Automatic oscilloscope signal analysis method based on double-pulse test

Abstract

The invention discloses an automatic oscilloscope signal analysis method based on double-pulse test, which is used for testing a power switch device and comprises the following steps of: S1, carrying out double-pulse test on the power switch device by an oscilloscope to obtain a corresponding time sequence signal and carrying out filtering processing to eliminate a high-frequency signal in the time sequence signal so as to obtain a filtered signal; s2, dividing the filtering signal into multiple sections of signals through a sliding window, and obtaining a trend interval of each signal channel of the oscilloscope after processing; and S3, respectively calculating each index of the power switch device according to the trend interval of each signal channel. According to the automatic oscilloscope signal analysis method based on the double-pulse test, parameters of various double-pulse tests do not need to be judged manually according to the signals of the oscilloscope, and a large amount of manpower and time are saved especially in the scene that a large number of power devices need to be tested in a factory.

Description

Technical Field

[0001] The invention belongs to the technical field of power switch device testing, and in particular relates to an automatic oscilloscope signal analysis method based on double pulse testing. Background Art

[0002] Power switching devices play a vital role in switching power supplies. They control the transfer and conversion of electrical energy by quickly turning on and off, thereby achieving efficient operation of the power supply. The performance of these devices directly determines the efficiency, output power and reliability of the switching power supply. High-quality power switching devices can maintain low conduction losses and fast switching characteristics under high-frequency working conditions, greatly improving the performance and stability of the power supply. Therefore, it is particularly important to evaluate the performance of power switching devices. Among them, double pulse testing is a basic method for analyzing the characteristics of power switching devices. This method can obtain parameters such as turn-on delay time (Tdon), turn-off delay time (Tdoff), rise time (Tr), and fall time (Tf), providing data support for optimizing product design.

[0003] Currently, double pulse testing is performed by manually observing oscilloscope signals and manually locating the time when calculating each parameter, which is inefficient and wastes manpower.

[0004] Therefore, further improvements are made to the above problems. Summary of the invention

[0005] The main purpose of the present invention is to provide an automated oscilloscope signal analysis method based on double pulse testing, which eliminates the need for manual judgment of the parameters of each double pulse test based on the oscilloscope signal, saving a lot of manpower and time, especially in scenarios where a large number of power devices need to be tested in a factory.

[0006] To achieve the above objectives, the present invention provides an automated oscilloscope signal analysis method based on double pulse testing, which tests a power switch device, comprising the following steps:

[0007] Step S1: an oscilloscope performs a double pulse test on a power switch device, obtains a corresponding timing signal, and performs filtering processing to eliminate high-frequency signals in the timing signal to obtain a filtered signal;

[0008] Step S2: Divide the filtered signal into multiple signal segments (i.e., multiple small windows) through a sliding window, and obtain the trend interval of each signal channel (CH1-CH4) of the oscilloscope after processing;

[0009] Step S3: Calculate various indicators of the power switch device according to the trend interval of each signal channel.

[0010] As a further preferred technical solution of the above technical solution, in step S1, a (Butterworth) low-pass filter is used to filter out high-frequency signals, and the formula is:

[0011]

[0012] Among them, ω c Represents the cutoff frequency, n is the filter order, and the filter is only allowed to be less than ω c The frequency of the original input is multiplied by A filtered signal is obtained by filtering out high frequency signals.

[0013] As a further preferred technical solution of the above technical solution, in step S2, the filtered signal of the current signal channel is segmented by a sliding window, and the waveform trend of each window is obtained after segmentation, and the differential mean of each window is calculated by the following formula:

[0014]

[0015] Among them, ch i represents the value of the current signal channel array (the signal channels are CH1-CH4, and each signal channel is calculated through step S2) at the i-th moment, and N represents the number of elements in the current signal channel array;

[0016] The difference mean of all windows is counted to obtain the difference mean sequence MD, and the outliers are calculated based on MD. The outliers are windows with an upward or downward trend (i.e., the following is greater than bound upper The value and value are less than bound lower The upper quartile q3 and lower quartile q1 of the difference mean sequence MD are calculated by the following formula:

[0017] q i =(x-floor(x))×MD ceil(x) +(ceil(x)-x)×MD floor(x) ;

[0018] in, Indicates the quantile position, floor(·) is the downward rounding function, ceil(·) is the upward rounding function, and then calculates the upper bound of the cluster upper and the lower bound lower :

[0019] bound upper =q3+λ(q3-q1);

[0020] boundlower =q1-λ(q3-q1);

[0021] Where λ is the scaling factor of the control boundary (set to 3 in the present invention), and the MD is larger than the bound upper The value is set to 1, which is less than bound lower The value of is set to -1, and the rest of the values ​​are set to 0, and the adjacent elements with the same value are merged and their first and last positions are recorded (w end 、w start ), keep w end -w start Elements with a value ≥ a preset value (the preset value is preferably 10) are selected to obtain the trend interval.

