A DSP-based current sharing control method for a board-level parallel power supply system

Abstract

The application relates to the technical field of power management, in particular to a board-level parallel power supply system current sharing control method based on DSP, which comprises the following steps: collecting output current and internal temperature, screening and activating a power supply branch to evaluate average temperature, combining bus voltage evaluation to evaluate equivalent line impedance, determining current sharing reference value and load reduction coefficient and generating a target current reference value, correcting a droop impedance parameter to generate an analog droop control voltage, obtaining a compensation voltage parameter through digital algorithm control and digital-analog conversion, and combining the analog droop control voltage to generate a board-level current sharing control regulation signal.In the application, by considering evaluation parameters such as branch current internal temperature and line impedance, proportional integral and droop control logic are introduced to jointly generate a control signal, impedance differences are compensated in real time, load distribution based on temperature is realized, a high-precision and rapid-dynamic-response process is achieved, defects that are difficult to balance are effectively solved, and the overall operation stability is improved.

Description

Technical Field

[0001] This invention relates to the field of power management technology, and in particular to a current sharing control method for a board-level parallel power supply system based on DSP. Background Technology

[0002] The field of power management technology mainly involves the overall planning and coordination of the conversion, distribution, monitoring, and protection of electrical energy within electronic devices. Core aspects of this field include voltage conversion, current regulation, power factor correction, and thermal management. Through operations such as voltage stabilization, buck-boosting, and inversion of DC or AC power supplies, it ensures that each electronic component receives a stable power supply that meets its operating specifications. Among these, the traditional board-level parallel power supply system current sharing control method refers to strategies for adjusting the uneven distribution of load current when multiple power supplies are connected in parallel on the same circuit board.

[0003] In existing technologies, when multiple modules operate in parallel, differences in component parameters, temperature drift, and differences in line impedance can all lead to deviations in the output current of each power module. This uneven distribution causes a certain branch to be in a local overload state for a long time, which can cause overheating and accelerate failure. At the same time, traditional solutions are difficult to fully adapt to the architecture of the motherboard with a signal processor for management in actual deployment, making it difficult to find an ideal balance between system reliability, current sharing accuracy, and dynamic response speed, thus affecting the operational stability of the entire power supply system. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a current sharing control method for board-level parallel power supply systems based on DSP.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a current sharing control method for a board-level parallel power supply system based on DSP, comprising the following steps:

[0006] S1: Collect the branch output current and internal temperature of the corresponding power supply branch, filter out the active power supply branches based on the branch output current to obtain the branch quantity characteristics, and use the internal temperature of the branch corresponding to the active power supply branch to evaluate the average temperature of the branch.

[0007] S2: For the activated power supply branch, collect the total output bus voltage and the branch output voltage, and evaluate the equivalent line impedance of the branch in combination with the branch output current. Evaluate the current sharing reference value based on the branch output current and the branch quantity characteristics.

[0008] S3: Calculate the branch load derating factor based on the branch internal temperature, the branch average temperature and the preset temperature adjustment weighting factor corresponding to the activated power supply branch, and generate the branch target current reference value based on the current sharing reference value and the branch load derating factor;

[0009] S4: The branch droop impedance parameter is generated by correcting the preset droop impedance reference value using the equivalent line impedance of the branch through a programmable gain device. The branch simulated droop control voltage is generated by using the branch output current, the branch droop impedance parameter and the preset branch reference voltage value.

[0010] S5: Based on the branch output current, the branch target current reference value and the historical branch current deviation term, a digital compensation bias voltage for the branch is generated through a digital proportional-integral control algorithm. The branch analog compensation voltage parameter is obtained by processing through a digital-to-analog converter. A board-level current sharing control signal is generated based on the branch analog compensation voltage parameter and the branch analog droop control voltage.

[0011] The present invention is improved in that the specific steps of S1 are as follows:

[0012] S111: According to the preset sampling period, the output current of each power supply branch and the internal temperature of each power supply branch are collected by the digital signal processor on the motherboard. Based on the output current of the branch, the active power supply branches in the active mode are selected from each power supply branch, and the number of active power supply branches is counted to obtain the branch quantity characteristics.

[0013] S112: Call the internal temperature of each activated power supply branch, calculate the sum of the internal temperatures of all branches, obtain the total temperature value, calculate the ratio of the total temperature value to the average of the branch quantity characteristics, and generate the branch average temperature.

[0014] The present invention is improved in that the specific steps of S2 are as follows:

[0015] S211: For each of the activated power supply branches, the digital signal processor collects the corresponding branch output voltage and the total output bus voltage, calculates the voltage difference between the total output bus voltage and the branch output voltage to obtain the branch voltage deviation characteristics, calls the corresponding branch output current, calculates the impedance mapping ratio between the branch voltage deviation characteristics and the branch output current, and obtains the branch equivalent line impedance.

