A power reference voltage suspension superposition method and a direct current stabilized power supply

Abstract

The present application relates to a kind of power reference voltage suspension superposition method and DC regulated power supply, the method includes: real-time acquisition and processing load current time series data, extract the reference current rate of change characterized by variation trend;Based on the preset alert threshold, intelligently determine that load is in steady state, slowly varying, mutation or energy feedback working condition;According to different working conditions, determine the suspension ground offset voltage in sampling period, and the voltage of power main power circuit output is subtracted by suspension ground offset voltage, generate the actual output voltage of power supply in sampling period;Utilize proportional-integral control algorithm to carry out closed-loop fine tuning to suspension ground offset voltage, until the error absolute value between actual output voltage and the rated working voltage of load is less than the voltage tolerance range allowed by load.The present application realizes the advanced perception and compensation to load mutation, greatly improves dynamic response speed and steady accuracy, and can safely handle energy feedback and other special working conditions, enhances the adaptability and reliability of system.

Description

Technical Field

[0001] This invention relates to the field of regulated power supplies, and in particular to a method for floating and superimposing a power supply reference voltage and a DC regulated power supply. Background Technology

[0002] As the core power supply unit of modern electronic devices, the dynamic response performance of DC regulated power supplies directly determines the stability and reliability of the entire system. Currently, mainstream DC regulated power supplies, whether linear or switching architectures, rely on a feedback closed loop of the output voltage for control. This feedback system has a fundamental limitation: it can only detect the error and initiate correction after the output voltage has actually deviated. This passive mode of delayed response inevitably leads to significant voltage drops or overshoots when faced with transient changes in load current, making it difficult to meet the stringent requirements of high-performance applications for voltage accuracy and dynamic response.

[0003] Furthermore, in traditional control schemes, in order to compensate for this voltage deviation, the system must ultimately correct the output voltage by adjusting the operating state of the main power stage (e.g., changing the pulse width modulation duty cycle of the switching power supply). However, the response speed of this direct control of the main power circuit is subject to multiple physical limitations, such as the switching delay of power devices, the characteristics of the drive circuit, and the bandwidth of the main control loop. When the load undergoes extremely rapid step changes, the calculation and execution speed of the main loop often cannot keep up with the transient change rate of the load, resulting in poor compensation effect and even potentially causing system oscillation, which in turn deteriorates the output quality. Summary of the Invention

[0004] The purpose of this invention is to provide a method for floating and superimposing a power supply reference voltage and a DC regulated power supply to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for floating and superimposing a power supply reference voltage includes the following steps:

[0007] Based on the current timing data of the power output circuit during the sampling period, the current change rate is analyzed and filtered to generate a reference current change rate that characterizes the load change trend during the sampling period.

[0008] Set a warning threshold to distinguish load operating conditions, and combine it with the reference current change rate to determine the operating condition type of the load operating condition within the sampling period. The operating condition types of the load operating condition include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition.

[0009] Based on the operating condition type of the load, the floating ground offset voltage within the sampling period is determined, and the voltage output by the main power circuit of the power supply is subtracted from the floating ground offset voltage to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative output terminal of the power supply and the common ground terminal.

[0010] Based on the real-time deviation between the actual output voltage and the rated operating voltage of the load within the sampling period, the proportional-integral control algorithm is used to iteratively update the control cycle until the absolute value of the error between the actual output voltage and the rated operating voltage of the load is less than the allowable voltage tolerance range of the load. The proportional coefficient and integral coefficient in the proportional-integral control algorithm are then determined to correct the floating ground offset voltage in the next sampling period.

[0011] Furthermore, generating a reference current change rate to characterize the load change trend within the sampling period includes the following steps:

[0012] Within the sampling period, sampling points are set according to the set sampling frequency to retrieve the load current under each sampling point, obtain the current time series data within the sampling period, calculate the current change between adjacent sampling points, and calculate the original current change rate of each sampling point in combination with the sampling time interval, and write the original current change rate into the data buffer.

[0013] The current change rate time series data of the power supply output circuit under steady-state conditions is obtained in multiple segments as reference time series data under steady-state conditions. After preprocessing each segment of reference time series data, the mean current change rate of all reference time series data is determined by combining statistical averaging algorithm. The preprocessing includes removing outliers based on the quartile method and filling missing values ​​based on the mean imputation method.

[0014] Each raw current change rate in the data buffer is compared with the mean current change rate. If the absolute value of the raw current change rate is less than or equal to the mean current change rate, the raw current change rate is marked as a normal target object. If the absolute value of the raw current change rate is greater than the mean current change rate, the raw current change rate is marked as an abnormal target object.

[0015] Calculate the ratio of the number of all abnormal target objects in the data buffer to the number of all original current change rates in the data buffer. If the ratio is less than 0.3, the average current change rate of all original current change rates in the data buffer is used as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.3 and less than 0.7, the average current change rate of all original current change rates in the data buffer is multiplied by an attenuation coefficient less than 1 to use as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.7, the average current change rate of all abnormal target objects in the data buffer is used as the reference current change rate characterizing the load change trend.

[0016] Furthermore, the warning thresholds include a first warning threshold and a second warning threshold, and the logic for determining the warning thresholds is as follows:

[0017] Based on the pre-processed reference timing data of each segment of the power output circuit under steady-state conditions, the overall standard deviation of all reference timing data is calculated.

[0018] Based on the overall standard deviation and mean current change rate of all reference time series data, the product of the overall standard deviation and the first preset margin factor is calculated, and the sum of the product and the mean current change rate is used as the first warning threshold. The first preset margin factor is determined to be 3 based on the three sigma criterion.

[0019] Calculate the product of the first warning threshold and the second preset margin factor, and use the product as the second warning threshold. The second preset margin factor is a positive number less than 1.

[0020] Furthermore, determining the operating condition type of the load during the sampling period includes the following steps:

[0021] If the rate of change of the reference current within the sampling period is greater than or equal to zero, and the rate of change of the reference current is less than the second warning threshold, then the load operation condition is defined as a steady-state condition.

[0022] If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the second warning threshold and less than the first warning threshold, then the load operation condition is defined as a slow-change condition.

[0023] If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the first warning threshold, then the load operating condition is defined as a sudden change condition.

[0024] If the rate of change of the reference current during the sampling period is less than zero, the load operation is defined as the energy feedback condition.

