Maintenance equipment, maintenance methods and maintenance procedures

Abstract

The present invention provides a maintenance device, comprising: an acquisition unit, which acquires, for each fixture, a plurality of test results when a plurality of different measured components are tested in sequence through a plurality of fixtures; a calculation unit, which uses the plurality of test results to calculate the change of the test results for each fixture; and a determination unit, which determines the maintenance timing of the fixture based on the change of the test results.

Description

[Technical field]

[0001] The invention relates to a maintenance device, a maintenance method and a maintenance program. [Background technology]

[0002] When a component to be measured is tested using a jig, the condition of the jig deteriorates as the test is repeated. Therefore, conventionally, the jig is maintained regularly at a predetermined timing.

[0003] Problems to be solved by the invention

[0004] However, depending on the situation of the component being tested, the periodic maintenance of the jig may be too much or too little. Therefore, when testing the component being tested with the jig, it is desirable to optimize the timing of jig maintenance. [Summary of the invention]

[0005] In order to solve the above-mentioned problem, the first aspect of the present invention provides a maintenance device. The maintenance device may include an acquisition unit, which acquires multiple test results when multiple different measured components are tested in sequence through multiple fixtures for each fixture. The maintenance device may include a calculation unit, which uses the multiple test results to calculate the change of the test results for each fixture. The maintenance device may include a determination unit, which determines the maintenance timing of the fixture based on the change of the test results.

[0006] The determination unit may compare the variation in the test result with the variation in the normal state to calculate a difference, and determine the maintenance timing based on the difference.

[0007] The variation in the normal state may be an average of the variation in the test results of a plurality of jigs in the normal state.

[0008] The decision unit may decide to perform maintenance on the jig when the difference exceeds a predetermined threshold value.

[0009] The predetermined threshold value may be determined by machine learning.

[0010] The determination unit can calculate the difference by machine learning.

[0011] The difference may be the Mahalanobis distance.

[0012] The maintenance device may further include a learning unit that uses a plurality of test results obtained in a normal state to learn a covariance matrix for calculating the Mahalanobis distance.

[0013] The maintenance device may further include a model generation unit that generates a model by machine learning, the model detecting a deviation from a change in a normal state based on a change in the test result; and the determination unit may calculate the difference based on an output of the model.

[0014] The calculation unit may calculate the variation of the test results by analyzing a plurality of test results in a predetermined period while sequentially changing the starting point of the period.

[0015] The calculation unit can calculate the change in the failure rate among the plurality of test results.

[0016] The failure rate may be an overall failure rate, which means that at least one item among a plurality of test items fails.

[0017] The failure rate may be a partial failure rate, which indicates that a specific item among a plurality of test items failed.

[0018] The decision section can decide the cleaning period of the jig as the maintenance opportunity.

[0019] The decision section can further decide the cleaning content of the jig.

[0020] The determination unit may determine the cleaning content of the jig based on the change in the test result of each of the plurality of test items.

[0021] A second aspect of the present invention provides a maintenance method. The maintenance method may include the following steps: obtaining multiple test results for each fixture when multiple different components are tested in sequence through multiple fixtures. The maintenance method may include the following steps: using the multiple test results, calculating the change in the test results for each fixture. The maintenance method may include the following steps: determining the maintenance timing of the fixture based on the change in the test results.

[0022] A third aspect of the present invention provides a storage medium storing a maintenance program. The maintenance program can be executed by a computer. The maintenance program can make the computer function as the following components: an acquisition unit that acquires, for each fixture, a plurality of test results when a plurality of different measured components are tested sequentially through a plurality of fixtures. The maintenance program can make the computer function as the following components: a calculation unit that uses a plurality of test results to calculate the change in the test results for each fixture. The maintenance program can make the computer function as the following components: a determination unit that determines the maintenance timing of the fixture based on the change in the test results.

[0023] In addition, the above invention summary does not list all the necessary features of the present invention. Moreover, sub-combinations of these feature groups can also constitute inventions.

Brief Description of the Drawings

[0024] Figure 1 The maintenance device 200 and the measurement system 10 of this embodiment are shown together.

[0025] Figure 2 A detailed example of the probe card 140 according to the present embodiment is shown.