[0016] S212: Call the branch output current corresponding to each of the activated power supply branches, calculate the sum of the currents of all the branch output currents, obtain the branch current accumulation feature, calculate the average reference ratio of the branch current accumulation feature and the branch quantity feature, and generate the current sharing reference value.

[0017] The present invention is improved in that the process of calculating the impedance mapping ratio between the branch voltage deviation characteristic and the branch output current is specifically as follows:

[0018] The product of the branch voltage deviation characteristic and the impedance correction weighting coefficient is calculated to obtain the corrected voltage deviation characteristic;

[0019] The quotient of the corrected voltage deviation characteristic and the branch output current is calculated to obtain the basic impedance mapping ratio.

[0020] Calculate the sum of the basic impedance mapping ratio and the line aging compensation benchmark value, and use the sum as the equivalent line impedance of the branch.

[0021] The impedance correction weighting coefficient is set as follows:

[0022] Extract the wiring length and resistivity of the activated power supply branch, calculate the correlation value of the product of the wiring length and the resistivity, and determine the correlation value of the product as the impedance correction weighting coefficient.

[0023] The method for setting the baseline value for line aging compensation is as follows:

[0024] Extract the cumulative operating time of the activated power supply branch, obtain the preset limit attenuation constant, calculate the ratio mapping value of the cumulative operating time to the limit attenuation constant, and determine the ratio mapping value as the line aging compensation benchmark value.

[0025] The present invention is improved in that the specific steps of S3 are as follows:

[0026] S311: For each of the activated power supply branches, call the corresponding branch internal temperature and branch average temperature, calculate the temperature deviation value between the branch internal temperature and the branch average temperature, obtain the branch temperature difference characteristic, and calculate the difference ratio between the branch temperature difference characteristic and the branch average temperature based on the branch temperature difference characteristic and the branch average temperature to generate a relative temperature deviation ratio.

[0027] S312: Obtain the preset temperature adjustment weight coefficient, calculate the product value of the relative temperature deviation ratio and the temperature adjustment weight coefficient, obtain the branch temperature compensation offset, call the branch temperature compensation offset, calculate the difference feature value between constant one and the branch temperature compensation offset, and generate the branch load derating coefficient.

[0028] S313: Call the current sharing reference value and the branch load derating factor, and perform nonlinear scaling calculation based on the current sharing reference value and the branch load derating factor to generate the branch target current reference value.

[0029] The present invention is improved in that the process of generating the target current reference value of the branch based on the current sharing reference value and the branch load derating factor through nonlinear scaling calculation specifically includes:

[0030] Obtain non-linear scaling weights;

[0031] The initial reference characteristics are obtained by calculating the product of the current sharing reference value and the branch load derating factor.

[0032] The target current characteristic is obtained by calculating the adjusted product of the initial reference characteristic and the nonlinear scaling weight.

[0033] The target current characteristic is determined as the reference value of the target current of the branch;

[0034] The temperature regulation weighting coefficient is set as follows:

[0035] Extract the thermal resistance parameter and ambient temperature parameter of the activated power supply branch;

[0036] Calculate the ratio of the thermal resistance parameter to the ambient temperature parameter;

[0037] The ratio value is determined as the temperature regulation weighting coefficient;

[0038] The nonlinear scaling weights are set as follows:

[0039] Extract the current fluctuation parameters and rated current parameters of the activated power supply branch;

[0040] Calculate the ratio of the current fluctuation parameter to the rated current parameter;

[0041] The ratio data is determined as the nonlinear scaling weight.

[0042] The present invention is improved in that the specific steps of S4 are as follows:

[0043] S411: For each of the activated power supply branches, a preset droop impedance reference value is obtained, and the droop impedance reference value is corrected by the equivalent line impedance of the branch through the internal programmable gain device to generate the branch droop impedance parameter.

[0044] S412: Calculate the product of the branch output current and the branch droop impedance parameter to generate the branch impedance voltage drop characteristic, obtain the preset branch reference set voltage value, use an operational amplifier to calculate the negative feedback voltage difference between the branch reference set voltage value and the branch impedance voltage drop characteristic, and generate the branch simulated droop control voltage.

[0045] The present invention is improved in that the process of calculating the negative feedback voltage difference between the branch reference set voltage value and the branch impedance voltage drop characteristic using an operational amplifier is specifically as follows:

[0046] Obtain the amplifier gain coefficient;

[0047] Calculate the direct difference between the reference set voltage value of the branch and the impedance voltage drop characteristic of the branch;

[0048] Calculate the mapping value of the product of the direct difference characteristic and the amplifier gain coefficient;

[0049] The product mapping value is determined as the negative feedback voltage difference;

[0050] The method for setting the droop impedance reference value is as follows:

[0051] Extract the design maximum voltage drop parameter and rated full-load current parameter of the activated power supply branch;

[0052] Calculate the quotient of the maximum design voltage drop parameter and the rated full-load current parameter;

[0053] The quotient parameter is determined as the droop impedance reference value;

[0054] The amplifier gain coefficient is set as follows:

[0055] Extract the local feedback loop bandwidth parameter and power supply switching frequency parameter of the activated power supply branch;

[0056] Calculate the proportional characteristic value of the local feedback loop bandwidth parameter to the power supply switching frequency parameter;

[0057] The proportional characteristic value is determined as the amplifier gain coefficient.

