Low-noise high-precision switching power supply

Abstract

The invention relates to a low-noise high-precision switching power supply, which comprises a hybrid topological structure, a control circuit, a combined control module and a multi-stage filter network, the hybrid topological structure comprises a step-down topological structure and a multi-winding transformer topological structure; the control circuit comprises a voltage sensor and a quick response switching circuit; the quick response switching circuit is used for switching the step-down topological structure and the multi-winding transformer topological structure according to an input voltage; the combined control module is based on improved PID control, introduces an adaptive control algorithm, and dynamically adjusts control parameters according to load change and input voltage fluctuation amplitude. According to the invention, a switchable mixed topological structure is adopted, a step-down topological structure and a multi-winding transformer topological structure are combined, and a high-precision voltage sensor and a quick response switching circuit are utilized, so that stable output in a wide input voltage range is realized; the combined control module is based on improved PID control, a self-adaptive control algorithm is introduced, and the output voltage of the system is stable under the dynamic working condition.

Description

Technical Field

[0001] This invention relates to the field of switching power supply technology, and in particular to a low-noise, high-precision switching power supply. Background Technology

[0002] Switching power supplies, as highly efficient power conversion devices, are widely used in electronic devices. Their main function is to convert alternating current (AC) or direct current (DC) into a stable and compliant DC output to meet the power supply needs of various electronic devices. Currently, switching power supplies have important applications in many fields, such as industrial equipment, communication base stations, medical equipment, electric vehicles, and consumer electronics.

[0003] However, existing switching power supplies have some shortcomings. Regarding circuit topology, a single buck converter topology suffers from insufficient duty cycle under high input voltage, leading to unstable output voltage; while multi-winding transformer topologies increase manufacturing cost and difficulty due to their complexity. In terms of control strategies, traditional PID control is slow to respond to load changes and prone to output voltage overshoot or undershoot; while adaptive control algorithms have advantages, their stability and convergence need improvement. Furthermore, existing power supplies also have deficiencies in electromagnetic compatibility and thermal management; for example, simple filtering circuits are insufficient to suppress high-frequency noise, and single heat dissipation methods cannot meet the heat dissipation requirements under high power density and high ambient temperature conditions.

[0004] In summary, existing switching power supply technology suffers from numerous shortcomings in circuit topology, control strategies, electromagnetic compatibility, and thermal management, limiting its performance and application scope. Therefore, a low-noise, high-precision switching power supply design method that can effectively address these issues is urgently needed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: in order to overcome the above-mentioned technical problems, the present invention provides a low-noise, high-precision switching power supply.

[0006] The technical solution adopted by the present invention to solve its technical problem is: a low-noise, high-precision switching power supply, including a hybrid topology, a control circuit, a joint control module, a multi-stage filter network, and a thermal management system; Hybrid topology: includes buck converter topology and multi-winding transformer topology; Control circuitry: Connected to the hybrid topology, including a voltage sensor and a fast-response switching circuitry. Voltage sensor: used for real-time monitoring of input voltage; Fast-response switching circuit: used to switch between buck topology and multi-winding transformer topology based on input voltage. The buck topology is enabled when the input voltage is < (80±5)V; When the input voltage is greater than (200±10)V, switch to the multi-winding transformer topology; When the input voltage is in the intermediate voltage range between (80±5)V and (200±10)V, the system will maintain the topology it used before entering this range until the input voltage crosses the switching threshold on the other side. To achieve this function, a hysteresis comparator can be preset in the control circuit. For example, when the input voltage rises from the low-voltage side and first exceeds (80±5)V, the system will not switch immediately but will continue to maintain the buck topology; only when the voltage rises further and exceeds the upper threshold of (200±10)V will it switch to the multi-winding transformer topology. Conversely, when the voltage drops from the high-voltage side to below (200±10)V, the system will maintain the multi-winding transformer topology until the voltage drops below (80±5)V before switching back to the buck topology. This switching mechanism with hysteresis effectively avoids frequent topology oscillations caused by small fluctuations in the input voltage near the threshold voltage, greatly improving the stability and reliability of the system and reducing losses during the switching process.

[0007] Joint control module: Based on improved PID control and introducing an adaptive control algorithm, the control parameters are dynamically adjusted according to load changes and input voltage fluctuations; The voltage sensor of the control circuit monitors the input voltage, and the output of its fast-response switching circuit is connected to the hybrid topology; the input of the joint control module is connected to the sampling circuit, and its output provides a control signal to the control circuit; the multi-stage filtering network is connected to the input and output of the hybrid topology. Multi-stage filtering network: EMI filter is set at the input end, LC filter is set at the output end, and two-stage RC auxiliary filter circuit is set at the key nodes; Thermal management system: includes temperature sensors, cooling fans, and heat sinks. Temperature sensor used to monitor the internal temperature of switching power supply; Cooling fan and heat sink are used to dissipate heat from the switching power supply; Key nodes include the gate drive node of the power switch, the primary and secondary connection node of the transformer, and the anode node of the output rectifier diode.

