Power supply detection device and test method thereof

Abstract

The invention discloses a power supply detection device and a test method thereof, particularly relates to the field of power electronic control, and is used for solving the problems of high-frequency phase distortion and incapability of reflecting a stability boundary under a real load dynamic state in an open-loop state caused by introduction of additional impedance due to disconnection of a loop in an existing open-loop injection test method. A switching node voltage ripple signal is synchronously collected as a disturbance input signal and an output voltage feedback signal is synchronously collected as a response output signal in a power supply closed-loop state; performing time domain alignment on the two signals to generate a time correlation signal pair; fourier transform is carried out on the signal pair to extract open-loop transfer function characteristics; mapping the characteristics to an intersection phase angle of a Nyquist plane identification curve and a unit circle; generating a chaos distortion index based on the local curvature change characteristics of the Nyquist curve of the intersection neighborhood; and correcting the phase angle of the intersection point by using a chaotic distortion index, and outputting a loop stability evaluation conclusion according to a correction result, thereby realizing high-precision stability evaluation of uninterrupted closed-loop control.

Description

Technical Field

[0001] The present invention relates to the field of power electronic control, and more particularly to a power supply detection device and a testing method thereof. Background Art

[0002] High-frequency switching power supplies are widely used in communications equipment and server power supply systems due to their superior power density. The stability of their control loops directly impacts system reliability. Industry standards require loop stability to be assessed through quantification of gain margin and phase margin. Existing testing methods require disconnecting the control loop and injecting a disturbance signal to measure the frequency domain response.

[0003] However, existing testing methods have defects in high-frequency working scenarios: the additional impedance introduced by the loop-breaking operation will change the original loop characteristics, resulting in phase measurement distortion in the high-frequency band; at the same time, the power supply in the open-loop state cannot reflect the stability boundary under the actual load dynamics, causing the margin assessment results to deviate from the actual working conditions. There is a conflict between the physical interference of the measurement intervention and the unobservability of the true state of the closed-loop system. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a power supply detection device and a testing method thereof to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A power supply testing method, comprising: S1. In a closed-loop state, a switch node voltage ripple signal of a high-frequency switching power supply is collected as a disturbance input signal, and an output voltage feedback signal of the high-frequency switching power supply is simultaneously collected as a response output signal; S2, aligning the disturbance input signal and the response output signal in the time domain to generate a time-correlated signal pair; S3, performing Fourier transform on the time-correlated signal pair to extract the amplitude-frequency characteristic and phase-frequency characteristic of the open-loop transfer function; S4, mapping the amplitude-frequency characteristic and the phase-frequency characteristic to the Nyquist plane, and identifying the phase angle of the intersection of the Nyquist curve and the unit circle; S5. Generate a chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; S6. Use the chaos distortion index to correct the intersection phase angle, and output the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

[0006] In a preferred embodiment, the method of collecting a switch node voltage ripple signal of a high-frequency switching power supply as a disturbance input signal in a closed-loop state and synchronously collecting an output voltage feedback signal of the high-frequency switching power supply as a response output signal includes: Use a differential voltage probe to connect the switch node of the high-frequency switching power supply to the reference ground, and capture the switch node voltage ripple signal at a sampling rate no less than ten times the switching frequency of the high-frequency switching power supply. At the same time, a voltage probe is used to connect the output voltage feedback node of the high-frequency switching power supply to capture the output voltage feedback signal at the same sampling rate as the switching node voltage ripple signal; The switch node voltage ripple signal and the output voltage feedback signal are respectively conditioned by an isolation amplifier, and after eliminating common-mode interference, they are transmitted to a digital storage oscilloscope; A synchronous trigger channel between the switch node voltage ripple signal and the output voltage feedback signal is established in the digital storage oscilloscope to align the acquisition time starting points and keep the sampling clocks from the same source.

[0007] In a preferred embodiment, performing time domain alignment on the disturbance input signal and the response output signal to generate a time-correlated signal pair includes: Extract the trigger timestamp of the switch node voltage ripple signal channel as the reference time; Based on the reference time, the waveform data time axis of the output voltage feedback signal channel is shifted so that the time coordinates of the switch node voltage ripple signal channel and the output voltage feedback signal channel at the reference time are aligned to zero; Identify the complete switching cycle boundary based on the switching node voltage waveform characteristics, and extract waveform segments corresponding to the number of consecutive complete switching cycles from the dual-channel waveforms after time domain alignment; A one-to-one mapping relationship is established between the switch node voltage ripple signal data points corresponding to all sampling time points in the intercepted waveform segment and the output voltage feedback signal data points to form a set of time-correlated signal pairs.

[0008] In a preferred embodiment, the trigger timestamp is generated when the trigger pulse arrives at the sample-and-hold circuit of the switch node voltage ripple signal channel.

[0009] In a preferred embodiment, Fourier transform is performed on the time-correlated signal pair to extract the amplitude-frequency characteristic and phase-frequency characteristic of the open-loop transfer function, including: Performing discrete Fourier transform on the switch node voltage ripple signal data and the output voltage feedback signal data in the time-correlated signal pair set to obtain a switch node voltage ripple signal spectrum component and an output voltage feedback signal spectrum component; In the complex domain, the spectral component of the output voltage feedback signal is divided by the spectral component of the switch node voltage ripple signal to obtain a complex result of the open-loop transfer function at the corresponding frequency point; Extracting amplitude-frequency characteristic data from the modulus of the complex result of the open-loop transfer function, and extracting phase-frequency characteristic data from the argument of the complex result of the open-loop transfer function; The number of operation points of the discrete Fourier transform is equal to the total number of sampling points of the set of time-correlated signal pairs, and the complex division operation is performed between spectral components corresponding to the same frequency point.

[0010] In a preferred embodiment, mapping the amplitude-frequency characteristic and the phase-frequency characteristic to the Nyquist plane and identifying the phase angle of the intersection of the Nyquist curve and the unit circle include: Convert the decibel values ​​in the amplitude-frequency characteristic data sequence into linear modulus values, and map the linear modulus values ​​and the phase angles in the phase-frequency characteristic data sequence to the Nyquist plane, and form a Nyquist curve consisting of the complex coordinates of discrete frequency points by calculating the real part of the complex coordinates of each frequency point as the linear modulus value multiplied by the cosine value of the phase angle and the imaginary part as the linear modulus value multiplied by the sine value of the phase angle; Identify the complex coordinate point with the modulus closest to 1 in the Nyquist curve as the intersection point of the Nyquist curve and the unit circle; The intersection phase angle is extracted from the phase angle in the phase-frequency characteristic data sequence corresponding to the complex coordinate point.

[0011] In a preferred embodiment, generating a chaotic distortion index based on the local curvature variation characteristics of the Nyquist curve in the neighborhood of the intersection point includes: After determining the intersection point with the unit circle in the Nyquist curve, a fixed number of frequency points are taken forward and backward with the frequency point index corresponding to the intersection point as the center to form a neighborhood range; For each frequency point within the neighborhood, the local curvature value at the corresponding frequency point is solved by using the complex coordinates corresponding to the frequency point and its two adjacent frequency points and calculating the geometric relationship between the vector angle and chord length between the adjacent complex coordinates. After traversing all frequency points in the neighborhood and completing the local curvature calculation, the maximum and minimum local curvature values ​​in the neighborhood are identified, and the difference between the maximum value and the minimum value is calculated as the curvature range; The curvature range is divided by the total number of frequency points in the neighborhood for normalization, and the resulting dimensionless scalar value is the chaos distortion index; When the neighborhood range boundary exceeds the Nyquist curve index range, normalization calculation is performed based on the total number of frequency points actually included.

