Electronic device thermal characteristic evaluation method based on Wiener filter

Abstract

The invention discloses an electronic device thermal characteristic evaluation method based on a Wiener filter. The method specifically comprises the steps that the junction temperature of an electronic device to be tested is obtained through thermal testing, a thermal impedance curve of the electronic device to be tested is obtained through calculation according to the junction temperature, and a thermal impedance element conversion derivative curve is obtained through element conversion processing and derivation processing; carrying out deconvolution on the thermal impedance transmutated derivative curve to obtain a time constant spectrum of the electronic device to be detected; obtaining a thermal characteristic evaluation result of the electronic device according to the time constant spectrum; in the deconvolution process, a Wiener filter is applied to the frequency domain form of a thermal impedance transmutated derivative curve, and a thermal characteristic representation function is introduced. Compared with two methods which are widely applied at the present stage, the modulation method is simpler, the accuracy is more close to the essence, and the obtained structure function can accurately reflect the thermal characteristics of each layer of the tested device.

Description

Technical Field

[0001] The invention relates to the field of thermal testing of power electronic devices, and in particular to a method for evaluating thermal characteristics of electronic devices based on a Wiener filter. Background Art

[0002] With the rapid development of wide bandgap semiconductor devices, power devices based on SiC and GaN materials have gradually occupied an important position in high-frequency, high-voltage, and high-current working environments. With the increase in working intensity and working time, the heat accumulation phenomenon inside the device becomes more and more severe, seriously affecting the working characteristics and service life of the device. Therefore, a test method is needed that can monitor the changes in the thermal conductivity characteristics of each layer inside the device caused by the heat accumulation phenomenon in real time.

[0003] Among the methods for monitoring the phenomenon of heat accumulation, the structure function method is the most widely used method in current research. This method is a non-destructive method. The junction temperature of the chip inside the device or module is extracted. After a series of mathematical changes, a time constant spectrum can be obtained through the deconvolution method. The time constant spectrum reflects the changes in the transfer of heat inside the device or module. Finally, a curve composed of cumulative thermal resistance and cumulative heat capacity can be obtained according to the calculation of the time constant spectrum. By analyzing this curve, the thermal characteristics of the internal structure of the device or module can be obtained. The time constant spectrum is an important factor in determining whether the method is reasonable or not. The widely used deconvolution methods at this stage are Fourier deconvolution and Bayesian deconvolution. The former achieves accurate solution of the time constant spectrum by adjusting the two parameters of the internal filter function and has a certain noise resistance; the latter solves the time constant spectrum by adjusting the number of iterations, but the results obtained by this method are sensitive to noise, and the accuracy of the obtained time constant spectrum needs to be improved. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention proposes a method for evaluating the thermal characteristics of electronic devices based on a Wiener filter. The method of the present invention is based on the principle of Wiener filtering and proposes a new solution method for solving the time constant spectrum in the structure function method, so that the obtained time constant spectrum has a certain noise resistance and can accurately characterize and evaluate the thermal characteristics of the device.

[0005] To achieve the above object, the technical solution of the present invention is as follows:

[0006] 1. A method for evaluating thermal characteristics of electronic devices based on Wiener filter

[0007] The method comprises the following steps:

[0008] 1) Obtaining the junction temperature of the electronic device under test through thermal testing, and calculating the thermal impedance curve of the electronic device under test according to the junction temperature of the electronic device under test;

[0009] In the step 1), the junction temperature of the electronic device to be tested refers to the junction temperature of the electronic device to be tested at different times during the cooling stage;

[0010] In the step 1), the thermal impedance curve of the electronic device to be tested is specifically:

[0011] a(t)=(T j (0)-T j (t)) / P

[0012] Where a(t) is the thermal impedance curve of the electronic device to be tested, T j (t) is the junction temperature of the electronic device under test at time t, T j (0) is the junction temperature of the electronic device under test at the initial moment, and P is the heating power of the electronic device under test during the measurement process.

[0013] 2) performing substitution and derivation on the thermal impedance curve to obtain a thermal impedance substitution derivative curve;

[0014] The specific steps of step 2) are:

[0015] The time variable t in the thermal impedance curve a(t) is replaced by the time logarithmic variable z to obtain the time axis logarithmic thermal impedance curve a(z); the time logarithmic variable z in the time axis logarithmic thermal impedance curve a(z) is derived to obtain the thermal impedance substitution derivative curve

[0016] The time logarithmic variable z and the time variable t satisfy the following relationship:

[0017] z=lnt

[0018] Among them, z is the time logarithm variable and t is the time variable.