[0008] The specific formula for the adaptive control algorithm in the joint control module is as follows: ; ; ; in, , , These are the basic parameter values ​​for the proportional, integral, and derivative coefficients of PID control; Kp, Ki, and Kd are the values ​​of the proportional, integral, and derivative coefficients of PID control, respectively. , , The adjustment amount of the PID parameters is calculated based on the load change rate and the input voltage fluctuation amplitude; The formula for calculating PID parameter adjustments is: ; ; ; in: , , , , , For adjustment coefficients; For the fuzzy control function of the load change rate; For the fuzzy control function of the input voltage fluctuation amplitude; Fuzzy control function for load change rate:

[0009] in, The load change rate is measured in real time using a high-bandwidth current sensor. Input voltage fluctuation amplitude fuzzy control function:

[0010] Among them, input voltage fluctuation The voltage is obtained using a differential voltage measurement circuit. PID basic parameter values: , , ; Adjustment factor: , , , , , .parameter =0.2、 =0.1 was determined through optimization using a genetic algorithm. The fuzzy control intervals (5±0.5)A / s and (15±1)A / s were derived statistically from experimental data.

[0011] Preferably, the joint control module outputs the result after limiting the real-time calculation result in the adaptive control algorithm to a preset range, wherein the preset range is adjustable and the preset range is: when At that time, Kp was limited to the range of 13.0 ± 2.0; when When Kp, Ki, and Kd are restricted to the ranges of 11.0±0.5, 0.095±0.005, and 0.115±0.005, respectively; The parameter refresh cycle of the joint control module is 10ms.

[0012] Preferably, in the multi-stage filtering network, the EMI filter includes a common-mode inductor of (1.2±0.05) mH, a differential-mode inductor of (0.6±0.03) mH, and an X capacitor of (0.12±0.01) μF, and the LC filter includes an inductor of (12±0.5) μH and a capacitor of (120±5) μF; the EMI filter suppresses interference in the 10kHz-30MHz frequency band by ≥35dB; the output ripple of the LC filter is ≤30mV; and the overall ripple coefficient is ≤0.2%.

[0013] Preferably, in the thermal management system, the cooling fan is configured as follows: When the temperature is >75℃, operate at (2800±224) rpm; When the temperature is <65℃, reduce the speed to (1200±96) rpm.

[0014] Preferably, in the thermal management system, the heat sink is made of aluminum alloy with a thickness of (1.2±0.1) mm, a fin spacing of (2.5±0.2) mm, and a thermal resistance coefficient of (0.3±0.03) ℃ / W.

[0015] Preferably, the thermal management system is configured to forcibly activate active cooling at an ambient temperature of (40±5)℃ and a power density of (10±1)W / cm³; and the heat sink has a heat dissipation efficiency of ≥92% under a heat flux density of (12±1.44)W / cm².

[0016] It also includes a magnetic shielding structure and a low-loss magnetic core, the magnetic shielding structure covering the magnetic components. The magnetic components include a main power transformer (using a TDK PC95 magnetic core), an output filter inductor (using a Magnetics Kool Mμ 77439), and a common-mode choke (using a Würth Elektronik WE-CMB series), and the magnetic shielding structure is a 0.2mm thick permalloy shield.

[0017] The main power transistor in the hybrid topology is a power semiconductor device, and the switching frequency of the power semiconductor device is based on the formula... The optimization setting is (100±5) kHz, where =0.82±0.05 is the experimental correction coefficient. The drain-source equivalent parasitic inductance of the power semiconductor device under 100kHz test conditions is measured to be (19±2)nH. The drain-source equivalent output capacitance under the same bias condition is measured to be [value missing]. .

[0018] Compared with the prior art, the present invention provides a low-noise, high-precision switching power supply, which has the following advantages: 1. This invention employs a switchable hybrid topology, combining a step-down topology and a multi-winding transformer topology. It utilizes a TI PGA411-Q1 high-precision voltage sensor (accuracy ±0.5%, bandwidth 500kHz) and a 16-bit ADC (ADS8860) to implement voltage sampling and a fast-response switching circuit (switching time ≤1μs). This achieves stable output over a wide input voltage range; when the input voltage varies within the range of 50V-400V, the output voltage fluctuation is controlled within ±1%. The joint control strategy is based on improved PID control and incorporates an adaptive control algorithm to dynamically adjust control parameters according to load changes and input voltage fluctuations. When the load change rate exceeds 15A / s, the proportional coefficient Kp is adjusted to 12.0±0.5; when the input voltage fluctuation exceeds ±(15±1)%, Kp is adjusted to 11.0±0.5, the integral coefficient Ki is adjusted to 0.095±0.005, and the derivative coefficient Kd is adjusted to 0.115±0.005. Parameter adjustment calculations are performed every 10ms to ensure that the output voltage stability of the system reaches ±0.5% under dynamic operating conditions.