[0022] As a further preferred technical solution of the above technical solution, for the signal channel:

[0023] CH1 represents the gate-source voltage signal of the power switching device;

[0024] CH2 represents the drain-source voltage signal of the power switch device;

[0025] CH3 represents the current signal of the power switch device;

[0026] CH4 represents the fast recovery diode voltage signal of the power switching device.

[0027] As a further preferred technical solution of the above technical solution, in step S3, the calculation of each index is specifically implemented as follows:

[0028] Calculation of Tdon indicator (on delay time): Take the rising trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the rising trend interval respectively, and then find the time when CH1 peak reaches 10% (industry standard value for double pulse test, the same below) in chronological order, and find the time when CH2 peak reaches 90%, and the difference between the two is obtained to get Tdon indicator;

[0029] Calculation of Tr indicator (rise time): Take the downward trend interval of CH2, find the peak value of CH2 in the downward trend interval, and then find the time when CH2 reaches 90% of the peak value and the time when CH2 reaches 10% of the peak value in chronological order, and make the difference between the two to get the Tr indicator;

[0030] Calculation of Tdoff indicator (turn-off delay time): Take the downward trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the downward trend interval respectively, then find the time when CH1 peak value reaches 90% and the time when CH2 peak value reaches 10% in chronological order, and make the difference between the two to get Tdoff indicator;

[0031] Calculation of Tf indicator (fall time): Take the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, and then find the time when CH2 reaches 10% of the peak value and the time when CH2 reaches 90% of the peak value in chronological order, and make the difference between the two to get the Tf indicator;

[0032] Didton indicator (turn-on current change rate) calculation: take out the rising trend interval of CH3, find the peak value of CH3 in the rising trend interval, then find the 10% value of the CH3 peak value and the corresponding time and the 90% value of the CH3 peak value and the corresponding time in chronological order, and finally calculate the slope, which is the Didton indicator;

[0033] Calculation of Dvdtoff indicator (turn-off voltage change rate): Take out the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, then find the 10% value of CH2 peak value and the corresponding time and the 90% value of CH2 peak value and the corresponding time in chronological order, and finally calculate the slope, which is the Dvdtoff indicator;

[0034] Eon indicator (opening loss) calculation: Take the rising trend interval of CH1 and the falling trend interval of CH2, find the peak value of CH1 in the rising trend interval and the peak value of CH2 in the falling trend interval, then find the time T1 corresponding to 10% of the peak value of CH1 and the time T2 corresponding to 2% of the peak value of CH2 in chronological order, and finally calculate the value according to the formula Get the Eon indicator (t represents the time interval from T1 to T2).

[0035] Eoff indicator (turn-off loss) calculation: Take the downward trend interval of CH1 and the downward trend interval of CH3, find the peak value of CH1 in the downward trend interval and the peak value of CH3 in the downward trend interval, and then find the time T1 corresponding to 90% of the peak value of CH1 and the time T2 corresponding to 2% of the peak value of CH3 in chronological order, and finally calculate according to the formula Get the Eoff indicator (t represents the time interval from T1 to T2).

[0036] The beneficial effects of the present invention are:

[0037] 1. It can realize double pulse testing automatically, saving a lot of manpower and time.

[0038] 2. Solved the problem of noise sensitivity in the original trend judgment algorithm

[0039] 3. Solved the problem that the original trend judgment algorithm cannot handle multi-peak waveforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1It is a schematic diagram of digitizing oscilloscope signals of the present invention.

[0041] Figure 2 It is a schematic diagram of the signal filtering result of the present invention.

[0042] Figure 3 It is a schematic diagram of the sliding window segmentation signal of the present invention.

[0043] Figure 4 It is a schematic diagram of trend intervals of each signal channel of the present invention. DETAILED DESCRIPTION

[0044] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0045] In the preferred embodiment of the present invention, those skilled in the art should note that the double pulse test, oscilloscope, etc. involved in the present invention can be regarded as prior art.

[0046] Preferred embodiments.