[0025] Furthermore, the processing logic for defining the load operation as a steady-state condition is as follows:

[0026] Based on the impedance measurement device connected to the output capacitor of the main power circuit, the complex impedance value of the output capacitor of the main power circuit is obtained at each frequency point, the impedance amplitude and frequency response curve of the output capacitor of the main power circuit is generated, the global minimum impedance value in the impedance amplitude and frequency response curve is identified, and the value of the minimum impedance value is determined as the equivalent series resistance of the output capacitor.

[0027] The equivalent series resistance is multiplied by the load current during the sampling period to generate the resistive voltage during the sampling period, and the difference between the common ground voltage and the resistive voltage is taken as the floating ground offset voltage during the sampling period.

[0028] Furthermore, the processing logic for load operation under gradually changing conditions is as follows:

[0029] A differential probe is used to determine the load's sag voltage during the sampling period between the power input terminal and the common ground terminal of the load device, and the ratio of the absolute value of the sag voltage to the rate of change of the reference current is calculated to generate the equivalent inductance during the sampling period.

[0030] The difference between the reference current change rate and the second warning threshold during the sampling period is multiplied by the equivalent inductance to generate the first prediction compensation voltage. The difference between the common ground terminal voltage and the first prediction compensation voltage is used as the floating ground offset voltage during the sampling period.

[0031] Furthermore, the handling logic for load operation conditions that exhibit abrupt changes is as follows:

[0032] Based on electromagnetic field simulation software, the parasitic inductance on the main power printed circuit board of the power supply is determined, and the parasitic inductance on the main power printed circuit board of the power supply is multiplied by the rate of change of the reference current during the sampling period to generate the inductive voltage during the sampling period.

[0033] Based on the inductive and resistive voltages within the sampling period, the sum of the inductive and resistive voltages is calculated to generate the second predictive compensation voltage within the sampling period. The difference between the common ground voltage and the second predictive compensation voltage is then used as the floating ground offset voltage within the sampling period.

[0034] Furthermore, the processing logic for the load operating condition as energy feedback mode is as follows: based on the type of semiconductor device in the load, determine the forward conduction voltage of the semiconductor device, and use the difference between the forward conduction voltage and the common ground voltage as the floating ground offset voltage within the sampling period.

[0035] Furthermore, the mathematical expression for the proportional-integral control algorithm is as follows:

[0036]

[0037] in, This represents the corrected floating ground offset voltage. This represents the floating ground offset voltage during the sampling period. This is the proportionality coefficient. This indicates the rated operating voltage of the load. This indicates the actual output voltage of the power supply during the sampling period. is the integral coefficient.

[0038] The present invention also provides a DC regulated power supply, comprising:

[0039] The main power circuit is used to provide output voltage to the load;

[0040] A floating ground drive unit is connected in series between the output negative terminal of the main power circuit and the system common ground, and is used to generate a controllable floating ground offset voltage.

[0041] The current sampling module is used to collect the current timing data of the power supply output circuit during the sampling period;

[0042] The data processing module is connected to the current sampling module and is used to analyze the rate of change of current in the current time series data and perform filtering to generate a reference rate of change of current to characterize the load change trend.

[0043] The operating condition judgment module is connected to the data processing module and has a preset warning threshold for distinguishing load operating conditions. It is used to determine the load operating condition type within the sampling period based on the comparison result of the reference current change rate and the warning threshold. The operating condition types include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition.

[0044] The voltage compensation module is connected to the output terminals of the operating condition judgment module and the main power circuit respectively. It is used to determine the floating ground offset voltage within the sampling period based on the operating condition type, and subtract the floating ground offset voltage from the voltage output by the main power circuit to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative terminal of the power supply output and the common ground terminal.

[0045] The control module is connected to the voltage compensation module and is used to iteratively update the control parameters using a proportional-integral control algorithm based on the real-time deviation between the actual output voltage and the rated operating voltage of the load, so as to correct the floating ground offset voltage in the next sampling time period until the absolute value of the error between the actual output voltage and the rated operating voltage is less than the allowable voltage tolerance range of the load.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] By analyzing the current change rate in real time during the sampling period and comparing it with a dynamic warning threshold, the system intelligently distinguishes between steady-state, slowly changing, abruptly changing, or energy feedback operating conditions of the load. Based on the determined operating condition, it dynamically adjusts the voltage difference between the negative terminal of the power supply output and the common ground terminal. The main power stage can operate stably in its inherent high-efficiency state. Fine-tuning of the voltage is achieved through an independent floating ground drive unit, which not only overcomes the physical response bottleneck of the main loop but also fundamentally eliminates the oscillation risk caused by rapid adjustment of the main power circuit, ensuring output quality. Furthermore, this predictive compensation strategy improves... It departs from the traditional reactive approach of post-event remediation, proactively intervening before the actual voltage drop occurs, greatly shortening the recovery time and effectively meeting the stringent requirements of high-performance applications for voltage accuracy and dynamic response. Furthermore, combined with a proportional-integral control algorithm, the system iteratively updates based on real-time deviations in each control cycle, continuously optimizing the proportional and integral coefficients to correct the floating ground offset voltage in the next sampling cycle. This ensures that the output voltage is firmly locked within the tolerance range of the rated operating voltage under various disturbances and parameter drifts, achieving high-precision output and excellent adaptive robustness under all operating conditions. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of the power supply reference voltage floating superposition method of the present invention;

[0050] Figure 2 This is a schematic diagram of the DC regulated power supply structure of the present invention.

[0051] Figure label:

[0052] 510. Main power circuit; 520. Floating ground drive unit; 530. Current sampling module; 540. Data processing module; 550. Operating condition judgment module; 560. Voltage compensation module; 570. Control module. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] The following is combined with Figures 1-2 This invention describes a method for floating and superimposing a power supply reference voltage and a DC regulated power supply.

[0055] Example:

[0056] Please see Figure 1 This invention provides a method for floating and superimposing a power supply reference voltage, comprising the following steps:

[0057] Step S100: Based on the current timing data of the power output circuit during the sampling period, analyze its current change rate, and filter the current change rate to generate a reference current change rate for characterizing the load change trend during the sampling period.

