num="0000">[Inspection Method]

FIG. 3 is a graph (an antilogarithm graph) showing an example of measurement results of the leakage current of electronic components W (i.e., capacitors). The X-axis in FIG. 3 shows the charging time (milliseconds (ms)), which is the time elapsed from the start of measurement (the origin (0)), and the Y-axis shows the measured values of the leakage current of electronic components W (amperes (A)).

FIG. 3 shows the measurement results of the leakage current of a normal electronic component (i.e., a “good component”) that contains no abnormalities and the measurement results of the leakage current of an electronic component having an anomaly (i.e., a “defective component”) in a case where the electronic components W (i.e., capacitors) are charged by applying a constant DC voltage.

The inventors of the present case have newly discovered, as a result that the inventors had continued to conduct their studies on the behavior of electrical characteristics (in particular, the leakage current characteristics) of capacitors (e.g., MLCCs) during charging, the fact that capacitors sometimes exhibit temporary abnormal behavior during charging (see the reference code of “Ab” in FIG. 3).

Such abnormal behavior can be ascertained by changes in the measured leakage current of a capacitor, but such abnormal behavior is not always constant and the amount of change may be small. Further, the inventors have newly discovered that there are some capacitors that return to normal behavior after exhibiting abnormal behavior for a certain period of time (e.g., tens of milliseconds to several seconds) and that thereafter behave like good capacitors.

A conventional inspection apparatus that detects capacitor abnormalities based on the measured peak value of the leakage current, is suitable for determining faults which causes the leakage current to increase in an unexpected timing, but is not necessarily suitable for determining faults which causes the leakage current to decrease. Further, conventional inspection apparatuses cannot always address in a sufficient manner capacitors that do not have deteriorated insulation but have anomalies that cause unstable leakage current for some reason.

Furthermore, the larger capacity of MLCCs have made progress in recent years, but the ceramic layers of MLCCs tend to become thinner as the capacity increases, making it difficult to apply a high voltage to MLCCs for screening inspections. Therefore, there is a need for a technology to achieve reliable screening inspections while keeping the applied voltage to MLCCs (capacitors) low.

For example, the device of Japanese patent application publication No. 9-152455 detects steep pulse-like abnormal currents (i.e., leakage currents) that occur in capacitors during charging. Although the device of Japanese patent application publication No. 9-152455 is effective when the leakage current during charging shows stable behavior, it is difficult for the device of Japanese patent application publication No. 9-152455 to reliably detect abnormal currents when the leakage current changes relatively rapidly per unit of time. Further, the device of Japanese patent application publication No. 9-152455 is unable to detect anomalies that cause the leakage current of a capacitor to decline.

Further, the device of Japanese patent application publication No. 2000-228337 determines whether a capacitor is good or bad based on whether the evaluation function that is determined from the capacitor's actual measurement current value and the standard screening current value, draws a curve protruding with respect to either the top or bottom. However, as also described in Japanese patent application publication No. 2000-228337 (see paragraph 0012), the device of Japanese patent application publication No. 2000-228337 treats temporary changes in the actual measurement current value of a capacitor as noise and does not detect them as abnormalities. Further, the device of Japanese patent application publication No. 2000-228337 is unable to detect anomalies that cause the leakage current of a capacitor to decline.

The inventors of the present case have discovered, as a result that the inventors had continued to conduct their studies in light of the above-mentioned knowledges, a new technology that can effectively and accurately detect even capacitor abnormalities that cannot be detected by conventional devices. Specifically, a new method has been found to accurately detect capacitor abnormalities based on a regression line derived from the logarithm of the measurement values of the current (leakage current) of an electronic component W (a capacitor) in the determination time range during charging of the electronic component W (the capacitor) and the logarithm of the measurement timing of said current. According to this method, it is possible to detect with high accuracy abnormal behavior that a capacitor having an anomaly temporarily exhibits during charging. In particular, even if the amount of change in the abnormal behavior is small or even if the abnormal behavior lasts only a limited time (e.g., tens of milliseconds to several seconds), such abnormal behavior can be detected with high accuracy.