[0047] like Figure 1-4 As shown, the software and library for implementing the present invention are preferably: IDE: pycharm (python); library: numpy, sklearn, scipy, pandas, matplotlib.

[0048] The timing signal of the oscilloscope is as follows Figure 1 As shown, Figure 1 The result after signal digitization is shown, and the signal trend judgment process is introduced by taking the gate-source voltage signal (CH1) as an example. The trend of the timing signal can be judged by calculating the difference between the previous and next moments in the sequence and averaging them, as shown in the formula:

[0049]

[0050] Among them ch i represents the value of the signal CH1 array at the i-th moment, and N represents the number of elements in the CH1 array. However, the formula has the following two problems: first, it is not robust to noise and is easy to identify noise as an upward or downward trend; second, it cannot identify waveforms with multiple peaks. To address the above two problems, filtering algorithms and sliding window methods are used.

[0051] The present invention discloses an automatic oscilloscope signal analysis method based on double pulse test, which tests a power switch device, comprising the following steps:

[0052] Step S1: The oscilloscope performs a double pulse test on the power switch device, obtains the corresponding timing signal and performs filtering processing (to prevent the trend judgment algorithm from identifying the high-frequency signal as an upward trend or a downward trend) to eliminate the high-frequency signal in the timing signal to obtain a filtered signal, such as Figure 2 As shown, take signal channel CH1 as an example;

[0053] Step S2: Divide the filtered signal into multiple segments (i.e., multiple small windows, such as Figure 3 After processing, the trend intervals of each signal channel (CH1-CH4) of the oscilloscope are obtained. The results are as follows Figure 4 As shown;

[0054] Step S3: Calculate various indicators of the power switch device according to the trend interval of each signal channel.

[0055] Specifically, in step S1, a (Butterworth) low-pass filter is used to filter out high-frequency signals, and the formula is:

[0056]

[0057] Among them, ω c represents the cutoff frequency, b is the filter order, and the filter is only allowed to be less than ω c The frequency of the original input is multiplied by A filtered signal is obtained by filtering out high frequency signals.

[0058] For step S1, the signal waveform is first filtered using a formula, specifically, the butter function in the scipy.signal library is called, and the parameter order (N) is set to 1, the 3dB bandwidth point (Wn) is set to 0.01, and the filter type (btype) is set to lowpass to obtain two parameters b and a. Then the parameters b, a and the original signal are sent to the filtfilt function in the scipy.signal library to obtain the filtered waveform.

[0059] More specifically, in step S2, Figure 3 The signal of each window in the middle red area, and the dark red area is the overlapping part of adjacent windows. The filtered signal of the current signal channel is segmented by the sliding window. After segmentation, the waveform trend of each window is obtained, and the differential mean of each window is calculated by the following formula:

[0060]

[0061] Among them, ch irepresents the value of the current signal channel array (the signal channels are CH1-CH4, and each signal channel is calculated through step S2) at the i-th moment, and N represents the number of elements in the current signal channel array;

[0062] The difference mean of all windows is counted to obtain the difference mean sequence MD, and the outliers are calculated based on MD. The outliers are windows with an upward or downward trend (i.e., the following is greater than bound upper The value and value are less than bound lower The upper quartile q3 and lower quartile q1 of the difference mean sequence MD are calculated by the following formula:

[0063] q i =(x-floor(x))×MD ceil(x) +(ceil(x)-x)×MD floor(x) ;

[0064] in, Indicates the quantile position, floor(·) is the downward rounding function, ceil(·) is the upward rounding function, and then calculates the upper bound of the cluster upper and the lower bound lower :

[0065] bound upper =q3+λ(q3-q1);

[0066] bound lower =q1-λ(q3-q1);

[0067] Where λ is the scaling factor of the control boundary (set to 3 in the present invention), and the MD is larger than the bound upper The value is set to 1, which is less than bound lower The value of is set to -1, and the rest of the values ​​are set to 0, and the adjacent elements with the same value are merged and their first and last positions are recorded (w end 、w start ), keep w end -w start ≥ the preset value (preferably 10), and finally get the trend interval, such as Figure 4 shown.

[0068] For step S2, use a sliding window to divide the filtered waveform into small windows with a window size of 500 and a step size of 20. Use the formula to calculate the differential mean of each small window. Calculate the upper quartile and lower quartile by the formula based on the differential mean of each small window, and then use the formula to calculate the upper and lower bounds. The values ​​in the small window greater than the upper bound are set to 1, the values ​​less than the lower bound are set to -1, and the rest are set to 0.