[0058] In this embodiment, generating the reference current change rate to characterize the load change trend within the sampling period specifically includes the following steps:

[0059] Step S110: Within the sampling period, set sampling points according to the set sampling frequency to retrieve the load current under each sampling point, obtain the current timing data within the sampling period, calculate the current change between adjacent sampling points, and calculate the original current change rate of each sampling point in combination with the sampling time interval, and write the original current change rate into the data buffer.

[0060] Set the sampling period (e.g.) ) and sampling frequency ( At each sampling point n, the voltage of the sampling resistor connected in series in the power output circuit is acquired by a high-precision analog-to-digital converter and converted into a load current value. (Unit: Ampere A) The load current values ​​are arranged in chronological order of sampling time to obtain the current timing data within the sampling period.

[0061] Assuming a CPU core is being powered, the current values ​​are continuously sampled at a certain stable moment: , , ...perform a first-order backward difference operation on the discrete current sequence to calculate the current change between adjacent sampling points: Divide the change by the sampling time interval The original rate of change of current is obtained, and its unit is... .

[0062] like , , The rate of change of the original current is This value indicates that the current is increasing at a rate of 200,000 amperes per second, suggesting that the load may be starting to increase.

[0063] The system maintains a first-in-first-out data buffer of length N (e.g., N=10). Each new raw current change rate is written to the end of the queue, while the oldest data is evicted. The data buffer always stores the raw current change rates of the most recent N sampling times within the sampling period (in this example, the most recent 11μs of data).

[0064] Step S120: Obtain time-series data of current change rate of multiple segments of power supply output circuit under steady-state conditions as reference time-series data under steady-state conditions. After preprocessing each segment of reference time-series data, and combining statistical averaging algorithm, determine the mean current change rate of all reference time-series data. The preprocessing includes removing outliers based on the quartile method and filling missing values ​​based on the mean imputation method.

[0065] Connect a purely resistive load (e.g., a power resistor) to the power supply output to simulate steady-state conditions, ensuring the load remains constant. Allow the power system to operate stably at the load's rated operating voltage for a sufficient time (e.g., several seconds) to achieve electrical equilibrium. This controls the high-precision analog-to-digital converter to operate at a fixed high-speed sampling rate (e.g., ...). The load current is collected by performing multiple samplings of finite duration to obtain multiple independent current time series data segments. For example, five data segments are collected, each containing 1000 consecutive current sampling points. Collecting multiple data segments instead of a single long data segment is to obtain statistical diversity and avoid accidental interference or specific patterns that may be encountered in a single sampling, making the final result more universal and robust. For each current data segment, the original current change rate sequence is calculated by applying the first-order backward difference formula.

[0066] It should be noted that the logic for outlier removal based on the quartile method is as follows: For each segment of reference time series data, its first quartile and third quartile are calculated, and the difference between the third quartile and the first quartile is calculated as the interquartile range. If there is a data point in the time series data that is less than the difference between the first quartile and 1.5 times the interquartile range, or greater than the sum of the third quartile and 1.5 times the interquartile range, then this data point is considered an outlier and removed. After removing outliers, the mean of the remaining data in the reference time series data, or the mean of the previous and next valid data points at the location of the missing value, is used to fill the missing value. The valid data points are the data points that were not removed. This is existing technology and will not be elaborated here.

[0067] The preprocessed current rate of change data are merged into a large reference time-series dataset, which may contain thousands of data points. These data points collectively characterize the system's noise variation characteristics under steady-state conditions, after removing occasional disturbances. The arithmetic mean of the reference time-series dataset is calculated. Assuming that the five data segments, after cleaning and merging, contain a total of 4995 data points, the sum is 250. Therefore, 250 / 4995 ≈ 0.05 This means that, under steady state, the rate of change (increase or decrease) of the current fluctuates randomly around an average level of 50,000 amperes per second, which is 0.05. The final value is the average rate of change of current for all reference timing data. This average rate of change of current becomes the baseline for the entire system to judge load dynamics. In subsequent processes, real-time data will be compared with this baseline to identify the real load activity signal.

[0068] Step S130: Compare each raw current change rate in the data buffer with the average current change rate. If the absolute value of the raw current change rate is less than or equal to the average current change rate, mark the raw current change rate as a normal target object. If the absolute value of the raw current change rate is greater than the average current change rate, mark the raw current change rate as an abnormal target object.

[0069] Set the data buffer length N = 10, and the sampling period... The data buffer contains the 10 most recently calculated raw current change rate data. The classification of normal and abnormal target objects is shown in Table 1 below:

[0070] Table 1. Comparison of Target Object Classification

[0071] 0.02 0.02 0.02 < 0.05 Normal target object -0.01 0.01 0.01 <0.05 Normal target object 0.08 0.08 0.08 > 0.05 Exception target object 0.15 0.15 0.15 > 0.05 Exception target object -0.03 0.03 0.03 < 0.05 Normal target object 0.28 0.28 0.28 > 0.05 Exception target object 0.12 0.12 0.12 > 0.05 Exception target object -0.2 0.2 0.20 > 0.05 Exception target object 0.05 0.05 0.05 = 0.05 Normal target object 0.1 0.1 0.10 > 0.05 Exception target object

[0072] Step S140: Calculate the ratio of the number of all abnormal target objects in the data buffer to the number of all original current change rates in the data buffer. If the ratio is less than 0.3, the average current change rate of all original current change rates in the data buffer is used as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.3 and less than 0.7, the average current change rate of all original current change rates in the data buffer is multiplied by an attenuation coefficient less than 1 to use as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.7, the average current change rate of all abnormal target objects in the data buffer is used as the reference current change rate characterizing the load change trend.

[0073] The cutoff points for the outlier ratio are set at 0.3 and 0.7, a comprehensive design based on engineering experience, statistical logic, and algorithm robustness. When the outlier ratio is less than 0.3, it means that within a sampling period, more than 70% of the data points do not exceed the average current change rate under steady-state conditions. There is sufficient statistical evidence to determine that no significant load change has occurred within the current sampling period. Those outliers (less than 30%) are more likely to be the upper limit of random noise fluctuations rather than the start of a systematic trend. Therefore, the system should adopt the most conservative strategy: ignore these few outliers and average the data across the entire window to smooth out noise to the greatest extent possible and output a baseline value. When the outlier ratio is greater than or equal to 0.7, it means that within a sampling period, more than 70% of the data points significantly exceed the average current change rate under steady-state conditions, indicating that the current sampling period... A clear and systematic load mutation is occurring within the system. Therefore, the system should adopt the most aggressive strategy, which is to fully trust the trend indicated by these outliers and use only the average of these outliers to calculate the baseline value, thereby amplifying the mutation signal. When the outlier ratio is greater than or equal to 0.3 and less than 0.7, it means that it is impossible to determine whether it is strong noise or weak mutation within a sampling period, and it is in an uncertain state. The algorithm adopts a compromise and conservative strategy, which is to still use the average of all data, but multiply it by an attenuation coefficient less than 1. The specific setting can be made by the staff according to the actual situation, and there is no restriction here. This part responds to the potential trend and avoids over-response due to misjudgment, which is equivalent to adding an observation period to the system.