[0019] 3) Deconvoluting the thermal impedance substitution derivative curve to obtain a time constant spectrum of the electronic device to be tested;

[0020] The specific steps of the deconvolution are: performing Fourier transform on the thermal impedance substitution derivative curve to obtain the frequency domain form of the thermal impedance substitution derivative curve, applying the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve, then introducing the thermal characteristic characterization function, and then obtaining the time constant spectrum by inverse Fourier transform;

[0021] The step 3) comprises the following steps:

[0022] 3.1) Through the preset weight function W z (z) and the time constant spectrum R z The convolution of (z) represents the thermal impedance substitution derivative curve;

[0023] The preset weight function W z (z) and the time constant spectrum R z The thermal impedance substitution derivative curve represented by the convolution of (z) is specifically:

[0024]

[0025] In the formula, is the thermal impedance substitution derivative curve, R z (z) is the time constant spectrum, W z (z) is the weight function;

[0026] 3.2) performing centering processing on the time logarithmic variable z in the weight function, and then performing Fourier transform on the thermal impedance substitution derivative curve after the weight function is centered, to obtain the frequency domain form of the thermal impedance substitution derivative curve;

[0027] The frequency domain form of the thermal impedance substitution derivative curve is specifically:

[0028]

[0029] Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x;

[0030] Among them, the central time logarithm variable x is expressed as:

[0031] x=zz m

[0032] In the formula, z is the logarithmic variable of time, m is the mean of the logarithm of time variable;

[0033] 3.3) Applying the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve to obtain the Wiener filter-thermal impedance substitution derivative curve, the Wiener filter-thermal impedance substitution derivative curve is expressed as:

[0034]

[0035] Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x, is the Wiener filter;

[0036] The Wiener filter Specifically:

[0037]

[0038] In the formula, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the weight function The conjugated form of is the frequency domain form of the centered time logarithmic variable x, and NSR is the derivative signal The signal-to-noise ratio parameter;

[0039] In the step 3.3), a Wiener filter is introduced in the deconvolution process to weaken the interference effect of noise on the solution process;

[0040] 3.4) introducing a thermal characteristic characterization function into the Wiener filter-thermal impedance substitution derivative curve to obtain a Wiener filter-thermal impedance substitution derivative adaptive curve, and then obtaining a time constant spectrum through inverse Fourier transform;

[0041] In step 3.4), a thermal characteristic characterization function is placed in the last step of deconvolution to screen and adjust the deconvolution result according to the principle of structure function to increase the accuracy of the obtained time constant spectrum. The thermal characteristic characterization function B(t) is specifically:

[0042]

[0043] Where B(t) is the thermal characteristic function, a(∞) is the thermal impedance when time tends to infinity, and R z (t) is the time constant spectrum after frequency domain is converted to time domain;

[0044] The time constant spectrum R(t) of the electronic device to be tested is expressed as:

[0045]

[0046] Where R(t) is the time constant spectrum of the electronic device under test, and F -1 () represents the inverse Fourier transform, B(t) is the thermal characteristic characterization function, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x;

[0047] Among them, the time logarithm variable z is expressed as:

[0048] z=lnt

[0049] In the formula, z is the time logarithm variable, and t is the time variable;

[0050] Among them, the central time logarithm variable x is expressed as:

[0051] x=zz m

[0052] In the formula, z is the logarithmic variable of time, m is the median value of the logarithmic time variable.

[0053] Furthermore, in the step 3), reasonable attempts are made on the noise-to-signal ratio in the Wiener filter until the filtering effect meets the requirements. The filtering effect meets the requirements specifically as follows: the obtained time constant spectrum has obvious fluctuations, which can characterize the heat transfer to different positions during the test period. Reasonable attempts are made on the noise-to-signal ratio in the Wiener filter specifically as follows: when the extracted junction temperature curve is relatively smooth, that is, the noise component is low, the NSR can be reduced to between 0 and 1, so that the Wiener filter plays a smaller role; when the noise component is high, the NSR can be expanded according to the required time constant spectrum to amplify the filtering effect of the Wiener filter.