[0012] In a preferred embodiment, the number of frequency points before and after the intersection point included in the neighborhood range is determined by a predefined neighborhood radius parameter.

[0013] In a preferred embodiment, the intersection phase angle is corrected using the chaos distortion index, and a loop stability evaluation conclusion of the high-frequency switching power supply is output based on the correction result, including: Obtain the original phase angle and the corresponding chaos distortion index determined by the intersection of the Nyquist curve and the unit circle; Determining a phase compensation amount based on a comparison result between a chaos distortion index and a preset threshold: generating a positive compensation amount when the chaos distortion index exceeds a first threshold, and generating a negative compensation amount when the chaos distortion index is lower than a second threshold; The phase compensation amount is algebraically superimposed on the original phase angle to generate a corrected phase angle; The corrected phase angle is compared with a preset phase tolerance threshold: when the corrected phase angle is less than the phase tolerance threshold, a stability qualified conclusion is output; otherwise, a stability defect conclusion is output.

[0014] In another aspect, the present invention provides a power supply detection device, comprising: Disturbance acquisition module: In a closed-loop state, it collects the switch node voltage ripple signal of the high-frequency switching power supply as the disturbance input signal, and synchronously collects the output voltage feedback signal of the high-frequency switching power supply as the response output signal; Time domain alignment module: performs time domain alignment on the disturbance input signal and the response output signal to generate a time-correlated signal pair; Frequency domain conversion module: performs Fourier transform on the time-correlated signal pair to extract the amplitude-frequency and phase-frequency characteristics of the open-loop transfer function; Intersection identification module: maps the amplitude-frequency characteristics and phase-frequency characteristics to the Nyquist plane and identifies the phase angle of the intersection of the Nyquist curve and the unit circle; Distortion quantification module: Generates chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; Evaluation and correction module: uses the chaos distortion index to correct the intersection phase angle, and outputs the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By directly acquiring the switch node voltage ripple as the disturbance input signal during the closed-loop operation of the power supply, and synchronously acquiring the output voltage feedback signal as the response signal, the high-frequency phase measurement distortion problem caused by the additional impedance introduced by disconnecting the control loop, which is common in traditional methods, is effectively avoided. Without interrupting the normal closed-loop control of the power supply, the measurement process is ensured to be performed under real load dynamics and complete loop characteristics. The resulting loop response characteristics are closer to the actual operating state of the power supply, significantly improving the consistency between the phase margin assessment results and the actual operating conditions.

[0016] 2. By analyzing the local curvature change characteristics of the Nyquist curve in the vicinity of the key intersection point, a chaotic distortion index is generated, and the identified intersection phase angle is corrected in a targeted manner using this index. This can effectively capture and quantify the impact of local curve distortion caused by complex factors such as high-frequency switching noise on phase measurement accuracy. Based on the dynamic correction of the distortion index, the accuracy of phase angle extraction in complex high-frequency working environments is significantly improved, making the final output loop stability assessment conclusion more reliable and valuable for engineering guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a power supply testing method of the present invention; Figure 2 The figure is a structural diagram of a power supply detection device of the present invention. DETAILED DESCRIPTION

[0018] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] Example 1: Figure 1 The present invention provides a power supply testing method, comprising: S1. In a closed-loop state, a switch node voltage ripple signal of a high-frequency switching power supply is collected as a disturbance input signal, and an output voltage feedback signal of the high-frequency switching power supply is simultaneously collected as a response output signal; S2, aligning the disturbance input signal and the response output signal in the time domain to generate a time-correlated signal pair; S3, performing Fourier transform on the time-correlated signal pair to extract the amplitude-frequency characteristic and phase-frequency characteristic of the open-loop transfer function; S4, mapping the amplitude-frequency characteristic and the phase-frequency characteristic to the Nyquist plane, and identifying the phase angle of the intersection of the Nyquist curve and the unit circle; S5. Generate a chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; S6. Use the chaos distortion index to correct the intersection phase angle, and output the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

[0020] S1. In a closed-loop state, the voltage ripple signal of the switch node of the high-frequency switching power supply is collected as a disturbance input signal, and the output voltage feedback signal of the high-frequency switching power supply is simultaneously collected as a response output signal. This can be specifically implemented as follows: In the closed-loop state, the switch node voltage ripple signal of the high-frequency switching power supply is collected as the disturbance input signal, and the output voltage feedback signal of the high-frequency switching power supply is synchronously collected as the response output signal. The specific implementation process is as follows: First, the positive probe of the differential voltage probe is connected to the connection point between the drain of the metal oxide semiconductor field effect transistor and the power inductor in the high-frequency switching power supply. This connection point is defined as the switching node, and the negative probe is connected to the negative terminal of the DC input bus of the high-frequency switching power supply as the measurement reference ground; the bandwidth of the differential voltage probe is set based on the harmonic components contained in the switching node voltage waveform, according to the nominal switching voltage specified in the high-frequency switching power supply specification. The switching frequency parameter is set, and the probe bandwidth is set to at least three times the switching frequency value to ensure that the high-frequency components of the switching transient process are captured. For example, when the switching frequency is 500kHz, a differential voltage probe with a bandwidth of at least 1.5MHz is selected. The sampling rate is set according to the requirements of the Nyquist sampling theorem for signal reconstruction. Considering that the main energy component of the switching node voltage ripple is distributed below the second harmonic of the switching frequency, the sampling rate is set to at least ten times the switching frequency to fully retain the second harmonic information. For example, a switching frequency of 500kHz corresponds to a sampling rate of 5MS / s. The sample-and-hold circuit of the digital storage oscilloscope is configured according to the sampling rate parameter.

[0021] The output voltage feedback signal is collected at the midpoint of the voltage divider of the high-frequency switching power supply's output voltage sampling resistor network. The ratio of the voltage at this node to the output voltage is determined by the resistor network's voltage divider ratio. A single-ended voltage probe is connected to the midpoint of the voltage divider, with the probe ground clip connected to the same DC bus reference ground as the differential voltage probe. In the digital storage oscilloscope's signal channel configuration interface, the sampling clock source of the output voltage feedback signal channel is set to the same clock source as the switching node voltage ripple signal channel. This clock source is provided by the digital storage oscilloscope's internal master clock generator. Clock coherence is achieved by transmitting the clock signal output by the master clock generator to the clock inputs of the analog-to-digital converters of both channels via equal-length traces on the printed circuit board. The trace length difference is controlled within 10 mm to ensure that the clock phase deviation is less than 60 picoseconds, which is less than one ten-thousandth of the switching period.