[0074] According to the classification results in Table 1 above, there are 4 normal target objects (0.02, -0.01, -0.03, 0.05) and 6 abnormal target objects (0.08, 0.15, 0.28, 0.12, -0.20, 0.10). The abnormal proportion is [missing information]. Calculate the mean of the entire data buffer:

[0075]

[0076] Applying an attenuation coefficient (assumed to be 0.5), the rate of change of the reference current is obtained. .

[0077] Step S200: Set a warning threshold for distinguishing load operating conditions, and combine it with the reference current change rate to determine the operating condition type of the load operating condition within the sampling period. The operating condition types of the load operating condition include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition.

[0078] The warning thresholds include a first warning threshold and a second warning threshold. The logic for determining the warning thresholds is as follows:

[0079] Step S210: Based on the preprocessed reference timing data of each segment of the power supply output circuit under steady-state conditions, calculate the overall standard deviation of all reference timing data.

[0080] The core purpose of calculating the overall standard deviation of all reference time series data is to quantify the dispersion of the inherent, random current fluctuations (i.e., background noise) of the system under steady-state conditions, thereby providing key statistical basis for scientifically setting warning thresholds.

[0081] Step S220: Based on the overall standard deviation and the mean current change rate of all reference time series data, calculate the product of the overall standard deviation and the first preset margin factor, and use the sum of the product and the mean current change rate as the first warning threshold. The first preset margin factor is determined to be 3 based on the three sigma criterion.

[0082] For a dataset that follows a normal distribution, the probability that a data point falls within the mean ± 3 standard deviations is approximately 99.7%. This is known as the three sigma criterion, which means that the system can only confidently determine that an anomaly has occurred when the detected rate of change of current significantly and continuously exceeds the safety boundary defined by "mean + 3 standard deviations" which covers the vast majority of noise. This triggers subsequent compensation, ensuring the stability and anti-interference capability of the system.

[0083] Step S230: Calculate the product of the first warning threshold and the second preset margin factor, and use the product as the second warning threshold. The second preset margin factor is a positive number less than 1.

[0084] By establishing a more sensitive second warning threshold below the first warning threshold, a three-level refined judgment of load dynamics (steady state, gradual change, sudden change) can be achieved, rather than a simple normal / abnormal dichotomy. This enables the system to respond to gradual load changes earlier and more gently, while retaining a strong response capability to drastic changes, thus improving the adaptability and smoothness of control. The second preset margin factor can be set by the staff according to the actual situation, and there are no restrictions here.

[0085] Step S300: Based on the operating condition type of the load, determine the floating ground offset voltage within the sampling period, and subtract the floating ground offset voltage from the voltage output by the main power circuit of the power supply to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative output terminal of the power supply and the common ground terminal.

[0086] It should be noted that the main power circuit of the power supply uses a Buck circuit with a fixed duty cycle. In this case, its output voltage... From input voltage The duty cycle D is determined by a fixed duty cycle, where D is the ratio of the switching time of the transistor to the total cycle, following an ideal relationship. Output voltage It will vary with the input voltage The voltage changes slowly and proportionally. In actual power supply main power circuits, the characteristics of inductors, capacitors, switching transistors and diodes (such as parasitic parameters and conduction voltage) will introduce losses, making the actual output voltage slightly lower than the ideal calculated value. The common ground terminal is defined as the 0V reference point.

[0087] The process of determining the operating condition type of the load within the sampling period includes the following steps:

[0088] Step S310: If the rate of change of the reference current within the sampling period is greater than or equal to zero, and the rate of change of the reference current is less than the second warning threshold, then the load operation condition is defined as a steady-state condition.

[0089] A reference current change rate greater than zero within the sampling period indicates that the current is increasing. If the reference current change rate is less than the second warning threshold, it means that the current load current change trend is very weak, and its intensity does not even exceed this lower second warning threshold. A reference current change rate equal to zero indicates that the current has not changed. The current either remains unchanged or is increasing at an extremely slow and weak rate. The degree of this change is so small that it can be considered as system background noise or normal small fluctuations, which is insufficient to trigger any warning or compensation response.

[0090] It should be noted that the processing logic for defining the load operation as a steady-state condition is as follows:

[0091] Step S311: Based on the impedance measuring device connected to the output capacitor of the main power circuit, obtain the complex impedance value of the output capacitor of the main power circuit at each frequency point, generate the impedance amplitude and frequency characteristic curve of the output capacitor of the main power circuit, identify the global minimum impedance value in the impedance amplitude and frequency characteristic curve, and determine the value of the minimum impedance value as the equivalent series resistance of the output capacitor.

[0092] Use an impedance measurement device (such as an impedance analyzer) that can output a sine wave sweep signal with controllable amplitude and accurately measure the ratio of voltage to current across the output capacitor of the main power circuit of the power supply, thereby directly obtaining the complex impedance. To ensure safety and accurate measurement of the output capacitor's characteristics, the main power circuit of the power supply must be completely disconnected from the input power supply, but the output capacitor should remain in its original position on the PCB board and should not be removed. Connect the excitation signal output terminal and the measurement input terminal of the impedance analyzer in parallel across the output capacitor under test through Kelvin test leads. The measurement point should be directly connected to the pads of the output capacitor, minimizing the lead length as much as possible so that the measurement result is closest to the actual performance of the output capacitor in the circuit.

[0093] Assuming the output filter circuit on the power board consists of a 470μF electrolytic capacitor C1 and a 10μF ceramic capacitor C2 connected in parallel, the requirement is to measure the overall impedance characteristics of this parallel combination. Therefore, the test clip should be clamped onto the pad of C1 or C2 to measure the parallel impedance of C1 and C2.

