding spatial physical coordinate data to construct a discrete spatial temperature field matrix that characterizes the heat distribution state inside the base at the current moment. Using the spatial finite difference algorithm, based on the temperature difference between adjacent temperature sensors and the physical distance in the corresponding spatial physical coordinates, the approximate values ​​of the partial derivatives of the discrete spatial temperature field matrix in each spatial dimension are calculated respectively. The approximate values ​​of the partial derivatives in each spatial dimension are synthesized as gradient components to obtain the real-time temperature spatial gradient.

[0011] In one embodiment, based on the predicted temperature distribution data, the power allocation weight of each power supply phase is dynamically adjusted by regulating the drive signals of each power device, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold, including: Extract the predicted temperature value corresponding to each power supply phase from the predicted temperature distribution data, and calculate the maximum predicted temperature difference between any two predicted temperature values ​​of the power supply phases; If the maximum predicted temperature difference is greater than the preset stability threshold, calculate the average predicted temperature of all the power supply phases, and calculate the dynamic power allocation weight of each power supply phase based on the average predicted temperature. The dynamic power allocation weights are converted into pulse width modulation duty cycles for controlling the on and off of the power devices through a preset mapping function, so as to generate the adjusted drive signal. The adjusted drive signal is synchronously applied to each of the power devices so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold.

[0012] This application also provides a power supply management system for a thermoelectric separation printed circuit board, wherein the printed circuit board includes multiple power supply phases, each power supply phase is provided with a power device, the printed circuit board is further provided with a base, and a temperature sensor array composed of multiple temperature sensors is embedded in the base, the system is used to implement the method described in any of the above claims, the system comprising: The data acquisition module is used to acquire the actual temperature inside the base collected by each of the temperature sensors, and simultaneously acquire the real-time output current of the power devices in each of the power supply phases. The gradient determination module is used to determine the real-time temperature spatial gradient of the base based on the actual temperatures. The loss determination module is used to determine the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device. The temperature prediction module is used to predict the predicted temperature distribution data of the base within a preset future time window based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board. The predicted temperature distribution data includes the predicted temperature value of each power supply phase at the corresponding position on the base. The power adjustment module is used to dynamically adjust the power allocation weight of each power supply phase by adjusting the drive signal of each power device according to the predicted temperature distribution data, so that the difference between the predicted temperature values ​​of any two power supply phases is kept within a preset stable threshold.

[0013] This application also provides a thermoelectric separation printed circuit board, including multiple power supply phases, an insulating dielectric layer and a base. Each power supply phase is provided with a power device. The base is a copper base embedded in the printed circuit board. The insulating dielectric layer surrounds the base and is formed by pressing a polymer resin containing aluminum nitride particles. The power device has a metal heat dissipation pad at its bottom, and the metal heat dissipation pad is metallurgically connected to the top of the copper base; The sidewall of the copper base is provided with a stepped pressure-reducing blind groove, and a laser-induced graphene temperature sensor is provided in the pressure-reducing blind groove. The pressure-reducing blind groove is filled with a polydimethylsiloxane flexible buffer layer covering the surface of the laser-induced graphene temperature sensor.

[0014] This application also provides a computer device including a memory and a processor, the memory storing a computer application program, and the processor executing the computer application program, wherein the computer application program is configured to perform the method as described in any of the preceding claims when executed by the processor.

[0015] The beneficial effects of this application are as follows: This application deploys a temperature sensor array within the base of a thermoelectrically separated printed circuit board to acquire temperature distribution data within the physical base. Based on the acquired temperature distribution data and the output current of each phase, real-time power loss and temperature spatial gradient are calculated, and thermal diffusion parameters are introduced to predict the temperature distribution data within a future time window. According to the predicted temperature distribution data, by changing the drive signal of the power devices, the load of phases with predicted temperatures exceeding a threshold is proportionally transferred to phases with temperatures below the threshold. This allows for load redistribution before the temperature rise exceeds the operating limit, suppressing the local hot spot effect caused by high power density, controlling the operating junction temperature of semiconductor devices within the rated range, avoiding uneven spatial heat load distribution, and maintaining the stability and safety of the system. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating a power supply management method for a thermoelectrically separated printed circuit board according to an embodiment of this application.

[0017] Figure 2 This is a cross-sectional view of a thermoelectrically separated printed circuit board according to an embodiment of the power supply management method of a thermoelectrically separated printed circuit board of this application.

[0018] Figure 3 This is a top-view structural diagram of the copper base and temperature sensor array in a thermoelectrically separated printed circuit board according to an embodiment of the power supply management method of the thermoelectrically separated printed circuit board of this application.