[0022] The first signal input port of the isolation amplifier receives the switch node voltage ripple signal output by the differential voltage probe, and the second signal input port receives the output voltage feedback signal output by the single-ended voltage probe. The common-mode rejection capability of the differential amplifier circuit inside the isolation amplifier is achieved through a precisely matched input resistor network, and its common-mode rejection ratio parameter is not less than 80dB. The gain coefficient is set by dynamically adjusting it based on the ratio of the expected signal amplitude range to the full-scale range of the subsequent analog-to-digital converter. The adjustment logic follows the dual principles of preventing signal clipping distortion and improving quantization accuracy. In specific implementation, the steady-state amplitude of the signal is first measured. When the peak-to-peak value of the switch node voltage ripple exceeds 80% of the analog-to-digital converter range, it is proportionally attenuated. When the output voltage feedback signal amplitude is less than 20% of the analog-to-digital converter range, it is proportionally amplified. For example, when the range of a 12-bit analog-to-digital converter is 5V, a 25V ripple signal is set to a gain of 0.2 times, and a 0.5V feedback signal is set to a gain of 10 times.

[0023] The establishment of a synchronous trigger channel is achieved through the trigger comparator circuit of a digital storage oscilloscope; the trigger level setting method is as follows: observe the switch node voltage ripple signal waveform of ten consecutive switching cycles on the oscilloscope, record the maximum instantaneous voltage value and the minimum instantaneous voltage value of each cycle, calculate the arithmetic mean of the maximum instantaneous voltage value and the arithmetic mean of the minimum instantaneous voltage value, and take half of the sum of the two as the trigger level; the positive input of the trigger comparator is connected to the switch node voltage ripple signal channel, and a trigger pulse is generated when the rising edge of the signal crosses the trigger level; the trigger pulse is simultaneously transmitted to the sample and hold circuits of the two signal channels through the trigger distribution bus, and the delay difference of the trigger pulse transmission path is calibrated to be less than 1 nanosecond; after receiving the trigger pulse, the sample and hold circuit immediately starts the analog-to-digital conversion process to ensure that the deviation of the acquisition time starting point of the two channels is less than a single sampling clock cycle.

[0024] The technical implementation of common-mode interference elimination in the signal conditioning process relies on the electromagnetic isolation barrier of the isolation amplifier; the electromagnetic isolation barrier is composed of a primary winding, a secondary winding and a magnetic core. When the input signal is applied to the primary winding, the electrical signal is converted into a magnetic field change through the principle of magnetic induction, and then the electrical signal is reconstructed in the secondary winding; this process cuts off the direct electrical connection between the two grounds and eliminates the common-mode interference introduced by the ground potential difference; the shielded twisted pair cable adopts a structure in which a copper wire braid layer covers the signal line, and the braid layer coverage rate is not less than 85%. The shielding layer is connected to the chassis ground terminal of the digital storage oscilloscope at one end.

[0025] The data storage process of a digital storage oscilloscope uses a segmented storage mechanism. Each channel is allocated an independent sampling memory. When a trigger event occurs, the analog-to-digital conversion result is written to the sampling memory of the corresponding channel. The storage depth is set according to the signal duration. The setting principle is to completely record the waveform data of at least 100 switching cycles. For example, a 500kHz switching frequency corresponds to a 200μs period. Recording 100 cycles requires 20ms. At a sampling rate of 5MS / s, each channel is allocated 100k sampling point memory.

[0026] S2. Align the disturbance input signal and the response output signal in the time domain to generate a time-correlated signal pair. This can be specifically implemented as follows: During the execution of the synchronized trigger mechanism in a digital storage oscilloscope, a hardware trigger pulse distribution circuit synchronously transmits the same trigger pulse to the sample-and-hold circuits of the switch node voltage ripple signal channel and the output voltage feedback signal channel. When the trigger pulse reaches the sample-and-hold circuit of the switch node voltage ripple signal channel, the oscilloscope's timebase system records the precise moment of occurrence and generates a trigger timestamp, which serves as the reference time. The time accuracy of the reference time is determined by the oscilloscope's internal clock resolution, for example, using an oven-controlled crystal oscillator clock source with a time resolution of 100 picoseconds. The physical significance of the reference time is to provide an absolute reference zero point for time-domain alignment of the dual-channel waveforms, ensuring that the sampling starting points of the two signal channels share the same time-domain coordinate origin.

[0027] Based on the previously generated reference time, a time axis shift operation is performed on the raw waveform data of the output voltage feedback signal channel. The specific implementation process involves reading the time coordinate value of each sampling point in the output voltage feedback signal channel waveform data and subtracting the absolute time value recorded at the reference time from all time coordinate values. For example, if the reference time is recorded as 153.2 microseconds, the time axis shift is -153.2 microseconds. After the shift operation is completed, the time coordinate corresponding to the reference time of the switch node voltage ripple signal channel is reset to zero, and the time coordinates of all sampling points in the output voltage feedback signal channel are synchronously offset by the same amount, forming dual-channel waveform data with strict time domain overlap. This operation relies on the oscilloscope firmware's time axis calibration algorithm, which is centered on vector addition and subtraction operations. The calculation formula is expressed as the post-shift time equals the pre-shift time minus the reference time value. In this formula, the pre-shift time is the original time coordinate of the i-th sampling point in the output voltage feedback signal channel, the reference time is the absolute time value generated in step 1, and the post-shift time is the calibrated time coordinate.

[0028] The complete switching cycle boundary is identified based on the characteristics of the switching node voltage waveform. The complete switching cycle boundary identification includes two judgment steps: first, the rising edge zero crossing point is detected. The judgment condition is that the voltage value crosses from a negative value to a positive value and the voltage change rate is greater than the set threshold, for example, greater than 5 million volts per second; second, the falling edge zero crossing point is detected. The judgment condition is that the voltage value crosses from a positive value to a negative value and the voltage change rate is less than the set threshold, for example, less than negative 5 million volts per second. The waveform interval between adjacent rising edge zero crossings is defined as a complete switching cycle. Based on the boundary identification result, starting from the starting rising edge of the first complete switching cycle, M consecutive waveform segments of complete switching cycles are intercepted. The value of M is not less than 3. This numerical lower limit ensures that the steady-state and typical transient processes of the switching power supply are covered. The interception operation strictly follows the boundary alignment principle. The interception start and end points are both located at the rising edge zero crossing point. The data length of the captured waveform segment is determined by the number of switching cycles, the average switching cycle duration, and the oscilloscope sampling rate. For example, when the average switching cycle duration is 2 microseconds, the sampling rate is 1 billion times per second, and 3 cycles are captured, the number of data points is 6000.

[0029] A data point matching operation is performed on all sampling time points within the captured waveform segment. Because both channels utilize the same timebase system for sampling, each sampling time point k corresponds to a unique timestamp. This timestamp is generated synchronously with the oscilloscope's sampling clock, for example, a uniform sampling sequence generated by dividing a 100 MHz reference clock. At these timestamps, the instantaneous voltage values ​​of the switch node voltage ripple signal channel and the output voltage feedback signal channel are extracted, both in volts, to form a signal pair. All sampling time points within the captured waveform segment are traversed to generate a set of time-correlated signal pairs containing all sampling point signal pairs. This set is stored in a two-dimensional array data structure, with the array row indexes strictly corresponding to the sampling time order. The first column stores the switch node voltage ripple signal data, and the second column stores the output voltage feedback signal data.