[0026] Figure 3 An example in which the probe card 140 of the present embodiment is brought into contact with the measurement object 20 is shown.

[0027] Figure 4 An example of a flow of determining the maintenance timing of the jig 148 by the maintenance device 200 according to the present embodiment is shown.

[0028] Figure 5 The variation of the test result and the Mahalanobis distance (Mahalanobis Distance) of each jig 138 calculated by the maintenance device 200 of the present embodiment are shown.

[0029] Figure 6 A maintenance device 200 according to a modified example of the present embodiment is shown.

[0030] Figure 7 An example of a computer 2200 is shown that can embody all or part of the plural aspects of the present invention. [Specific implementation method]

[0031] The present invention will be described below by way of the embodiments of the invention, but the following embodiments do not limit the invention in the scope of the claims. In addition, the solution means of the invention do not necessarily require all combinations of the features described in the embodiments.

[0032] Figure 1 The maintenance device 200 and the measuring system 10 of this embodiment are shown together. The maintenance device 200 of this embodiment can be applied to various tests performed in the measuring system 10, such as tests performed on a wafer having a plurality of semiconductors or electronic components such as micro-electromechanical systems (MEMS), tests performed simultaneously on a plurality of bare chips after slicing the wafer, or tests performed simultaneously on a plurality of packages after sealing the chips, etc. That is, the maintenance device 200 can be applied to tests performed in either the so-called front step or the back step.

[0033] The example shown in this figure is a case where the maintenance device 200 is applied to a wafer test, in which a tester is used to test a wafer having a plurality of components 30 (eg, chips) to be measured as measurement objects 20. This case is described below.

[0034] In the measurement system 10, a test device 100 tests a measurement object 20. The test device 100 may be, for example, a component test device such as a system LSI (large scale integrated circuit) tester, an analog tester, a logic tester, or a memory tester. The test device 100 applies various test signals to the measurement object 20 via a fixture, and obtains a response signal from the measurement object 20.

[0035] The test apparatus 100 includes a tester body 110 , a test head 120 , a performance board 130 , a probe card 140 and a probe tester 150 .

[0036] The tester body 110 is the main body of the test device 100 and controls various tests. The tester body 110 may also have the function of outputting the test results obtained by various tests to the maintenance device 200 of this embodiment via wired or wireless means.

[0037] The test head 120 is connected to the tester body 110 via a cable, and is configured to be drivable between a test position for testing the measurement object 20 and a retreat position. When the test head 120 is to perform a test, based on the control performed by the tester 110, it transmits a test signal to the measurement object 20 in the test position, receives a response signal from the measurement object 20, and relays the response signal to the tester body 110.

[0038] The performance board 130 is detachably mounted on the test head 120 and is electrically connected to the test head 120 .

[0039] The probe card 140 is detachably mounted on the performance board 130 and is electrically connected to the performance board 130. The probe card 140 has a substrate 142, a probe housing 144, and a plurality of probes 146, and the plurality of probes 146 are brought into contact with a plurality of electrode pads in the measurement object 20 to achieve electrical contact. That is, the probe card 140 functions as an interface portion that links the test function of the test device 100 with the measurement object 20 when the test device 100 tests the measurement object 20. The probe card 140 may use, for example, a vertical probe type probe or a cantilever type probe. In addition, the structure of the probe card 140 may be appropriately changed in accordance with the type of the measurement object 20. For example, in the case where the maintenance device 200 is targeted for testing in a later step, the probe card 140 may also have a socket, etc. The details of the probe card 140 will be described later.

[0040] The substrate 142 may be, for example, a printed circuit board. A probe housing 144 is provided on one surface of the substrate 142. Also, a plurality of connection pads (not shown) for contacting the performance board 130 are provided on the other surface of the substrate 142.

[0041] The probe housing 144 holds a plurality of probes 146 therein such that the tips of the probes are exposed.

[0042] Each of the plurality of probes 146 has one end exposed outside the probe housing 144 and the other end electrically connected to a connection pad provided on the other surface of the substrate 142 via the probe housing 144 and the inside of the substrate 142 .