[0058] The present invention is improved in that the specific steps of S5 are as follows:

[0059] S511: For each of the activated power supply branches, call the current output current of the branch and the reference value of the target current of the branch at the current moment, calculate the current deviation value between the reference value of the target current of the branch and the output current of the branch, obtain the branch current deviation term at the current moment, obtain the historical branch current deviation term cached at the previous moment, construct the parameter mapping relationship of the digital proportional-integral control algorithm based on the current branch current deviation term and the historical branch current deviation term, and generate discrete error input variables;

[0060] S512: Obtain the voltage conversion proportional gain parameter and voltage conversion integral gain parameter corresponding to the digital proportional-integral control algorithm; calculate the product value of the discrete error input variable and the voltage conversion proportional gain parameter based on the discrete error input variable and the voltage conversion proportional gain parameter to obtain the proportional response bias voltage; call the discrete error input variable and the voltage conversion integral gain parameter to calculate the cumulative product value of the discrete error input variable and the voltage conversion integral gain parameter to obtain the integral cumulative bias voltage; calculate the sum of the voltage of the proportional response bias voltage and the integral cumulative bias voltage to generate the branch digital compensation bias voltage;

[0061] S513: Process the branch digital compensation bias voltage using a digital-to-analog converter, extract the continuous voltage parameters corresponding to the branch digital compensation bias voltage, obtain the branch analog compensation voltage parameters, call the branch analog compensation voltage parameters and the branch analog droop control voltage, calculate the sum of the fused voltage of the branch analog compensation voltage parameters and the branch analog droop control voltage, and generate the board-level current sharing control signal.

[0062] The present invention is improved in that the process of calculating the sum of the fused voltage of the branch simulated compensation voltage parameter and the branch simulated droop control voltage to generate the board-level current sharing control signal is specifically as follows:

[0063] Obtain the voltage fusion weighting coefficients;

[0064] Calculate the adjusted compensation voltage based on the branch simulation compensation voltage parameters and the voltage fusion weighting coefficient;

[0065] Calculate the difference between constant 1 and the voltage fusion weighting coefficient;

[0066] Calculate the proportional droop voltage between the simulated droop control voltage of the branch and the difference value;

[0067] Calculate the sum of the adjusted compensation voltage and the proportional droop voltage;

[0068] The sum of the voltage values ​​is determined as the board-level current sharing control signal;

[0069] The voltage fusion weighting coefficient is set as follows:

[0070] Extract the digital control sampling frequency and analog feedback bandwidth frequency of the activated power supply branch;

[0071] Calculate the frequency ratio parameter between the digital control sampling frequency and the analog feedback bandwidth frequency;

[0072] The frequency ratio parameter is determined as the voltage fusion weighting coefficient.

[0073] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0074] In this invention, the power supply branches are screened and the average temperature is evaluated by collecting branch current and internal temperature. The equivalent line impedance is evaluated by collecting the total output bus voltage and branch output voltage to determine the current sharing reference value. The load derating factor is calculated based on the temperature parameter to generate the target current reference value. The droop impedance parameter is obtained by correcting the droop impedance reference value using the line impedance. A simulated droop control voltage is generated by combining the reference set voltage. A digital compensation bias voltage is generated by proportional-integral control based on the output current and target parameter. The simulated droop control voltage works together to generate a board-level current sharing control signal, realizing the fusion of fast response and accurate compensation, and avoiding local overheating when dealing with load changes. Attached Figure Description

[0075] Figure 1 This is the main flowchart of the current sharing control method for the board-level parallel power supply system of the present invention;

[0076] Figure 2 This is a flowchart of the activation branch and temperature equalization process of the present invention;

[0077] Figure 3 This is a flowchart of the line impedance correction and simulated droop control of the present invention;

[0078] Figure 4 This is a flowchart of the digital compensation and fusion control process of the present invention. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0080] Please see Figures 1 to 4 This embodiment provides a current sharing control method for a board-level parallel power supply system based on a DSP. In practical applications, such as when a power supply motherboard carries multiple parallel power supply branches and supplies power to the output bus, each power supply branch is configured with a current sampling channel, a voltage sampling channel, a temperature sampling channel, a local analog droop adjustment link, a programmable gain device, a digital-to-analog converter, and a data buffer connected to the DSP. The DSP reads the current, voltage, and temperature fields of each branch according to the sampling period, converts the digital compensation results into analog control quantities, and then superimposes them onto the local control link. The method includes the following steps:

[0081] S1: Collect the branch output current and internal temperature of the corresponding power supply branch. Based on the branch output current, filter out active power supply branches to obtain the branch quantity characteristics. Use the internal temperature of the active power supply branch to evaluate the average branch temperature. The branch output current is the current status field sent to the DSP after sampling and conditioning from the current sampling channel. The branch internal temperature is the thermal status field output from the temperature detection channel located in the branch power device, magnetic device, or heat dissipation coupling area. Active power supply branches are those whose current status field meets the power supply participation judgment rules and are not in the protection exit state within the current sampling period. The branch quantity characteristic is the active branch count field generated by the DSP in this period. This field is stored in the periodic operation cache and called by the current sharing reference value generation process.

[0082] S111: Based on a preset sampling period, the digital signal processor on the motherboard collects the output current and internal temperature of each power supply branch. Based on the output current, it filters out active power supply branches in active mode and counts the number of active power supply branches to obtain branch quantity characteristics. The sampling period is a configurable field in the DSP timing task table. Its engineering boundaries are jointly defined by the power control loop update requirements, sampling interface conversion capabilities, and board-level communication scheduling relationships. The update condition is the arrival of a periodic task or a protection status refresh.

[0083] For non-numerical activation determination features, this embodiment establishes quantization rules: the "hardware enable high-level signal" read by the system is quantized as a logic value of 1, and the "hardware overcurrent protection not triggered" is quantized as a logic value of 1. When the branch current sampling value is greater than the noise floor determination threshold set based on the sensor zero-point drift characteristics (set to 0.5A in this embodiment) and both of the above two logic quantization values ​​are 1, the branch is determined to be activated. In a system containing 4 parallel power supply branches, if the output currents of branches 1, 2, and 3 are collected as 10.2A, 9.8A, and 10.0A respectively with a logic value of 1, and the current of branch 4 is 0.1A and protection is triggered, then based on the above rules, branches 1, 2, and 3 are selected as activated power supply branches, and the branch quantity feature is statistically obtained as 3.

[0084] S112: The internal temperatures of each activated power supply branch are retrieved, and the sum of all branch internal temperatures is calculated to obtain a total temperature value. The ratio of this total temperature value to the average of the branch quantity characteristics is then calculated to generate the branch average temperature. The internal temperatures of the branches involved in the processing are first validated using a temperature channel. In specific implementation, a surface-mount NTC thermistor combined with a 12-bit ADC is used to acquire the internal temperature, and the valid internal temperatures of branches 1, 2, and 3 are extracted to be 65.5℃, 68.2℃, and 64.3℃, respectively.

[0085] The calculation process is as follows:

[0086] The total temperature value = 65.5℃ + 68.2℃ + 64.3℃ = 198.0℃;

[0087] Average temperature of branch circuit = 198.0℃ / 3 = 66.0℃;

[0088] The results indicate that the overall heat load of the current system is within the safe temperature rise range for normal operation (the preset safe reference average temperature is usually between 45.0℃ and 75.0℃), and provide a reliable dynamic reference line for subsequent determination of local hot spots in each individual branch.

[0089] S2: For the activated power supply branch, collect the total output bus voltage and the branch output voltage, and evaluate the equivalent line impedance of the branch in combination with the branch output current. Evaluate the current sharing reference value based on the branch output current and the characteristics of the number of branches.

[0090] S211: For each activated power supply branch, the digital signal processor acquires the corresponding branch output voltage and the total output bus voltage, calculates the voltage difference between the total output bus voltage and the branch output voltage to obtain the branch voltage deviation characteristics, calls the corresponding branch output current, and calculates the impedance mapping ratio between the branch voltage deviation characteristics and the branch output current to obtain the branch equivalent line impedance. Through the precision voltage divider network at the power output terminal, the output voltage of branch 2 is acquired to be 12.05V, the total output bus voltage is 12.00V, and the corresponding branch output current is 9.8A.

[0091] The calculation process is as follows:

[0092] Branch voltage deviation characteristic = 12.05V - 12.00V = 0.05V = 50mV;

[0093] The wiring length of branch 2 extracted from the non-volatile configuration area is 120mm. Based on the copper cross-sectional area, the conductivity resistivity is 0.01 (dimensionless conversion ratio). The calculation process is as follows:

[0094] Impedance correction weighting factor = 120 × 0.01 = 1.2;

[0095] Corrected voltage deviation characteristic = 50mV × 1.2 = 60mV;

[0096] The basic impedance mapping ratio = 60mV / 9.8A = 6.12mOhm;

[0097] The effects of line aging are calculated by reading the cumulative running time (10,000 hours) and the limiting attenuation constant calibrated by physical experiments (50,000 hours):

[0098] Line aging compensation baseline value = 10000 / 50000 = 0.2mOhm;

[0099] The equivalent line impedance of the branch is 6.12 mOhm + 0.2 mOhm = 6.32 mOhm.