[0030] S3. Performing Fourier transform on the time-correlated signal pair to extract the amplitude-frequency characteristic and phase-frequency characteristic of the open-loop transfer function can be specifically implemented as follows: When performing frequency-domain characteristic extraction of the open-loop transfer function on a set of time-correlated signal pairs, the switch node voltage ripple signal data sequence and the output voltage feedback signal data sequence are first separated from the set. These two data sequences have exactly the same number of sampling points and identical sampling time coordinates, a property guaranteed by the synchronous sampling mechanism used to generate the set of time-correlated signal pairs. This separation is achieved by reading a two-dimensional array data structure: the switch node voltage ripple signal data sequence corresponds to all elements in the first column of the two-dimensional array, and the output voltage feedback signal data sequence corresponds to all elements in the second column of the two-dimensional array. The data sequence length is equal to the total number of sampling points in the set of time-correlated signal pairs, which is determined by the product of the number of complete switching cycles captured, the average switching cycle duration, and the oscilloscope sampling rate. For example, if three switching cycles are captured, the average cycle duration is 2 microseconds, and the sampling rate is 1 billion times per second, the data sequence length is fixed at 6000 sampling points.

[0031] A discrete Fourier transform (DFT) is performed on the isolated switch node voltage ripple signal data sequence, with the number of transform points strictly equal to the length of the data sequence. The mathematical expression of the DFT is to convert the time-domain voltage signal sequence into a complex frequency-domain representation. Its physical essence is the projective decomposition of the signal onto orthogonal basis functions. The implementation utilizes a fast Fourier transform (FFT) algorithm. The input is an array of voltage values ​​from the switch node voltage ripple signal data sequence, and the output is an array of spectral components containing the same number of complex elements. Each spectral component corresponds to a signal component at a specific frequency point. The frequency point spacing is determined by the sampling rate and the number of transform points. For example, at a 1 GHz sampling rate and a 6000-point transform, the frequency resolution is 166.7 kHz. The modulus of a spectral component represents the amplitude of that frequency component, and the argument represents the phase offset. The DFT operation on the output voltage feedback signal data sequence follows the same process, generating an array of spectral components with identical frequency point coordinates.

[0032] After obtaining the spectral component arrays for the two channels, a division operation is performed in the complex domain between the spectral components. This division operation is limited to spectral components corresponding to the same frequency point: that is, the elements of the output voltage feedback signal spectral component array are divided by the corresponding elements of the switching node voltage ripple signal spectral component array. This operation is mathematically represented as the division of two complex numbers. The calculation formula is: the real part of the result is equal to the real part of the dividend multiplied by the real part of the divisor plus the imaginary part of the dividend multiplied by the imaginary part of the divisor, divided by the square of the divisor modulus. The imaginary part of the result is equal to the real part of the divisor multiplied by the imaginary part of the dividend minus the real part of the dividend multiplied by the imaginary part of the divisor, divided by the square of the divisor modulus. The division result at each frequency point generates the corresponding complex open-loop transfer function, forming an array of open-loop transfer function frequency responses. The physical meaning of complex division is to eliminate the spectral characteristics of the switching disturbance signal at the system input, retaining only the response characteristics affected by the system transfer function. For example, in a buck converter with a switching frequency of 500 kHz, this operation can effectively isolate the resonant characteristics of the power stage inductor and capacitor filter.

[0033] When extracting amplitude-frequency characteristic data from the complex result array of the open-loop transfer function, the modulus of each complex element in the array is calculated. This modulus calculation uses the mathematical method of converting rectangular coordinates to polar coordinates, with the formula stating that the modulus is equal to the square root of the sum of the squares of the real and imaginary parts of the complex number. The results form an amplitude-frequency characteristic data sequence, with the sequence index strictly corresponding to the frequency point. The values ​​are expressed in decibels, using the conversion formula of 20 multiplied by the base-10 logarithm of the modulus value. Phase-frequency characteristic data is extracted simultaneously, calculating the argument of each complex element. A four-quadrant inverse tangent function is used to ensure that the phase angle output ranges from -180 degrees to +180 degrees. The calculated results are expressed in degrees. Together, the amplitude-frequency and phase-frequency characteristic data sequences form a complete description of the frequency domain characteristics of the open-loop transfer function. For example, 600 characteristic data points are generated at 166.7 kHz intervals within the 100 Hz to 10 MHz frequency band.

[0034] To ensure operational feasibility, the discrete Fourier transform process declares the use of a radix-2 fast Fourier transform algorithm, with implementation relying on floating-point arithmetic units or graphics processor hardware acceleration. When the total number of sampling points is not an integer power of 2, zeros are automatically padded to the nearest integer power of 2, for example, 5992 points are padded to 8192 points. An exception handling mechanism is defined for complex division operations: when the modulus of the spectral component of the switch node voltage ripple signal falls below the noise threshold, the data at that frequency point is deemed invalid and the calculation is skipped. The noise threshold is set to one thousandth of the full-scale voltage, based on the engineering practice requirement of a signal-to-noise ratio greater than 60 decibels. For example, the lower threshold of a 5V range is 5 millivolts. Boundary condition handling includes separate processing of the DC component to avoid a zero denominator, and limiting the frequency index range to the Nyquist interval of 0 to half the sampling rate.

[0035] The decibel conversion process for the amplitude-frequency characteristic data series clearly defines the conversion formula as 20 multiplied by the base-10 logarithm. This logarithm is implemented using the natural logarithm and the common logarithm conversion coefficient, which is the approximate value of 2.302585, the natural logarithm 10 divided by the natural base of the common logarithm. The phase angle calculation for the phase-frequency characteristic data series uses a four-quadrant inverse tangent function. The function input parameter is the ratio of the complex imaginary part to the real part. The output angle range covers the entire circumference. The specific implementation determines the quadrant affiliation by determining the sign combination of the real and imaginary parts. The storage format of the open-loop transfer function complex result array is declared as a double-precision floating-point complex array, with each element occupying 16 bytes of memory. The array length is equal to the number of valid frequency points. The mapping between frequency point index and physical frequency is determined by the frequency resolution, which is equal to the sampling rate divided by the number of transformation points. For example, at a sampling rate of 1 GHz, the frequency resolution for an 8192-point transformation is 122.07 kHz.

[0036] The generation of the spectral component array includes a window preprocessing step. A Hanning window is used to reduce spectral leakage. The window coefficient array is element-by-element multiplied with the original voltage signal array. Window normalization ensures signal energy conservation; the normalization factor is equal to the square root of the sum of the squares of the window coefficients. The spectral components output by the discrete Fourier transform require amplitude correction. The correction factor is the number of transformation points multiplied by the window energy compensation factor. For example, the Hanning window energy compensation factor is 2. Data validity is verified before complex division. Verification conditions include ensuring that the modulus of the divisor spectral component is greater than the noise threshold and that the frequency point is within the Nyquist frequency range. Frequency points that do not meet these conditions are marked as invalid data points. The real and imaginary components of the complex result of the open-loop transfer function are stored in separate arrays to facilitate subsequent modulus and argument calculations.

[0037] The modulus calculation process uses square and square root operations, implemented via library functions declared as standard double-precision floating-point arithmetic. The argument calculation uses the four-quadrant inverse tangent function, implemented based on the atan2 function from the standard math library. The decibel conversion of the amplitude-frequency characteristic data series is performed in a step-by-step manner: first, the natural logarithm of the modulus is calculated, then multiplied by 20 and the approximate reciprocal of the natural base of the common logarithm, 8.6858896380650368. The phase angle output units of the phase-frequency characteristic data series are uniformly expressed in degrees. The conversion formula is radians multiplied by 180 divided by the approximate value of pi, 3.141592653589793. The resulting amplitude-frequency and phase-frequency characteristic data series are stored as two-dimensional floating-point arrays, with the frequency value in the first column and the corresponding characteristic value in the second column. The frequency value is obtained by multiplying the frequency index by the frequency resolution.