[0019] Figure 4 This is a schematic diagram of the structure of a thermoelectric separation printed circuit board power supply management system according to another embodiment of this application.

[0020] Figure 5 This is a block diagram of a computer device according to an embodiment of this application. Detailed Implementation

[0021] The terms "first," "second," and "third," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the term "comprising," and any variations thereof, is intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or modules is not limited to the listed steps or modules, but may optionally include steps or modules not listed, or may optionally include other steps or modules inherent to these processes, methods, products, or apparatuses.

[0022] Please see Figure 1-5 This application provides a power supply management method for thermoelectrically separated printed circuit boards, which can be applied to the fields of new energy vehicles and high-performance computing, such as the CPU / GPU core power supply module of an artificial intelligence server. The main body executing this method is a high-frequency thermoelectrically separated printed circuit board power supply management system, which includes at least: Printed circuit board: It adopts a thermoelectric separation structure, which includes a circuit layer 2 for electrical wiring and a base (such as a copper base 4) for heat dissipation. Power supply phases and power devices: Multiple power supply phases are arranged on the printed circuit board (such as a multi-phase parallel buck circuit), and each phase has at least one power device (such as a SiC MOSFET or IGBT) as the main heat source. For example... Figure 2 In the context, the first power supply phase 11 and the second power supply phase 12 of the power device; Temperature sensor array: Multiple temperature sensors are embedded in the base in a preset matrix to sense the real-time temperature at different coordinate points within the base; Current acquisition unit: used to acquire current data of power devices in each power supply phase in real time; Control unit (digital signal processor / microcontroller): electrically connected to the temperature sensor, current acquisition unit, and drive terminals of power devices, used to perform logic operations and adjust the drive signals of power devices.

[0023] The method includes: S1. Obtain the actual temperature inside the base collected by each of the temperature sensors and simultaneously obtain the real-time output current of the power devices in each of the power supply phases; S2. Determine the real-time temperature spatial gradient of the base based on the actual temperatures described above; S3. Determine the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device. S4. Based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, predict the predicted temperature distribution data of the base within a preset future time window, wherein the predicted temperature distribution data includes the predicted temperature value of each power supply phase at the corresponding position on the base. S5. Based on the predicted temperature distribution data, the power allocation weight of each power supply phase is dynamically adjusted by adjusting the drive signal of each power device, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold.

[0024] As described above, the control unit synchronously acquires real-time temperatures from temperature sensors embedded in the base, and simultaneously acquires the real-time output current of power devices in each power supply phase through a current acquisition unit. The real-time temperature characterizes the heat dissipation distribution within the base, and the output current characterizes the real-time load intensity of each electrical phase. The control unit determines the real-time temperature spatial gradient of the base based on the actual temperatures, thus understanding the rate and direction of heat diffusion from the area below the power devices to the surrounding areas within the base. Using the real-time temperatures and pre-stored internal resistance characteristics of the power devices (such as a curve or function of on-resistance versus temperature), the control unit calculates the real-time internal resistance value of the power devices at the current temperature. Subsequently, combined with the real-time output current, it calculates the real-time power loss of each power device. Based on the actual temperatures, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, the control unit predicts the predicted temperature distribution data of the base within a preset future time window. Finally, the control unit performs real-time comparisons with the predicted temperature values ​​of each power supply phase. If a deviation is detected between the predicted temperature values ​​corresponding to different power supply phases, the control unit dynamically adjusts the power distribution weight of each power supply phase by adjusting the drive signals of each power device (such as changing the duty cycle of the PWM signal). For example, it reduces the load proportion borne by the phase with the higher predicted temperature and correspondingly increases the load proportion borne by the phase with the lower predicted temperature. This ensures that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold, thereby ensuring that the thermal load of each phase of the entire printed circuit board is in a long-term balanced state.

[0025] In one embodiment, see Figure 2 This is a cross-sectional view of the thermoelectric separation printed circuit board in this application.

