阅读说明：本技术 建立功率半导体装置中的电连接的劣化状态的诊断装置和方法 (Diagnostic device and method for establishing a degraded state of an electrical connection in a power semiconductor device ) 是由 N·德格雷纳 J·C·布兰德勒罗 于 2018-10-30 设计创作，主要内容包括：一种建立功率半导体装置(1)中的电连接的劣化状态的方法包括：-针对相同的温度值在两个相应电流值下测量至少两个电压降值。两个电流值严格不同,或者在晶体管的两个不同的栅极电平(G)下进行测量；-保存所测量的值作为校准数据；-监测所述功率半导体装置的操作条件；-在相应与前述相同的电流值下并且在所监测的操作条件对应于与操作状态和公共温度有关的两个相应预定义的标准集合的两个相应时刻测量至少两个电压降值；-保存至少两个值作为操作数据；-以估计所述功率半导体装置(1)的劣化状态的方式计算数值指标。(A method of establishing a degraded state of an electrical connection in a power semiconductor device (1) comprises: -measuring at least two voltage drop values at two respective current values for the same temperature value. The two current values are strictly different or are measured at two different gate levels (G) of the transistor; -saving the measured values as calibration data; -monitoring an operating condition of the power semiconductor device; -measuring at least two voltage drop values at respective same current values as previously and at two respective instants at which the monitored operating conditions correspond to two respective predefined sets of criteria relating to the operating state and the common temperature; -saving at least two values as operation data; -calculating a numerical indicator in a manner that estimates the state of degradation of the power semiconductor device (1).)

1. A method of establishing a degraded state of an electrical connection in a power semiconductor device, the method comprising the steps of:

a) when the power semiconductor device is subjected to a first current value (I)_{on_1}) And a first temperature value (T)_{1_calibration}) At least a first voltage drop value (V) of the power semiconductor device is measured in a first state_{on_1_calibration})；

b) When the power semiconductor device is subjected to a second current value (I)_{on_2}) And a second temperature value (T)_{2_calibration}) In a second state of (a), at least the power semiconductor device is measuredSecond voltage drop value (V)_{on_2_calibration}) Said second temperature value (T)_{2_calibration}) Equal to said first temperature value (T)_{1_calibration}) ± 5 ℃, satisfying at least one of the following two conditions:

-said first current value (I)_{on_1}) Equal to the first threshold (X% × I)_nom) And the second current value (I)_{on_2}) Equal to the second threshold (Y% × I)_nom) The second threshold value (Y% × I)_nom) Strictly greater than the first threshold (X% × I)_nom)，

-said first state and said second state differ from each other at least with respect to a gate level (G) of a transistor or a group of transistors of said power semiconductor device;

c) saving at least two values (V)_{on_1_calibration},V_{on_2_calibration}) As calibration data;

d) monitoring a current (I) and at least one parameter (T, F) representative of the operating state of the power semiconductor device under operating conditions_mod,G)；

e) Measuring at least a third voltage drop value (V) of the power semiconductor device under the following conditions_{on_1_test})：

-at a value equal to said first current value (I)_{on_1}) Plus or minus nominal current (I) of the power semiconductor device_nom) 5% of current (I)_{1_test}) Down, and

-monitoring said at least one parameter (T, F)_modG) corresponds to an operating state of the power semiconductor device and a third temperature value (T) of the power semiconductor device_{1_test}) A time of day of a first predefined set of criteria in question;

f) measuring at least a fourth voltage drop value (V) of the power semiconductor device under the following conditions_{on_2_test})：

-at a value equal to said second current value (I)_{on_2}) + -said nominal current (I) of said power semiconductor device_nom) 5% of current (I)_{2_test}) Down, and

-monitoring said at least one parameter (T, F)_modG) corresponds toThe operating state of the power semiconductor device and the temperature value (T) of the power semiconductor device equal to the third temperature value_{1_test}) Fourth temperature value (T) of + -5 deg.C_{2_test}) The time of day of the relevant second predefined set of criteria;

g) saving at least two values (V)_{on_1_test},V_{on_2_test}As operational data;

h) according to the calibration data (V) in a manner that the degradation state of the power semiconductor device is estimated_{on_1_calibration},V_{on_2_calibration}) And said operation data (V)_{on_1_test}；V_{on_2_test}) To calculate a numerical index (Δ R).

2. The method of claim 1, wherein the power semiconductor device comprises an Insulated Gate Bipolar Transistor (IGBT), a Metal Oxide Semiconductor Field Effect Transistor (MOSFET), a diode component, or a combination of these elements.

3. Method according to one of the preceding claims, wherein a voltage drop value (V) of the power semiconductor device is measured_{on_1_test},V_{on_2_test}) Is carried out continuously or at a predetermined frequency, not only at the at least one parameter (F) monitored_modG) is carried out at a time corresponding to a predefined operating state of the power semiconductor device.

4. Method according to one of the preceding claims, wherein the second threshold value (Y% × I)_nom) Greater than or equal to the first threshold (X% × I)_nom) Twice as much.

5. Method according to one of the preceding claims, wherein the numerical indicator (Δ R) is calculated by extrapolation according to the following assumptions: two of the calibration data (V)_{on_1_calibration},V_{on_2_calibration}) Each corresponds to a case where the numerical index has a value (Δ R ═ 0) corresponding to complete absence of deterioration of the electrical connection of the power semiconductor device.

6. Method according to one of the preceding claims, wherein the monitoring step comprises monitoring at least the frequency (F) of the current (I)_mod)。

7. The method of claim 6, wherein the first and second predefined sets of criteria comprise a frequency (F) greater than or equal to 20Hz_mod) And measuring at least said third voltage drop value and said fourth voltage drop value (V)_{on_1_test},V_{on_2_test}) Is carried out at two respective instants of a common cycle of said current (I).

8. Method according to one of the preceding claims, wherein the monitored parameter corresponds to a state of the power semiconductor device related to a gate level (G) of a transistor or a group of transistors of the power semiconductor device, the first and second predefined set of criteria comprising criteria different from each other for the state of the power semiconductor device related to the gate level (G).