[0094] Based on the power supply's operating characteristics, the noise and load dynamic frequency of a switching power supply are typically in the range of tens of Hz to several MHz. The sweep frequency range can be set from 10 Hz to 10 MHz. When selecting to measure complex impedance, the device will directly output the real part (R) and imaginary part (X) of the impedance at each frequency point, and a small AC voltage (e.g., 50V) can be set. This is to avoid damaging the capacitor or introducing nonlinearity.

[0095] Activate the frequency sweep measurement function of the impedance analyzer. The device will generate sinusoidal excitation point by point from the starting frequency to the ending frequency, and simultaneously measure the response. It will calculate the complex impedance value at each frequency point and use the measurement data, plotting the frequency f on the x-axis and the impedance amplitude on the y-axis. Plot the impedance amplitude and frequency response curves with the vertical axis as the ordinate.

[0096] The impedance amplitude and frequency response curves show a distinct V-shaped or U-shaped trough. In the low-frequency range, due to the large capacitive reactance, the impedance amplitude is very high and decreases as the frequency increases. In the high-frequency range, due to the parasitic inductance effect of the capacitor, the impedance increases as the frequency increases. At a certain frequency point in the middle, the capacitive reactance and inductive reactance cancel each other out, and the impedance reaches its minimum value.

[0097] On the impedance amplitude-frequency response curve, find the point with the lowest amplitude. The frequency corresponding to this point is the self-resonant frequency of the output capacitor. At the self-resonant frequency, the capacitive reactance and parasitic inductive reactance are equal in magnitude and opposite in direction, thus canceling each other out. At this point, the imaginary part of the impedance is zero, and the total impedance is zero. It reaches its minimum value, and its value is exactly equal to the equivalent series resistance of the output capacitor.

[0098] Step S312: Multiply the equivalent series resistance by the load current during the sampling period to generate the resistive voltage during the sampling period, and use the difference between the common ground voltage and the resistive voltage as the floating ground offset voltage during the sampling period.

[0099] By calculating the resistive voltage within the sampling period, the fixed voltage drop caused by the load current flowing through the output capacitor can be easily offset. In each control cycle, the current sampling module (such as a high-precision sampling resistor and analog-to-digital converter) measures and outputs a filtered and calibrated current value that represents the average load of the current cycle.

[0100] Assuming the system measures a load current of 10.0 A during the current sampling period, The resistive voltage is The floating ground offset voltage is In other words, in order to offset the output voltage loss caused by the output capacitor, the potential of the negative terminal of the power supply output needs to be lower than the actual system ground.

[0101] Assume that the main power circuit outputs a voltage of 5.0V and the load's rated operating voltage is 5.0V.

[0102] Before compensation (if no adjustment): The voltage across the load will be lost due to the voltage drop across the equivalent series resistance of the output capacitor. The actual output voltage of the power supply during the sampling period is... This will cause the voltage across the load to be lower than the rated operating voltage;

[0103] After compensation: the floating ground potential is set to The actual output voltage of the power supply during the sampling period is Because the actual output voltage is raised in advance, this raised voltage can offset the fixed voltage drop caused by the load current flowing through the equivalent series resistance of the output capacitor, thereby stabilizing the voltage across the load at near the rated operating voltage.

[0104] Under steady-state conditions, this process achieves a feedforward compensation: it does not wait for the output voltage to have an error, but actively predicts the voltage loss that will inevitably be caused by the equivalent series resistance of the known output capacitor and the real-time measured load current, and cancels it out by setting a floating ground potential. This ensures that even under steady-state conditions without drastic changes, the power supply can achieve accuracy and response speed superior to traditional single feedback loops.

[0105] Step S320: If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the second warning threshold and less than the first warning threshold, then the load operation condition is defined as a slow-change condition.

[0106] A reference current change rate greater than zero within the sampling period indicates that the current is increasing. A reference current change rate greater than or equal to the second warning threshold and less than the first warning threshold describes a clear, continuous but relatively controllable load increase process. For example, scenarios such as a CPU gradually increasing its frequency from a low frequency, multiple modules being powered on sequentially, or a motor accelerating smoothly, often have their current change rate fall within this range. During this stage, partial pre-compensation is mainly performed on the relatively gentle induced voltage generated on the equivalent inductance of the circuit due to the current change trend.

[0107] Specifically, the processing logic for a slowly changing load condition is as follows:

[0108] Step S321: Use a differential probe to determine the load's sag voltage during the sampling period between the power input terminal and the common ground terminal of the load device, and calculate the ratio of the absolute value of the sag voltage to the rate of change of the reference current to generate the equivalent inductance during the sampling period.

[0109] The positive terminal of the high-precision differential probe is clamped to the power input terminal of the load device, and the negative terminal of the probe is clamped to the common ground terminal of the load device. This measurement point is directly located at both ends of the load, and the measured voltage is the actual voltage received by the load. It includes the voltage drop caused by the combined effect of all parasitic parameters (line resistance, inductance) of the power output circuit, and is the most realistic affected voltage.

[0110] Assuming the load's rated operating voltage is 5.0V, and during a certain period of gradual current increase, the measured voltage received by the load is 4.95V, then the voltage drop is... ,this It is the relatively gentle real-time voltage drop generated by the change in current across the equivalent inductance of the circuit.

[0111] The equivalent inductance during the sampling period includes all inductive components that affect dynamic response, such as PCB trace inductance, cable inductance, and device pin inductance. It is assumed that the reference current changes at a rate of 0.2% during the sampling period. The equivalent inductance is .

[0112] Step S322: Multiply the difference between the reference current change rate and the second warning threshold within the sampling period by the equivalent inductance to generate the first prediction compensation voltage, and use the difference between the common ground terminal voltage and the first prediction compensation voltage as the floating ground offset voltage within the sampling period.

[0113] Assume the second warning threshold is 0.1. The first predicted compensation voltage is The floating ground offset voltage is In order to offset the relatively gentle real-time voltage drop caused by current changes across the equivalent inductance of the circuit, we need to ensure that the potential of the negative terminal of the power supply output is lower than the actual system ground, thus raising the load voltage. The trend, which is precisely used to offset the calculated It predicts voltage drops, thereby stabilizing the load voltage near the rated operating voltage.