[0026] In this embodiment, the printed circuit board adopts a highly integrated thermoelectric separation physical architecture, and the bottom support of the printed circuit board, which serves as the main heat dissipation channel, is a copper base 4. To achieve an unobstructed direct thermal path (DTP), the top of the copper base 4 is integrally formed with an upwardly extending thermally conductive protrusion, which physically penetrates the insulating dielectric layer 3 in the vertical direction. Above the thermally conductive protrusion of the copper base 4, a first power supply phase 11 and a second power supply phase 12 for the power device are distributed. The power devices of each power supply phase adopt a thermoelectric separation bridging design. The large area of ​​metal heat dissipation pad at the bottom center is directly metallurgically connected to the top of the heat-conducting boss of the copper base 4 using high-temperature solder. The thermal conductivity of copper is about 398 W / mK, which is more than 100 times that of ordinary aluminum substrate. Compared with the traditional transition through insulating dielectric layer 3, this embodiment constructs a pure metal heat conduction channel in the Z-axis direction. At the same time, the electrical pins on both sides of the power device extend downward across the heat-conducting boss and precisely connect to the circuit layer 2 located on the periphery of the boss to independently carry control signals and electrical circuits.

[0027] The insulating dielectric layer 3 is disposed around the sidewalls of the thermally conductive protrusions of the copper base 4 and directly below the circuit layer 2. This not only provides electrical insulation support for the circuit layer 2 but also ensures that the insulating resin with a heat-insulating effect is not located in the vertical core heat dissipation path at the bottom of the device, thus achieving physical and thermodynamic thermoelectric separation. Therefore, rapid vertical heat dissipation is possible; that is, the actual temperature generated by each power device during operation is directly conducted to the top of the copper base via the bottom metal heat dissipation pads. Through this direct heat conduction path without insulating resin obstruction, the high-density heat flow generated by the power devices during operation can be conducted vertically without loss to the base, significantly reducing the overall thermal resistance from the junction to the environment, thereby improving the overall heat dissipation efficiency of the system.

[0028] Furthermore, to ensure electrical safety while achieving rapid heat conduction, the insulating dielectric layer 3 is formed by laminating a polymer resin containing a specific proportion of aluminum nitride particles. This highly thermally conductive insulating dielectric layer 3 physically separates the copper base 4, which bears a high current thermal load, from the low-power signal path that carries control signals; by utilizing the thermally conductive but non-conductive properties of aluminum nitride, it ensures that sensitive signal circuits are not directly affected by the high temperature of the power path.

[0029] In one embodiment, the temperature sensor is a laser-induced graphene temperature sensor 5, which is disposed in a stepped pressure-reducing blind groove 6 on the sidewall of the base. The pressure-reducing blind groove 6 is filled with a polydimethylsiloxane flexible buffer layer covering the surface of the laser-induced graphene temperature sensor 5, which is used to protect the porous structure of the laser-induced graphene temperature sensor 5 through hydrostatic pressure during the lamination of the printed circuit board.

[0030] As described above, the temperature sensor array consists of multiple laser-induced graphene temperature sensors 5. In order to obtain the most accurate heat distribution data inside the base and protect the precision sensing structure, a decompression blind groove 6 with a microscopic stepped shape is processed on the side wall of the base, and the laser-induced graphene temperature sensor 5 is generated in situ on the inner wall of the decompression blind groove 6 using a carbon dioxide laser.

[0031] In the high-temperature, high-pressure lamination process of multilayer printed circuit boards (pressures typically reach 300 psi to 500 psi), the micro / nano sensor structure is easily damaged. Therefore, in this embodiment, a polydimethylsiloxane flexible buffer layer is dispensing and potting into the pressure-reducing blind groove 6, covering the surface of the laser-induced graphene temperature sensor 5. During lamination, the polydimethylsiloxane flexible buffer layer acts as an incompressible fluid support, thereby converting the vertical rigid mechanical compressive stress from the press surface into isotropic hydrostatic pressure. This hydrostatic pressure acts uniformly on the porous graphene framework, thus effectively protecting the thermistor characteristics and the integrity of the conductive path of the sensor. The pins of the laser-induced graphene temperature sensor 5 penetrate this buffer layer and are electrically connected to an independent signal acquisition plane inside the printed circuit board via laser-drilled blind holes, ensuring that the actual temperature data can be accurately and in real-time transmitted back to the control unit.

[0032] For example, the sensor integration potting process used in this embodiment specifically includes the following steps: ① Blind groove excavation: On the side wall of the copper base 4, a pressure-reducing blind groove 6 with a microscopic stepped shape is pre-processed using a CNC machine tool or laser milling. It should be noted that the number and position of the pressure-reducing blind groove 6 correspond to the arrangement of the temperature sensor array.

[0033] ② In-situ graphene generation: Polyimide films are attached to the inner walls of each of the vacuum blind grooves 6, and laser-induced graphene temperature sensors 5 with three-dimensional porous structures are generated in situ by carbon dioxide laser irradiation.