[0019] 2. This invention incorporates an EMI filter at the power supply input, comprising a common-mode inductor of (1.2±0.05) mH, a differential-mode inductor of (0.6±0.03) mH, and an X capacitor of (0.12±0.01) μF, and an LC filter at the output, comprising an inductor of (12±0.5) μH and a capacitor of (120±5) μF. An auxiliary filter circuit with a two-stage RC filter structure is also incorporated at key nodes. The EMI filter achieves a suppression effect of over 35 dB on high-frequency interference signals in the 10 kHz-30 MHz frequency range. The LC filter controls the output voltage ripple to within 30 mV, and the auxiliary filter circuit reduces the ripple coefficient to 0.2%. The switching frequency of the power semiconductor devices is optimized to (100±5) kHz, effectively controlling the electromagnetic interference level of the power supply and reducing noise intensity by over 20%. The magnetic components employ a magnetic shielding structure (shielding efficiency ≥30 dB) and low-loss magnetic core material (loss ≤0.1 W / kg), reducing magnetic leakage by over 40%. The intelligent cooling system combines active and passive cooling methods. It uses a temperature sensor (accuracy ±0.8℃) to monitor the temperature of key components inside the power supply in real time and intelligently adjusts the cooling fan speed based on the temperature signal and preset thresholds. Under high power density and high ambient temperature conditions, the active cooling system automatically activates to ensure cooling efficiency; under low load or low ambient temperature conditions, passive cooling is sufficient to meet the requirements, reducing energy consumption and noise. The heat sink is made of aluminum alloy with a thickness of (1.2±0.1)mm, a fin spacing of (2.5±0.2)mm, and a thermal resistance coefficient of (0.3±0.03)℃ / W. Under a heat flux density of (12±1.44)W / cm², the cooling efficiency is (92±2)%. Attached Figure Description

[0020] Figure 1 This is a flowchart of the hybrid topology switching control proposed in this invention; Figure 2 This is a flowchart of the adaptive PID control of the present invention; Figure 3 This is a flowchart of the signal processing of the multi-stage filtering network proposed in this invention. Detailed Implementation

[0021] The invention will now be described in further detail with reference to the accompanying drawings. It should be emphasized that the following description is merely exemplary and not intended to limit the scope or application of the invention.

[0022] This invention provides a low-noise, high-precision switching power supply, comprising: 1. Hybrid Topology: A switchable hybrid topology is adopted, combining a step-down topology and a multi-winding transformer topology. The step-down topology is suitable for scenarios where the input voltage is below the low voltage threshold of (80±5)V, while the multi-winding transformer topology is suitable for scenarios where the input voltage is above the high voltage threshold of (200±10)V.

[0023] 2. Control Circuit: The input voltage is monitored in real time by using a TI PGA411-Q1 high-precision voltage sensor (accuracy ±0.5%, bandwidth 500kHz). It works with a 16-bit ADC (ADS8860) to sample the voltage and uses a fast-response switching circuit (switching time ≤1μs) to automatically switch the topology according to the set threshold. This ensures that the topology switching process is seamless when the input voltage changes abruptly, and the output voltage fluctuation is controlled within ±1%, achieving stable output over a wide input voltage range (e.g., 50V-400V).

[0024] The fast-response switching circuit is used to switch between buck topology and multi-winding transformer topology according to the input voltage. It adopts a high-speed MOSFET switching array (Infineon IPD90N04S4-04) and a low parasitic capacitance drive circuit (drive current ≥5A), and works with an FPGA controller (Xilinx Spartan-6 XC6SLX9) to achieve a switching time of ≤1μs. During the switching process, the output capacitor voltage is maintained by a pre-charge circuit to ensure seamless switching.

[0025] To ensure that the hybrid topology does not experience current backflow, voltage spikes, or brief open / short circuit states during switching, thus achieving truly seamless switching, the fast-response switching circuit of this invention employs a collaborative design that includes precise dead-time control, pre-charge compensation, and fast gate drive. Its specific connection relationships and working mechanism are as follows: The high-speed MOSFET switch array (Infineon IPD90N04S4-04) is configured as mutually exclusive complementary switch groups, where one group must reliably turn off when the other is on, physically eliminating the risk of shoot-through short circuits. The low-side gate driver (TI UCC27524) directly drives the gates of these MOSFETs, and its high pull-sink current of up to 5A ensures rapid charging and discharging of the gate charge, significantly shortening the rise / fall time of the switching transistors, thereby reducing switching losses and voltage / current stress.

[0026] To achieve an uninterrupted current path during switching, a pre-charge branch is introduced. This branch (consisting of a 50V / 1A constant current source and a 22Ω current-limiting resistor) is activated by the controller (Xilinx Spartan-6 XC6SLX9) 1μs before the switching command is executed, pre-charging the output capacitor to compensate its voltage to the target value of 200mV. This measure effectively offsets the output voltage drop that may be caused by the brief dead time inherent in topology switching, maintaining the stability of the load voltage.

[0027] To prevent backflow of current due to the MOSFET body diode conducting during the dead time, the drive signal generated by the controller includes a precisely controllable dead time. The analog switch (ADI ADG1414) plays a crucial role here, rapidly acting at the start of the dead time (t_ON = 25ns) to quickly discharge the gate charge of the MOSFET about to be turned off to the source through a 10Ω resistor, ensuring it is completely and reliably turned off before the other set of switches turns on. This "gate charge pre-discharge" mechanism, combined with the driver's own fast turn-off capability, minimizes the turn-off delay, ensuring the accuracy and minimization of the dead time.

[0028] The timing of the entire switching process is uniformly scheduled by the FPGA controller. The FPGA receives the digitized voltage signal from the voltage detection loop (INA240 + ADS8860 ADC), and after detecting the threshold for switching, issues a decision command within 30ns. Subsequently, the FPGA executes a strict timing sequence: first, pre-charging is initiated, then the current conduction path is forcibly turned off, a dead time is entered, and the gate charge is discharged using an analog switch. After the dead time ends, the target conduction path is immediately turned on. This series of actions is completed within microseconds. Through parallel processing of hardware logic and precise delay control, the energy transfer between different topologies is ensured to be smooth and orderly, ultimately achieving seamless switching with output voltage disturbance ≤120mV and recovery time ≤150μs.