[0100] The advantage of the formula is that by introducing the physical parameters of PCB wiring and the historical operating time of the equipment, it breaks through the limitations of the traditional method of simply relying on electrical differences to calculate impedance.

[0101] S212: Call the branch output current corresponding to each activated power supply branch, calculate the sum of the currents of all branch output currents, obtain the branch current accumulation characteristic, calculate the average reference ratio of the branch current accumulation characteristic and the branch quantity characteristic, and generate the current sharing reference value. Call the currents of the valid branch 1 (10.2A), branch 2 (9.8A) and branch 3 (10.0A), and combine them with the branch quantity characteristic (3).

[0102] The calculation process is as follows:

[0103] The cumulative characteristic of branch current = 10.2A + 9.8A + 10.0A = 30.0A;

[0104] Reference value for current sharing = 30.0A / 3 = 10.0A;

[0105] This result indicates that, under the current total load demand, the ideal current rating that each active branch should share should serve as the absolute starting reference for subsequent derating calculations based on the differences in the characteristics of each branch.

[0106] S3: Calculate the branch load derating factor based on the branch internal temperature, branch average temperature and preset temperature adjustment weighting factor corresponding to the activated power supply branch, and generate the branch target current reference value based on the current sharing reference value and the branch load derating factor.

[0107] S311: For each activated power supply branch, retrieve the corresponding branch internal temperature and branch average temperature, calculate the temperature deviation between the branch internal temperature and the branch average temperature, obtain the branch temperature difference characteristic, and calculate the ratio of the branch temperature difference characteristic to the branch average temperature based on the branch temperature difference characteristic and the branch average temperature to generate the relative temperature deviation ratio. Taking branch 2 as an example, retrieve its internal temperature (68.2℃) and branch average temperature (66.0℃).

[0108] The calculation process is as follows:

[0109] Branch temperature difference characteristic = 68.2℃ - 66.0℃ = 2.2℃;

[0110] Relative temperature deviation ratio = 2.2℃ / 66.0℃ = 0.0333;

[0111] S312: Obtain the preset temperature regulation weight coefficient, calculate the product of the relative temperature deviation ratio and the temperature regulation weight coefficient to obtain the branch temperature compensation offset, call the branch temperature compensation offset, calculate the difference characteristic quantity between constant 1 and the branch temperature compensation offset, and generate the branch load derating coefficient. The thermal resistance parameter is extracted from the thermal design calibration file as 2.5℃ / W, and the ambient temperature parameter is 25.0℃.

[0112] The calculation process is as follows:

[0113] Temperature regulation weighting coefficient = 2.5 / 25.0 = 0.1;

[0114] Branch temperature compensation offset = 0.0333 × 0.1 = 0.00333;

[0115] Branch load derating factor = 1 - 0.00333 = 0.99667;

[0116] S313: Call the current sharing reference value and the branch load derating factor, and perform nonlinear scaling calculation based on the current sharing reference value and the branch load derating factor to generate the target current reference value of the branch. Extract the effective steady-state continuous current parameter (19.8A) within the recent sliding window of the branch, replace the transient current to suppress noise, and extract the design rated current parameter (20.0A).

[0117] The calculation process is as follows:

[0118] Initial reference feature = 10.0A × 0.99667 = 9.9667A;

[0119] Nonlinear scaling weight = 19.8A / 20.0A = 0.99;

[0120] The target current reference value for the branch is 9.9667A × 0.99 = 9.867A;

[0121] The advantage of this formula lies in its ability to proactively reduce the expected load on a single tube before an abnormal temperature rise occurs by integrating cross-domain thermal resistance data, ambient temperature, and steady-state load characteristics. The calculation results show that, because branch 2's temperature is higher than the system average temperature, its target current is reasonably reduced from the absolute average of 10.0A to 9.867A, ensuring a thermal balance effect.

[0122] S4: By using a programmable gain device, the branch equivalent line impedance is used to correct the preset droop impedance reference value to generate the branch droop impedance parameter. The branch output current, the branch droop impedance parameter and the preset branch reference voltage value are used to generate the branch simulated droop control voltage.

[0123] S411: For each active power supply branch, a preset droop impedance reference value is obtained. The internal programmable gain device corrects the droop impedance reference value using the branch's equivalent line impedance, generating the branch droop impedance parameter. The maximum design voltage drop parameter is extracted from the system specifications as 0.6V, and the rated full-load current parameter is 20.0A.