[0034] ③ Polydimethylsiloxane Flexible Buffer Layer Encapsulation: To prevent the fragile graphene from collapsing and breaking during the printed circuit board lamination process, a low-modulus, high-thermal-conductivity polydimethylsiloxane flexible buffer layer is applied to the surface of the generated graphene. Under high temperature and pressure, the polydimethylsiloxane flexible buffer layer acts as an incompressible fluid support, uniformly converting the vertical rigid mechanical lamination stress into hydrostatic pressure, thereby effectively protecting the thermistor properties and conductive pathways of the internal porous graphene framework from damage.

[0035] ④ Pin interconnection: The pins at both ends of each graphene sensor penetrate the polydimethylsiloxane flexible buffer layer and are electrically connected to the independent signal acquisition plane on the bottom layer through laser drilling blind holes and copper plating process.

[0036] It should be noted that, Figure 2 Only a single / row of decompression blind slots 6 are shown in a two-dimensional cross-section. In the actual three-dimensional physical structure, the decompression blind slots 6 and the laser-induced graphene temperature sensors 5 inside them are distributed in a two-dimensional matrix along the horizontal direction (X-axis and Y-axis) at the bottom of the copper base 4, thus forming the hardware basis for obtaining the spatial discrete temperature matrix.

[0037] For example, please refer to Figure 3 , Figure 3 This is a top-view perspective structural diagram of the copper base 4 and temperature sensor array described in a preferred embodiment of this application. In order to accurately obtain the true and comprehensive spatial heat distribution characteristics inside the copper base 4, the pressure-reducing blind groove 6 and the laser-induced graphene temperature sensor 5 generated in situ inside it are arranged in a two-dimensional matrix pattern along the horizontal direction (i.e., the XY coordinate plane shown in the figure) inside the copper base 4, thus jointly constituting the aforementioned temperature sensor array.

[0038] In the XY plane coordinate system, each of the laser-induced graphene temperature sensors 5 in the matrix array has unique corresponding spatial physical coordinate data (e.g., coordinate node X). i ,Y j This discrete, gridded physical node layout allows the system to simultaneously acquire the actual temperature at different horizontal coordinate positions inside the copper base 4 at any given time cross-section. The control unit can be based on... Figure 3 The coordinates of each sensor node shown are used to map the collected actual temperatures to their corresponding spatial physical coordinates, thereby constructing a discrete spatial temperature field matrix that characterizes the heat distribution at the current moment. Since there is a fixed physical distance between adjacent sensors, physical sampling points can be provided to calculate approximate values ​​of the partial derivatives of the discrete spatial temperature field matrix in each spatial dimension, ensuring that the synthesized real-time temperature spatial gradient has high spatial resolution and accuracy.

[0039] In one embodiment, step S3, which determines the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device, includes: S301, Preloaded space thermal impedance and thermal diffusion characteristic matrix; As described above, the control unit is preloaded with spatial thermal impedance and thermal diffusion characteristic matrices between each phase extracted through three-dimensional finite element analysis (FEA). These characteristic matrices characterize the physical topological constraints and thermal conduction coupling relationships between the base, power devices, and conductive networks in a thermoelectric separation structure.

[0040] It should be noted that the extraction and preloading process of the spatial thermal impedance and thermal diffusion characteristic matrix specifically includes: (1) Offline thermodynamic modeling: During the equipment manufacturing or R&D stage, a three-dimensional finite element (FEA) digital twin model of the thermoelectric separation printed circuit board is established using an external computer; the model pre-sets the actual physical dimensions and intrinsic thermal conductivity parameters of the copper base 4, the polymer resin containing aluminum nitride particles, and the laser-induced graphene sensor.

[0041] (2) Matrix order reduction extraction: In the three-dimensional finite element model, a unit step power excitation is applied to each power supply phase in sequence, and the steady-state temperature response and transient thermal impedance curves corresponding to the remaining phases are extracted by simulation calculation; using the model order reduction algorithm, the complex partial differential heat conduction equations are simplified into linear constant matrices that are only related to the coordinates of each phase node, namely the spatial thermal impedance and thermal diffusion characteristic matrix.

[0042] (3) Online firmware loading: The extracted thermal diffusion feature matrix is ​​used as a static firmware constant and preloaded into the non-volatile memory (such as Flash or EEPROM) of the control unit. During the actual operation control cycle of the device, the control unit does not need to perform high-computation-power three-dimensional partial differential equation solving, but only needs to call the matrix and perform low-computation-power matrix multiplication and addition operations with real-time power loss, thereby meeting the microsecond-level ultra-high frequency active thermal equalization scheduling requirements of the power supply management system.