[0029] High-speed MOSFET switch array: Infineon IPD90N04S4-04, 4 in total, RDS(on) = 4.2mΩ, COss = 1.4nF (typical); Low-side gate driver: TI UCC27524, peak pull / sink current 5A, propagation delay 12ns; Analog switch: ADI ADG1414, t ON = 25ns, used for gate charge pre-release; Where RDS(on) represents the drain-source on-resistance of the MOSFET in the fully on (conducting) state. The smaller this value, the lower the power loss and voltage drop when the MOSFET is on, and the higher the efficiency.

[0030] COSS represents the output capacitance of the MOSFET. It is one of the key parameters affecting switching speed; the smaller the capacitance value, the faster the MOSFET turns on and off, and the lower the switching losses.

[0031] tON: This represents the turn-on time required for the analog switch to fully conduct from receiving a control signal to its output channel being fully turned on. The shorter this time, the faster the circuit responds to the control signal, which is crucial for achieving nanosecond-level timing control.

[0032] Pre-charge branch: 50V / 1A constant current source and 22Ω current limiting resistor, to compensate the output capacitor voltage to the target value of 200mV 1μs in advance; Controller: Xilinx Spartan-6 XC6SLX9, 100MHz clock, generating complementary PWM and 1μs feedforward trigger pulse; Voltage detection: The INA240 output is sampled by a 16-bit SAR ADC (ADS8860) with a conversion time of 1μs. The results are compared in parallel by the FPGA. The threshold is reached and the output switching command is issued in 3 clock cycles (30ns).

[0033] Switching timing: a) 1μs before the switching command arrives: FPGA shuts down the gate drive of the transistor to be turned off and starts pre-charging at the same time; b) At the moment the switching command arrives: The FPGA first sets the ADG1414 bit, shorts the gate of the transistor to be turned off to the source through a 10Ω resistor, and the gate charge is discharged within ≤20ns. c) After a 40ns delay (to ensure a dead time), the FPGA sends a turn-on pulse, and the UCC27524 charges the gate of the transistor about to be turned on with a peak current of 5A. d) The measured drain-source voltage VDS falling edge time from 10% to 90% is 0.78μs, the sampling and display disturbance of the whole machine output voltage is ≤120mV, and the recovery time is 150μs.

[0034] By combining the above hardware with precise timing control, the topology switching time can be ≤1μs.

[0035] Control Strategy Integration and Innovation: This invention proposes a joint control strategy based on improved PID control and adaptive control. Based on PID control, initial PID parameter values ​​are set as follows: proportional coefficient Kp = 10, integral coefficient Ki = 0.1, and derivative coefficient Kd = 0.05. An adaptive control algorithm is introduced to form the adaptive control algorithm in the joint control strategy of the joint control module. The specific formula is as follows: ; ; ; in, , , These are the basic parameter values ​​for the proportional, integral, and derivative coefficients of PID control; Kp, Ki, and Kd are the values ​​of the proportional, integral, and derivative coefficients of PID control, respectively. , , The adjustment amount of the PID parameters is calculated based on the load change rate and the input voltage fluctuation amplitude.

[0036] Formula for calculating PID parameter adjustment: ; ; ; in: , , , , , The specific values ​​for the adjustment coefficients are determined based on experiments and simulations. For the fuzzy control function of the load change rate; This is a fuzzy control function for the amplitude of input voltage fluctuations.

[0037] Fuzzy control function for load change rate:

[0038] in, The load change rate is measured in real time using a high-bandwidth current sensor (such as LEM LAH 100-P, bandwidth 1MHz); Input voltage fluctuation amplitude fuzzy control function:

[0039] Among them, input voltage fluctuation The voltage is obtained using a differential voltage measurement circuit (such as TI INA826); PID basic parameter values: , , .

[0040] Adjustment factor: , , , , , .parameter =0.2、 =0.1 was determined through optimization using a genetic algorithm. The fuzzy control intervals (5±0.5)A / s and (15±1)A / s were derived statistically from experimental data.

[0041] This algorithm is based on fuzzy control rules and monitors the load change rate and input voltage fluctuation amplitude in real time. When the load change rate exceeds 15A / s, that is... When the input voltage fluctuation exceeds ±(15±1)%, the adaptive control algorithm adjusts the proportional coefficient Kp to 13.0±2.0, decreases the integral coefficient Ki by 10%, and increases the derivative coefficient Kd by 15%. When the load change rate exceeds 15A / s or the input voltage fluctuation exceeds ±15%, the adaptive control algorithm dynamically adjusts the PID control parameters according to the above formula to optimize the system response. For example, when the load change rate exceeds 15A / s, Kp is adjusted to 11.0±0.5, Ki is adjusted to 0.095±0.005, and Kd is adjusted to 0.115±0.005. Parameter adjustment calculations are performed every 10ms to ensure that the output voltage stability reaches ±0.5% under dynamic operating conditions, thereby improving the system's dynamic performance and output accuracy. This is suitable for applications with high dynamic response requirements, such as server power supplies and communication base station power supplies. When the load change rate exceeds 15A / s, the adaptive control algorithm dynamically adjusts the PID control parameters according to the above formula to optimize the system response. ,at this time proportionality coefficient Increase the input voltage to accelerate system response; when the input voltage fluctuation exceeds ±15%, ,at this time Integral coefficient Reduce and suppress output voltage overshoot.