[0038] S4. Mapping the amplitude-frequency characteristic and the phase-frequency characteristic to the Nyquist plane and identifying the phase angle of the intersection of the Nyquist curve and the unit circle can be specifically implemented as follows: After obtaining the amplitude-frequency characteristic data sequence and phase-frequency characteristic data sequence of the open-loop transfer function generated by the previous step, the conversion operation from decibel value to linear modulus value is performed. This conversion process traverses each data point in the amplitude-frequency characteristic data sequence and applies the conversion relationship to each decibel value: the linear modulus value is equal to the power of 10, and the exponent of the power is the quotient of the decibel value divided by 20. This calculation is implemented by the floating-point arithmetic unit of the processor performing double-precision floating-point operations, and the conversion formula is reflected in the linear modulus value being equal to the decibel value of 10 divided by the power of 20. For example, when the decibel value is negative 6.0206, the linear modulus value is 0.5, when the decibel value is 0, the linear modulus value is 1, and when the decibel value is positive 6.0206, the linear modulus value is 2. The converted linear modulus value is stored as a new data sequence, and the sequence length is exactly the same as the amplitude-frequency characteristic data sequence. The phase angle data in the phase-frequency characteristic data sequence maintains its original value. The phase angle is in degrees, and its numerical range is between negative 180 degrees and positive 180 degrees. This range is guaranteed by the four-quadrant inverse tangent function calculation in the previous step and is consistent with the output range of the inverse tangent function of the standard mathematical library.

[0039] After the conversion is completed, the linear modulus data sequence is mapped to the phase-frequency characteristic data sequence. The mapping process generates complex coordinates on the Nyquist plane for each frequency point: the real part of the complex coordinate is calculated by multiplying the linear modulus corresponding to the frequency point by the cosine value of the phase angle, and the imaginary part is calculated by multiplying the same linear modulus by the sine value of the phase angle. The cosine and sine functions are implemented by calling the double-precision trigonometric functions of the standard mathematical library. The function input angle unit is degree, and the calculation process automatically handles the angle periodicity problem. For example, when the phase angle is 0 degrees, the cosine value is 1 and the sine value is 0. When the phase angle is 90 degrees, the cosine value is 0 and the sine value is 1. When the phase angle is negative 45 degrees, the cosine and sine values ​​are both negative 0.7071. The complex coordinates of all frequency points are arranged in order from low to high frequency to form a discrete Nyquist curve. The curve data is stored as a two-dimensional double-precision floating-point array. The number of array rows is equal to the total number of frequency points. The first column stores the real part value, and the second column stores the imaginary part value. The array index increases from 0 to correspond to the order of increasing frequency. This order is exactly the same as the frequency point scanning order in the previous step.

[0040] When identifying the intersection of the Nyquist curve and the unit circle, the first step is to traverse each complex coordinate point in the curve data. The modulus is calculated for each complex coordinate point. This modulus calculation uses the square root of the real part plus the square of the imaginary part of the complex number. This calculation is performed by the processor's floating-point unit, performing double-precision square root calculations. For example, a real part of 0.6 and an imaginary part of 0.8 results in a modulus of 1. After obtaining the modulus, the absolute deviation of the modulus from 1 is calculated. The absolute deviation is equal to the absolute value of the modulus minus 1. This calculation is performed using the floating-point absolute value function for double precision. For example, a modulus of 1.05 results in an absolute deviation of 0.05, and a modulus of 0.97 results in an absolute deviation of 0.03. After the traversal is complete, the absolute deviation values ​​of all frequency points are compared, and the complex coordinate point with the smallest absolute deviation is selected as the intersection of the Nyquist curve and the unit circle. When multiple complex coordinate points share the same minimum absolute deviation, the point with the lowest frequency is selected as the intersection point. The frequency is determined by the order of the point's index in the Nyquist curve data, with smaller indexes indicating lower frequencies. Boundary condition handling includes the following: If the absolute deviation of all modulo values ​​is greater than 0.5, it is determined that there is no intersection and an exception handling process is triggered. This exception handling process includes returning a predefined error code 255 and skipping subsequent steps. The threshold of 0.5 is set based on engineering experience that system open-loop gain typically ranges from 0.5 to 2 times, which covers the stability margin analysis requirements of common control systems.

[0041] After determining the complex coordinate point corresponding to the intersection, the phase angle data point at the same index position in the phase-frequency characteristic data sequence is located based on the index position of the complex coordinate point in the Nyquist curve data. This phase angle value is directly read as the intersection phase angle output, and the read operation is an array index access operation. Because the Nyquist curve data and the phase-frequency characteristic data sequence have exactly the same frequency point order and index numbering, this order is determined by the discrete Fourier transform frequency point order in the previous step. The relationship between the frequency index number and the physical frequency is determined by the frequency resolution, which is equal to the sampling rate divided by the number of Fourier transform points. Therefore, the read operation can accurately obtain the phase angle at the intersection without introducing additional calculation errors. For example, when the complex coordinate of the intersection point is located at the 30th data point in the Nyquist curve data, the frequency of this point is 30 multiplied by the frequency resolution. If the phase angle value of the 30th data point in the phase-frequency characteristic data sequence is directly read and the value is negative 120 degrees, the output intersection phase angle is negative 120 degrees. The phase angle output value retains the original accuracy without any rounding or truncation processing to ensure the integrity of the phase information. The output data format is completely consistent with the phase-frequency characteristic data sequence storage format of the previous step.

[0042] The cosine and sine values ​​of the phase angle are calculated using double-precision trigonometric functions provided by the standard math library. The function calls automatically convert the angle input from degrees to radians for calculation. When the phase angle is at a boundary value, such as negative 180 degrees or positive 180 degrees, the cosine function outputs negative 1 and the sine function outputs zero. This result conforms to the mathematical definition and forms a closed loop with the phase angle range of the previous step. Nyquist curve data is stored using contiguous memory allocation, with each complex coordinate occupying two double-precision floating-point memory spaces. The storage order strictly corresponds to the frequency point index. During unit circle intersection identification, modulus calculation follows the standard square sum-then-square root calculation process. The square operation uses a floating-point multiplication instruction, while the square root operation uses a hardware-optimized square root instruction. The calculation accuracy meets the IEEE 754 double-precision floating-point standard. The exception handling threshold of 0.5 was determined by analyzing the open-loop gain fluctuation range of a typical control system. For example, if the system gain error exceeds 50%, it is considered to have significantly deviated from the design value, and the analysis process is interrupted. The intersection phase angle is directly referenced when outputting the original phase-frequency characteristic data to avoid precision loss caused by repeated calculations. The data reference process is implemented through memory address offset, and the offset is equal to the index value multiplied by the single-precision floating-point storage length.