[0114] Under gradually changing operating conditions, by using the ratio of voltage to current change rate measured in real time, the voltage deviation trend that is occurring but has not yet caused serious consequences can be finely corrected, realizing a leap from post-event remediation to in-event control, and significantly improving the smoothness and accuracy of dynamic response.

[0115] Step S330: If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the first warning threshold, then the load operation condition is defined as a sudden change condition.

[0116] A reference current change rate greater than zero within the sampling period indicates that the current is increasing. A reference current change rate greater than or equal to the first warning threshold describes a catastrophic or quasi-catastrophic scenario in which the load current undergoes a step-like, explosive increase in a very short time (microseconds or nanoseconds). The rate of change is so fast that it causes a catastrophic drop in output voltage before the feedback loop of a conventional power supply can make any effective response.

[0117] The handling logic for a sudden change in load operation is as follows:

[0118] Step S331: Based on electromagnetic field simulation software, determine the parasitic inductance on the main power printed circuit board of the power supply, and multiply the parasitic inductance on the main power printed circuit board of the power supply with the rate of change of the reference current during the sampling period to generate the inductive voltage during the sampling period.

[0119] For a typical main power Buck circuit, the core loop consists of the positive terminal of the input filter capacitor, the high-side MOSFET, the power inductor, the output capacitor or load, the low-side MOSFET or freewheeling diode, and the negative terminal of the return input capacitor. The parasitic inductance of this loop is the main cause of the instantaneous voltage drop at the load.

[0120] Import the complete printed circuit board design file (including all layer information) directly into the electromagnetic field simulation software. To improve simulation efficiency, the model can be reasonably simplified: retain all traces, vias, pads, and related capacitors and metal-oxide-semiconductor field-effect transistor package models of the target loop; retain adjacent power and ground layers, as they serve as current return paths and significantly affect loop inductance; remove components and lines irrelevant to the analysis; ensure that the printed circuit board stack-up structure, copper thickness (e.g., 35μm for 1 ounce of copper), and dielectric material (e.g., FR-4, dielectric constant approximately 4.4) parameters in the electromagnetic field simulation software are set correctly.

[0121] In electromagnetic field simulation software, set two ports: one at the beginning of the loop (such as the positive terminal of the input capacitor) and the other at the end of the loop (the negative terminal of the input capacitor). Select either lumped port or wave port as the port type, and the software will automatically calculate the characteristic impedance of the ports.

[0122] Select the S-parameter sweep frequency and set a reasonable sweep range according to the switching frequency and harmonics of the power supply. For example, for a 500kHz switching power supply, in order to cover its higher harmonics and higher frequency components introduced by load changes, the sweep range can be set to 1 kHz to 100 MHz or higher. Run the simulation, and the software will calculate the S-parameter matrix of the circuit at each frequency point in the set frequency band based on methods such as the finite element method. The S-parameter matrix is ​​a matrix used to describe the linear relationship between the incident wave and the reflected wave or the incident wave and the transmitted wave of a multi-port network under high frequency signal excitation.

[0123] After the simulation is completed, the S-parameter results of the port are converted into Z-parameters (complex impedance parameters), and then converted using the formula... Calculate the total parasitic inductance of the loop at each frequency f, where, The total parasitic inductance of the loop at a specific frequency f. For frequency, The imaginary part of the complex impedance is given, and the curve of parasitic inductance as a function of frequency f is plotted. The average value of the inductance value is taken in the range of 1-10 times the power switching frequency, where the inductance value is relatively stable. For example, if the switching frequency is 500kHz, the parasitic inductance value at f = 1MHz can be taken as the compensation model parameter. The parasitic inductance value of the critical loop determined by simulation is written as a constant into the non-volatile memory of the power controller.

[0124] Step S332: Based on the inductive voltage and resistive voltage within the sampling period, calculate the sum of the inductive voltage and resistive voltage to generate the second predictive compensation voltage within the sampling period, and use the difference between the common ground terminal voltage and the second predictive compensation voltage as the floating ground offset voltage within the sampling period.

[0125] Inductive voltage represents the instantaneous high-voltage spike induced on the parasitic inductance of the circuit due to drastic changes in load current. This is the most critical and fastest source of voltage disturbance during sudden changes. Resistive voltage represents the ohmic voltage drop generated on the parasitic resistance of the circuit due to the instantaneous value of the load current. This is a relatively slow but certain voltage loss. Adding the two together means that the system has completely predicted the total voltage drop that will occur at the load end in the next instant due to the combined effect of the circuit parasitic parameters (parasitic inductance and equivalent series resistance). The second predictive compensation voltage is the total voltage drop that the system predicts will occur in the next instant. The goal of the compensation is to actively inject a voltage of equal magnitude and opposite direction to completely cancel it out. This allows the system to actively and forcefully adjust the potential of the floating ground at the same moment when the load undergoes a drastic change, thereby canceling out the disturbance that is inevitably caused by parasitic parameters at the source of the output voltage waveform.

[0126] Step S340: If the rate of change of the reference current within the sampling period is less than zero, the load operation condition is defined as energy feedback condition.

[0127] A reference current change rate less than zero indicates that the instantaneous change trend of the load current is decreasing. This usually occurs when the energy stored at the load end (such as the back electromotive force of a motor, the residual charge of a battery, or the energy stored in a supercapacitor) needs to be released, causing the current to flow from the load to the power source.

[0128] The processing logic for the load operation under energy feedback mode is as follows: Based on the type of semiconductor device in the load, determine the forward conduction voltage of the semiconductor device, and use the difference between the forward conduction voltage and the common ground voltage as the floating ground offset voltage within the sampling period.

[0129] When the load is operating under energy feedback conditions, in order to prevent the feedback energy from flooding into the working power stage and causing more serious damage such as bus voltage surge and overvoltage breakdown of switching transistors, all active switching transistors of the main power circuit (such as the high-side and low-side metal-oxide-semiconductor field-effect transistors of the Buck circuit) will be immediately turned off. After being turned off, the main power circuit will no longer perform power conversion, and its ideal output voltage will tend to 0V.