[0043] S302. Based on the actual temperature of each power device, dynamically update the real-time equivalent impedance corresponding to each power supply phase.

[0044] Specifically, the control unit retrieves pre-stored multi-material reference electrical parameters from its local memory. These reference electrical parameters include: the nominal internal resistance of each power device at a standard reference temperature, the nonlinear temperature coefficient matrix of the semiconductor material, and the reference line impedance and linear temperature impedance coefficient of the pure metal for each phase circuit layer 2 wiring at the standard reference temperature. The control unit then extracts the actual temperature T corresponding to the current phase. kAnd by combining the nonlinear temperature coefficient matrix of the semiconductor material, the real-time internal resistance R of the power device is calculated. DS(on),k (T). Since the semiconductor carrier mobility decreases nonlinearly with increasing temperature, the control unit uses a preset second-order polynomial approximation function (or a three-dimensional high-precision lookup table interpolation algorithm) to solve for it. An example of the calculation formula is as follows: R DS(on),k (T)=R DS(on),25 ×[1+γ1×(T k -25)+γ2×(T k -25) 2 ]; Among them, R DS(on),25 The nominal internal resistance is γ1 and γ2, which are the first and second temperature coefficients of the semiconductor material inside the power device, respectively.

[0045] Then, the control unit calculates the actual temperature T extracted from the data. k Simultaneously calculate the real-time impedance R of the circuit layer 2 wiring corresponding to the k-th phase. trace,k (T). Based on the linear expansion and lattice scattering characteristics of the metal conductor (such as copper foil layer) in circuit layer 2, a linear temperature compensation model is used for solution, and the calculation formula is as follows: R trace,k (T) = R trace,25 ×[1+α metal ×(T k -25)]; Among them, R trace,25 The baseline impedance for routing on circuit layer 2; α metal The linear temperature impedance coefficient of the wiring metal material for this circuit layer 2.

[0046] The control unit, based on the circuit topology corresponding to the power supply phase, determines the real-time internal resistance R of the power device. DS(on),k (T) and the real-time impedance R of the wiring in the circuit layer 2 trace,k (T) Perform equivalent superposition to generate accurate real-time equivalent impedance R for each phase under the current thermal field state. eq (T): R eq,k (T) =R DS(on),k (T)+R trace,k (T); The real-time equivalent impedance R eq,k (T) is then used to calculate the real-time power loss in step S303, thereby eliminating the loss calculation error caused by treating the impedance as a constant in the traditional algorithm.

[0047] S303, the control unit calculates the real-time power loss of the power device and associated wiring in the k-th phase at the current time based on the real-time output current and the real-time equivalent impedance. The calculation formula is as follows: ; in, P is the square of the real-time output current of the k-th phase. sw,k The preset switching loss component for the k-th phase.

[0048] In one embodiment, step S4, which predicts the predicted temperature distribution data of the base within a preset future time window based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, includes: S401. Integrate the actual temperatures into a spatial discrete temperature matrix; simultaneously, the control unit arranges the real-time power loss scalars of each power supply phase calculated in step S303 according to the physical spatial topology order to construct a real-time power loss matrix P characterizing the global heat intensity. loss ; S402. Obtain the real-time temperature spatial gradient; The real-time temperature spatial gradient is generated as follows: First, the control unit acquires the three-dimensional or two-dimensional spatial physical coordinate data of each temperature sensor in the temperature sensor array within the base. Since the temperature sensors are generated in situ within the pressure-reducing blind groove 6 on the side wall of the base, this spatial physical coordinate data precisely corresponds to the actual thermal sampling nodes inside the base. The control unit maps each actual temperature to its corresponding spatial physical coordinate data one-to-one, thereby constructing a discrete spatial temperature field matrix representing the current heat distribution state inside the base. The control unit uses a spatial finite difference algorithm to calculate the approximate values ​​of the partial derivatives of the discrete spatial temperature field matrix in each spatial dimension (such as the X-axis and Y-axis directions) based on the temperature difference between adjacent temperature sensors and their absolute distance in the spatial physical coordinates. These approximate values ​​of the partial derivatives are then used as the gradient components of the real-time temperature spatial gradient, thereby synthesizing the real-time temperature spatial gradient.