[0069] For adjacent elements with the same merge value, the DBSCAN clustering algorithm in the sklearn.cluster library is used to cluster adjacent elements with the same value into one category. The parameters eps and min_samples in the DBSCAN function are set to 1 and 2, respectively. Then, small windows with the same category are merged and their first and last positions (w start 、w end ), keep w end -w start Elements with values ​​≥10 get the final trend interval.

[0070] Furthermore, for the signal channel:

[0071] CH1 represents the gate-source voltage signal of the power switching device;

[0072] CH2 represents the drain-source voltage signal of the power switch device;

[0073] CH3 represents the current signal of the power switch device;

[0074] CH4 represents the fast recovery diode voltage signal of the power switching device.

[0075] Furthermore, in step S3, the calculation of each indicator is specifically implemented as follows:

[0076] Calculation of Tdon indicator (on delay time): Take the rising trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the rising trend interval respectively, and then find the time when CH1 peak reaches 10% (industry standard value for double pulse test, the same below) in chronological order, and find the time when CH2 peak reaches 90%, and the difference between the two is obtained to get Tdon indicator;

[0077] Calculation of Tr indicator (rise time): Take the downward trend interval of CH2, find the peak value of CH2 in the downward trend interval, and then find the time when CH2 reaches 90% of the peak value and the time when CH2 reaches 10% of the peak value in chronological order, and make the difference between the two to get the Tr indicator;

[0078] Calculation of Tdoff indicator (turn-off delay time): Take the downward trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the downward trend interval respectively, then find the time when CH1 peak value reaches 90% and the time when CH2 peak value reaches 10% in chronological order, and make the difference between the two to get Tdoff indicator;

[0079] Calculation of Tf indicator (fall time): Take the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, and then find the time when CH2 reaches 10% of the peak value and the time when CH2 reaches 90% of the peak value in chronological order, and make the difference between the two to get the Tf indicator;

[0080] Didton indicator (turn-on current change rate) calculation: take out the rising trend interval of CH3, find the peak value of CH3 in the rising trend interval, then find the 10% value of the CH3 peak value and the corresponding time and the 90% value of the CH3 peak value and the corresponding time in chronological order, and finally calculate the slope, which is the Didton indicator;

[0081] Calculation of Dvdtoff indicator (turn-off voltage change rate): Take out the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, then find the 10% value of CH2 peak value and the corresponding time and the 90% value of CH2 peak value and the corresponding time in chronological order, and finally calculate the slope, which is the Dvdtoff indicator;

[0082] Eon indicator (opening loss) calculation: Take the rising trend interval of CH1 and the falling trend interval of CH2, find the peak value of CH1 in the rising trend interval and the peak value of CH2 in the falling trend interval, then find the time T1 corresponding to 10% of the peak value of CH1 and the time T2 corresponding to 2% of the peak value of CH2 in chronological order, and finally calculate the value according to the formula Get the Eon indicator (t represents the time interval from T1 to T2).

[0083] Eoff indicator (turn-off loss) calculation: Take the downward trend interval of CH1 and the downward trend interval of CH3, find the peak value of CH1 in the downward trend interval and the peak value of CH3 in the downward trend interval, and then find the time T1 corresponding to 90% of the peak value of CH1 and the time T2 corresponding to 2% of the peak value of CH3 in chronological order, and finally calculate according to the formula Get the Eoff indicator (t represents the time interval from T1 to T2).

[0084] It is worth mentioning that the technical features such as double pulse testing and oscilloscope involved in the patent application of this invention should be regarded as prior art. The specific structure, working principle and possible control method and spatial layout method of these technical features can be selected according to the conventional selection in the field, and should not be regarded as the inventive point of the patent of this invention. The patent of this invention will not be further elaborated.

[0085] For those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An automated oscilloscope signal analysis method based on double pulse testing for testing power switching devices, characterized in that: The following steps are involved: Step S1: an oscilloscope performs a double pulse test on a power switch device, obtains a corresponding timing signal, and performs filtering processing to eliminate high-frequency signals in the timing signal to obtain a filtered signal; Step S2: Divide the filtered signal into multiple signal segments through a sliding window, and obtain the trend interval of each signal channel of the oscilloscope after processing; Step S3: Calculate various indicators of the power switch device according to the trend interval of each signal channel.