[0054] 4) evaluating the thermal characteristics inside the electronic device to be measured according to the time constant spectrum to obtain a thermal characteristics evaluation result of the electronic device;

[0055] The step 4) is specifically as follows: performing network model conversion on the time constant spectrum to obtain an integral structure function; differentiating the thermal capacity of the integral structure function with respect to the thermal resistance to obtain a differential structure function, and obtaining a thermal characteristic evaluation result of the electronic device according to the differential structure function.

[0056] 2. A Thermal Characteristics Evaluation System for Electronic Devices

[0057] The electronic device thermal characteristic evaluation system is applied to the electronic device thermal characteristic evaluation method.

[0058] The electronic device thermal characteristic evaluation system comprises a thermal test module, which is used to perform a thermal test on the electronic device to be tested and obtain the junction temperature of the electronic device to be tested;

[0059] The electronic device thermal characteristic evaluation system also includes a data processing module, which is used to obtain a thermal impedance curve according to the junction temperature of the electronic device to be tested, and to obtain a time constant spectrum of the electronic device to be tested according to the thermal impedance curve; the time constant spectrum is obtained by the following formula:

[0060]

[0061] Where R(t) is the time constant spectrum of the electronic device to be tested, F -1 () is the inverse Fourier transform, B(t) is the thermal characteristic characterization function, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x;

[0062] The electronic device thermal characteristic evaluation system further comprises an evaluation module, which is used to evaluate the thermal characteristics inside the electronic device to be tested according to the time constant spectrum to obtain a thermal characteristic evaluation result of the electronic device.

[0063] The thermal characteristic evaluation results of the electronic device include, but are not limited to, parameters such as the cumulative thermal resistance and cumulative thermal capacity of the electronic device.

[0064] The electronic device is mainly a device with a single heat source and a heat transfer direction that is mainly longitudinal.

[0065] The thermal testing method includes a temperature-sensitive parameter method and the like.

[0066] The beneficial effects of the present invention are:

[0067] (1) The principle of the present invention is derived from the principle of Wiener filtering, that is, to ensure that the mean square error of the derivative curve under noise interference and the derivative curve under ideal state is minimized at all times, which effectively suppresses a certain degree of noise interference in the thermal impedance curve, so that the obtained time constant spectrum more accurately reflects the thermal characteristics of the internal structure of the device.

[0068] (2) Compared with the traditional Fourier deconvolution method, the present invention makes the solution process of the time constant spectrum under different situations more flexible by flexibly adjusting the noise-to-signal ratio parameter NSR; compared with the traditional Bayesian deconvolution method, due to the existence of the Wiener filter, the present invention always ensures that the mean square error between the derivative curve with noise and the derivative curve without noise is minimized. By adjusting the NSR, a time constant spectrum with certain noise resistance can be obtained, and the thermal characteristics of the details of the device under test can be more accurately reflected. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 is the thermal impedance curve used in Example 1 of the present invention;

[0070] Figure 2 The time constant spectra obtained in Example 1 and Comparative Example 1 of the present invention; (A) is Example 1, and (B) is Comparative Example 1;

[0071] Figure 3The structure function curves obtained in Example 1 and Comparative Example 1 of the present invention; (A) is Example 1, and (B) is Comparative Example 1;

[0072] Figure 4 Derivative curves under different noise levels in Example 2 of the present invention and Comparative Example 2;

[0073] Figure 5 1 and 2. These are the time constant spectra obtained in Example 2 and Comparative Example 2 of the present invention; (A) is Example 2, and (B) is Comparative Example 2. DETAILED DESCRIPTION

[0074] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0075] The specific embodiments of the present invention are as follows:

[0076] Example 1

[0077] This embodiment includes the following steps:

[0078] In the present invention, the electronic device to be tested includes a device having only a single heat source and a heat transfer direction that is mainly longitudinal. In this embodiment, MOSFET is used as the electronic device to be tested. In this embodiment, the thermal impedance data ( Figure 1 ) as the thermal impedance curve.

[0079] 1) Through thermal testing, the junction temperature of the electronic device under test that changes with time during the cooling stage is obtained, and the thermal impedance curve is obtained according to the junction temperature of the electronic device under test. The thermal impedance curve of the electronic device under test is specifically:

[0080] a(t)=(T j (0)-T j (t)) / P

[0081] Where a(t) is the thermal impedance curve of the electronic device to be tested, T j is the junction temperature of the electronic device under test, T j (t) is the junction temperature curve of the electronic device under test, T j (0) is the junction temperature of the electronic device under test at the initial moment, and P is the heating power of the electronic device under test during the measurement process.

