Probabilistic physics-based fast life prediction method for voltage regulator diodes

Abstract

The application discloses a kind of based on probability physics's zener diode service life fast prediction method, utilizes probability physics method according to accelerated degradation test and accelerated life modeling fast prediction power supply product in zener diode service life.Product is placed in higher than rated working stress level environment and is tested, then modeling analysis is carried out to test data, extrapolation is carried out by suitable acceleration model and the life distribution of product under rated working stress is obtained.The degradation track of product performance will change with stress, the form of degradation track is related to stress parameter, after the expression of degradation track is determined under high stress, the degradation track of product performance under rated working stress level can be obtained by acceleration model, and life prediction is carried out by reliability function.

Description

Technical Field

[0001] This invention relates to the field of reliability technology, and more specifically to a method for rapidly predicting the lifespan of Zener diodes based on probabilistic physics. Background Technology

[0002] A Zener diode, also known as a voltage regulator diode, is a surface-contact crystal diode made of silicon. These diodes have a high impurity concentration, a large charge density in their space charge region, and a narrow charge region, making them prone to generating a strong electric field. They are semiconductor devices that maintain high resistance until the critical reverse breakdown voltage. During reverse breakdown, within a certain current range (or a certain power loss range), the terminal voltage of a Zener diode remains almost constant; this is known as its voltage regulation characteristic. It is widely used in regulated power supplies and limiting output circuits, and its performance significantly impacts the overall circuit output. With the increasing demands for performance and reliability in power supply products, research on the reliability of Zener diodes is essential.

[0003] Currently, the lifespan of electronic components and products is continuously increasing. Traditional classical lifespan prediction methods, under large sample conditions, cannot obtain sufficient failure data within a limited time, or even lack such data altogether, making it impossible to establish an effective lifespan distribution model. Modern industrial production is characterized by "multiple varieties, small batches, and rapid production." Limited by budget and time constraints, products can only be tested on a small number of samples, resulting in very little failure data and casting doubt on the reliability of traditional lifespan prediction theories. Furthermore, most products have their own performance characteristics, which gradually degrade over time during operation or storage, eventually rendering them unusable. By starting with product performance parameters and continuously measuring certain quantities characterizing product function to obtain degradation data, the rich lifespan information provided by this degradation data can compensate for the insufficient information in lifespan prediction for long-life products.

[0004] However, the extended product lifespan results in a slower degradation rate under normal stress conditions, which means that obtaining complete performance degradation data under normal use becomes difficult, making it hard to make useful inferences within a reasonable timeframe.

[0005] Therefore, how to quickly predict the lifespan of Zener diodes is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a method for rapid prediction of the lifespan of Zener diodes based on probabilistic physics. This method analyzes the degradation process of product functions using degradation data, combines probabilistic physics to determine product lifespan characteristic parameters, establishes a degradation model, and provides estimated values ​​for the model parameters. Based on this, by setting a failure threshold standard and establishing a product lifespan distribution model, the product lifespan can be predicted. Furthermore, the degradation process of performance parameters can be accelerated by increasing stress, thereby obtaining usable degradation information in a shorter time and achieving rapid prediction of product lifespan.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A method for rapidly predicting the lifespan of a Zener diode based on probabilistic physics includes the following steps:

[0009] Accelerated testing was conducted on the Zener diode sample under test, and degradation data was collected.

[0010] A degradation distribution model is established based on degradation data, and the parameters are estimated by maximum likelihood estimation using the reliability function. The estimated values ​​of the degradation distribution model parameters of the sample at different measurement times under different test stresses are obtained. The average value of the estimated values ​​of the degradation distribution model parameters of the sample under different test stresses is linearly fitted over time to obtain the degradation model under different test stresses.

[0011] Based on the degradation model and parameter estimates under different test stresses, the lifetime acceleration model is fitted to construct the lifetime acceleration equation and obtain the fitting curve.

[0012] Obtain the operating stress to be predicted, solve the corresponding performance degradation rate using the lifetime acceleration equation, and construct the corresponding reliability function; use the lifetime acceleration equation to predict the degradation distribution model under normal operating stress.

[0013] The lifespan of the Zener diode under the predicted operating stress is obtained by combining the predicted reliability with the corresponding reliability function.

[0014] Preferably, in the accelerated test, the input voltage is used as the test stress condition, and the data of the change of the sample's stable voltage value over time under different test stresses are used as degradation data; the test is ended by setting a fixed test duration or a fixed failure parameter threshold.