[0042] 3. Low-noise design: Multi-stage filtering networks are used at the power supply input, output, and key nodes, combined with optimized magnetic component design to reduce electromagnetic interference. Key nodes include the power switch gate drive node, the transformer primary-secondary connection node, and the output rectifier diode anode node. The two-stage RC auxiliary filter circuits at these nodes adopt a two-stage RC structure, with R=10Ω±5% and C=100nF±5%. Simultaneously, the switching frequency of the power semiconductor devices is optimized to reduce high-frequency noise. The hybrid topology main power transistor is a power semiconductor device, model Infineon-IPT65R033G7, and the optimized switching frequency formula is: ; in, =0.82±0.05 is the experimental correction coefficient. The switching frequency was optimized and set to (100±5) kHz according to the above formula to minimize the electromagnetic interference level of the power supply. The switching frequency was set to (100±5) kHz according to the formula to reduce the noise intensity by ≥20%. The drain-source equivalent parasitic inductance of the power semiconductor device under 100kHz test conditions is measured to be (19±2)nH. The drain-source equivalent output capacitance under the same bias condition is measured to be [value missing]. .

[0043] 4. Thermal Management System Design: A combination of active and passive cooling methods is employed, intelligently adjusting the system's operation based on power supply status and ambient temperature. Under high power density and high ambient temperature conditions, the active cooling system automatically activates; under low load or low ambient temperature conditions, passive cooling is sufficient to meet cooling requirements.

[0044] The control circuit of this invention employs a high-precision voltage sensor with an accuracy of ±0.5%; simultaneously, it utilizes a fast-response switching circuit with a switching time not exceeding 1μs. This design ensures seamless transition during topology switching, avoiding fluctuations in output voltage.

[0045] In the joint control strategy, the adaptive control algorithm dynamically adjusts the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd of the PID control based on the load change rate (exceeding the set threshold of 10A / s) and the input voltage fluctuation amplitude (exceeding the set range of ±10%) to adapt to different operating conditions.

[0046] The multi-stage filtering network includes an EMI filter at the power input, comprising a common-mode inductor of 1.2 ± 0.05 mH, a differential-mode inductor of 0.5 mH, and an X capacitor of 0.1 μF; and an LC filter at the output, comprising an inductor of 10 μH and a capacitor of 100 μF. Furthermore, a two-stage RC auxiliary filter circuit is incorporated at critical nodes to filter out high-frequency interference signals in the range of 10 kHz to 30 MHz, controlling the output voltage ripple to within 50 mV.

[0047] The optimized design of magnetic components includes the use of a magnetic shielding structure with a shielding efficiency of 30dB or higher, and low-loss magnetic core materials with a loss of no more than 0.1W / kg. This design can reduce magnetic leakage and electromagnetic radiation, reducing magnetic leakage by more than 40%.

[0048] The switching frequency of the power semiconductor device has been optimized. Through experiments and simulation analysis, it has been determined that within a specific frequency range of (100±5) kHz, the electromagnetic interference level of the power supply is minimized and the noise intensity is reduced by more than 20%.

[0049] The thermal management system includes a temperature sensor, a cooling fan, and a heatsink. The temperature sensor monitors the temperature of key components inside the power supply in real time and sends the temperature signal to the control circuit. The control circuit controls the speed of the cooling fan based on the temperature signal and a preset temperature threshold. The temperature sensor has an accuracy of ±0.8℃, the cooling fan has an airflow of (90 CFM ± 8%), and the heatsink has a heat dissipation area of ​​(0.48 ± 0.02) m². Based on the temperature signal, the control circuit operates the cooling fan at a speed of (2800 ± 224) rpm when the monitored temperature exceeds 75℃, and reduces the fan speed to 1200 rpm ± 8% when the temperature is below 65℃. The heatsink is made of aluminum alloy with a thickness of 1.2 mm ± 0.1 mm and a fin spacing of 2.5 mm ± 0.2 mm. Its thermal resistance coefficient is (0.3 ± 0.03)℃ / W, ensuring a heat dissipation efficiency of (92 ± 2)% under a heat flux density of 12 W / cm² ± 12%.

[0050] The design of the heat sink can be optimized based on the heat flux density, which is (10±1)W / cm², to ensure that heat can be evenly conducted from the heat-generating element to the surrounding environment, thereby improving the heat dissipation efficiency to 90% or more.

[0051] Under high power density and high ambient temperature conditions, such as ambient temperature (40±5)℃ and power density (10±1)W / cm³, the active cooling system automatically starts; under low load or low ambient temperature conditions, such as ambient temperature 20℃ and load power 50W, passive cooling can meet the heat dissipation requirements to reduce energy consumption and noise.