[0082] 2) Perform substitution processing and derivation on the thermal impedance curve to obtain the thermal impedance substitution derivative curve. The specific steps are: replace the time variable t in the thermal impedance curve a(t) with the time logarithmic variable z to obtain the time axis logarithmic thermal impedance curve a(z); derive the time logarithmic variable z in the time axis logarithmic thermal impedance curve a(z) to obtain the thermal impedance substitution derivative curve;

[0083] The time logarithmic variable z and the time variable t satisfy the following relationship:

[0084] z=lnt

[0085] Among them, z is the time logarithm variable and t is the time variable.

[0086] Step 2) specifically includes the following steps:

[0087] 2.1) Use RC network (thermal resistance and heat capacitance network) to characterize the thermal impedance of the device under test. The front of each thermal resistor R is connected to its corresponding heat capacitor C. One end of the heat capacitor C is connected to the thermal resistor R, and the other end is grounded. Therefore, in order to simply characterize the overall thermal impedance of the electronic device under test, the thermal impedance curve of the electronic device under test under unit response is calculated using the Laplace form:

[0088]

[0089] Where Z(s) is the thermal impedance curve of the electronic device under test under unit response, n is the order of the RC network, R i is the i-th order thermal resistance, s is the complex variable in Laplace transform, τ i It is the product of the thermal resistance and thermal capacity of the i-th order, where i is the current order.

[0090] Among them, τ i Obtained by the following formula:

[0091] τ i =R i ·C i

[0092] In the formula, τ i is the product of the i-th order thermal resistance and heat capacity, R i is the i-th order thermal resistance, C i is the i-th order heat capacity.

[0093] The thermal impedance curve Z(s) of the electronic device under test under unit response is converted back to the discrete form Z(t) in the time domain. The expression of the thermal impedance curve Z(t) of the electronic device under test in the time domain is as follows:

[0094]

[0095] Where Z(t) is the thermal impedance curve of the electronic device under test in the time domain, n is the order of the RC thermal network, R i is the i-th order thermal resistance, s is the complex variable in Laplace transform, τ i is the product of the thermal resistance and heat capacity of the i-th order, i is the current order

[0096] The thermal impedance curve Z(t) of the electronic device to be tested in the time domain is replaced by variables z=lnt and ξ=lnτ to obtain a logarithmic thermal impedance curve a(z) with a time axis, and the logarithmic thermal impedance curve a(z) with respect to the variable z is derived to obtain a thermal impedance replacement derivative curve. Thermal impedance substitution derivative curve It is expressed as:

[0097]

[0098] Where z is the logarithmic time variable and ζ is the logarithmic form of τ.

[0099] In step 3.1), the preset weight function is:

[0100] W z (z) = e^[ze^(z)]

[0101] Where z is the logarithmic variable of time.

[0102] Substitute the preset weight function into the thermal impedance substitution derivative curve In the expression, we get:

[0103]

[0104] It can be seen that the thermal impedance substitution derivative curve Can be based on a preset weight function W z The convolution representation of (z) and the time constant spectrum R(z) is:

[0105]

[0106] In the formula, is the thermal impedance substitution derivative curve, R z (z) is the time constant spectrum, W z (z) is the weight function.

[0107] 3) Deconvolute the thermal impedance substitution derivative curve to obtain the time constant spectrum of the electronic device to be tested.

[0108] The specific process of deconvolution is: perform Fourier transform on the thermal impedance substitution derivative curve to obtain the frequency domain form of the thermal impedance substitution derivative curve, apply the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve, then introduce the thermal characteristic characterization function, and then convert the frequency domain form after the joint processing of the Wiener filter and the thermal characteristic characterization function back to the time domain through inverse Fourier transform to obtain the time constant spectrum. Specifically, it includes the following steps:

[0109] 3.1) Through the preset weight function W z (z) and the time constant spectrum Rz The convolution of (z) represents the thermal impedance substitution derivative curve;

[0110] Through the preset weight function W z (z) and time constant spectrum R z The thermal impedance substitution derivative curve represented by the convolution of (z) is specifically:

[0111]