[0130] Since the main power stage is off, the only low-impedance path between the power output and the input bus is the freewheeling diodes or metal-oxide-semiconductor field-effect transistors. If the load voltage is high at this time, current will flow back to the input bus through these diodes, which is still dangerous. Therefore, by actively adjusting the floating ground potential, the actual output voltage of the power supply becomes a very low negative value, pulling the output voltage down to a negative value. This is equivalent to creating a huge reverse potential difference between the external load (high potential) and the power supply output (low / negative potential). This will immediately cause the freewheeling diodes or body diodes that are naturally present or added inside the power supply to enter the reverse bias cutoff state. By utilizing their unidirectional conductivity, they physically block the reverse current from flowing into the power supply, thereby protecting the sensitive power switches and control circuits from damage.

[0131] Among them, the forward conduction voltage of silicon PN junction diodes (such as ordinary freewheeling diodes) is 0.7 V, the forward conduction voltage of Schottky barrier diodes is 0.3 V to 0.5 V, and the forward conduction voltage of metal oxide semiconductor field-effect transistors is 0.7 V to 1.0 V. The forward conduction voltage of semiconductor devices is used as a system constant and is read from non-volatile memory during controller initialization.

[0132] After successfully blocking a large current, the system can safely trigger a bleed resistor or a controllable load connected between the output terminal and ground to dissipate the residual energy at the load terminal in a controllable manner, causing its voltage to drop smoothly to zero and completing the safe shutdown.

[0133] Step S400: Based on the real-time deviation between the actual output voltage and the rated operating voltage of the load within the sampling period, the proportional-integral control algorithm is used to iteratively update the control cycle until the absolute value of the error between the actual output voltage and the rated operating voltage of the load is less than the allowable voltage tolerance range of the load. The proportional coefficient and integral coefficient in the proportional-integral control algorithm are then determined to correct the floating ground offset voltage in the next sampling period.

[0134] It should be noted that the mathematical expression for the proportional-integral control algorithm is as follows:

[0135]

[0136] in, This represents the corrected floating ground offset voltage. This represents the floating ground offset voltage during the sampling period. This is the proportionality coefficient. This indicates the rated operating voltage of the load. This indicates the actual output voltage of the power supply during the sampling period. is the integral coefficient.

[0137] proportionality coefficient The product of the current error and the proportional term produces an immediate response when the error occurs. The larger the error, the greater the adjustment. It allows the system to react quickly to new load changes and reduces the depth of voltage drops or overshoots.

[0138] Integral coefficient The product of the current error and the total historical integral is added to the total integral. As long as the error exists (even if it is very small), the integral term will continue to accumulate and gradually increase the adjustment strength. Its task is to reduce the error until the absolute value of the error between the actual output voltage and the rated working voltage is less than the voltage tolerance range allowed by the load.

[0139] Please see Figure 2 The present invention also provides a DC regulated power supply, including a main power circuit 510, a floating ground drive unit 520, a current sampling module 530, a data processing module 540, a working condition judgment module 550, a voltage compensation module 560, and a control module 570.

[0140] The main power circuit 510 is used to provide output voltage to the load.

[0141] The floating ground drive unit 520 is connected in series between the output negative terminal of the main power circuit 510 and the system common ground, and the floating ground drive unit 520 is used to generate a controllable floating ground offset voltage.

[0142] The current sampling module 530 is used to collect the current timing data of the power supply output circuit during the sampling period.

[0143] The data processing module 540 is connected to the current sampling module 530, and the data processing module 540 is used to analyze the current change rate of the current time series data and perform filtering processing to generate a reference current change rate for characterizing the load change trend.

[0144] The operating condition judgment module 550 is connected to the data processing module 540. The operating condition judgment module 550 has a preset warning threshold for distinguishing the load operating conditions. The operating condition judgment module 550 is used to determine the load operating condition type within the sampling period based on the comparison result between the reference current change rate and the warning threshold. The operating condition types include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition.

[0145] The voltage compensation module 560 is connected to the output terminals of the operating condition judgment module 550 and the main power circuit 510 respectively. The voltage compensation module 560 is used to determine the floating ground offset voltage within the sampling period based on the operating condition type, and subtract the floating ground offset voltage from the voltage output by the main power circuit 510 to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative terminal of the power supply output and the common ground terminal.

[0146] The control module 570 is connected to the voltage compensation module 560, and the control module 570 is used to iteratively update the control parameters using a proportional-integral control algorithm based on the real-time deviation between the actual output voltage and the rated operating voltage of the load, so as to correct the floating ground offset voltage in the next sampling time period until the absolute value of the error between the actual output voltage and the rated operating voltage is less than the allowable voltage tolerance range of the load.

[0147] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0148] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0149] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for floating and superimposing a power supply reference voltage, characterized in that, Includes the following steps: Based on the current timing data of the power output circuit during the sampling period, the current change rate is analyzed and filtered to generate a reference current change rate that characterizes the load change trend during the sampling period. Set a warning threshold to distinguish load operating conditions, and combine it with the reference current change rate to determine the operating condition type of the load operating condition within the sampling period. The operating condition types of the load operating condition include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition. Based on the operating condition type of the load, the floating ground offset voltage within the sampling period is determined, and the voltage output by the main power circuit of the power supply is subtracted from the floating ground offset voltage to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative output terminal of the power supply and the common ground terminal. Based on the real-time deviation between the actual output voltage and the rated operating voltage of the load within the sampling period, the proportional-integral control algorithm is used to iteratively update the control cycle until the absolute value of the error between the actual output voltage and the rated operating voltage of the load is less than the allowable voltage tolerance range of the load. The proportional coefficient and integral coefficient in the proportional-integral control algorithm are then determined to correct the floating ground offset voltage in the next sampling period.

2. The method for floating and superimposing power supply reference voltage according to claim 1, characterized in that, Generating a reference current change rate to characterize the load change trend within the sampling period includes the following steps: Within the sampling period, sampling points are set according to the set sampling frequency to retrieve the load current under each sampling point, obtain the current time series data within the sampling period, calculate the current change between adjacent sampling points, and calculate the original current change rate of each sampling point in combination with the sampling time interval, and write the original current change rate into the data buffer. The current change rate time series data of the power supply output circuit under steady-state conditions is obtained in multiple segments as reference time series data under steady-state conditions. After preprocessing each segment of reference time series data, the mean current change rate of all reference time series data is determined by combining statistical averaging algorithm. The preprocessing includes removing outliers based on the quartile method and filling missing values ​​based on the mean imputation method. Each raw current change rate in the data buffer is compared with the mean current change rate. If the absolute value of the raw current change rate is less than or equal to the mean current change rate, the raw current change rate is marked as a normal target object. If the absolute value of the raw current change rate is greater than the mean current change rate, the raw current change rate is marked as an abnormal target object. Calculate the ratio of the number of all abnormal target objects in the data buffer to the number of all original current change rates in the data buffer. If the ratio is less than 0.3, the average current change rate of all original current change rates in the data buffer is used as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.3 and less than 0.7, the average current change rate of all original current change rates in the data buffer is multiplied by an attenuation coefficient less than 1 to use as the reference current change rate characterizing the load change trend. If the ratio is greater than or equal to 0.7, the average current change rate of all abnormal target objects in the data buffer is used as the reference current change rate characterizing the load change trend.