9. The method according to one of the preceding claims,

-said first and second voltage drop values (V)_{on_1_calibration},V_{on_2_calibration}) The measuring steps a) and b) of (a) are carried out at two respective times separated from each other by at most 50 milliseconds; and/or

-said third and fourth voltage drop values (V)_{on_1_test},V_{on_2_test}) The measuring steps e) and f) of (a) are carried out at two respective times which are at most 50 milliseconds apart from each other.

10. Method according to one of the preceding claims, wherein at least the first voltage drop value (Vfirst) is measured_{on_1_calibration}) Further comprises measuring said first temperature value (T)_{1_calibration}) And/or

Wherein at least the second voltage drop value (V) is measured_{on_2_calibration}) Further comprises measuring said second temperature value (T)_{2_calibration})；

Wherein the step of storing further comprises storing the corresponding voltage drop value (V)_{on_1_calibration},V_{on_2_calibration}) Associated respective said temperature value (T)_{1_calibration},T_{2_calibration}) As calibration data, and

wherein the step of calculating a numerical indicator further comprises calculating a numerical indicator based on said calibration data (V)_{on_1_calibration},T_{1_calibration}；V_{on_2_calibration},T_{2_calibration}) And corresponding operation data (V)_{on_1_test},V_{on_2_test}) To calculate the temperature (T)_{1_test}；T_{2_test})。

11. Method according to one of claims 1 to 9, wherein the at least one parameter (F) representative of the operating state of the power semiconductor device monitored during the monitoring step_modG) the measured temperature (T) is removed; the operation data (V)_{on_1_test}；V_{on_2_test}) A step of removing the measured temperature value for calculating the numerical indicator (Δ R).

12. Method according to one of the preceding claims, further comprising determining from a current value (I) when said numerical indicator (AR) is equal to zero_{on_1},I_{on_2}) To configure the at least two voltage drops (V)_{on_1_calibration},V_{on_2_calibration}) A step of modeling the relationship between them, said modeling then being used to determine the calibration data (V) from said calibration data (V)_{on_1_calibration},V_{on_2_calibration}) And said operation data (V)_{on_1_test}；V_{on_2_test}) To calculate said numerical indicator (Δ R).

13. The method according to one of the preceding claims, further comprising the step of:

i) constructing a degradation model of the power semiconductor device by extrapolating the evolution of the numerical indicator (Δ R) in a manner that estimates the end of life of the electronic device.

14. A diagnostic device comprising at least one processor operatively associated with at least one storage medium and a set of plugs that can be plugged into a power semiconductor device, the diagnostic device being arranged to implement a method according to one of the preceding claims when plugged into the power semiconductor device.

15. A non-transitory computer-readable storage medium comprising a program product stored thereon and executable by a processor in the form of a software agent, the program product comprising at least one software module comprising instructions to implement a method according to one of claims 1 to 13.

Technical Field

The present invention relates to the technical field of power semiconductor devices. More particularly, the invention relates to a method of assessing the state of such a power semiconductor device and a diagnostic device arranged to implement such a method.

Background

Typically, during operation of power semiconductor devices, wire bonds and other interconnects (e.g., metallization and solder) are subject to thermo-mechanical stresses including high thermal variations and vibrations. Degradation of the interconnects, wire bonds and/or metallization is manifested as an increase in resistance (Δ R) in the current path. The resistance can be measured as an increase in voltage drop. In the event of a wire bond failure, the voltage drop may be gradual and/or sudden. The voltage drop may be detected by measuring a voltage external to the power electronics or directly at the interconnect. Therefore, measuring this increase in voltage can be used to estimate the health of the wire bond and/or metallization.

However, such voltages are highly dependent on operating conditions including temperature. It is difficult to distinguish/isolate the effect of operating conditions on the measured voltage from the effect of degradation.

Additionally, these power electronics are often difficult to access in use (e.g., when sealed into a non-removable sealed housing). Physical inspection of individual devices is economically unwise, particularly when many devices are equipped with entire fleets of vehicles in use (like trains) or inaccessible windmills. In general, the theoretical life cycle is arbitrarily fixed for all devices in the set. They are only replaced after a valid and unexpected failure or when any life cycle is exceeded, regardless of the true state of the device.

Some existing methods of testing devices are planned to be implemented on a test stand under stable and/or controlled conditions. These methods cannot be transferred to the power electronics in use. This means that the power electronics operation is interrupted to test it. This is not satisfactory.

Disclosure of Invention

The present invention improves this situation.

The applicant has proposed a method of establishing a degraded state of an electrical connection in a power semiconductor device. The method comprises the following steps:

a) when the power semiconductor device is subjected to a first current value (I)_{on_1}) And a first state of a first temperature value, measuring at least a first voltage drop value of the power semiconductor device;

b) measuring at least a second voltage drop value of the power semiconductor device when the power semiconductor device is in a second state subjected to a second current value and a second temperature value, the second temperature value being equal to the first temperature value ± 5 ℃, at least one of the following two conditions being satisfied:

-the first current value is equal to the first threshold value and the second current value (I)_{on_2}) Equal to a second threshold value, strictly greater than the first threshold value,

-the first state and the second state differ from each other at least with respect to the gate level of a transistor or a group of transistors of the power semiconductor device;

c) saving at least two values as calibration data;

d) monitoring the current and at least one parameter representative of the operating state of the power semiconductor device under operating conditions;

e) measuring at least a third voltage drop value of the power semiconductor device under the following conditions

-at a current equal to the first current value ± 5% of the nominal current of the power semiconductor device, and

-at a moment when the monitored at least one parameter corresponds to a first predefined set of criteria relating to an operating state of the power semiconductor device and a third temperature value of the power semiconductor device;

f) measuring at least a fourth voltage drop value of the power semiconductor device under the following conditions

-at a current equal to a second current value ± 5% of the nominal current of the power semiconductor device, and

-at a moment when the monitored at least one parameter corresponds to a second predefined set of criteria relating to the operating state of the power semiconductor device and a fourth temperature value of the power semiconductor device equal to the third temperature value ± 5 ℃;

g) saving at least two values as operation data;

h) a numerical index is calculated from calibration data and operation data in such a manner that the state of degradation of the power semiconductor device (1) is estimated.