[0112] In the formula, is the thermal impedance substitution derivative curve, R z (z) is the time constant spectrum, W z (z) is the weight function;

[0113] 3.2) Centralizing the time logarithmic variable z in the weight function, and then performing Fourier transform on the thermal impedance substitution derivative curve after the weight function is centered to obtain the frequency domain form of the thermal impedance substitution derivative curve;

[0114] The frequency domain form of the thermal impedance substitution derivative curve is as follows:

[0115]

[0116] Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x;

[0117] Among them, the central time logarithm variable x is expressed as:

[0118] x=zz m

[0119] In the formula, z is the logarithmic variable of time, m is the mean of the logarithm of time variable;

[0120] 3.3) Apply the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve to obtain the Wiener filter-thermal impedance substitution derivative curve. The Wiener filter-thermal impedance substitution derivative curve is expressed as:

[0121]

[0122] Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x, is the Wiener filter;

[0123] Among them, the Wiener filter Specifically:

[0124]

[0125] In the formula, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the weight function The conjugated form of is the frequency domain form of the centered time logarithmic variable x, and NSR is the thermal impedance substitution derivative curve The signal-to-noise ratio parameter.

[0126] 3.4) Introducing a thermal characteristic characterization function into the Wiener filter-thermal impedance substitution derivative curve to obtain a Wiener filter-thermal impedance substitution derivative adaptive curve, and then obtaining a time constant spectrum through inverse Fourier transform;

[0127] The time constant spectrum R(t) of the electronic device to be tested is expressed as:

[0128]

[0129] Where R(t) is the time constant spectrum of the electronic device under test related to the time variable t, and F -1 () is the inverse Fourier transform, B(t) is the thermal characteristic characterization function, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x;

[0130] Among them, the thermal characteristic characterization function B(t) is specifically:

[0131]

[0132] Where B(t) is the thermal characteristic function, a(∞) is the thermal impedance when time tends to infinity, and R z (t) is the time constant spectrum after the frequency domain is converted to the time domain.

[0133] Among them, the time constant spectrum after frequency domain conversion to time domain can be expressed as:

[0134]

[0135] In the formula, F-1 () is the inverse Fourier transform, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x;

[0136] In step 3.2), the time logarithmic variable z in the weight function is centered, that is, the centralized time logarithmic variable x replaces the time logarithmic variable z in the weight function, so that the obtained time constant spectrum can accurately describe the thermal characteristics of the device under test from the inside to the outside.

[0137] In step 3.3), a Wiener filter is introduced in the deconvolution process to reduce the interference effect of noise on the solution process. Further, reasonable attempts are made on the modulation parameters in the Wiener filter until the filtering effect meets the requirements.

[0138] In the present invention, the optimization goal is to minimize the mean square error between the time constant spectrum when carrying noise and the time constant spectrum in the ideal state, and the number of parameters in the Wiener filter is the optimization goal. When the signal-to-noise ratio is used as the only parameter, the filtering performance of the Wiener filter reaches the best.

[0139] In step 3.4), a thermal characteristic characterization function is introduced in the last step of deconvolution, and the deconvolution result is screened and adjusted according to the principle of structure function to increase the accuracy of the obtained time constant spectrum.

[0140] 4) Evaluate the thermal characteristics inside the electronic device to be tested based on the time constant spectrum to obtain the thermal characteristics evaluation results of the electronic device.

[0141] Step 4) is specifically as follows: convert the time constant spectrum into a network model to obtain an integral structure function; differentiate the thermal capacity of the integral structure function with respect to the thermal resistance to obtain a differential structure function, and obtain the thermal characteristic evaluation result of the electronic device according to the differential structure function.

[0142] The thermal characteristics evaluation results of electronic devices include but are not limited to parameters such as cumulative thermal resistance and cumulative thermal capacity.

[0143] In this embodiment, the time constant spectrum obtained in step 3) is as follows Figure 2 As shown in (A), the integral structure function curve obtained in step 4) is as follows Figure 3 As shown in (A).

[0144] Example 2

[0145] In this embodiment, different degrees of Gaussian white noise are added to the same derivative curve to obtain the following Figure 4Multiple derivative curves with different noises as shown. In this embodiment, each derivative curve is processed according to the same procedures as steps 3) and 4) in Embodiment 1 to obtain the time constant spectrum corresponding to each derivative curve.