[0052] Example Implementation of the circuit topology: The control circuit monitors the input voltage in real time and compares it with preset high and low voltage thresholds. When the input voltage is lower than the set low voltage threshold (80±5V), the control circuit triggers the buck topology to operate, utilizing its high efficiency to reduce energy loss. When the input voltage exceeds (200±10)V, the control circuit switches to a multi-winding transformer topology. By extending the duty cycle, the stability of the output voltage is ensured. The control circuit employs a high-precision voltage sensor (accuracy ±0.5%) and a fast-response switching circuit (switching time ≤1μs) to ensure seamless switching of the topology and avoid fluctuations in the output voltage. This hybrid topology is particularly suitable for applications with a wide range of input voltage variations, such as on-board charging systems for new energy vehicles and power supply systems for industrial equipment.

[0053] The joint control strategy of the joint control module is implemented as follows: Based on PID control, initial PID parameter values ​​are set (proportional coefficient Kp=10, integral coefficient Ki=0.1, derivative coefficient Kd=0.05). Simultaneously, an adaptive control algorithm is introduced to monitor load changes and input voltage fluctuations in real time. When the load change rate exceeds a set threshold (5±1 A / s) or the input voltage fluctuation exceeds ±(10±1)%, the adaptive control algorithm dynamically adjusts the PID control parameters according to preset control rules.

[0054] For example, when the load suddenly increases, the adaptive control algorithm will increase the proportional coefficient of the PID control (Kp increases to 15) to speed up the system's response to load changes and reduce the undershoot of the output voltage; when the input voltage suddenly rises, the algorithm will adjust the integral coefficient (Ki decreases to 0.05) to suppress the overshoot of the output voltage.

[0055] In this way, the joint control strategy can maintain high accuracy and stability of power output under various operating conditions, making it suitable for applications with high dynamic response requirements, such as server power supplies and communication base station power supplies.

[0056] Implementation of low-noise design: An EMI filter is set at the input of the power supply, including a common-mode inductor (1.2±0.05)mH, a differential-mode inductor (0.6±0.03)mH and an X capacitor of 0.1μF, to filter out high-frequency interference signals from the power grid, with a frequency range of 10kHz-30MHz; An LC filter is installed at the output end, including a 10μH inductor and a 100μF capacitor, to further reduce the ripple and noise of the output voltage. The ripple voltage is ≤30mV. At the same time, the magnetic components such as transformers and inductors are optimized by adopting a magnetic shielding structure (shielding efficiency ≥30dB) and low-loss magnetic core material (loss ≤0.1W / kg) to reduce magnetic leakage and electromagnetic radiation (magnetic leakage is reduced by ≥40%).

[0057] In addition, the switching frequency of the power semiconductor device is optimized (set to (100±5)kHz), and a suitable switching frequency point is selected to minimize the high-frequency noise generated by the device during the switching process, reducing the noise intensity by ≥20%. This low-noise design can effectively improve the electromagnetic compatibility of the power supply, enabling it to meet the strict electromagnetic interference requirements of medical equipment and aerospace electronic equipment.

[0058] Implementation of the thermal management system design: The intelligent heat dissipation system includes a temperature sensor (accuracy (±1)℃), a cooling fan with an airflow of (100±10)CFM, and a heat sink (heat dissipation area (0.50±0.025)㎡). The temperature sensor monitors the temperature of key components inside the power supply in real time (such as the temperature of power semiconductor devices and magnetic components) and sends the temperature signal to the control circuit.

[0059] The control circuit controls the speed of the cooling fan based on the received temperature signal and a preset temperature threshold (e.g., 70°C). Under high power density and high ambient temperature conditions (e.g., ambient temperature 40°C, power density 10W / cm³), the cooling fan runs at full speed (2800±224 rpm) to quickly dissipate heat; under low load or low ambient temperature conditions, such as ambient temperature 20°C and load power 50W, the cooling fan runs at low speed (1000±100 rpm) or stops to reduce energy consumption and noise. Meanwhile, the heat sink design is optimized based on heat flux density (10W / cm²±10%) to ensure that heat can be evenly conducted from the heat-generating element to the surrounding environment, thereby improving heat dissipation efficiency to ≥90%.

[0060] This intelligent thermal management system not only improves the heat dissipation efficiency of the power supply, but also extends the service life of heat dissipation components such as cooling fans and reduces maintenance costs.

[0061] The following explanation is based on testing and verification of key performance indicators: 1. Circuit Construction and Component Selection: The buck topology in the hybrid topology is implemented using the TI TPS54260 chip, while the multi-winding transformer topology is built using a custom transformer (450μH primary inductance, turns ratio 1:1.2:1.5) and an Infineon IRS2505L driver chip. The core of the control circuit is the STMicroelectronics STM32G474RET6 microcontroller (with built-in high-precision 16-bit ADC) and the Analog Devices ADG1414 analog switch chip, which together form a fast-response switching circuit. The voltage sensor uses the Texas Instruments INA240 high-precision current sensing amplifier (accuracy ±0.5%). The algorithm of the joint control module is integrated into the aforementioned STM32G474RET6 and runs. The EMI filter in the multi-stage filtering network uses TDK series components for its common-mode inductor (1.2mH), differential-mode inductor (0.6mH), and X capacitor (0.12μF), while the output LC filter uses Murata low-ESR components for its inductor (12μH) and capacitor (120μF). The thermal management system uses Maxim's DS18B20 temperature sensor (accuracy ±0.5℃), a Sunon KD1208PTS1 cooling fan (rated airflow 90CFM±8%), and an aluminum alloy heatsink with an optimized thickness of 1.2mm and a fin spacing of 2.5mm.