This approach allows monitoring of a numerical health indicator (e.g., resistance increase Δ R) to estimate the state of health and remaining life of the power semiconductor device. The result of this method is independent of the operating temperature. This method can be implemented using a single voltage sensor. This method is low cost. This method does not require off-line calibration. Therefore, older power semiconductor devices may be equipped with a system according to the invention to implement the method. Due to the high influence of the used high currents and operating parameters like the operating temperature on the measured voltage, the method is particularly suitable for implementation on power semiconductor devices comprising diodes, IGBTs and/or MOSFETs.

In a second aspect of the invention, the applicant proposes a diagnostic device comprising at least one processor operatively associated with at least one storage medium and a set of plugs that can be plugged into a power semiconductor device. The diagnostic device is arranged to implement the method according to the preceding when inserted into the power semiconductor device.

In another aspect of the invention, the applicant proposes a program product executable by a processor in the form of a software agent, comprising at least one software module comprising instructions implementing a method according to the above, and a non-transitory computer-readable storage medium comprising a program product stored thereon and executable by a processor in the form of a software agent, the program product comprising at least one software module comprising instructions implementing a method according to the above.

The method and/or apparatus may include the following features (alone or in combination with one another):

the power semiconductor device comprises an Insulated Gate Bipolar Transistor (IGBT), a Metal Oxide Semiconductor Field Effect Transistor (MOSFET), a diode assembly or a combination of these elements. For these devices, the method gives particularly good results.

The step of measuring the voltage drop value of the power semiconductor device is carried out continuously or at a predetermined frequency, not only at the moment when the monitored at least one parameter corresponds to a predefined operating state of the power semiconductor device. This enables to ensure a continuous monitoring of the health evolution and a rapid detection of any abnormal situation.

-the second threshold value is greater than or equal to the first threshold value (X% × I)_nom) Twice as much. Measuring the voltage drop at two very different current values enables a more realistic model to be constructed. The accuracy of index calculation increases.

-calculating the numerical indicator by extrapolation according to the following assumptions: both calibration data correspond to the case where the numerical index has a value corresponding to no degradation at all of the electrical connection of the power semiconductor device. Thus, the calibration data measurement may be made at the beginning of the life of the device, while the operational data is later measured under the actual operating conditions of the device.

-the monitoring step comprises monitoring at least the frequency of the current. This makes it possible to ensure that the temperature is substantially the same for both voltage drop measurements even when the temperature itself cannot be measured directly.

-the first and second predefined sets of criteria comprise frequencies greater than or equal to 20 Hz. And the step of measuring at least the third voltage drop value and the fourth voltage drop value is performed at two respective instants of a common period of current. This makes it possible to ensure that the temperature is substantially the same for both voltage drop measurements even when the temperature itself cannot be measured directly due to the thermal inertia of the device.

-the monitored parameter corresponds to a state of the power semiconductor device related to a gate level of a transistor or a group of transistors of said power semiconductor device, the first and second predefined set of criteria comprising criteria different from each other for the state of the power semiconductor device related to said gate level. This makes it possible to ensure that the two states of the device are different for the two voltage drop measurements. In addition, if two measurements are taken immediately before and after the state change with respect to the gate level, it can be determined that the temperature is substantially the same for both voltage drop measurements even when the temperature itself cannot be directly measured.

-in the method, the step of applying a voltage,

-the measurement steps a) and b) of the first and second voltage drop values are carried out at two respective instants at most 50 milliseconds apart from each other; and/or

-said measuring steps e) and f) of the third and fourth voltage drop values are carried out at two respective instants at most 50 milliseconds apart from each other. This makes it possible to ensure that the temperature is substantially the same for both voltage drop measurements even when the temperature itself cannot be measured directly due to the thermal inertia of the device.

-in the method, the step of applying a voltage,

-the step of measuring at least the first voltage drop value further comprises measuring a first temperature value, and/or

-the step of measuring at least a second voltage drop value further comprises measuring a second temperature value;

-the step of saving further comprises saving each of said temperature values associated with a respective voltage drop value as calibration data, and

the step of calculating a numerical indicator further comprises calculating a temperature from the calibration data and the corresponding operational data. In the case where the temperature can be measured directly, the configuration of the model and/or index calculation is more accurate.

-said at least one parameter representative of the operating state of said power semiconductor device monitored during the monitoring step is removed from the measured temperature. And the operating data is removed from the measured temperature values for calculating the numerical indicator. Other available parameters may be used, such as the current frequency and/or the state of the power semiconductor device with respect to the gate level. Thus, even if temperature measurements are difficult or impossible to perform during the operational life of the device, the method may be improved.

The method further comprises a step of constructing a model of the relationship between at least two voltage drops from the current value when the numerical indicator is equal to zero, said model then being used to calculate said numerical indicator from the calibration data and the operating data. Constructing such a model makes it possible to avoid any direct measurement of the temperature. This method and results are achievable even if temperature measurements are difficult or impossible to perform during the operational life of the device.

-the method further comprises the steps of:

i) a degradation model of the power semiconductor device is constructed by extrapolating the evolution of the numerical index in a manner that estimates the end of life of the electronic device. This complementary step enables to facilitate the maintenance scheduling of the device or the whole cluster of similar devices.

Other features, details, and advantages will be apparent from the following detailed description and drawings.

Drawings

[ FIG. 1]

Fig. 1 is a logic diagram of the interrelated technical effects in a power semiconductor device.

[ FIG. 2]

Fig. 2 is a graph showing the emitter current according to the emitter-collector voltage of a diode placed anti-parallel to the IGBT.

[ FIG. 3]

Fig. 3 is a graph showing the collector-emitter saturation voltage according to the collector current of the IGBT.

[ FIG. 4]

Fig. 4 shows the behavior of an IGBT.

[ FIG. 5]

Fig. 5 shows the behavior of the diode.

[ FIG. 6]

Fig. 6 shows the behavior of the diode.

[ FIG. 7]

Fig. 7 shows the behavior of an IGBT.

[ FIG. 8]

Fig. 8 shows the behavior of a diode or IGBT.

[ FIG. 9]

Fig. 9 is a full-bridge configuration consisting of two half-bridge legs.

[ FIG. 10]

Figure 10 shows a system according to the invention.

[ FIG. 11]

Fig. 11 is an example of the measurement of two current values.