[0049] Specifically, taking the gradient component in the X-axis direction as an example, the calculation formula is as follows: ; in, T is the partial derivative of temperature with respect to position along the X-axis; i+1,j,k With T i,j,kΔx represents the actual temperature collected by the temperature sensor at two adjacent spatial nodes (node ​​i and node i+1) in the X-axis direction within the spatial physical coordinate grid of the base, where the Y-axis (j) and Z-axis (k) coordinate levels are the same; Δx represents the absolute spatial physical distance between the two adjacent nodes. The control unit obtains the real-time temperature spatial gradient by integrating approximate values ​​of partial derivatives in each spatial dimension. This real-time temperature spatial gradient has vector properties, pointing in the direction of the fastest temperature increase, and its magnitude characterizes the transient rate of heat diffusion from the region below the power device to the surrounding region within the base. Through this real-time temperature spatial gradient, the system can quantify the dynamic characteristics of heat diffusion within the base. Furthermore, since the sensor array is embedded within the pressure-reducing blind slot 6, the real-time temperature spatial gradient calculated by the control unit represents the true three-dimensional voxel heat flux gradient within the copper base 4, rather than the traditional two-dimensional thermal gradient on the PCB surface.

[0050] S403. Extract the preset thermal diffusion parameters of the printed circuit board; As described above, the thermal diffusion parameters include spatial thermal impedance and thermal diffusion characteristic matrix, a first adjustment coefficient, and a second adjustment coefficient. The thermal diffusion characteristic matrix is ​​a physical constant unique to the copper base 4-aluminum nitride dielectric layer composite architecture, containing the lower vertical thermal impedance parameter of the copper base 4 and the high in-plane thermal conductivity determined by the polymer resin containing aluminum nitride particles. This allows the algorithm to correspond to the physical environment of the thermoelectrically separated printed circuit board of this application at the underlying algorithm level. The adjustment coefficient is used to characterize the discrete mapping ratio between material heat capacity and time step.

[0051] S404. Based on the actual temperature matrix, real-time power loss matrix, real-time temperature spatial gradient, and preset thermal diffusion parameters of the printed circuit board, predict the predicted temperature distribution data of the base within a preset future time window, wherein the calculation formula is: ; Among them, T predict This refers to the predicted temperature distribution data of the base after a preset future time window; T actual C represents the spatially discrete temperature matrix at the current moment, where Δt is the preset time step. thermal P is the spatial thermal impedance and thermal diffusion characteristic matrix; loss This is the real-time power loss matrix for each of the power devices. The discrete temperature field Laplace operator matrix is ​​derived from the real-time temperature spatial gradient, representing the spatial heat flow divergence rate; α is a preset first adjustment coefficient, representing the conversion gain of energy loss to temperature rise, with dimensions in K·W. -1 ·s -1β is a preset second adjustment coefficient, characterizing the intrinsic thermal diffusivity of the copper base material, with dimensions in m. 2 ·s -1 Because the thermal time constant delay caused by the traditional FR-4 dielectric layer is eliminated, the system can perform the above calculations based on a very small prediction step size, ensuring that the predicted temperature distribution data matches the extremely fast thermal conductivity of the underlying physical structure. Thus, the predicted temperature distribution can accurately reflect the actual temperature distribution of the base.

[0052] In one embodiment, step S5, which involves dynamically adjusting the power allocation weight of each power supply phase by adjusting the drive signals of each power device based on the predicted temperature distribution data, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold, includes: S501. Based on the predicted temperature distribution data, extract the predicted temperature value T corresponding to the spatial physical coordinates of each power supply phase. predict,k (k is the sequence number of the power supply phase), the control unit calculates the absolute difference between any two predicted temperature values ​​of the power supply phase in real time, and extracts the maximum predicted temperature difference ΔT. max Subsequently, the maximum predicted temperature difference ΔT max Compared with the preset stability threshold T in the memory th Compare them, if ΔT max >T th This will trigger the dynamic balancing mode.

[0053] S502. In dynamic balancing mode, the control unit calculates the new power allocation weight W for each power supply phase based on the principle of total load current conservation and using a preset negative feedback balancing algorithm. k Specifically, for power supply phases with higher predicted temperature values, the system will proportionally reduce their allocation weight; for power supply phases with lower predicted temperature values, the system will correspondingly increase their allocation weight. An example of the calculation formula is as follows: ; Among them, W k The dynamic power allocation weight for the k-th power supply phase; W nom K represents the average weight of each phase under the current total load demand. balance This is the preset closed-loop thermal equilibrium proportional gain coefficient. The predicted temperature is the average value for all power supply phases. Simultaneously, the control unit enforces a constraint that the sum of the power allocation weights for all power supply phases is always equal to 1, ensuring that the total electrical output power meets the system load requirements without any drop.

[0054] S503, Map the power allocation weights to the underlying drive signals, and the control unit will calculate the power allocation weights W.k This is converted into physical electrical signals that specifically control the power devices to turn on and off.