[0043] S5. Generate a chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the neighborhood of the intersection point, which can be specifically implemented as follows: After accurately determining the intersection of the Nyquist curve and the unit circle and its corresponding frequency point index through the above steps, the intersection frequency point index value is used as the central reference point, and a fixed number of continuous frequency points are selected in the forward and backward directions along the frequency index axis to form a neighborhood range. The number of forward frequency points and backward frequency points of the intersection included in the neighborhood range is directly determined by the predefined neighborhood radius parameter, which is set to an integer greater than or equal to 5 and less than or equal to 20 based on the control system frequency resolution and engineering experience. For example, when the frequency resolution is 0.1 Hz, the neighborhood radius parameter is typically set to 10, which means that 10 frequency points are taken forward and 10 frequency points are taken backward with the intersection frequency point index as the center, forming a neighborhood window with a total width of 21 frequency points. The neighborhood radius parameter is stored in the configuration file and loaded into the memory variable through the file reading function. Its value setting needs to ensure that the frequency bandwidth covered by the neighborhood can reflect the dynamic characteristics of the system: when the frequency resolution is lower than 0.5 Hz, a smaller radius parameter such as 5 is used to avoid cross-modal interference. When the frequency resolution is higher than 0.01 Hz, a larger radius parameter such as 15 is used to capture the complete distortion characteristics.

[0044] For each frequency point within the neighborhood, the local curvature is calculated using the complex coordinate data corresponding to the frequency point and its immediately preceding and succeeding frequency points. The complex coordinates are derived from the discrete point set of the Nyquist curve calculated in the previous step and are stored as double-precision floating-point numbers for the real and imaginary parts. The specific calculation process is as follows: Get the complex coordinates corresponding to three consecutive frequency points, marked as complex coordinate A (index i-1), complex coordinate B (index i), and complex coordinate C (index i+1) in index order; Calculate vector AB: Subtract the real part of complex coordinate A from the real part of complex coordinate B to get the horizontal component, and subtract the imaginary part from the imaginary part to get the vertical component; Calculate vector BC: Subtract the real part of complex coordinate B from the real part of complex coordinate C to get the horizontal component, and subtract the imaginary part from the imaginary part to get the vertical component; To calculate the cosine of the angle θ between two vectors, multiply the horizontal component of vector AB by the horizontal component of vector BC, then add the vertical component of vector AB multiplied by the vertical component of vector BC. Multiply the modulus of vector AB (the square root of the sum of the squares of the horizontal and vertical components) by the modulus of vector BC. Finally, divide the dot product by the product of the moduli. Calculate the value of θ using inverse trigonometric functions: perform the arccosine operation on the cosine value obtained in step 4, and the result is in radians; Calculate the chord length L: Add the square of the difference between the horizontal components of the complex coordinates A and C to the square of the difference between the vertical components, and take the square root of the sum. Calculate the local curvature K: multiply the sine of the angle θ by 2 and divide it by the chord length L, where the sine is calculated using the trigonometric function library; This process is implemented by calling the standard math library. When it is detected that the modulus of vector AB or vector BC is less than 10 to the power of -10, it is considered a zero vector and a predefined error code is returned. When the three points are collinear and the angle θ approaches 0, the curvature value is set to 0.

[0045] After calculating the local curvature values ​​for all frequency points within the neighborhood, the resulting curvature values ​​are stored in an array. A traversal comparison algorithm is then used to identify the maximum and minimum values. The maximum value identification process initializes the variable to the theoretical minimum floating-point number, compares the array elements one by one, and updates the maximum value. Similarly, the minimum value identification process initializes the variable to the theoretical maximum floating-point number and updates the minimum value. The curvature range, R, is calculated as the difference between the maximum value and the minimum value, maintaining double precision. The traversal process uses a single loop structure, with the number of iterations equal to the total number of valid frequency points.

[0046] Normalization is performed by dividing the curvature range R by the total number of effective frequency points N actually involved in the calculation within the neighborhood. The normalized calculation expression is that the chaos distortion index is equal to the curvature range divided by the number of effective points, and the output result is a dimensionless scalar. The determination of the total number of effective frequency points N requires handling boundary conditions: When the center point index minus the neighborhood radius parameter is less than 0, the actual number of points taken in the forward direction is the current index value; When the center point index plus the neighborhood radius parameter exceeds the maximum index of the Nyquist curve, the actual number of points taken in the backward direction is the maximum index value minus the current index value; The final effective points N is equal to the forward effective points plus the backward effective points plus 1; For example, if the neighborhood radius parameter is 10: if the intersection index is 50 and the maximum index is 100, then N = 21; if the intersection index is 5, then the number of forward points = 5, the number of backward points = 10, and N = 16; if the intersection index is 95, then the number of forward points = 10, the number of backward points = 5, and N = 16. Division uses floating-point division instructions. If N is 0, a predefined invalid value is returned and an error is logged.

[0047] At the implementation constraint level: the neighborhood radius parameter range [5,20] is set based on the following: when the system frequency resolution is in the range of 0.01 to 1 Hz, this parameter can ensure that the neighborhood covers the key frequency band of 0.1 to 20 Hz. A value lower than 5 will miss high-frequency distortion features, and a value higher than 20 will introduce low-frequency interference; the angle processing in the curvature calculation uses radians, and the arc cosine and sine operations are implemented through the standard mathematical function library; the boundary index processing is implemented through conditional judgment: the forward point count takes the smaller value of the neighborhood radius parameter and the current index value, and the backward point count takes the smaller value of the neighborhood radius parameter and the maximum index minus the current index value; exception handling covers three scenarios: zero vector (returning a specific error code), division by zero error (returning an invalid value), and index out of bounds (automatically truncated to the valid range); the actual measured range of the chaos distortion index is usually lower than 0.05 in a stable system and exceeds 0.25 in a chaotic distortion system. This threshold is statistically derived from thousands of test data sets.

[0048] S6. Using the chaos distortion index to correct the intersection phase angle, and outputting the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result, can be specifically implemented as follows: The operation to obtain the original phase angle and corresponding chaotic distortion index determined at the intersection of the Nyquist curve and the unit circle is performed by extracting the index value corresponding to the intersection point from the array storing the discrete point data of the Nyquist curve. This index value is used to locate the original phase angle value stored in the phase angle data set, which is a floating-point number expressed in radians. The original phase angle is calculated by performing a standard four-quadrant inverse tangent operation on the ratio of the imaginary and real parts of the complex coordinate at the intersection frequency index, and then multiplying the result by the conversion factor 180 / π to convert it to a degree value. The chaotic distortion index is generated by dividing the curvature range by the number of valid neighborhood points and is stored in double-precision floating-point format. If the complex coordinate is located on the negative half of the real axis, a 180-degree offset is added to the original phase angle calculation to conform to the phase definition specification.

[0049] The operation method for determining the phase compensation amount based on the comparison result of the chaotic distortion index and the preset threshold is as follows: the first threshold and the second threshold are statistically derived from thousands of sets of historical test data. For example, the first threshold is set to 0.25 to identify a significant distortion state, and the second threshold is set to 0.05 to identify a stable state. The threshold setting is based on the following: in thousands of high-frequency switching power supply test samples, when the chaotic distortion index is higher than 0.25, the probability of system instability exceeds 95%, and when it is lower than 0.05, the probability of instability is less than 5%. The comparison logic is implemented through a conditional judgment structure: when the chaotic distortion index is greater than the first threshold, the difference between the chaotic distortion index and the first threshold is multiplied by the proportional coefficient to generate a positive compensation amount; when the chaotic distortion index is less than the second threshold, the difference between the chaotic distortion index and the second threshold is multiplied by the proportional coefficient to generate a negative compensation amount; if the chaotic distortion index is between the first and second thresholds, the phase compensation amount is set to zero. The scaling factor is set based on the power supply topology: for example, a flyback topology uses a scaling factor of 5, while a half-bridge topology uses a scaling factor of 3. These values ​​are determined through circuit simulation. The scaling factor is stored in a configuration file and loaded during program initialization.