Detailed Description

The drawings and the following detailed description contain substantially the exact elements. They may be used to enhance the understanding of the present invention and, if desired, to define the invention. It will be understood that some details of the three-dimensional structure of the device are difficult to describe in detail outside the figures.

Hereinafter, the word "[ two values]Substantially equal "must be interpreted according to the technical context. For both current values, the error range depends on the device and its power environment. For example, "[ two current values]Substantially equal "may be understood as a reference current or nominal current I of the power semiconductor device in use for which two current values are present_nomWithin + -5%, + -4%, + -3%, + -2% or + -1% of the same. For both temperature values, the error range depends on the desired measurement accuracy. For example, "[ two temperature values]Substantially equal "may be understood as two temperature values being equal within ± 5 ℃, + -4 ℃, + -3 ℃, + -2 ℃ or ± 1 ℃.

The method described below solves the problem of monitoring the condition and health of a power semiconductor device for a state of health and/or a remaining useful life estimation, in particular for a power semiconductor device. For example, the method may be implemented on a device comprising an Insulated Gate Bipolar Transistor (IGBT), a Metal Oxide Semiconductor Field Effect Transistor (MOSFET), a diode, or a combination of these elements. As a non-limiting example, the following modules commercialized by the applicant have been used: "CM 150TX-24S1-high power switching use insulated type". Technical announcements including detailed features of this module are publicly available for free.

Hereinafter, reference is made to V_onThe "voltage drop value" of (1) corresponds to the "on-state voltage drop" (equivalent to the collector-emitter voltage V of the IGBT_ceAnode to cathode voltage V of the diode_akDrain-source voltage V of MOSFET in forward mode_dsOr drain-source voltage V of MOSFET in reverse mode_sd) Or "wire bond voltage drop" (Kelvin to power emitter V of IGBT_keAnd Kelvin to power supply V_ksIn between) depending on the nature of the electronic components tested in the power switching device/module (diode assembly, MOSFET and/or IGBT). "Voltage drop" must be understood as a predefined input V of the power semiconductor device_inVoltage measurement and predefined output V of the power semiconductor device_outThe difference between the voltage measurements at (a). In this context, the input and output are not necessarily general inputs and outputs of the entire power semiconductor device, but may be intermediate inputs/outputs of electronic components included in the power semiconductor device. Preferably, the general inputs and outputs of the device are used, especially when the power semiconductor device is a packaged device. For various embodiments, the inputs and outputs are the same for various parts of the method, so that any differences between the measured values are not attributed to the measurement locations. Hereinafter, "voltage drop" is referred to as "V_on", and must be understood as V_outAnd V_inDifference of difference (V)_on＝V_out-V_in)。

Fig. 1 is a positive feedback loop related to wire bond failure and solder failure in a power semiconductor device. According to fig. 1, a wire bond fault can be detected by monitoring the resistance evolution Δ R during converter operation. Can be monitored by monitoring the voltage drop V_onThe resistance evolution deltar is monitored. However, the voltage drop V_onNot only on joint failure, but also on conditions such as temperature T_jAnd current I_onOther parameters of (2). Therefore, the voltage drop V is known_onAccurate resistance and therefore accurate wire bond degradation and related health information about the power semiconductor device cannot be inferred directly.

A method of establishing a degraded state of electrical connection in the power semiconductor device 1 will be described below. Such a method may be implemented using a diagnostic device 100 as represented on fig. 10. The diagnostic device 100 comprises at least one processor operatively associated with at least one storage medium and a set of plugs that can be plugged into the input and output of the power semiconductor device 1 to monitor it. The diagnostic device 100 is arranged to implement the method described below when inserted into the input and output of the power semiconductor device 1.

The method comprises the following steps:

a) when the power semiconductor device is subjected to a first current value (I)_{on_1}) And a first temperature value (T)_{1_calibration}) At least a first voltage drop value (V) of the power semiconductor device is measured in a first state_{on_1_calibration})；

b) When the power semiconductor device is subjected to a second current value (I)_{on_2}) And a second temperature value (T)_{2_calibration}) At least a second voltage drop value (V) of the power semiconductor device is measured in a second state_{on_2_calibration}) Second temperature value (T)_{2_calibration}) Is substantially equal to the first temperature value (T)_{1_calibration}) At least one of the following two conditions is satisfied:

-a first current value (I)_{on_1}) Equal to the first threshold (X% × I)_nom) And a second current value (I)_{on_2}) Equal to the second threshold (Y% × I)_nom) Second threshold (Y% × I)_nom) Strictly greater than a first threshold (X% × I)_nom)，

The first state and the second state differ from each other at least with respect to the gate level of a transistor or a group of transistors of the power semiconductor device;

c) saving at least two values (V)_{on_1_calibration},V_{on_2_calibration}) As calibration data;

d) monitoring the current (I) and at least one parameter (T, F) representative of the operating state of the power semiconductor device under operating conditions_mod,G)；

e) Measuring at least a third voltage drop value (V) of the power semiconductor device under the following conditions_{on_1_test})

-at a value substantially equal to the first current value (I)_{on_1}) Current (I) of_{1_test}) Down, and

-at least one parameter (T, F) being monitored_modG) corresponds to the operating state of the power semiconductor device and a third temperature value (T) of the power semiconductor device_{1_test}) A time of day of a first predefined set of criteria in question;

f) measuring at least a fourth voltage drop value (V) of the power semiconductor device under the following conditions_{on_2_test})

-at a value equal to the second current value (I)_{on_2}) Current (I) of_{2_test}) Down, and

-at least one parameter (T, F) being monitored_modG) corresponds to an operating state of the power semiconductor device and a value (T) of the power semiconductor device (1) which is substantially equal to the third temperature value_{1_test}) Is measured at a fourth temperature value (T)_{2_test}) The time of day of the relevant second predefined set of criteria;

g) saving at least two values (V)_{on_1_test},V_{on_2_test}As operational data;

h) based on the calibration data (V) in a manner to estimate the degradation state of the power semiconductor device_{on_1_calibration},V_{on_2_calibration}) And operation data (V)_{on_1_test}；V_{on_2_test}) To calculate a numerical index (Δ R).