[0062] 2. The control circuit of this invention works in conjunction with the joint control module to ensure the stability of the output voltage during topology switching. Its specific workflow and verification results are as follows: (1) Pre-adjustment before switching, 10μs before switching: Before issuing the switching command, the control circuit pre-adjusts the PID parameters to the optimal anti-disturbance values ​​Kp=15, Ki=0.05, Kd=0.2 to enhance the robustness of the system.

[0063] (2) Instantaneous Feedforward Compensation During Switching: To compensate for the estimated 200mV output voltage drop caused by topology switching, output voltage feedforward compensation is enabled 1μs before the switching command is issued. The compensation amount ΔV is calculated using the formula ΔV = Kf·Vcomp. Wherein, Vcomp is a compensation reference value preset based on the circuit model and experimental data, which is taken as 200mV in this example; Kf is a compensation intensity coefficient between 0 and 1, and its specific value is used to finely adjust the compensation depth to avoid overcompensation or undercompensation. In this embodiment, Kf is determined to be 0.2 by least squares fitting, which is dimensionless. This compensation amount ΔV is superimposed on the DAC output of the voltage loop through the FPGA to achieve early intervention.

[0064] (3) Adaptive parameter tuning after switching: Within 100μs after switching: After switching, the system starts the adaptive parameter search mode. Within the next 100μs, Kp=15, Ki=0.05, Kd=0.2, and the output voltage error sequence e(k) is quickly recorded at a sampling rate of 1MHz. Then, a 16-point recursive least squares (RLS) algorithm is run once to identify the system and obtain the optimal PID parameter triplet (Kp, Ki, Kd) under the current operating condition. The result is written back to the hardware register to complete the online optimization of the parameters. The whole process is completed within 100μs.

[0065] To verify the collaborative control effect of the joint control module in handling input disturbances within a single topology, the power supply module was tested: The system was stably operated in a multi-winding transformer topology at room temperature (25℃) and a full load of 10A. Using a Keysight N6705B DC power analyzer, the input voltage was stepped from 200V to 250V. The measured output voltage waveform showed that the output voltage was 12V before the switch, the maximum overshoot during the control strategy adjustment was 120mV, the fluctuation was ≤±1%, and the voltage stabilized within 150μs. This result demonstrates that even when the input voltage experiences a large step change within the same topology operating region, the joint control module can ensure high output stability.

[0066] 3. To verify the low-noise characteristics of the power supply, EMI testing was conducted according to the CISPR 22 standard. The testing equipment used was a Rohde & Schwarz ESR7 EMI receiver and a 50Ω / 50μH LISN (Line Impedance Stabilization Network). Test results showed that, within the frequency range of 10kHz to 30MHz, the quasi-peak detection value of conducted interference was far below the standard limit. At the worst frequency point of 24MHz, the measured interference level was 37dBμV, more than 23dB below the standard limit (60dBμV), proving that the multi-stage filtering network of this invention has a suppression effect of ≥35dB on high-frequency interference signals, demonstrating excellent EMI performance.

[0067] 4. To verify the effectiveness of the thermal management system, performance tests were conducted on the heat dissipation system. A Fluke Ti480 PRO infrared thermal imager and K-type thermocouples were used to monitor the heat sink temperature. Under harsh conditions of an ambient temperature of 40℃ and a heat flux density of 12W / cm², the heat sink substrate temperature rise was measured to be 36℃. Its calculated thermal resistivity was 0.30℃ / W, consistent with the design value of (0.3±0.03)℃ / W, achieving a heat dissipation efficiency of 92%. When the core temperature exceeded 75℃, the cooling fan automatically operated at 2800rpm to ensure that the junction temperature of the power devices remained within a safe range.

[0068] 5. Noise intensity and magnetic leakage test (1) Noise intensity Test criteria: CISPR 22:2008, Section 6.2 Conducted emission method; Test setup: 50 Ω / 50 μH LISN + R&S ESR7 receiver, RBW=9 kHz, peak + quasi-peak detection; Test sample: Input 250 V, full load 10 A, switching frequency 100 kHz; Results: The maximum quasi-peak value of 37 dBμV was observed at 24 MHz, which is 23 dB lower than the standard limit of 60 dBμV; Conversion relationship: A 23 dB reduction in electric field strength corresponds to a linear amplitude attenuation of ≈14.1 times. Based on this, the conducted noise intensity of the prototype of this invention is reduced by ≥20% compared with the control prototype of the same power (without multi-stage filtering and frequency optimization).

[0069] (2) Magnetic leakage Test standard: IEC 62333-1:2006, Clause 5, "Proximity magnetic field leakage" method; Test setup: The prototype was placed on a 3 m non-magnetic test bench, using a Lakeshore 460 triaxial gaussmeter, with the probe 10 mm away from the magnetic component housing; Tested prototype: 10 A full load, 0.2 mm magnetic shield, fully assembled with permalloy alloy; Results: The maximum leakage point was located directly above the main power transformer, and was measured at 0.42 μT; Control group: After removing the magnetic shield and replacing it with an unshielded magnetic core, the leakage at the same point was 0.70 μT; Decrease: (0.70) 0.42) / 0.70=40%, that is, magnetic leakage is reduced by ≥40%.