[0050] The algebraic addition of the phase compensation value and the original phase angle is defined as a scalar addition operation: the original phase angle's angular value is directly added to the phase compensation value, with the result rounded to two decimal places. Data type verification is performed during the operation: if the original phase angle is not a floating-point number or the phase compensation value is not a numeric value, an exception handling process is triggered and an error code is returned. A unit consistency check is performed before the addition operation to confirm that both the original phase angle and the phase compensation value are in angular units.

[0051] The implementation details of comparing the corrected phase angle with the preset phase tolerance threshold and outputting the conclusion include: the phase tolerance threshold is set to 45 degrees, which is determined according to the IEEE stability criterion standard. The comparison process uses floating-point comparison instructions: when the corrected phase angle is less than 45 degrees, a stability qualified conclusion is sent to the output interface; when the corrected phase angle is greater than or equal to 45 degrees, a stability defect conclusion is sent. Data validity verification is performed before output: if the corrected phase angle exceeds the physical range of 0 to 360 degrees, the value is reset to zero and a warning message is added; if the input value is an error identification code, the data invalid conclusion is output.

[0052] The raw phase angle is derived from the intersection point's complex coordinate calculation process, and its accuracy is determined by the Nyquist curve sampling density. The chaotic distortion index is derived from the neighborhood curvature processing process, with the neighborhood range set to 10 data points before and after the intersection frequency index. The scaling factor is related to the power supply topology: the flyback topology has higher loop gain and is more sensitive to phase fluctuations, so a larger scaling factor of 5 is used; the half-bridge topology has lower gain and uses a scaling factor of 3. Boundary handling mechanisms include: when the chaotic distortion index is greater than 1.0, it is calculated as 1.0; when it is less than 0, it is calculated as 0; and the phase compensation output value is limited to the range of -10 degrees to +10 degrees. The phase tolerance threshold of 45 degrees aligns with the lower phase margin requirement in the IEEE Std 519 specification.

[0053] The following are examples of key parameter implementation: Example of obtaining the Chaos Distortion Index: When the number of valid points in the neighborhood is 20 and the curvature range is 4.0, the calculated value of the Chaos Distortion Index is 4.0 / 20=0.2; Compensation amount generation example: When the chaotic distortion index in the flyback topology is 0.3, since 0.3>0.25, the positive compensation amount = (0.3-0.25)×5=0.25 degrees; Phase correction example: The original phase angle of 44.5 degrees is added with a positive compensation of 0.25 degrees to generate a corrected phase angle of 44.75 degrees. Conclusion output example: The corrected phase angle of 44.75 degrees is less than 45 degrees, and the output stability is qualified.

[0054] All angle data are stored and calculated in a unified angle system; Exception identification code definition: -999 is returned for data type errors, and -998 is returned for out-of-range data; Compensation limit mechanism: When the calculated value exceeds +10 degrees, +10 degrees are forced to be output, and when it is lower than -10 degrees, -10 degrees are output; Threshold application order: First determine whether the chaos distortion index is greater than the first threshold, and then determine whether it is less than the second threshold.

[0055] In the technical field of evaluating the loop stability of high-frequency switching power supplies, traditional methods fail to recognize the systematic interference that chaotic distortion generated by switching power supplies under specific operating conditions can cause on phase measurement accuracy. This embodiment, through the technical chain constructed through steps S1 to S6, reveals the hidden correlation mechanism between chaotic distortion and phase estimation error and establishes a compensation system that can be implemented in an engineering manner. Specifically: First, the quantitative method for constructing a chaotic distortion index based on the extreme curvature of a frequency neighborhood in steps S1 to S3 breaks through the limitation of traditional stability analysis, which focuses solely on single-point phase information, and incorporates the local distortion degree of the curve into the evaluation system. Second, steps S4 to S5 establish a dynamic mapping rule between the chaotic distortion index and the phase compensation amount, and achieve intelligent switching of the compensation direction through a dual-threshold judgment mechanism. This design is not a conventional method of simple threshold comparison, but is based on in-depth exploration of the nonlinear relationship between the degree of distortion and the probability of system instability in thousands of test data sets. Finally, step S6, through the algebraic superposition of the corrected phase angle generated, substantially eliminates the evaluation bias introduced by chaotic noise, resulting in a step-by-step improvement in the reliability of the stability conclusion.

[0056] Example 2: Figure 2 A schematic structural diagram of a power supply detection device of the present invention is provided, wherein the power supply detection device comprises: Disturbance acquisition module: In a closed-loop state, it collects the switch node voltage ripple signal of the high-frequency switching power supply as the disturbance input signal, and synchronously collects the output voltage feedback signal of the high-frequency switching power supply as the response output signal; Time domain alignment module: performs time domain alignment on the disturbance input signal and the response output signal to generate a time-correlated signal pair; Frequency domain conversion module: performs Fourier transform on the time-correlated signal pair to extract the amplitude-frequency and phase-frequency characteristics of the open-loop transfer function; Intersection identification module: maps the amplitude-frequency characteristics and phase-frequency characteristics to the Nyquist plane and identifies the phase angle of the intersection of the Nyquist curve and the unit circle; Distortion quantification module: Generates chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; Evaluation and correction module: uses the chaos distortion index to correct the intersection phase angle, and outputs the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

[0057] The power supply detection device synchronously acquires the switching node voltage ripple signal and the output voltage feedback signal through the disturbance acquisition module to form the original input-output data pair; the time domain alignment module receives the data pair, eliminates the transmission delay by triggering edge alignment and sampling clock synchronization, and generates a strictly synchronized time-correlated signal pair; the frequency domain conversion module performs a windowed Fourier transform on the aligned signal pair to extract the amplitude-frequency characteristic curve and phase-frequency characteristic curve of the open-loop transfer function; the intersection identification module receives the frequency domain characteristic data, maps the complex coordinates to the Nyquist plane, locates the intersection of the Nyquist curve and the unit circle through the minimum Euclidean distance search algorithm, and records its phase angle; the distortion quantification module generates a chaotic distortion index based on the curvature extremes of the intersection neighborhood and the density of valid data points. The index quantifies the local distortion intensity using a predetermined formula; the evaluation and correction module dynamically calculates the phase compensation amount based on the distortion index, corrects the original phase angle through algebraic superposition, and finally compares the correction value with the preset phase tolerance threshold to output a stability conclusion.

[0058] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to actual conditions.