Optionally, the method further comprises the steps of:

i) the degradation model of the power semiconductor device 1 is constructed by extrapolating the evolution of the numerical indicator (Δ R) in such a way as to estimate the end of life of the electronic device.

Steps a) to c) may be considered as a calibration phase of the method applied to the power semiconductor device 1, while steps e) to h) may be considered as an active phase of the diagnosis. Steps a) to c) may be performed once for each power semiconductor device 1 (e.g. at the beginning of the operational life of the power semiconductor device 1 or even on a test bench before its operational life).

After saving the calibration data (step c), steps d) to h) may be performed continuously or repeatedly without repeating steps a) to c). For example, the monitoring step d) may be performed continuously throughout the lifetime of the power semiconductor device 1. In this case, the measuring steps e) and f) can be carried out continuously or repeatedly, not only during the instants in time at which the predefined operating state of the power semiconductor device 1 is detected. In this case, the step g) of saving the correlation value as the operation data may be regarded as including a step of selecting among all available measurement data to record only the correlation data as the operation data. The measurements of steps a) and b) may also be performed continuously or repeatedly. In this case, the step c) of saving the correlation value as the calibration data may be regarded as including a step of selecting among all available measurement data to record only the correlation data as the operation data.

The saving step c) of the calibration data can be seen as selecting at least two voltage drop values corresponding to two different operating modes of the power semiconductor device 1 while the temperatures are approximately the same. The saving step g) of the operating data can be considered as selecting the best moment to acquire the data during the operational functioning of the power semiconductor device 1, the operating conditions corresponding as much as possible to the conditions under which the calibration data are measured.

Steps a) and b) enable at least two values to be obtained: for example I_{on_1}V at_{on_1_calibration}And I_{on_2}V at_{on_2_calibration}. In some examples, I_{on_1}Is a low current, and I_{on_2}Is a high current. In this context, "low" and "high" must be understood to correspond to two different operating modes of the power semiconductor device 1. Several criteria may be selected to ensure that there are two different modes of operation:

-a first current value I_{on_1}Equal to the first threshold value and a second current value I_{on_2}Equal to the second threshold. The second threshold value is strictly greater than the first threshold value (I)_{on_2}>I_{on_1})。

In a more precise embodiment, the first threshold is equal to the reference current (e.g. the nominal current value I)_nom) X%, the second threshold being equal to Y% of said reference current, wherein Y is strictly greater than X (Y)>X). For example, X is 20% and Y is 80%, or X is 30% and Y is 60%, or X is 20% and Y is 60%.

-a second current value I_{on_2}Greater than or equal to twice (I) the first current value_{on_2}/I_{on_1}≥2)。

A person skilled in the art will know how to adapt the criterion regarding the threshold current to correspond to two different operating modes of the power semiconductor device 1 depending on the application. It is preferable to make the two current values different as much as possible.

In other examples, the calibration data is obtained in two different states at least with respect to the gate level G of a transistor or a group of transistors of the power semiconductor device 1. In these examples, calibration data V is acquired_{on_1_calibration}And V_{on_2_calibration}Current value of_{on_1}And I_{on_2}May be equal or different from each other. For example, the pass/block state (or behavior) of an electronic element can be easily defined with respect to the gate level G. In other words, at least two calibration data V may be obtained when the control currents of the elements of the power semiconductor device 1 are larger and smaller than the gate level G, respectively_{on_1_calibration}And V_{on_2_calibration}。

The calibration values may be obtained prior to operation of the power semiconductor device 1, such as during a conformance test after manufacture. In this case, the power semiconductor device 1 can be provided with a current I at the input voluntarily_{on_1}And I_{on_2}Two different selected values of (a).

By measuring V_inAnd V_outTo determine the voltage drop value V_{on_1_calibration}、V_{on_2_calibration}. A pair of voltage drop values V can be repeated_{on_1_calibration}、V_{on_2_calibration}The measurement of (2). In this case, a plurality of pairs of voltage drop values V are obtained_{on_1_calibration_i}、V_{on_2_calibration_i}Which isWhere "i" is the iteration index. If multiple pairs are measured, then preferably at high frequency F_modMeasurements are made but other operating conditions are different, e.g. different heat sink (or housing) temperatures, different load currents I_load(but still I)_load≥max(I_{on_1},I_{on_2}) Different switching frequencies and/or different modulation frequencies. Two voltage drop values V of which the temperature is preferably in the same pair_{on_1_calibration_i}、V_{on_2_calibration_i}As constant as possible, but the temperature is preferably in two pairs V_{on_1_calibration_i}、V_{on_2_calibration_i}；V_{on_1_calibration_j}、V_{on_2_calibration_j}(i ≠ j) differs as much as possible.

The testing phases (steps d) to h)) can be carried out with different equipment and/or by different persons for a long time after the calibration phases (steps a) to c)). In other words, the two phases can be viewed as two different methods. In this case, calibration data may be provided from another source to enable the testing phase.

In a further embodiment of the method, steps a) and/or b) can be implemented when the power semiconductor device 1 is in operation (for example as soon as the power semiconductor device 1 is on-line).

In embodiments of the method that require a small amount of calibration data, only two values may be measured. Hereinafter, two voltage drop values V_{on_1_calibration}、V_{on_2_calibration}The relationship between has a curved form. Using these two values to complete curve V_{on_2}＝f(V_{on_1}And Δ R ═ 0). If multiple pairs are measured at different temperatures, then curves are constructed for as many pairs as measured. In an example corresponding to calibration data obtained at high and low currents, a low current I may be obtained_{on_1}Voltage drop V_{on_1_calibration}And high current I_{on_2}Voltage drop V_{on_2_}Theoretical relationships between calibrmation.

In a preferred embodiment of the method, which allows to establish the calibration accurately, easily and quickly, a plurality of points are measured and filtered to establish the curve V_{on_2}＝f(V_{on_1}Δ R ═ 0, variable T). E.g. number of measurements and performing the measurementsMay be of fixed duration or of a duration long enough to obtain sufficient data. The more measurements, the more precisely the relationship can be defined in a way that characterizes the nominal operation of the power semiconductor device 1.

Then, the measured value is registered (step c)). Which may include directly saving pairs of measured values. In other embodiments, the measured values may be processed/analyzed prior to saving. Thus, the calibration curve V resulting from the processing/analysis is saved_{on_2}＝f(V_{on_1}Δ R — 0, variable T), thus saving calibration data memory.