Although there are many possible variations of such an anomaly detection method, all of them can detect abnormal behavior of a capacitor based on a regression line. A typical example of such an anomaly detection method is shown below; however, anomaly detection methods that utilize regression lines in a form different from the typical example described below are also included in the contents of the present disclosure.

[First Anomaly Detection Method]

FIG. 4 is a graph (i.e., a double logarithm graph) that shows an example of measurement results of the leakage current of electronic components W, for a graph for explaining a first anomaly detection method. The X and Y axes in FIG. 4 are based on a logarithmic scale, the X axis in FIG. 4 shows the charging time (milliseconds), which is the time elapsed from the start of measurement (the origin (0)) using a logarithm, and the Y axis shows measurement values (amperes) of the leakage current of electronic components W using a logarithm.

FIG. 4 shows measurement results of the leakage current of a normal electronic component that contains no abnormalities (i.e., a “good component”) and measurement results of the leakage current of an electronic component with an anomaly (i.e., a “defective component”) when a constant DC voltage was applied to the electronic components W (capacitors) for charging. Further, in FIG. 4, a “regression line Lr” obtained based on the least-squares method from the measurement results of the good component and the defective component is shown. According to the inspection method based on the first anomaly detection method, a charging step, a current measurement step and an anomaly detection step are performed under the control of the control unit 55.

In the charging step, a voltage is applied to an electronic component W (a capacitor) by the electric circuit 50 (the voltage applying unit 51) in a condition where the electronic component W is sandwiched by a probe 31 and the electrode 36 shown in FIG. 2 in such a manner that the charging of the electronic component W is performed. In the current measurement step, the current of an electronic component W (the leakage current of a capacitor) that is being charged is continuously measured by the electric circuit 50 (the current measurement unit 52). In this manner, the current measurement step and the charging step are performed in parallel.

Then, in the anomaly detection step, the control unit 55 (the anomaly detection unit) detects an anomaly of an electronic component W (a capacitor) based on a regression line Lr derived from the logarithm of measured values of the leakage current in a determination time range Tv during charging and the logarithm of measurement timing of the leakage current. More specifically, the control unit 55 detects an anomaly of an electronic component W based on the degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr.

The “determination time range Tv” here is a time range suitable for determining whether or not an electronic component W has an anomaly, and more specifically, is set to a time range where the degree of the agreement of the measurement results of a good component with respect to the regression line Lr is high. The optimal determination time range Tv is determined under the influence of the leakage current characteristics of an electronic component W and/or the applied voltage to an electronic component W, and may be set in the time range of 200 ms to 1500 ms after the start of charging (see the origin “0” in FIG. 4) as an example.

The “degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr” is typically calculated as a coefficient of determination based on the least-squares method. The specific method of calculating the coefficient of determination is not limited, and the coefficient of determination can be calculated using a known arithmetic formula.

For example, if an electronic component W is a good component, the “degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr” is relatively high, and the value of the coefficient of determination (a contribution ratio) based on the least-squares method is also relatively high. If the coefficient of determination for a case in which the logarithm of the measured value of the leakage current and the logarithm of the timing of measurement coincide perfectly with the regression line Lr is expressed as “100%”, the coefficient of determination for a good electronic component W may indicate, for example, 95.0% or more (as an example, 99.0% or more).

On the other hand, if an electronic component W is a defective component, the “degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr” is relatively low, and the value of the determination coefficient is also relatively low and may indicate a value smaller than 95.0% (as an example, a value smaller than 99.0%) for example.

Although electronic components W to be inspected can be classified as good components or defective components based on the determination coefficient as described above, the threshold value of the determination coefficient that serves as the classification criteria for good components and defective components, is set appropriately according to the actual tolerance of abnormalities. In other words, the closer the threshold value is to “100%,” the stricter the determination of the presence or absence of an anomaly becomes and even minor abnormalities are not tolerated, and consequently the tendency to increase the percentage of electronic components W determined to be defective components is created.