[0015] Preferably, the stable voltage values ​​of each sample at the same measurement time under the same test stress conform to the Weibull distribution, and the degradation model is constructed using the Weibull-type degradation distribution model; the performance degradation distribution is the probability distribution of the performance degradation amount y being less than or greater than the failure threshold;

[0016] If the performance degradation y follows a shape parameter of m y (t), with scale parameter ηy If (t) follows a Weibull distribution, then when the product failure criterion is y≤D f At that time, D f The reliability function, which represents the relationship between reliability and the distribution of performance degradation, is shown in equation (1), where the failure threshold is used.

[0017]

[0018] When the failure criterion is y≥D f At that time, the reliability function is as shown in equation (2):

[0019]

[0020] The stabilized voltage values ​​of each sample at different measurement times under different test stresses are used as performance degradation values ​​and substituted into equation (1) or equation (2) to perform parameter maximum likelihood estimation fitting, thereby obtaining the parameter estimates of the degradation distribution model at different measurement times under different test stresses, including the estimated values ​​of scale parameters and shape parameters. These are then substituted into equations (1) and (2) as shape parameters and scale parameters, respectively, to obtain the reliability function of the degradation distribution model and evaluate the product reliability; the shape parameters (generally independent of time) and scale parameters of the product's performance degradation at time t;

[0021] The degradation model is obtained by linearly fitting the average of the scale parameter estimates. The degradation model expression is as follows:

[0022] η y (t)=Kt+b (3)

[0023] Where, η y (t) represents the scale parameter; t represents time; K represents the slope; b represents the intercept; and m represents the shape parameter. y (t) does not change over time.

[0024] Preferably, the lifetime acceleration model is an inverse power-law model, expressed as:

[0025] R = aV -b (6)

[0026] Where R is the performance degradation rate; V is the test stress; and a and b are unknown constants.

[0027] Taking the logarithm of the inverse power-law model, it is expressed as:

[0028] lnR=lna-blnV (7)

[0029] The slope K of the degradation model of the scale parameter η and the test stress are analyzed, as well as the correlation between the shape parameter m and the test stress. The slope K and the estimated value of the shape parameter are substituted into equation (7) as the performance degradation rate to obtain the acceleration equation and the fitting curve. The life acceleration equation includes the scale parameter acceleration equation and the shape parameter acceleration equation. K is used as the degradation rate parameter. Using the K value data under different stresses V, the a and b values ​​of the scale parameter acceleration equation are obtained by fitting the equation (7) (equations (6) and (7) are equivalent). Similarly, since the shape parameter m y The value of (t) does not change with time under a specific stress, so m is directly selected as the performance degradation rate for the shape parameter m. y The acceleration equation for (t) is fitted and solved.

[0030] Preferably, the obtained working stress to be predicted is substituted into the lifetime acceleration equation to obtain the corresponding performance degradation rate, including the new slope and shape parameters of the degradation model; the shape parameters are used as parameters of the degradation distribution model.

[0031] The new slope is substituted into the degradation model to obtain the scale parameter;

[0032] Substitute the shape parameters and scale parameters into formulas (1) and (2) to obtain the corresponding reliability functions;

[0033] The service life under the predicted working stress is obtained by predicting the reliability and the corresponding reliability function; a predicted reliability is input into the corresponding reliability function to obtain a corresponding time t, which is the service life.

[0034] Equation (6) is the lifetime acceleration model (for the scale parameter η). y (t), shape parameter m y (t) does not change with time and is not considered. When a specific experimental stress V is known, the K value and shape parameter m can be obtained according to the acceleration equation. y The scale parameter at this time can be obtained from the K value according to the degradation model formula (3) of the scale parameter. The scale parameter function η is obtained from the scale parameter. y The reliability function corresponding to the degradation distribution model can be obtained from the shape parameter value (t), and the corresponding service life can be predicted according to the reliability requirements.