[0070] In summary, through the synergistic optimization of hardware topology design, control algorithm innovation and thermal management, and after rigorous testing and verification, this invention has indeed achieved the design goals of low noise, high precision and high reliability, and all performance indicators have met or exceeded the expected design requirements.

[0071] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A low-noise, high-precision switching power supply, characterized in that, This includes a hybrid topology, control circuitry, joint control modules, multi-stage filter networks, and a thermal management system. Hybrid topology: includes buck converter topology and multi-winding transformer topology; Control circuitry: Connected to the hybrid topology, including a voltage sensor and a fast-response switching circuitry. Voltage sensor: used for real-time monitoring of input voltage; Fast-response switching circuit: used to switch between buck topology and multi-winding transformer topology based on input voltage. The buck topology is enabled when the input voltage is < (80±5)V; When the input voltage is greater than (200±10)V, switch to the multi-winding transformer topology; When the input voltage is in the middle voltage range between (80±5)V and (200±10)V, the system will maintain the topology it used before entering the range until the input voltage crosses the switching threshold on the other side. Joint control module: Based on improved PID control and introducing an adaptive control algorithm, the control parameters are dynamically adjusted according to load changes and input voltage fluctuations; The voltage sensor of the control circuit monitors the input voltage, and the output of its fast-response switching circuit is connected to the hybrid topology; the input of the joint control module is connected to the sampling circuit, and its output provides a control signal to the control circuit; the multi-stage filtering network is connected to the input and output of the hybrid topology. Multi-stage filtering network: EMI filter is set at the input end, LC filter is set at the output end, and two-stage RC auxiliary filter circuit is set at the key nodes; Thermal management system: includes temperature sensors, cooling fans, and heat sinks. Temperature sensor used to monitor the internal temperature of switching power supply; Cooling fan and heat sink are used to dissipate heat from the switching power supply; The key nodes include the power switch gate drive node, the transformer primary and secondary side connection node, and the output rectifier diode anode node.

2. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, The specific formula for the adaptive control algorithm in the joint control module is as follows: ； ； ； in, , , These are the basic parameter values ​​for the proportional, integral, and derivative coefficients of PID control, respectively; Kp, Ki, and Kd are the proportional, integral, and derivative coefficients of PID control, respectively. , , The adjustment amount of the PID parameters is calculated based on the load change rate and the input voltage fluctuation amplitude; The formula for calculating PID parameter adjustments is: in: , , , , , For adjustment coefficients; For the fuzzy control function of the load change rate; For the fuzzy control function of the input voltage fluctuation amplitude; Fuzzy control function for load change rate: in, The load change rate is measured in real time using a high-bandwidth current sensor. Input voltage fluctuation amplitude fuzzy control function: Among them, input voltage fluctuation The voltage is obtained using a differential voltage measurement circuit. PID basic parameter values: , , ; Adjustment factor: , , , , , .

3. The low-noise, high-precision switching power supply as described in claim 2, characterized in that, The joint control module outputs the result after limiting the real-time calculation result in the adaptive control algorithm to a preset range, wherein the preset range is: when At that time, Kp was limited to the range of 13.0 ± 2.0; when When Kp, Ki, and Kd are restricted to the ranges of 11.0±0.5, 0.095±0.005, and 0.115±0.005, respectively; The parameter refresh cycle of the joint control module is 10ms.

4. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, In the multi-stage filtering network, the EMI filter includes a common-mode inductor of (1.2±0.05) mH, a differential-mode inductor of (0.6±0.03) mH, and an X capacitor of (0.12±0.01) μF, and the LC filter includes an inductor of (12±0.5) μH and a capacitor of (120±5) μF.

5. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, The EMI filter suppresses interference in the 10kHz-30MHz frequency band by ≥35dB; the output ripple of the LC filter is ≤30mV; and the overall ripple coefficient is ≤0.2%.

6. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, In the thermal management system, the cooling fan is configured as follows: When the temperature is >75℃, operate at (2800±224) rpm; When the temperature is <65℃, reduce the speed to (1200±96) rpm.

7. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, In the thermal management system, the heat sink is made of aluminum alloy with a thickness of (1.2±0.1) mm, a fin spacing of (2.5±0.2) mm, and a thermal resistance coefficient of (0.3±0.03) ℃ / W.

8. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, The thermal management system is designed to automatically activate active cooling at an ambient temperature of (40±5)℃ and a power density of (10±1)W / cm³; the heat sink has a heat dissipation efficiency of ≥92% under a heat flux density of (12±1.44)W / cm².

9. The low-noise, high-precision switching power supply as described in claim 1, characterized in that, It also includes a magnetic shielding structure and a low-loss magnetic core, wherein the magnetic shielding structure covers the magnetic elements.

10. The low-noise, high-precision switching power supply as described in claim 8, characterized in that, The main power transistor in the hybrid topology is a power semiconductor device, and the switching frequency of the power semiconductor device is based on the formula... The optimization setting is (100±5) kHz, where =0.82±0.05 is the experimental correction coefficient. The drain-source equivalent parasitic inductance of a power semiconductor device under 100kHz test conditions. This is the drain-source equivalent output capacitance under the same bias condition.

Images