2. The automatic oscilloscope signal analysis method based on double pulse test according to claim 1, characterized in that: In step S1, high-frequency signals are filtered out by a low-pass filter, and the formula is: Among them, ω c Represents the cutoff frequency, n is the filter order, and the filter is only allowed to be less than ω c The frequency of the original input is multiplied by A filtered signal is obtained by filtering out high frequency signals.

3. The automatic oscilloscope signal analysis method based on double pulse test according to claim 2, characterized in that: In step S2, the filtered signal of the current signal channel is segmented by a sliding window, and the waveform trend of each window is obtained after segmentation, and the differential mean of each window is calculated by the following formula: Among them, ch i represents the value of the current signal channel array at the i-th moment, and N represents the number of elements in the current signal channel array; The difference mean of all windows is counted to obtain the difference mean sequence MD. The outliers are calculated based on MD. The outliers are windows with an upward or downward trend. The upper quartile q3 and lower quartile q1 of the difference mean sequence MD are calculated by the following formula: q i =(x-floor(x))×MD ceil(x) +(ceil(x)-x)×MD floor(x) ; in, Indicates the quantile position, floor(·) is the downward rounding function, ceil(·) is the upward rounding function, and then calculates the upper bound of the cluster upper and the lower bound lower : bound upper =q3+λ(q3-q1); bound lower =q1-λ(q3-q1); Among them, λ is the scaling factor of the control boundary, which is greater than the bound in MD. upper The value is set to 1, which is less than bound lower The value of is set to -1, and the rest of the values ​​are set to 0, and the adjacent elements with the same value are merged and their first and last positions are recorded (w end 、w start ), keep w end -w start Elements ≥ the preset value, and finally the trend interval is obtained.

4. The automatic oscilloscope signal analysis method based on double pulse test according to claim 3, characterized in that: For the signal channel: CH1 represents the gate-source voltage signal of the power switching device; CH2 represents the drain-source voltage signal of the power switch device; CH3 represents the current signal of the power switch device; CH4 represents the fast recovery diode voltage signal of the power switching device.

5. The method for automatic oscilloscope signal analysis based on double pulse test according to claim 4, characterized in that: In step S3, the calculation of each indicator is specifically implemented as follows: Tdon indicator calculation: Take the rising trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the rising trend interval respectively, then find the time when CH1 reaches 10% of its peak value and the time when CH2 reaches 90% of its peak value in chronological order, and make the difference between the two to get the Tdon indicator; Calculation of Tr indicator: Take out the downward trend interval of CH2, find the peak value of CH2 in the downward trend interval, and then find the time when CH2 reaches 90% of the peak value and the time when CH2 reaches 10% of the peak value in chronological order, and make the difference between the two to get the Tr indicator; Tdoff indicator calculation: Take the downward trend interval of CH1 and CH2 in the corresponding time period, find the peak values ​​of CH1 and CH2 in the downward trend interval respectively, then find the time when the peak value of CH1 reaches 90% and the time when the peak value of CH2 reaches 10% in chronological order, and make the difference between the two to get the Tdoff indicator; Tf indicator calculation: take out the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, and then find the time when CH2 reaches 10% of the peak value and the time when CH2 reaches 90% of the peak value in chronological order, and make the difference between the two to get the Tf indicator; Didton indicator calculation: take out the rising trend interval of CH3, find the peak value of CH3 in the rising trend interval, then find the 10% value and corresponding time of the CH3 peak value and the 90% value and corresponding time of the CH3 peak value in chronological order, and finally calculate the slope, which is Didton's indicator; Dvdtoff indicator calculation: take out the rising trend interval of CH2, find the peak value of CH2 in the rising trend interval, then find the 10% value of the CH2 peak value and the corresponding time and the 90% value of the CH2 peak value and the corresponding time in chronological order, and finally calculate the slope, which is the Dvdtoff indicator; Eon indicator calculation: Take the rising trend interval of CH1 and the falling trend interval of CH2, find the peak of CH1 in the rising trend interval and the peak of CH2 in the falling trend interval, then find the time T1 corresponding to 10% of the peak of CH1 and the time T2 corresponding to 2% of the peak of CH2 in chronological order, and finally calculate according to the formula Get the Eon indicator; Eoff indicator calculation: Take the downward trend interval of CH1 and the downward trend interval of CH3, find the peak of CH1 in the downward trend interval and the peak of CH3 in the downward trend interval, then find the time T1 corresponding to 90% of the peak of CH1 and the time T2 corresponding to 2% of the peak of CH3 in chronological order, and finally calculate according to the formula Get the Eoff index.

Images