[0055] For example, the control unit uses a pulse width modulation module to adjust the power allocation weight W. k The linear or nonlinear mapping is represented by the duty cycle D of each power device drive terminal (such as the gate of a MOSFET). k : W k =f map (W k ); Among them, f map This represents a mapping function.

[0056] S504 The control unit synchronously applies the adjusted drive signal to each of the power devices.

[0057] Since the copper base 4 of this application is a direct thermal path structure, the changes in transient heating power (i.e., real-time power loss) of each phase caused by changes in the driving signal can be directly reflected in the thermal conduction network of the base with a low thermal time constant. Compared with traditional PCBs with insulating resin barriers, this step can reshape the three-dimensional spatial distribution of the heat source in advance using algorithms before the actual physical high temperature (local hot spot) is formed. This allows the heat flow to be pre-dissipated and neutralized within the copper base 4, thereby ensuring that the difference between the predicted temperature values ​​of any two power supply phases is quickly converged and maintained within the preset stable threshold, achieving long-term thermoelectric steady-state operation of the power supply system.

[0058] It is understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0059] Based on the same inventive concept, this application also provides a system for implementing the above-described thermoelectric separation printed circuit board power supply management method. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations in one or more system embodiments provided below can be found in the limitations of the method described above, and will not be repeated here.

[0060] In one exemplary embodiment, such as Figure 4 As shown, a power supply management system for a thermoelectrically separated printed circuit board is provided, including: a data acquisition module 61, a gradient determination module 62, a loss determination module 63, a temperature prediction module 64, and a power adjustment module 65.

[0061] The above system corresponds to the aforementioned method embodiments and can implement the corresponding method steps. Its implementation principle and technical effect are similar, and will not be repeated here.

[0062] Please see Figure 4 This application also provides a computer device. The computer device 90 may include a processor 91, a memory 92, and computer applications, wherein: The memory 92 is used to store the computer application, and the memory may also be flash memory. The computer application is, for example, an application that implements the various method embodiments described above.

[0063] Processor 91 is configured to execute the computer application stored in the memory to implement the steps in the various method embodiments described above. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0064] Alternatively, the memory 92 can be either standalone or integrated with the processor 91.

[0065] When the memory 92 is a device independent of the processor 91, the computer device 90 may further include: Bus 93 is used to connect the memory 92 and the processor 91.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein, and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A power supply management method for a thermoelectrically separated printed circuit board, characterized in that, The printed circuit board includes multiple power supply phases, each power supply phase is provided with a power device, and the printed circuit board also has a base, in which a temperature sensor array composed of multiple temperature sensors is embedded. The method includes: The actual temperature inside the base is obtained from each of the temperature sensors, and the real-time output current of the power devices in each power supply phase is obtained simultaneously. The real-time temperature spatial gradient of the base is determined based on the actual temperatures described above. Based on the actual temperature and the real-time output current of each power device, the real-time power loss of each power device is determined. Based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, the predicted temperature distribution data of the base within a preset future time window is predicted, wherein the predicted temperature distribution data includes the predicted temperature value of each power supply phase at the corresponding position on the base. Based on the predicted temperature distribution data, the power allocation weight of each power supply phase is dynamically adjusted by regulating the drive signals of each power device, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold.

2. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, The base is a copper base embedded in the printed circuit board, and the power device has a metal heat dissipation pad at the bottom; the printed circuit board also includes an insulating dielectric layer surrounding the base, which is formed by pressing a polymer resin containing aluminum nitride particles. The step of obtaining the actual temperature inside the base collected by each of the temperature sensors includes: The actual temperature generated by the operation of each power device and directly conducted to the top of the copper base via the bottom metal heat dissipation pad is obtained.

3. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, The temperature sensor is a laser-induced graphene temperature sensor. The step of obtaining the actual temperature inside the base collected by each of the temperature sensors includes: The actual temperature inside the base is obtained by the laser-induced graphene temperature sensor disposed in the stepped pressure-reducing blind groove on the side wall of the base; wherein, the pressure-reducing blind groove is filled with a polydimethylsiloxane flexible buffer layer covering the surface of the laser-induced graphene temperature sensor, so as to protect the porous structure of the laser-induced graphene temperature sensor by hydrostatic pressure during the pressing process.

4. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, The step of determining the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device includes: Based on the actual temperature of each power device, the real-time equivalent impedance corresponding to each power supply phase is dynamically updated. Based on the real-time output current and the real-time equivalent impedance, calculate the real-time power loss of each power device at the current time.

5. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, The prediction of the base's temperature distribution data within a preset future time window, based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board, includes: The actual temperatures described are integrated into a spatially discrete temperature matrix; A real-time power loss matrix is ​​constructed based on the real-time power loss of each power device. Extract the preset thermal diffusion parameters of the printed circuit board, wherein the thermal diffusion parameters include spatial thermal impedance and thermal diffusion characteristic matrix, a first adjustment coefficient and a second adjustment coefficient; Based on the actual temperature matrix, real-time power loss matrix, real-time temperature spatial gradient, and preset thermal diffusion parameters of the printed circuit board, the predicted temperature distribution data of the base within a preset future time window is predicted.

6. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, Determining the real-time temperature spatial gradient of the base based on the actual temperatures described includes: Acquire the spatial physical coordinate data of each temperature sensor in the temperature sensor array within the base; The actual temperatures are correlated and mapped with the corresponding spatial physical coordinate data to construct a discrete spatial temperature field matrix that characterizes the heat distribution state inside the base at the current moment. Using the spatial finite difference algorithm, based on the temperature difference between adjacent temperature sensors and the physical distance in the corresponding spatial physical coordinates, the approximate values ​​of the partial derivatives of the discrete spatial temperature field matrix in each spatial dimension are calculated respectively. The approximate values ​​of the partial derivatives in each spatial dimension are synthesized as gradient components to obtain the real-time temperature spatial gradient.

7. The power supply management method for thermoelectrically separated printed circuit boards according to claim 1, characterized in that, Based on the predicted temperature distribution data, the power allocation weight of each power supply phase is dynamically adjusted by regulating the drive signals of each power device, so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold, including: Extract the predicted temperature value corresponding to each power supply phase from the predicted temperature distribution data, and calculate the maximum predicted temperature difference between any two predicted temperature values ​​of the power supply phases; If the maximum predicted temperature difference is greater than the preset stability threshold, calculate the average predicted temperature of all the power supply phases, and calculate the dynamic power allocation weight of each power supply phase based on the average predicted temperature. The dynamic power allocation weights are converted into pulse width modulation duty cycles for controlling the on and off of the power devices through a preset mapping function, so as to generate the adjusted drive signal. The adjusted drive signal is synchronously applied to each of the power devices so that the difference between the predicted temperature values ​​of any two power supply phases remains within a preset stable threshold.

8. A power supply management system for a thermoelectrically separated printed circuit board, characterized in that, The printed circuit board includes multiple power supply phases, each power supply phase is provided with a power device, and the printed circuit board also has a base, in which a temperature sensor array composed of multiple temperature sensors is embedded. The system is used to implement the method as described in any one of claims 1-7, and the system includes: The data acquisition module is used to acquire the actual temperature inside the base collected by each of the temperature sensors, and simultaneously acquire the real-time output current of the power devices in each of the power supply phases. The gradient determination module is used to determine the real-time temperature spatial gradient of the base based on the actual temperatures. The loss determination module is used to determine the real-time power loss of each power device based on the actual temperature and the real-time output current of each power device. The temperature prediction module is used to predict the predicted temperature distribution data of the base within a preset future time window based on the actual temperature, real-time temperature spatial gradient, real-time output current, real-time power loss, and preset thermal diffusion parameters of the printed circuit board. The predicted temperature distribution data includes the predicted temperature value of each power supply phase at the corresponding position on the base. The power adjustment module is used to dynamically adjust the power allocation weight of each power supply phase by adjusting the drive signal of each power device according to the predicted temperature distribution data, so that the difference between the predicted temperature values ​​of any two power supply phases is kept within a preset stable threshold.

9. A thermoelectrically separated printed circuit board, characterized in that, It includes multiple power supply phases, an insulating dielectric layer, and a base. Each power supply phase is equipped with a power device. The base is a copper base embedded in the printed circuit board. The insulating dielectric layer surrounds the base and is formed by pressing a polymer resin containing aluminum nitride particles. The power device has a metal heat dissipation pad at its bottom, and the metal heat dissipation pad is metallurgically connected to the top of the copper base; The sidewall of the copper base is provided with a stepped pressure-reducing blind groove, and a laser-induced graphene temperature sensor is provided in the pressure-reducing blind groove. The pressure-reducing blind groove is filled with a polydimethylsiloxane flexible buffer layer covering the surface of the laser-induced graphene temperature sensor.

10. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer application, and the processor executing the computer application, wherein the computer application is used to implement the method as described in any one of claims 1-7 when executed by the processor.

Images