[0124] The calculation process is as follows:

[0125] Droop impedance reference value = 0.6V / 20.0A = 0.03 Ohm = 30.0mOhm;

[0126] Branch droop impedance parameter = 30.0 mOhm + 6.32 mOhm = 36.32 mOhm;

[0127] S412: Calculate the product of the branch output current and the branch droop impedance parameter to generate the branch impedance voltage drop characteristic, obtain the preset branch reference set voltage value, use an operational amplifier to calculate the negative feedback voltage difference between the branch reference set voltage value and the branch impedance voltage drop characteristic, and generate the branch simulated droop control voltage. The extracted system reference set voltage value for the branch is 12.00V.

[0128] The calculation process is as follows:

[0129] Branch impedance voltage drop characteristic = 9.8A × 36.32mOhm = 355.936mV ≈ 0.356V;

[0130] Direct difference characteristic = 12.00V - 0.356V = 11.644V;

[0131] Read the ADC register and PWM configuration, extract the local feedback loop bandwidth parameter as 10kHz, and the power supply switching frequency parameter as 100kHz:

[0132] Amplifier gain factor = 10kHz / 100kHz = 0.1;

[0133] The simulated droop control voltage of the branch circuit = 11.644V × 0.1 = 1.164V;

[0134] S5: Based on the branch output current, the branch target current reference value and the historical branch current deviation term, the branch digital compensation bias voltage is generated through the digital proportional-integral control algorithm. The branch analog compensation voltage parameter is obtained by the digital-to-analog converter. The board-level current sharing control signal is generated based on the branch analog compensation voltage parameter and the branch analog droop control voltage.

[0135] S511: For each activated power supply branch, call the current branch output current and the branch target current reference value, calculate the current deviation value between the branch target current reference value and the branch output current, and obtain the branch current deviation term at the current moment.

[0136] The calculation process is as follows:

[0137] The branch current deviation at the current moment is 9.867A - 9.8A = 0.067A;

[0138] The historical branch current deviation term (set to 0.05A) of the previous cycle is obtained from the DSP timing buffer and combined with the current deviation to construct the discrete error input variable of the digital proportional-integral algorithm.

[0139] S512: Obtain the voltage conversion proportional gain parameter and voltage conversion integral gain parameter corresponding to the digital proportional-integral control algorithm. Based on the discrete error input variable and the voltage conversion proportional gain parameter, calculate the product value of the discrete error input variable and the voltage conversion proportional gain parameter to obtain the proportional response bias voltage. Call the discrete error input variable and the voltage conversion integral gain parameter to calculate the accumulated product value of the discrete error input variable and the voltage conversion integral gain parameter to obtain the integral accumulated bias voltage. Calculate the sum of the superimposed voltages to generate the branch digital compensation bias voltage. Based on the loop Bode plot, the pre-matched proportional gain parameter is 0.5 and the integral gain parameter is 0.01, and the accumulated historical integral is read as 0.15V.

[0140] The calculation process is as follows:

[0141] Proportional response bias voltage = 0.067A × 0.5 = 0.0335V;

[0142] Integral cumulative bias voltage = 0.15V + (0.067A × 0.01) = 0.15067V;

[0143] Branch digital compensation bias voltage = 0.0335V + 0.15067V = 0.18417V;

[0144] S513: The branch digital compensation bias voltage is processed by a digital-to-analog converter, the continuous voltage parameters corresponding to the branch digital compensation bias voltage are extracted, the branch analog compensation voltage parameters are obtained, the branch analog compensation voltage parameters and the branch analog droop control voltage are called, the sum of the fused voltage of the branch analog compensation voltage parameters and the branch analog droop control voltage is calculated, and the board-level current sharing control signal is generated.

[0145] The actual analog compensation voltage parameter of the branch obtained by mapping through the 12-bit DAC output and subtracting link losses is 0.184V. The digital control sampling frequency is extracted to be 20kHz, and the analog feedback bandwidth frequency is 100kHz.

[0146] The calculation process is as follows:

[0147] Voltage fusion weighting coefficient = 20kHz / 100kHz = 0.2;

[0148] Adjust the compensation voltage = 0.184V × 0.2 = 0.0368V;

[0149] The difference is 1 - 0.2 = 0.8;

[0150] The equal-gradient droop voltage = 1.164V × 0.8 = 0.9312V;

[0151] The total voltage is calculated as follows: 0.0368V + 0.9312V = 0.968V.

[0152] 0.968V was determined as the final board-level current sharing control signal sent to the voltage regulation interface. This result demonstrates that the system successfully superimposed a precise and smooth digital closed-loop intervention on top of the fast-response analog droop. The advantage of this formula is that, compared to the delay oscillations caused by purely digital loops or the steady-state error problems existing in purely analog loops, this weighted fusion mechanism based on digital and analog frequency band characteristics shortens the dynamic recovery time of the power supply system during load steps.