[0035] As can be seen from the above technical solution, compared with the prior art, this invention discloses a method for rapid prediction of the lifespan of Zener diodes based on probabilistic physics. It utilizes probabilistic physics methods, based on accelerated degradation testing and accelerated lifespan modeling, to rapidly predict the lifespan of Zener diodes in power supply products. The product is placed in an environment with a higher stress level than its rated operating stress for testing. The test data is then modeled and analyzed, and the lifespan distribution of the product under rated operating stress is obtained through extrapolation using a suitable accelerated lifespan model. Different stresses result in different degradation trajectories for the product's performance. The form of the degradation trajectory is related to the stress parameters. After determining the expression for the degradation trajectory under high stress, the degradation trajectory of the product's performance at the rated operating stress level can be obtained through the accelerated lifespan model. Lifespan prediction can be initiated based on the reliability function of the degradation distribution model. The scale parameter η, the slope K of the degradation model, and the shape parameter m under a specific stress are obtained by fitting and estimating the degradation data. Then, the functional relationship between the slope K and the shape parameter m and the voltage stress V is obtained through data fitting and estimation, i.e., the accelerated equation. Thus, the reliability function of the Zener diode under different voltage stresses can be determined, and its lifespan can be determined according to the reliability requirements. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0037] Figure 1 The attached figure is a flowchart of the method for rapid prediction of the lifespan of Zener diodes based on probabilistic physics provided by the present invention.

[0038] Figure 2 The attached figure is a schematic diagram of the accelerated testing circuit provided by the present invention;

[0039] Figure 3 The attached figure is a schematic diagram of the change of η value under 20V input voltage stress provided by the present invention;

[0040] Figure 4 The attached figure is a schematic diagram of the change of η value under 25V input voltage stress provided by the present invention;

[0041] Figure 5 The attached figure is a schematic diagram of the change of η value under 30V input voltage stress provided by the present invention;

[0042] Figure 6 The attached figure is a schematic diagram of the change of η value under 35V input voltage stress provided by the present invention;

[0043] Figure 7 The attached figure is a schematic diagram of the change of η value under 40V input voltage stress provided by the present invention;

[0044] Figure 8 The attached figure is a schematic diagram showing the relationship between the slope K and the input voltage V of the η-degeneracy model provided by this invention;

[0045] Figure 9 The attached figure is a schematic diagram showing the relationship between the value of m and the input voltage provided by the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] This invention discloses a method for rapidly predicting the lifespan of a Zener diode based on probabilistic physics. An accelerated degradation test is designed for predicting the lifespan of a Zener diode in actual use scenarios. A performance degradation distribution model, a degradation model, and an accelerated lifespan model are established. Parameter estimation is performed using experimental data. The relationship between different voltage stress conditions and the reliability function of the degradation distribution model of the Zener diode is obtained through probabilistic physics methods. The lifespan is predicted according to different reliability requirements, enabling rapid prediction of lifespan. Simultaneously, the test time and sample size can be determined based on correlation requirements and time and budget constraints. The specific process is as follows:

[0048] S1: Perform accelerated testing on the Zener diode sample under test and collect degradation data;

[0049] In accelerated testing, the input voltage is used as the test stress condition, and the data of the change of the sample's stable voltage value over time under different test stresses are used as degradation data; the test is ended by setting a fixed test duration or a fixed failure parameter threshold.

[0050] S2: Based on the degradation data, a degradation distribution model is established and the parameters are estimated by maximum likelihood estimation to obtain the estimated values ​​of scale parameters and shape parameters of the sample at different measurement times under different test stresses; the estimated values ​​of scale parameters and shape parameters of the sample under different test stresses are linearly fitted over time to obtain the degradation model; the degradation distribution model adopts the Weibull type degradation distribution model;

[0051] S21: Based on the degradation data, establish a degradation distribution model and perform maximum likelihood estimation of the parameters to obtain the estimated values ​​of the scale parameters and shape parameters of the sample at different times under different experimental stresses;

[0052] S22: The estimated scale parameters of the sample at different times under each experimental stress are linearly fitted to obtain the degradation model; the expression of the degradation model is:

[0053] η y (t)=Kt+b (3)

[0054] Where, η y (t) represents the scale parameter; t represents time; K represents the slope; b represents the intercept;

[0055] The scale parameter is obtained by modeling and solving it as a function of time; commonly used models for scale parameter variation over time include linear models, logarithmic models, and exponential models.

[0056] Linear model:

[0057] η y (t)=Kt+b (3)

[0058] Logarithmic model:

[0059] η y (t)=Klnt+b (4)

[0060] Exponential model:

[0061] η y (t)=be Kt (5)

[0062] The parameter estimates are linearly fitted over time to determine that the scale parameter η satisfies a linear relationship, and the slope K and intercept b of the linear equation are solved; the shape parameter m does not change with time.