In some embodiments, the measuring step a) and/or b) further comprises measuring or deducing the temperature (respectively T_{1_calibration}And T_{2_calibration}) To check that the temperatures are substantially equal in nature (T)_{1_calibration}≈T_{2_calibration}). Preferably, T is measured_{1_calibration}Since temperature is easier to measure at low currents. In these cases, the temperature value T is saved_{1_calibration}And/or T_{2_calibration}As part of the calibration data. In other words, the calibration data may have the following form, for example:

-V_{on_1_calibration_i}(one or more of the measured values),_{Ion_1}(one predefined value); v_{on_2_calibration_i}(value(s) measured), I_{on_2}(one predefined value);

-V_{on_1_calibration}，I_{on_1}，T_{1_calibration}；V_{on_2_calibration}，I_{on_2}，T_{2_calibration}；

model V_{on_2_calibration}＝f(V_{on_1_calibration}Δ R ═ 0, the T variable)

Model V_{on_2_calibration}＝f(V_{on_1_calibration}Δ R ═ 0, T variable) and model V_{on_1_calibration}＝f(T,ΔR＝0)

The calculation step of the numerical index (step h)) therefore comprises the calculation of the resistance Δ R and optionally may comprise the calculation of the temperature T, as will be described hereinafter.

Initially, when the power semiconductor device 1 is new (unused), the connections and wire bonds may be considered intact. The resistance of the connection may be defined as a reference and virtually set to Δ R — 0. The low current I can be established, for example, in the following manner_{on_1}Voltage drop V_{on_1_calibration}And high current I_{on_2}Voltage drop V_{on_2_calibration}The theoretical relationship between the two is as follows: v_{on_2_calibration}＝f(V_{on_1_calibration}). Any such relationship for Δ R ≧ 0 can be obtained by solving a simple analytical equation:

V_{on_2_calibration}＝V_{on_2_calibration}(Δ R ═ 0, delta T) + Δ R × I_{on_2}

V_{on_1_calibration}＝V_{on_1_calibration}(Δ R ═ 0, delta T) + Δ R × I_{on_1}，

Then, the following relationships may be established: Δ R ═ f (V)_{on_1},V_{on_2})

These relationships may have a curvilinear form (see, e.g., fig. 4 and 5). The calibration phase allows this relationship to be determined in the form of a look-up table or as an analytical (physical or empirical) equation (dashed lines on fig. 4-7). For example, FIG. 4 shows an IGBT at low current I_{on_1}Voltage drop V_{on_1}And at high current I_{on_2}V at_{on_2}. Fig. 5 shows the same for the diode assembly.

The measured values forming the calibration data may be saved directly and/or in the form of established relationships (step c)).

The on-state voltages of the diodes, IGBTs and MOSFETs and the voltage drop across the wire bond connection are typically dependent on the current I and the temperature T. In the case of wire bond degradation, the increase in resistance also results in a total voltage drop V_onAnd (4) increasing. Fig. 2 (diode assembly) and fig. 3(IGBT) are examples of characterizing the behavior of the power semiconductor device. For example, fig. 2 shows for a diode in an IGBT power module, for a plurality of temperatures T_jAccording to the anode-cathode saturation voltage V_ACAnodic current I in volts_A(amperes). In an example, FIG. 2 shows a diode in Applicant' S device referenced as "CM 150TX-24S1Negative Temperature Coefficient (NTC) behavior of the tube arrangement. For example, FIG. 3 shows for a plurality of temperatures T_jAccording to collector current I_CCollector-emitter saturation voltage V in amperes_ECsat(volt), gate-emitter voltage V_GEFixed at 15 volts. In an example, fig. 3 shows a current I below 40A_cNegative Temperature Coefficient (NTC) ("Low Current") and Current I above 40A_c(high current) Positive Temperature Coefficient (PTC). This corresponds to the behavior of the IGBTs in applicants' device referenced as "CM 150TX-24S 1".

Resistance due to wire bonding (quoted as R)_wb) Increasing the temperature at which this can form a wire bond (referenced as T)_wb) Is locally elevated. The net resistivity of the wire bonded material (e.g., aluminum) may increase slightly and result in a significantly higher increase in resistance being observed. This self-amplifying effect of the resistance increase ar is generally negligible, since it is relatively small and does not spread over the entire device. The following relationships may be formed:

V_{on_measured}＝(R_wb(T_wb)+ΔR)×I_on+V_{on_DIE}(T_j,I_on)

according to the power semiconductor device 1, the on-state current I_onCan be as follows:

the collector (or emitter) current of the IGBT,

anode (or cathode) current of diode assembly, or

The drain (or source) current of the MOSFET.

The measured data may be obtained from a sensor (e.g., a hall effect sensor or a shunt sensor). The measured data may be obtained from a controller of the power semiconductor device (e.g., a reference current of the controller). In some existing power semiconductor devices, there are sensors in the converter for control purposes. The measured data may be derived/estimated from the available data (optionally by improving processing, e.g. by applying gain or offset removal).

If the calibration data includes temperature data or voltage drop V_{on_1_calibration}Associated with a temperature T at which Δ R-0Data, then the temperature T can also be estimated. Without a specific sensor for temperature, such calibration is difficult to achieve online. Without a specific sensor, such calibration may be performed offline prior to power semiconductor device assembly. The resistivity ar may be estimated from time to time. When modulating frequency F_modThis is preferred high (for IGBT and diode assemblies). In this case, the temperature T may be considered to be in a quarter cycle (T) of an Alternating Current (AC) due to thermal inertia_modThe period/4) is constant. FIGS. 6, 7 and 8 show temperature T and voltage drop V_onExamples of correlation data in the form of a cross-curve. For example, fig. 6 and 7 show the diode assembly and IGBT at high current I, respectively_{on_2}Voltage drop V_on. For example, fig. 8 shows a diode assembly and IGBT at low current I_{on_1}Voltage drop V_on。

In some embodiments where the power semiconductor device is a packaged device, the on-state voltage drop V_onIs the voltage drop across the packaged device in the on state. For example, the on-state voltage drop V is measured using a circuit consisting of a voltage clamp, a signal conditioner and an analog-to-digital converter (ADC)_on。