[0153] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A current sharing control method for a DSP-based board-level parallel power supply system, characterized in that, Includes the following steps: S1: Collect the branch output current and internal temperature of the corresponding power supply branch, filter out the active power supply branches based on the branch output current to obtain the branch quantity characteristics, and use the internal temperature of the branch corresponding to the active power supply branch to evaluate the average temperature of the branch. S2: For the activated power supply branch, collect the total output bus voltage and the branch output voltage, and evaluate the equivalent line impedance of the branch in combination with the branch output current. Evaluate the current sharing reference value based on the branch output current and the branch quantity characteristics. S3: Calculate the branch load derating factor based on the branch internal temperature, the branch average temperature and the preset temperature adjustment weighting factor corresponding to the activated power supply branch, and generate the branch target current reference value based on the current sharing reference value and the branch load derating factor; S4: The branch droop impedance parameter is generated by correcting the preset droop impedance reference value using the equivalent line impedance of the branch through a programmable gain device. The branch simulated droop control voltage is generated by using the branch output current, the branch droop impedance parameter and the preset branch reference voltage value. S5: Based on the branch output current, the branch target current reference value and the historical branch current deviation term, a digital compensation bias voltage for the branch is generated through a digital proportional-integral control algorithm. The branch analog compensation voltage parameter is obtained by processing through a digital-to-analog converter. A board-level current sharing control signal is generated based on the branch analog compensation voltage parameter and the branch analog droop control voltage.

2. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 1, characterized in that, The specific steps of S1 are as follows: S111: According to the preset sampling period, the output current of each power supply branch and the internal temperature of each power supply branch are collected by the digital signal processor on the motherboard. Based on the output current of the branch, the active power supply branches in the active mode are selected from each power supply branch, and the number of active power supply branches is counted to obtain the branch quantity characteristics. S112: Call the internal temperature of each activated power supply branch, calculate the sum of the internal temperatures of all branches, obtain the total temperature value, calculate the ratio of the total temperature value to the average of the branch quantity characteristics, and generate the branch average temperature.

3. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 2, characterized in that, The specific steps of S2 are as follows: S211: For each of the activated power supply branches, the digital signal processor collects the corresponding branch output voltage and the total output bus voltage, calculates the voltage difference between the total output bus voltage and the branch output voltage to obtain the branch voltage deviation characteristics, calls the corresponding branch output current, calculates the impedance mapping ratio between the branch voltage deviation characteristics and the branch output current, and obtains the branch equivalent line impedance. S212: Call the branch output current corresponding to each of the activated power supply branches, calculate the sum of the currents of all the branch output currents, obtain the branch current accumulation feature, calculate the average reference ratio of the branch current accumulation feature and the branch quantity feature, and generate the current sharing reference value.

4. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 3, characterized in that, The process of calculating the impedance mapping ratio between the branch voltage deviation characteristic and the branch output current is as follows: The product of the branch voltage deviation characteristic and the impedance correction weighting coefficient is calculated to obtain the corrected voltage deviation characteristic; The quotient of the corrected voltage deviation characteristic and the branch output current is calculated to obtain the basic impedance mapping ratio. Calculate the sum of the basic impedance mapping ratio and the line aging compensation benchmark value, and use the sum as the equivalent line impedance of the branch. The impedance correction weighting coefficient is set as follows: Extract the wiring length and resistivity of the activated power supply branch, calculate the correlation value of the product of the wiring length and the resistivity, and determine the correlation value of the product as the impedance correction weighting coefficient. The method for setting the baseline value for line aging compensation is as follows: Extract the cumulative operating time of the activated power supply branch, obtain the preset limit attenuation constant, calculate the ratio mapping value of the cumulative operating time to the limit attenuation constant, and determine the ratio mapping value as the line aging compensation benchmark value.

5. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 3, characterized in that, The specific steps of S3 are as follows: S311: For each activated power supply branch, call the corresponding branch internal temperature and branch average temperature, calculate the temperature deviation value between the branch internal temperature and the branch average temperature, obtain the branch temperature difference characteristic, and calculate the difference ratio between the branch temperature difference characteristic and the branch average temperature based on the branch temperature difference characteristic and the branch average temperature to generate a relative temperature deviation ratio. S312: Obtain the preset temperature adjustment weight coefficient, calculate the product value of the relative temperature deviation ratio and the temperature adjustment weight coefficient, obtain the branch temperature compensation offset, call the branch temperature compensation offset, calculate the difference feature value between constant one and the branch temperature compensation offset, and generate the branch load derating coefficient. S313: Call the current sharing reference value and the branch load derating factor, and perform nonlinear scaling calculation based on the current sharing reference value and the branch load derating factor to generate the branch target current reference value.

6. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 5, characterized in that, The process of generating the target current reference value for the branch based on the current sharing reference value and the branch load derating factor through nonlinear scaling calculation specifically includes: Obtain non-linear scaling weights; The initial reference characteristics are obtained by calculating the product of the current sharing reference value and the branch load derating factor. The target current characteristic is obtained by calculating the adjusted product of the initial reference characteristic and the nonlinear scaling weight. The target current characteristic is determined as the reference value of the target current of the branch; The temperature regulation weighting coefficient is set as follows: Extract the thermal resistance parameter and ambient temperature parameter of the activated power supply branch; Calculate the ratio of the thermal resistance parameter to the ambient temperature parameter; The ratio value is determined as the temperature regulation weighting coefficient; The nonlinear scaling weights are set as follows: Extract the current fluctuation parameters and rated current parameters of the activated power supply branch; Calculate the ratio of the current fluctuation parameter to the rated current parameter; The ratio data is determined as the nonlinear scaling weight.

7. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 5, characterized in that, The specific steps of S4 are as follows: S411: For each of the activated power supply branches, a preset droop impedance reference value is obtained, and the droop impedance reference value is corrected by the equivalent line impedance of the branch through the internal programmable gain device to generate the branch droop impedance parameter. S412: Calculate the product of the branch output current and the branch droop impedance parameter to generate the branch impedance voltage drop characteristic, obtain the preset branch reference set voltage value, use an operational amplifier to calculate the negative feedback voltage difference between the branch reference set voltage value and the branch impedance voltage drop characteristic, and generate the branch simulated droop control voltage.

8. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 7, characterized in that, The process of using an operational amplifier to calculate the negative feedback voltage difference between the branch reference set voltage value and the branch impedance voltage drop characteristic is as follows: Obtain the amplifier gain coefficient; Calculate the direct difference between the reference set voltage value of the branch and the impedance voltage drop characteristic of the branch; Calculate the mapping value of the product of the direct difference characteristic and the amplifier gain coefficient; The product mapping value is determined as the negative feedback voltage difference; The method for setting the droop impedance reference value is as follows: Extract the design maximum voltage drop parameter and rated full-load current parameter of the activated power supply branch; Calculate the quotient of the maximum design voltage drop parameter and the rated full-load current parameter; The quotient parameter is determined as the droop impedance reference value; The amplifier gain coefficient is set as follows: Extract the local feedback loop bandwidth parameter and power supply switching frequency parameter of the activated power supply branch; Calculate the proportional characteristic value of the local feedback loop bandwidth parameter to the power supply switching frequency parameter; The proportional characteristic value is determined as the amplifier gain coefficient.

9. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 7, characterized in that, The specific steps of S5 are as follows: S511: For each of the activated power supply branches, call the current output current of the branch and the reference value of the target current of the branch at the current moment, calculate the current deviation value between the reference value of the target current of the branch and the output current of the branch, obtain the branch current deviation term at the current moment, obtain the historical branch current deviation term cached at the previous moment, construct the parameter mapping relationship of the digital proportional-integral control algorithm based on the current branch current deviation term and the historical branch current deviation term, and generate discrete error input variables; S512: Obtain the voltage conversion proportional gain parameter and voltage conversion integral gain parameter corresponding to the digital proportional-integral control algorithm; calculate the product value of the discrete error input variable and the voltage conversion proportional gain parameter based on the discrete error input variable and the voltage conversion proportional gain parameter to obtain the proportional response bias voltage; call the discrete error input variable and the voltage conversion integral gain parameter to calculate the cumulative product value of the discrete error input variable and the voltage conversion integral gain parameter to obtain the integral cumulative bias voltage; calculate the sum of the voltage of the proportional response bias voltage and the integral cumulative bias voltage to generate the branch digital compensation bias voltage; S513: Process the branch digital compensation bias voltage using a digital-to-analog converter, extract the continuous voltage parameters corresponding to the branch digital compensation bias voltage, obtain the branch analog compensation voltage parameters, call the branch analog compensation voltage parameters and the branch analog droop control voltage, calculate the sum of the fused voltage of the branch analog compensation voltage parameters and the branch analog droop control voltage, and generate the board-level current sharing control signal.

10. The current sharing control method for a DSP-based board-level parallel power supply system according to claim 9, characterized in that, The process of calculating the sum of the branch simulated compensation voltage parameter and the branch simulated droop control voltage to generate the board-level current sharing control signal is as follows: Obtain the voltage fusion weighting coefficients; Calculate the adjusted compensation voltage based on the branch simulation compensation voltage parameters and the voltage fusion weighting coefficient; Calculate the difference between constant 1 and the voltage fusion weighting coefficient; Calculate the proportional droop voltage between the simulated droop control voltage of the branch and the difference value; Calculate the sum of the adjusted compensation voltage and the proportional droop voltage; The sum of the voltage values ​​is determined as the board-level current sharing control signal; The voltage fusion weighting coefficient is set as follows: Extract the digital control sampling frequency and analog feedback bandwidth frequency of the activated power supply branch; Calculate the frequency ratio parameter between the digital control sampling frequency and the analog feedback bandwidth frequency; The frequency ratio parameter is determined as the voltage fusion weighting coefficient.

Images