[0063] S3: The effect of electrical stress on product degradation rate is described using an inverse power law model, expressed as follows:

[0064] R = aV -b

[0065] Where R is the performance degradation rate; V is the electrical stress; and a and b are unknown constants.

[0066] Taking the logarithm of the inverse power-law model is

[0067] lnR=lna-blnV (7)

[0068] The slope K and shape parameter of the degradation model of the scale parameter η are selected as degradation rate parameters. The correlation between the slope K of the degradation model of the scale parameter η and the test stress, and the correlation between the shape parameter m and the test stress are analyzed. They are substituted into equation (7) to obtain the acceleration equation and the fitting curve, thereby predicting the degradation rate of performance parameters under normal working stress.

[0069] S4: Based on the predicted degradation rate data, a predicted degradation model is derived, and the reliability function of the Zener diode is obtained. Under the given reliability requirements, the predicted service life is calculated by combining the reliability function.

[0070] Example

[0071] In one embodiment, the process includes three steps: accelerated testing, degradation modeling, and lifetime acceleration model construction.

[0072] 1. Accelerated testing

[0073] Design as Figure 2 The test circuit shown has a power supply U as the test voltage source with adjustable output voltage; resistor R1 is a current-limiting resistor; and D1 is the Zener diode under test. The circuit component parameters are shown in Table 1.

[0074] Table 1 Component Parameter Table

[0075] Adjustable power supply Voltage adjustable from 0-50V, maximum output current 5A resistance Resistance 1KΩ Zener diode The voltage regulation value is 7.5V, and the failure threshold is 8.5V.

[0076] The test stress is the power supply output voltage U, and the characteristic quantity is the diode voltage regulation value U0. U0 is continuously measured, and the test voltage and test time of each sample are shown in Table 2 below.

[0077] Table 2. Test voltage and test time for each sample.

[0078] #1～#30 20 180 #31～#60 25 180 #61～#90 30 180 #91～#120 35 180 #121～#150 40 180

[0079] 2. Degeneracy Modeling

[0080] The Weibull distribution parameter estimation results of the stabilized voltage values ​​at each measurement time under various voltage stresses are shown in Table 3.

[0081] Table 3. Estimation results of Weibull distribution parameters for Zener diode voltage regulation value.

[0082]

[0083]

[0084] As shown in Table 3, the scale parameter η tends to increase over time, while the shape parameter m is not significantly correlated with time. Figure 3-7 The graphs show the variation of the scale parameter η over time when the input voltage stress is 20V, 25V, 30V, 35V, and 40V, respectively. The horizontal axis represents time, and the vertical axis represents the scale parameter value.

[0085] Analysis shows that a linear model is more suitable for the degradation model of η under each stress level, and the mean of the estimated values ​​of m at all times is taken. The degradation model of η and the value of m under each stress level are shown in Table 4.

[0086] Table 4. Degeneracy model of η and m value at various stress levels.

[0087] 20 <![CDATA[η1(t)=6.43×10 -4 t+2.52]]> 5.75 25 <![CDATA[η2(t)=14×10 -4 t+2.52]]> 4.91 30 <![CDATA[η3(t)=12×10 -4 t+2.52]]> 4.60 35 <![CDATA[η4(t)=26×10 -4 t+2.52]]> 4.44 40 <![CDATA[η5(t)=40×10 -4 t+2.52]]> 4.15

[0088] 3. Lifetime Acceleration Model Construction

[0089] As shown in Table 4, the slope K and parameter m of the Weibull parameter η degradation model are strongly correlated with the electrical stress. Substituting them into equation (7) yields the acceleration equation and correlation coefficients, as shown in Table 5. The fitting curve is as follows: Figure 8-9 As shown, Figure 8 The curve is the fitted curve of the slope K of the size parameter η degradation model versus the input voltage V, with lnV on the x-axis and lnK on the y-axis. Figure 9 This is a fitted curve showing the relationship between the shape parameter m and the input voltage, with the horizontal axis representing lnV and the vertical axis representing 1nm.

[0090] Table 5. Slope and parameter m values ​​of the Weibull parameter η degeneracy model, and acceleration equations.