In the context of step d), the monitored value representing the operating state of the power semiconductor device 1 may comprise, for example, the load current I_loadAnd/or modulating the frequency F_mod. In some embodiments, the modulation frequency F_modThus receiving or estimating the voltage drop V_onUseful parameters of (2). Depending on the operational application of the power semiconductor devices, some of the devices are used with Alternating Current (AC), e.g., in combination with an electric motor. In this case, the load current I_loadHaving a periodic shape (e.g., sinusoidal behavior). Modulation frequency F_modIs variable, for example between 1Hz and 1 kHz. For example, in the case of a device used as a converter connected to a synchronous motor, the modulation frequency is proportional to the rotational speed of the motor. One example is an electric vehicle. In this application, the modulation frequency F is such that it is applied when the vehicle is travelling slowly_modLow, when the vehicle is travelling rapidly, modulation frequency F_modHigh. Modulation frequency F_modCan be derived from the current and voltage values. It may also be provided by an external controller.

The steps a), b), c), e), f) and g) are arranged to be performed at optimal times to obtain at least one pair of two correlated voltage drops (V)_{on_1_calibration}And V_{on_2_calibration}) And (V)_{on_1_test}And V_{on_2_test}) The correlation measurement of (2). The first and second predefined set of criteria of the measuring steps e) and f) are selected accordingly. Preferably, the first and second predefined sets of criteria are selected in correspondence with each other.

For example, the optimal time may preferably be when two voltage drops V are measured_{on_1}And V_{on_2}The temperatures T are substantially the same. For example, two voltage drops V are measured at successive times_{on_1}And V_{on_2}. For example, the first set of criteria and the second set of criteria may be selected to improve two measurements at two time instances separated by at most 5 milliseconds. For example, during the same quarter period (T) of the AC current_modIn/4) and for high modulation frequencies F_mod(e.g., a modulation frequency greater than Z, where Z equals 20Hz, 30Hz, or 50Hz) two voltage drops V are measured_{on_1}And V_{on_2}. FIG. 11 shows two different current values I at two successive times separated from each other by less than a quarter-cycle T_{on_1}And I_{on_2}Measuring different voltage drop values V_{on_1}And V_{on_2}。

The monitored at least one value representative of the operating state of the power semiconductor device 1 may comprise the current I experienced by the power semiconductor device 1_onItself. In this case, the set of criteria may include information about the current I_onThe standard itself. This enables improvements in terms of current I_onFor example, at low and high currents. For example, the first set of predefined criteria includes the monitored current value I being less than a first threshold value. The second set of predefined criteria includes the monitored current value I being greater than a second threshold value. For example, the second threshold is selected to be greater than the first threshold. For example, the threshold value is defined as a nominal current value I_nomIs a part of: the first threshold value being equal to the nominal current value I_nomX% of the nominal current value I_nomY% of (C). For example, X equals 20, 30 or 40%. For example, Y equals 60, 70 or 80%. These values can be defined as the nominal current values I of the power semiconductor device 1_nomAnd also on the type of application and the characteristics (precision and accuracy) of the current sensor. At the first current value I_{on_1}And a second current value I_{on_2}In embodiments where the conditions of (one is strictly greater than the other), the set of criteria of steps e), f) may be the same as the conditions of steps a), b).

The monitored at least one value representative of the operating state of the power semiconductor device 1 may comprise the current I experienced by the power semiconductor device 1_onItself and the gate level G of a transistor or a group of transistors of the power semiconductor device 1. In this case, the set of criteria may be as follows:

-the first predefined set of criteria comprises a low level of the gate level, and

-the second set of predefined criteria comprises a high level of the gate level.

The different current values I between steps a) and b) can be combined in the same embodiment_{on_1}And I_{on_2}With respect to the two sets of criteria of steps e) and f) relating to the operating state defined with respect to the gate level, and vice versa.

Embodiments comprising the conditions (steps a) and b)) and/or the criteria regarding the gate level (steps e) and f)) are particularly suitable for MOSFETs without external anti-parallel diodes. For current I when the MOSFET is on (high gate level)_{on_1}Measuring voltage (V)_{on_1_calibration}And/or V_{on_1_test}). For current I when the MOSFET is off (low gate level) and the internal body diode is on_{on_2}Measuring voltage (V)_{on_2_calibration}And/or V_{on_2_test}). Current value I_{on_1}And I_{on_2}May be equal and negative (flowing from source to drain).

Since the state of health may not need to be estimated very frequently, for at least some embodiments, it may be decided to deactivate the health estimation for a particular operating time.

In other embodiments, the voltage drop value V of the power semiconductor device 1_{on_1}And V_{on_2}And the current value I to which the power semiconductor device 1 is subjected_{on_1}And I_{on_2}Is carried out continuously or at a predetermined frequency, not only at the at least one value (e.g. F) monitored_modAnd/or I_load) The instants corresponding to the first or second predefined operating states of the power semiconductor device 1 (when complying with the set of standards) are implemented. In other words, the voltage drop V can be continuously monitored_onAnd current I_onAnd defining the operation data V by post-processing_{on_1_test}、I_{on_1}、V_{on_2_test}、I_{on_2}Or when the current is equal to I_{on_1}And I_{on_2}The monitoring is triggered.

Including only when the current is equal to I_{on_1}And I_{on_2}The only-time-triggered embodiment is particularly suitable for bipolar devices like IGBTs and diodes, including body diode MOSFETs. These embodiments and low current I_{on_1}And high current I_{on_2}The following measurements (see examples above) are particularly advantageous in combination.

The above examples of criteria may be combined with each other.

In the step h) of calculating the numerical index, the measured voltage V is included_{on_1_test}And V_{on_2_test}Operating data and calibration data V_{on_1_calibration}、I_{on_1}And/or T_{1_calibration}、V_{on_2_calibration}、I_{on_2}And/or T_{2_calibration}Used together to estimate the resistance increase Δ R and optionally the temperature T (if the calibration data includes a drop V from the current)_onData of estimated temperature).