[0059] It should be noted that the present invention can be deployed on the device itself to implement embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0060] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0061] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0062] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0063] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0064] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0065] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0066] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0067] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A power supply testing method, characterized in that: include: S1. In a closed-loop state, a switch node voltage ripple signal of a high-frequency switching power supply is collected as a disturbance input signal, and an output voltage feedback signal of the high-frequency switching power supply is simultaneously collected as a response output signal; S2, aligning the disturbance input signal and the response output signal in the time domain to generate a time-correlated signal pair; S3, performing Fourier transform on the time-correlated signal pair to extract the amplitude-frequency characteristic and phase-frequency characteristic of the open-loop transfer function; S4, mapping the amplitude-frequency characteristic and the phase-frequency characteristic to the Nyquist plane, and identifying the phase angle of the intersection of the Nyquist curve and the unit circle; S5. Generate a chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; S6. Use the chaos distortion index to correct the intersection phase angle, and output the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

2. A power supply testing method according to claim 1, characterized in that: The method collects the switch node voltage ripple signal of the high-frequency switching power supply as a disturbance input signal in a closed-loop state, and synchronously collects the output voltage feedback signal of the high-frequency switching power supply as a response output signal, including: Use a differential voltage probe to connect the switch node of the high-frequency switching power supply to the reference ground, and capture the switch node voltage ripple signal at a sampling rate no less than ten times the switching frequency of the high-frequency switching power supply. At the same time, a voltage probe is used to connect the output voltage feedback node of the high-frequency switching power supply to capture the output voltage feedback signal at the same sampling rate as the switching node voltage ripple signal; The switch node voltage ripple signal and the output voltage feedback signal are respectively conditioned by an isolation amplifier, and after eliminating common-mode interference, they are transmitted to a digital storage oscilloscope; A synchronous trigger channel between the switch node voltage ripple signal and the output voltage feedback signal is established in the digital storage oscilloscope to align the acquisition time starting points and keep the sampling clocks from the same source.

3. A power supply testing method according to claim 1, characterized in that: Perform time domain alignment on the disturbance input signal and the response output signal to generate a time-correlated signal pair, including: Extract the trigger timestamp of the switch node voltage ripple signal channel as the reference time; Based on the reference time, the waveform data time axis of the output voltage feedback signal channel is shifted so that the time coordinates of the switch node voltage ripple signal channel and the output voltage feedback signal channel at the reference time are aligned to zero; Identify the complete switching cycle boundary based on the switching node voltage waveform characteristics, and extract waveform segments corresponding to the number of consecutive complete switching cycles from the dual-channel waveforms after time domain alignment; A one-to-one mapping relationship is established between the switch node voltage ripple signal data points corresponding to all sampling time points in the intercepted waveform segment and the output voltage feedback signal data points to form a set of time-correlated signal pairs.

4. A power supply testing method according to claim 3, characterized in that: The trigger timestamp is generated when the trigger pulse arrives at the sample-and-hold circuit of the switch node voltage ripple signal path.

5. A power supply testing method according to claim 1, characterized in that: Perform Fourier transform on the time-correlated signal pair to extract the amplitude-frequency and phase-frequency characteristics of the open-loop transfer function, including: Performing discrete Fourier transform on the switch node voltage ripple signal data and the output voltage feedback signal data in the time-correlated signal pair set to obtain a switch node voltage ripple signal spectrum component and an output voltage feedback signal spectrum component; In the complex domain, the spectral component of the output voltage feedback signal is divided by the spectral component of the switch node voltage ripple signal to obtain a complex result of the open-loop transfer function at the corresponding frequency point; Extracting amplitude-frequency characteristic data from the modulus of the complex result of the open-loop transfer function, and extracting phase-frequency characteristic data from the argument of the complex result of the open-loop transfer function; The number of operation points of the discrete Fourier transform is equal to the total number of sampling points of the set of time-correlated signal pairs, and the complex division operation is performed between spectral components corresponding to the same frequency point.

6. A power supply testing method according to claim 1, characterized in that: Map the amplitude-frequency characteristics and phase-frequency characteristics to the Nyquist plane and identify the phase angle of the intersection of the Nyquist curve and the unit circle, including: Convert the decibel values ​​in the amplitude-frequency characteristic data sequence into linear modulus values, and map the linear modulus values ​​and the phase angles in the phase-frequency characteristic data sequence to the Nyquist plane, and form a Nyquist curve consisting of the complex coordinates of discrete frequency points by calculating the real part of the complex coordinates of each frequency point as the linear modulus value multiplied by the cosine value of the phase angle and the imaginary part as the linear modulus value multiplied by the sine value of the phase angle; Identify the complex coordinate point with the modulus closest to 1 in the Nyquist curve as the intersection point of the Nyquist curve and the unit circle; The intersection phase angle is extracted from the phase angle in the phase-frequency characteristic data sequence corresponding to the complex coordinate point.

7. A power supply testing method according to claim 1, characterized in that: The chaotic distortion index is generated based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood, including: After determining the intersection point with the unit circle in the Nyquist curve, a fixed number of frequency points are taken forward and backward with the frequency point index corresponding to the intersection point as the center to form a neighborhood range; For each frequency point within the neighborhood, the local curvature value at the corresponding frequency point is solved by using the complex coordinates corresponding to the frequency point and its two adjacent frequency points and calculating the geometric relationship between the vector angle and chord length between the adjacent complex coordinates. After traversing all frequency points in the neighborhood and completing the local curvature calculation, the maximum and minimum local curvature values ​​in the neighborhood are identified, and the difference between the maximum value and the minimum value is calculated as the curvature range; The curvature range is divided by the total number of frequency points in the neighborhood for normalization, and the resulting dimensionless scalar value is the chaos distortion index; When the neighborhood range boundary exceeds the Nyquist curve index range, normalization calculation is performed based on the total number of frequency points actually included.

8. A power supply testing method according to claim 7, characterized in that: The number of frequency points before and after the intersection point included in the neighborhood range is determined by the predefined neighborhood radius parameter.

9. A power supply testing method according to claim 1, characterized in that: The chaotic distortion index is used to correct the intersection phase angle, and the loop stability evaluation conclusion of the high-frequency switching power supply is output based on the correction result, including: Obtain the original phase angle and the corresponding chaos distortion index determined by the intersection of the Nyquist curve and the unit circle; Determining a phase compensation amount based on a comparison result between a chaos distortion index and a preset threshold: generating a positive compensation amount when the chaos distortion index exceeds a first threshold, and generating a negative compensation amount when the chaos distortion index is lower than a second threshold; The phase compensation amount is algebraically superimposed on the original phase angle to generate a corrected phase angle; The corrected phase angle is compared with a preset phase tolerance threshold: when the corrected phase angle is less than the phase tolerance threshold, a stability qualified conclusion is output; otherwise, a stability defect conclusion is output.

10. A power supply detection device, used to implement a power supply testing method according to any one of claims 1 to 9, characterized in that: include: Disturbance acquisition module: In a closed-loop state, it collects the switch node voltage ripple signal of the high-frequency switching power supply as the disturbance input signal, and synchronously collects the output voltage feedback signal of the high-frequency switching power supply as the response output signal; Time domain alignment module: performs time domain alignment on the disturbance input signal and the response output signal to generate a time-correlated signal pair; Frequency domain conversion module: performs Fourier transform on the time-correlated signal pair to extract the amplitude-frequency and phase-frequency characteristics of the open-loop transfer function; Intersection identification module: maps the amplitude-frequency characteristics and phase-frequency characteristics to the Nyquist plane and identifies the phase angle of the intersection of the Nyquist curve and the unit circle; Distortion quantification module: Generates chaotic distortion index based on the local curvature change characteristics of the Nyquist curve in the intersection neighborhood; Evaluation and correction module: uses the chaos distortion index to correct the intersection phase angle, and outputs the loop stability evaluation conclusion of the high-frequency switching power supply based on the correction result.

Images