3. The method for floating and superimposing power supply reference voltage according to claim 2, characterized in that, The warning thresholds include a first warning threshold and a second warning threshold. The logic for determining the warning thresholds is as follows: Based on the pre-processed reference timing data of each segment of the power output circuit under steady-state conditions, the overall standard deviation of all reference timing data is calculated. Based on the overall standard deviation and mean current change rate of all reference time series data, the product of the overall standard deviation and the first preset margin factor is calculated, and the sum of the product and the mean current change rate is used as the first warning threshold. The first preset margin factor is determined to be 3 based on the three sigma criterion. Calculate the product of the first warning threshold and the second preset margin factor, and use the product as the second warning threshold. The second preset margin factor is a positive number less than 1.

4. The method for floating and superimposing power supply reference voltage according to claim 1, characterized in that, Determining the load condition type within the sampling period includes the following steps: If the rate of change of the reference current within the sampling period is greater than or equal to zero, and the rate of change of the reference current is less than the second warning threshold, then the load operation condition is defined as a steady-state condition. If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the second warning threshold and less than the first warning threshold, then the load operation condition is defined as a slow-change condition. If the rate of change of the reference current within the sampling period is greater than zero, and the rate of change of the reference current is greater than or equal to the first warning threshold, then the load operating condition is defined as a sudden change condition. If the rate of change of the reference current during the sampling period is less than zero, the load operation is defined as the energy feedback condition.

5. The method for floating and superimposing a power supply reference voltage according to claim 4, characterized in that, The processing logic for defining the load operation as a steady-state condition is as follows: Based on the impedance measurement device connected to the output capacitor of the main power circuit, the complex impedance value of the output capacitor of the main power circuit is obtained at each frequency point, the impedance amplitude and frequency response curve of the output capacitor of the main power circuit is generated, the global minimum impedance value in the impedance amplitude and frequency response curve is identified, and the value of the minimum impedance value is determined as the equivalent series resistance of the output capacitor. The equivalent series resistance is multiplied by the load current during the sampling period to generate the resistive voltage during the sampling period, and the difference between the common ground voltage and the resistive voltage is taken as the floating ground offset voltage during the sampling period.

6. The method for floating and superimposing power supply reference voltage according to claim 4, characterized in that, The processing logic for a slowly changing load condition is as follows: A differential probe is used to determine the load's sag voltage during the sampling period between the power input terminal and the common ground terminal of the load device, and the ratio of the absolute value of the sag voltage to the rate of change of the reference current is calculated to generate the equivalent inductance during the sampling period. The difference between the reference current change rate and the second warning threshold during the sampling period is multiplied by the equivalent inductance to generate the first prediction compensation voltage. The difference between the common ground terminal voltage and the first prediction compensation voltage is used as the floating ground offset voltage during the sampling period.

7. The method for floating and superimposing a power supply reference voltage according to claim 5, characterized in that, The handling logic for a sudden change in load operation is as follows: Based on electromagnetic field simulation software, the parasitic inductance on the main power printed circuit board of the power supply is determined, and the parasitic inductance on the main power printed circuit board of the power supply is multiplied by the rate of change of the reference current during the sampling period to generate the inductive voltage during the sampling period. Based on the inductive and resistive voltages within the sampling period, the sum of the inductive and resistive voltages is calculated to generate the second predictive compensation voltage within the sampling period. The difference between the common ground voltage and the second predictive compensation voltage is then used as the floating ground offset voltage within the sampling period.

8. The method for floating and superimposing power supply reference voltage according to claim 4, characterized in that, The processing logic for the load operating in energy feedback mode is as follows: Based on the type of semiconductor device in the load, determine the forward conduction voltage of the semiconductor device, and use the difference between the forward conduction voltage and the common ground voltage as the floating ground offset voltage within the sampling period.

9. The method for floating and superimposing power supply reference voltage according to claim 1, characterized in that, The mathematical expression for the proportional-integral control algorithm is as follows: in, This represents the corrected floating ground offset voltage. This represents the floating ground offset voltage during the sampling period. This is the proportionality coefficient. This indicates the rated operating voltage of the load. This indicates the actual output voltage of the power supply during the sampling period. is the integral coefficient.

10. A DC regulated power supply, characterized in that, include: The main power circuit is used to provide output voltage to the load; A floating ground drive unit is connected in series between the output negative terminal of the main power circuit and the system common ground, and is used to generate a controllable floating ground offset voltage. The current sampling module is used to collect the current timing data of the power supply output circuit during the sampling period; The data processing module is connected to the current sampling module and is used to analyze the rate of change of current in the current time series data and perform filtering to generate a reference rate of change of current to characterize the load change trend. The operating condition judgment module is connected to the data processing module and has a preset warning threshold for distinguishing load operating conditions. It is used to determine the load operating condition type within the sampling period based on the comparison result of the reference current change rate and the warning threshold. The operating condition types include steady-state operating condition, sudden change operating condition, slow change operating condition and energy feedback operating condition. The voltage compensation module is connected to the output terminals of the operating condition judgment module and the main power circuit respectively. It is used to determine the floating ground offset voltage within the sampling period based on the operating condition type, and subtract the floating ground offset voltage from the voltage output by the main power circuit to generate the actual output voltage of the power supply within the sampling period. The floating ground offset voltage is the voltage difference between the negative terminal of the power supply output and the common ground terminal. The control module is connected to the voltage compensation module and is used to iteratively update the control parameters using a proportional-integral control algorithm based on the real-time deviation between the actual output voltage and the rated operating voltage of the load, so as to correct the floating ground offset voltage in the next sampling time period until the absolute value of the error between the actual output voltage and the rated operating voltage is less than the allowable voltage tolerance range of the load.

Images