Viewed from another perspective, the process can be decomposed as follows:

1) receiving or generating calibration data;

2) receiving or measuring information about operating conditions;

3) determining a time of a state of health estimate and/or a temperature estimate;

4) at a current I_on1Lower reception or measurement of a first voltage drop measurement V_on1；

5) At a current I_on2Lower reception or measurement of a second voltage drop measurement V_on2；

6) Processing data, including:

a. estimating junction temperature using previously estimated Δ R, and/or

b. The resistance increase Δ R is estimated.

In the example tested by the applicant, the model was established after calibration for temperature T. Convergence yields the following relationship:

V_on＝(0.001+ΔR)×I_on+[(1–0.004×T)+0.015×I_on]

in such an equation, the first component is the connection resistance (wire bond) and the second component is the die voltage drop.

Hereinafter, this relationship is used to estimate Δ R and T

1. The purpose is as follows: t estimation

We consider Δ R as 0 because the module is new (unused).

F_modNot required high, therefore T1 ≠ T2.

The above equation can be used with the measured V_{on_1_test}And V_{on_2_test}To estimate T1 and T2.

Assuming that the true temperatures T1real and T2real are 80 and 90 c, V is the correct equation_{on_1_test}And V_{on_2_test}Allows accurate estimation of T1 and T2

a. Low current (I)_{on_1}2A), measured V_{on_1_test}The estimated temperature of 80 ℃ was obtained at 0.712V.

b. High current (I)_{on_2}100A), measured V_{on_2_test}The estimated temperature of 90 ℃ was obtained at 2.24V.

Therefore, the equation is correct.

2. The purpose is to; t estimation

Due to degradation, the resistance increases by 5m Ω, but this is unknown (assuming Δ R is 0m Ω)

Errors due to erroneous estimates of Δ R may affect the estimate of T2.

Δ Rreal 5m Ω, estimated (true 80 ℃ and 90 ℃)

a. Low current (I)_{on_1}2A), measured V_on0.713V gave an estimated temperature of 79.75 deg.c instead of 80 deg.c.

b. High current (I)_{on_2}100A), measured V_onThe estimated temperature of 77.5 deg.c was obtained instead of 90 deg.c 2.29.

Therefore, the value of Δ R must be updated.

3. The purpose is to; high F_modDelta R estimation of

Δ Rreal 5m Ω, Δ R estimate (true 80 ℃)

a. Low current (I)_{on_1}2A), measured V_on0.713V gave an estimated temperature of 79.75 deg.c instead of 80 deg.c. Here we only estimate the temperature.

b. High current (I)_{on_2}100A), measured V_on2.33V (at true 80 ℃ because of high F)_mod) An estimated Δ R of 4.9m Ω (79.75 ℃ C. according to the model) was obtained. Here, Δ R is estimated.

After Δ R estimation, it is used again to estimate temperature and damage, resulting in convergence of both T and Δ R measurements.

Hereinafter, some examples of the embodiments are described according to various types/configurations of power semiconductor devices.

In a first example, the power semiconductor arrangement 1 has a half-bridge configuration. For example, fig. 9 shows a full-bridge configuration consisting of two half-bridge legs. The diode is connected in anti-parallel with the IGBT. A voltage is sensed across the combination. The current in the load is sensed. The measured voltage drop is the voltage drop across the IGBT (positive voltage) or across the diode (negative voltage) depending on the load current. For example, the monitoring of the respective devices is performed during the subsequent half modulation period.

In the second example, the power semiconductor device 1 is used as the gate voltage sensitivity. For some IGBTs and MOSFETs, V_on(I_on) The temperature sensitivity of the curve depends on the gate voltage. This modifies the curve V_{on_2}(V_{on_1}) And V_on(T_j) As shown in fig. 4 to 8. Thus for the method of estimating only Δ R and estimating Δ R and T_jThe gate voltage can be controlled to increase the sensitivity. This involves a special means of controlling the gate voltage. In this case, it is necessary to establish V with an appropriate gate voltage_{on_1}And V_{on_2}And (Δ R or V)_on) And (T)_j、I_onAnd Δ R).

In a third example, the method is used to estimate the state of health of the power semiconductor device 1. The resistance increase Δ R can be directly used to estimate the state of health. For example, an increase of 1m Ω may indicate a 50% damage level. Detailed pre-calibration data.

In a fourth example, the method is for estimating the state of health of the power semiconductor device 1 using stress counting. The online estimated junction temperature may be used as an input to a temperature cycle count algorithm, such as the "rainflow" algorithm. So-called damage models, such as the Coffin-Manson model or the combined Coffin-Manson and Arrhenius model, can be used to estimate the damage caused by each cycle. The total damage can then be estimated using a damage accumulation law, such as a linear accumulation law.

In a fifth example, the method is for estimating the state of health of the power semiconductor device 1 using a thermal resistance parameter. Can utilize the input I_on、V_onTo estimate the power loss P in the die_loss. Can use T_jEstimated and casing temperature T_cTo estimate the temperature difference at between the die and the substrate. Can use P_lossAnd Δ T ═ T_j-T_cTo estimate the thermal resistance R_th。

In a sixth example comprising the optional step i), the method is used to estimate the end of life (or remaining useful life RU L) of the power semiconductor device 1.

This is a simplified model assuming a linear increase in resistivity of the metallization and linear crack propagation in the wire bond. Such a model may be identified/fitted and then extrapolated. The intersection with the threshold provides the end of life.

The foregoing examples may be combined with each other and with the above-described embodiments. The method of establishing the state of degradation of the electrical connections and their health in the power semiconductor device 1 can be applied in various power modules in various technical fields, in particular in the field of traction equipment, wind energy, factory automation or in the field of automobiles.

The invention has been described as a method. In another aspect, the invention may also be seen as a diagnostic device 100 (shown in fig. 10) comprising at least one processor operatively associated with at least one storage medium and a set of plugs that can be plugged into the power semiconductor device 1. Such a diagnostic device 100 is arranged to implement the method according to the foregoing when inserted into the power semiconductor device 1. The invention may also have the form of a kit comprising, for example, a power semiconductor device 1 and a diagnostic device 100 that can be operatively coupled to each other to form an operating assembly.

The present invention is not limited to the devices, assemblies, kits, and processes described herein, which are examples only. The invention covers each and every alternative that may occur to those skilled in the art within the scope of the following claims.