[0146] The time constant spectra corresponding to the derivative curves obtained in this embodiment are shown in Figure 5 (A).

[0147] Comparative Example 1

[0148] This comparative example is used to verify the accuracy of the method of the present invention. This comparative example is the time constant spectrum and the structure function curve exported from the authoritative software TDIM-Master. The time constant spectrum of this comparative example is as shown in Figure 2 (B), and the structure function curve is as shown in Figure 3 (B).

[0149] By comparing the time constant spectra and the structure function curves obtained in Embodiment 1 and Comparative Example 1, it can be seen that both the time constant spectrum obtained by the present invention and the structure function curve obtained from the time constant spectrum have a high similarity with the curves provided in the software, which fully shows that the results obtained by the present invention have the potential for wide application and verifies the accuracy of the present invention.

[0150] Comparative Example 2

[0151] This comparative example is used to verify the noise resistance of the method of the present invention. In this comparative example, the Bayesian deconvolution method is used to solve the time constant spectrum under different noises, that is, the Bayesian deconvolution method is used to process each derivative curve obtained in Embodiment 2 to obtain the time constant spectrum corresponding to each derivative curve.

[0152] The time constant spectra corresponding to the derivative curves obtained in this comparative example are shown in Figure 5 (B).

[0153] Generally, the comparison range of signals is as follows:

[0154] SNR < 10 dB: The quality of the signal is extremely poor;

[0155] 10 < SNR < 20 dB: The quality of the signal is poor;

[0156] 20 < SNR < 30 dB: The quality of the signal is good;

[0157] SNR > 30 dB: The quality of the signal becomes better and better as the decibel increases.

[0158] By comparing the results of Example 2 and Comparative Example 2, it can be found that when the derivative signal is 20 dB or above, the signal quality of the time constant spectra obtained by the two methods is almost the same, both in a good range; when the derivative signal is only 10 dB, the noise influence has a slight interference on the signal quality of the time constant spectrum obtained by Example 2 of the present invention, while the quality obtained by the Bayesian deconvolution method is very seriously interfered, as shown in Figure 5 The contrast effect is very obvious. Therefore, compared with the traditional Bayesian deconvolution method, the present invention has better anti-noise performance. When the thermal impedance derivative is seriously disturbed by noise, the method of the present invention has a better effect on solving the time constant spectrum.

[0159] The above is the conceptual principle and experimental effect of the present invention. By comparing with authoritative data, the credibility of the present invention is increased. Many modifications can be made to the exemplary embodiments, and functions to implement different functions can be added before and after the solution process, as long as they do not deviate from the spirit and scope of the present invention as defined by the attached claims.

[0160] The above specific implementation modes are used to explain the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for evaluating thermal characteristics of electronic devices based on Wiener filter, characterized in that: The method comprises the following steps: 1) Obtaining the junction temperature of the electronic device under test through thermal testing, and calculating the thermal impedance curve of the electronic device under test according to the junction temperature of the electronic device under test; 2) performing substitution and derivation on the thermal impedance curve to obtain a thermal impedance substitution derivative curve; 3) Deconvoluting the thermal impedance substitution derivative curve to obtain a time constant spectrum of the electronic device to be tested; The specific steps of the deconvolution are: performing Fourier transform on the thermal impedance substitution derivative curve to obtain the frequency domain form of the thermal impedance substitution derivative curve, applying the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve, then introducing the thermal characteristic characterization function, and then obtaining the time constant spectrum by inverse Fourier transform; 4) Evaluating the thermal characteristics inside the electronic device to be tested according to the time constant spectrum to obtain a thermal characteristics evaluation result of the electronic device.

2. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 1, characterized in that: In step 3), the time constant spectrum R(t) of the electronic device to be tested is expressed as: Where R(t) is the time constant spectrum of the electronic device to be tested, F -1 () represents the inverse Fourier transform, B(t) is the thermal characteristic characterization function, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x; The time logarithm variable z is expressed as: z=lnt In the formula, z is the time logarithm variable, and t is the time variable; Among them, the central time logarithm variable x is expressed as: x=z-z m In the formula, z is the logarithmic variable of time, m is the median value of the logarithmic time variable.

3. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 1, characterized in that: In the step 1), the thermal impedance curve of the electronic device to be tested is specifically: a(t)=(T j (0)-T j (t)) / P Where a(t) is the thermal impedance curve of the electronic device to be tested, T j (t) is the junction temperature of the electronic device under test at time t, T j (0) is the junction temperature of the electronic device under test at the initial moment, and P is the heating power of the electronic device under test during the measurement process.

4. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 1, characterized in that: The specific steps of step 2) are: The time variable t in the thermal impedance curve a(t) is replaced by the time logarithmic variable z to obtain the time axis logarithmic thermal impedance curve a(z); the time logarithmic variable z in the time axis logarithmic thermal impedance curve s(z) is derived to obtain the thermal impedance substitution derivative curve The time logarithmic variable z and the time variable t satisfy the following relationship: z=lnt Among them, z is the time logarithm variable and t is the time variable.

5. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 1, characterized in that: The step 3) comprises the following steps: 3.1) Through the preset weight function W z (z) and time constant spectrum R z The convolution of (z) represents the thermal impedance substitution derivative curve; The preset weight function W z (z) and time constant spectrum R z The thermal impedance substitution derivative curve represented by the convolution of (z) is specifically: In the formula, is the thermal impedance substitution derivative curve, R z (z) is the time constant spectrum, W z (z) is the weight function; 3.2) performing a centering process on the weight function, and then performing a Fourier transform on the thermal impedance substitution derivative curve after the centering process of the weight function, to obtain a frequency domain form of the thermal impedance substitution derivative curve; The frequency domain form of the thermal impedance substitution derivative curve is specifically: Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x; Among them, the central time logarithm variable x is expressed as: x=z-z m In the formula, z is the logarithmic variable of time, m is the mean of the logarithm of time variable; 3.3) Applying the Wiener filter to the frequency domain form of the thermal impedance substitution derivative curve to obtain the Wiener filter-thermal impedance substitution derivative curve, the Wiener filter-thermal impedance substitution derivative curve is expressed as: Where D(f) is the frequency domain form of the thermal impedance substitution derivative curve, R z (f) is the frequency domain form of the time constant spectrum, is the frequency domain form of the weight function; f is the frequency domain form of the time logarithmic variable z, is the frequency domain form of the centered time logarithmic variable x, is the Wiener filter; 3.4) Introducing a thermal characteristic characterization function into the Wiener filter-thermal impedance substitution derivative curve to obtain a Wiener filter-thermal impedance substitution derivative adaptive curve, and then obtaining a time constant spectrum through inverse Fourier transformation.

6. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 2, characterized in that: The Wiener filter Specifically: In the formula, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the weight function The conjugated form of is the frequency domain form of the centered time logarithmic variable x, and NSR is the noise-to-signal ratio parameter.

7. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 2 or 5, characterized in that: The thermal characteristic characterization function B(t) is specifically: Where B(t) is the thermal characteristic function, a(∞) is the thermal impedance when time tends to infinity, and R z (t) is the time constant spectrum after the frequency domain is converted to the time domain.

8. The method for evaluating thermal characteristics of electronic devices based on Wiener filter according to claim 1, characterized in that: The step 4) is specifically as follows: performing network model conversion on the time constant spectrum to obtain an integral structure function; differentiating the thermal capacity of the integral structure function with respect to the thermal resistance to obtain a differential structure function, and obtaining a thermal characteristic evaluation result of the electronic device according to the differential structure function.

9. An electronic device thermal characteristics evaluation system applied to the electronic device thermal characteristics evaluation method according to any one of claims 1 to 8, characterized in that: include: A thermal test module, used to perform thermal testing on the electronic device to be tested and obtain the junction temperature of the electronic device to be tested; The data processing module is used to obtain a thermal impedance curve according to the junction temperature of the electronic device to be tested, and to obtain a time constant spectrum of the electronic device to be tested according to the thermal impedance curve; the time constant spectrum is obtained by the following formula: Where R(t) is the time constant spectrum of the electronic device to be tested, F -1 () is the inverse Fourier transform, B(t) is the thermal characteristic characterization function, t is the time variable, D(f) is the frequency domain form of the thermal impedance substitution derivative curve, f is the frequency domain form of the time logarithmic variable z, is the Wiener filter, is the frequency domain form of the weight function, is the frequency domain form of the centered time logarithmic variable x; The evaluation module is used to evaluate the thermal characteristics inside the electronic device to be tested according to the time constant spectrum to obtain the evaluation result of the thermal characteristics of the electronic device.

Images