Therefore, the higher the requirement for reliability of the normality of the electrical characteristics (leakage current characteristics) of electronic components W, the closer to “100%” is set the threshold value of the determination coefficient that serves as the classification criteria for good components and defective components. To meet the high level of reliability required by recent advanced technologies, the threshold value of the determination coefficient may be required to be set to “99.0%” or higher, for example.

In this manner, the “degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr” represents the linearity of the logarithm of the measured values of the leakage current and the logarithm of the measurement timing, and can be expressed as a coefficient of determination based on the least-squares method. However, the “degree of the agreement of the logarithm of the measurement timing and the logarithm of the measured values of the leakage current with respect to the regression line Lr” may be obtained based on any other method that can directly or indirectly evaluate the linearity on a double logarithm graph of the “CHARGING TIME-MEASUREMENT CURRENT”.

For example, the control unit 55 (the anomaly detection unit) may determine the “degree of the discrepancy” of the logarithm of the measured values of the leakage current and the logarithm of the measurement timing with respect to the regression line Lr using any method, and may detect an anomaly of an electronic component W based on the degree of the discrepancy. Further, the control unit 55 (the anomaly detection unit) may extract the data of the measured values of the leakage current of an electronic component W to be inspected only for a part of the time range (the determination time range Tv) during the charging time. In this case, the control unit 55 may detect an anomaly of an electronic component W based on the regression line Lr derived from the logarithm of the measured values of the extracted leakage current and the logarithm of the measurement timing of said leakage current.

The control unit 55 controls the electronic component discharge unit 14 based on the results of the inspection including the series of steps described above to discharge electronic components W that are determined to have an anomaly and electronic components W that are not determined to have an anomaly into separate collection boxes of the collection device 16 via the discharge paths 15.

[Second Anomaly Detection Method]

In the following description about a second anomaly detection method, detailed descriptions are omitted for processes similar to the first anomaly detection method described above (see FIG. 4).

FIG. 5 is a graph (i.e., a double logarithm graph) that shows an example of measurement results of the leakage current of electronic components W, for explaining a second anomaly detection method. The X and Y axes in FIG. 5 are represented by logarithmic scales, the X axis in FIG. 5 shows the logarithm of the charging time (milliseconds), which is the time elapsed from the start of measurement (origin (0)), and the Y axis shows the logarithm of the measured values (amperes) of the leakage current of electronic components W.

FIG. 5 shows the measurement results of an electronic component being a good component and the measurement results of an electronic component being a defective component, as well as the regression line Lr obtained based on the measurement results of the good component and the defective component. In addition, FIG. 5 shows the upper limit B1 and the lower limit B2 of the current tolerance range.

Also in the inspection method based on the second anomaly detection method, a charging step, a current measurement step and an anomaly detection step are performed, and in particular, the charging step and the current measurement step are performed in a similar manner to those in the inspection method based on the first anomaly detection method described above.

However, in the anomaly detection step based on the second anomaly detection method, the control unit 55 (the anomaly detection unit) determines whether or not the logarithm of the measured values of the leakage current of an electronic component W in the determination time range Tv is included within the current tolerance range defined based on the regression line Lr. In other words, an anomaly of an electronic component W (a capacitor) is detected based on the relation of the logarithm of the measured values of the leakage current with respect to at least one of the upper limit B1 and the lower limit B2 of the current tolerance range which are defined based on the regression line Lr over the determination time range Tv.

In the logarithmic graph in FIG. 5, the area between the line indicated by the reference sign “B1” (i.e., the upper limit) and the line indicated by the reference sign “B2” (i.e., the lower limit) corresponds to the current tolerance range, and the regression line Lr exists in this area (i.e., the current tolerance range) in the determination time range TV.