[0091]

[0092] 4. Service life prediction

[0093] If the failure threshold of the regulated voltage is set at 8.5V, then the lifetime prediction is as follows:

[0094] (1) 12V input voltage life assessment

[0095] When the input voltage is 12V, η y (t)=2.32×10 -4 t+2.52, m y (t) = 7.07. Substituting this into the formula, we obtain the reliability life assessment results with reliability of 0.99, 0.95, 0.9, 0.8, and 0.5, as shown in Table 6.

[0096] Table 6. Reliability lifetime assessment results of Zener diodes at 12V.

[0097] 0.99 <![CDATA[1.01×10 5 ]]> 0.95 <![CDATA[1.10×10 5 ]]> 0.9 <![CDATA[1.16×10 5 ]]> 0.8 <![CDATA[1.26×10 5 ]]> 0.5 <![CDATA[1.49×10 5 ]]>

[0098] (2) 15V input voltage life assessment

[0099] When the input voltage is 15V, η y (t) = 3.2 × 10 -4 t+2.52, m y (t) = 6.41. Substituting this into the formula, we obtain the reliability life assessment results with reliability of 0.99, 0.95, 0.9, 0.8, and 0.5, as shown in Table 7.

[0100] Table 7. Reliability lifetime assessment results of Zener diodes at 15V.

[0101]

[0102]

[0103] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0104] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for rapidly predicting the lifespan of a Zener diode based on probabilistic physics, characterized in that, Includes the following steps: Accelerated testing was conducted on the Zener diode sample under test, and degradation data was collected. A degradation distribution model is established based on degradation data, and the parameters are estimated by maximum likelihood estimation using the reliability function. The estimated values ​​of the degradation distribution model parameters, including the estimated values ​​of scale parameters and shape parameters, are obtained at different measurement times under different test stresses. The average values ​​of the estimated values ​​of the degradation distribution model parameters of the samples under different test stresses are linearly fitted over time to obtain the degradation model under different test stresses. The steady-state voltage values ​​of each sample at the same measurement time under the same test stress conform to the Weibull distribution. A degradation model is constructed using a Weibull-type degradation distribution model. If the performance degradation y follows a shape parameter of m y (t), with scale parameter η y If (t) follows a Weibull distribution, then when the product failure criterion is y≤D f At that time, D f The failure threshold is represented by the reliability function, which expresses the relationship between reliability and the distribution of performance degradation, as shown in equation (1). ; When the failure criterion is y > D f When the reliability function is as shown in equation (2), （2）; The stable voltage values ​​of each sample at different measurement times under different test stresses are used as performance degradation values ​​and substituted into equation (1) or equation (2) for parameter maximum likelihood estimation fitting to obtain the parameter estimates of the degradation distribution model at different measurement times under different test stresses. The degradation model is obtained by linearly fitting the average of the scale parameter estimates. The degradation model expression is as follows: η y (t)=Kt+b (3) Where, η y (t) represents the scale parameter; t represents time; K represents the slope; b represents the intercept; Based on the degradation model and parameter estimates under different test stresses, the lifetime acceleration model is fitted to construct the lifetime acceleration equation and obtain the fitting curve. The lifetime acceleration model is an inverse power-law model, and its expression is: R=aV -b (6) Where R is the performance degradation rate; V is the test stress; and a and b are unknown constants. Taking the logarithm of the inverse power-law model, it is expressed as: lnR=lna - blnV (7) Substitute the slope and shape parameter estimates of the degradation model into equation (7) as the performance degradation rate to obtain the lifetime acceleration equation and the fitting curve; Obtain the operating stress to be predicted, use the lifetime acceleration equation to solve for the corresponding performance degradation rate, and construct the corresponding reliability function; The lifespan of the Zener diode under the predicted operating stress is obtained by combining the predicted reliability with the corresponding reliability function. Substitute the obtained working stress to be predicted into the lifetime acceleration equation to obtain the corresponding performance degradation rate, including the new slope and shape parameters of the degradation model; The scale parameters are calculated based on the new slope and degradation model; Substitute the shape parameters and scale parameters into formulas (1) and (2) to obtain the corresponding reliability functions; The service life under the predicted working stress is obtained based on the reliability to be predicted and the corresponding reliability function.

2. The method for rapid prediction of Zener diode lifespan based on probabilistic physics according to claim 1, characterized in that, In accelerated testing, the input voltage is used as the test stress condition, and the data of the change of the sample's stable voltage value over time under different test stresses are used as degradation data; the test is ended by setting a fixed test duration or a fixed failure parameter threshold.

Images