As described above, electronic components W to be inspected can be classified as good components or defective components based on the current tolerance range (the upper limit B1 and the lower limit B2), and the upper limit B1 and the lower limit B2 that define the current tolerance range correspond to threshold values that serve as classification criteria for good components and defective components and are appropriately set according to the actual tolerance of abnormalities. In other words, the closer the values of the upper limit B1 and the lower limit B2 of the current tolerance range are to the regression line Lr, the stricter the determination of the presence or absence of an anomaly becomes, the less acceptable even minor anomalies become, and the greater the percentage of electronic components W determined to be defective. Therefore, the higher the demand for reliability of the normality of the electrical characteristics (leakage current characteristics) of electronic components W becomes, the closer to the regression line Lr the values are set as the upper limit B1 and the lower limit B2 of the current tolerance range, which are the classification criteria for good and defective components.

The specific method of determining the upper limit B1 and the lower limit B2 of the current tolerance range is not limited. For example, between the upper limit B1 and the lower limit B2 of the current tolerance range, the differences from the regression line Lr regarding the measurement current (see Y-axis in FIG. 5) may be the same value or different values. Further, the difference with respect to the measurement current between the regression line Lr and the upper limit B1 or the lower limit B2 of the current tolerance range may be the same or may not be the same over the entire determination time range Tv. Therefore, in a double logarithm graph of “CHARGING TIME-MEASUREMENT CURRENT” (see FIG. 5), the line indicating the upper limit B1 of the current tolerance range and the line indicating the lower limit B2 of the current tolerance range may be parallel or non-parallel to the regression line Lr.

Further, the upper limit B1 and the lower limit B2 of the current tolerance range may be determined from the values of the regression line Lr (Y-axis values with respect to the measurement current) at multiple discrete time timings selected from the determination time range Tv. For example, the upper limit B1 and the lower limit B2 of the current tolerance range may be defined based on the regression line Lr with respect to the two points of the “start time timing (start point)” and the “end time timing (end point)” of the determination time range Tv. In this case, the upper limit B1 of the current tolerance range over the entire determination time range Tv may be defined by a straight line passing through these two points (i.e., the start point and the end point) regarding the upper limit B1 of the current tolerance range in a double logarithmic graph of “CHARGING TIME-MEASUREMENT CURRENT” (see FIG. 5). Similarly, the lower limit B2 of the current tolerance range over the entire determination time range Tv may be defined by a straight line passing through these two points (the start point and the end point) regarding the lower limit B2 of the current tolerance range in a double logarithm graph of “CHARGING TIME-MEASUREMENT CURRENT”.

Further, the control unit 55 (the anomaly detection unit) may set only one of the upper limit B1 and the lower limit B2 of the current tolerance range to a value different from the regression line Lr. In this case, the other of the upper limit B1 and the lower limit B2 of the current tolerance range may be set the same value as the regression line Lr.

In the anomaly detection step described above, if the entire measurement current of an electronic component W in the determination time range Tv is within the current tolerance range (including, for example, the boundaries (the upper limit B1 and the lower limit B2)), the control unit 55 can determine that there is no anomaly in the electronic component W. On the other hand, if at least a part of the measurement current of an electronic component W in the determination time range Tv is in a range outside the current tolerance range (which does not include, for example, the boundaries (the upper limit B1 and the lower limit B2) of the current tolerance range), the electronic component W can be determined to have an anomaly.

Then, the control unit 55 controls the electronic component discharge unit 14 based on the results of the inspection including the series of steps described above in such a manner that the electronic components W that are determined to have an anomaly and the electronic components W that are not determined to have an anomaly are discharged into separate collection boxes of the collection device 16 via the discharge paths 15.

[Examples of Application of Anomaly Detection Methods]

The first anomaly detection method and the second anomaly detection method described above (see FIGS. 4 and 5) or other detection methods based on the regression line Lr can be applied to a method of inspecting electronic components W as follows to accurately detect anomalies in electronic components W.

[Application 1]

FIG. 6 is a double-logarithmic graph of “CHARGING TIME-MEASUREMENT CURRENT” that shows an example of the general electrical characteristics of a capacitor.

The correlation between the logarithm of the charging time of a capacitor and the logarithm of the leakage current of the capacitor when a constant DC voltage is applied to the capacitor for charging is not necessarily constant over the entire charging time, and there may be multiple phases (see “P