[0043] The plurality of probes 146 are arranged to correspond to each of the plurality of electrode pads in several (four in the present embodiment) of the plurality of components 30 to be measured formed on the measurement object 20. Here, a group of probes 146 corresponding to each of the plurality of electrode pads in one component 30 to be measured forms a probe unit. In the present embodiment, the jig 148 may be such a probe unit as an example. Therefore, the probe card 140 may have a plurality of jigs 148 corresponding to several of the plurality of components 30 to be measured. In this way, each of the plurality of jigs 148 may be constructed by functionally separated regions provided on one part (e.g., one probe card 140). As an alternative to the above, each of the plurality of jigs 148 may also be constructed by one independent part.

[0044] The probe tester 150 carries and places the measurement object 20 on the platform, and aligns the position of the measurement object 20 with the probe card 140. In addition, the probe tester 150 has a cleaning unit (not shown), which is used to clean the plurality of probes 146 in the fixture 148. When the test device 100 is electrically connected to the component 30 to be measured via the fixture 148, the surface of the electrode pad in the component 30 to be measured is scratched by the probe 146, that is, the contact is achieved by touchdown in an overdrive manner. At this time, oxides or garbage on the electrode pad will adhere to the tip of the probe 146. Therefore, the attachment will accumulate on the tip of the probe 146 with each touchdown, causing the contact resistance to gradually increase, so that the test device 100 will gradually fail to perform a correct test. Therefore, the test device 100 is provided with a cleaning unit on the probe tester 150, and the tip of the probe 146 is ground down by the probe stroke on the cleaning unit to clean the probe 146 and remove the attached matter accumulated on the tip.

[0045] The maintenance device 200 obtains and analyzes a plurality of test results, which are obtained when the test device 100 tests a plurality of different measured components 30 in sequence through a plurality of fixtures 148 in the measurement system 10. In addition, the test device 100 uses the analyzed information to determine the maintenance timing of the fixture 148. The maintenance device 200 can be a computer such as a PC (personal computer), a tablet computer, a smart phone, a workstation, a server computer, or a general-purpose computer, or a computer system in which a plurality of computers are connected together. Such a computer system is also a computer in a broad sense. In addition, the maintenance device 200 can also be constructed in the computer by one or more executable virtual computer environments. As an alternative to the above, the maintenance device 200 can be a dedicated computer designed for the maintenance of the fixture 148, or it can be dedicated hardware implemented by a dedicated circuit. As an example, the maintenance device 200 may be a web server connected to the Internet. In this case, the user can access the maintenance device 200 on the cloud from various environments that can be connected to the Internet and receive various services. In addition, the maintenance device 200 may be configured as a separate device that is connected to the measurement system 10 directly or via a network such as a local area network (LAN), or may be configured to be integrally formed with the measurement system 10 and implemented as a part of the functional block of the measurement system 10. For another example, in the case where multiple test results can be obtained by direct input from the user or a storage medium such as a USB (universal serial bus) memory, the maintenance device 200 may not be connected to the measurement system 10, but may be configured as a device independent of the measurement system 10.

[0046] The maintenance device 200 includes an input unit 210, an acquisition unit 220, a learning unit 230, a calculation unit 240, a determination unit 250, and an output unit 260. In addition, these blocks represent functional blocks and may not necessarily be consistent with the actual component configuration. In other words, even if it is represented as one block, it is not necessarily composed of only one component. Furthermore, even if it is represented as separate blocks, it is not necessarily composed of different components.

[0047] The input unit 210 is an interface unit for inputting a plurality of test results. The input unit 210 is connected to the tester body 110 of the test device 100, for example, directly or via a network, and receives input of a plurality of test results obtained by the test device 100. In addition, the input unit 210 may also be a user interface for receiving direct input from a user via a keyboard or a mouse, or may also be a component interface for connecting a USB memory or an optical disk drive to the maintenance device 200, and may receive input of a plurality of test results obtained by the test device 100 via such interfaces.

[0048] The acquisition unit 220 is connected to the input unit 210, and acquires a plurality of test results for each fixture 148. The plurality of test results are obtained by the test device 100 in the measurement system 10 sequentially testing a plurality of different components 30 to be measured through the plurality of fixtures 148. In addition, the acquisition unit 220 supplies the plurality of test results obtained in the normal state to the learning unit 230. In addition, the acquisition unit 220 supplies the plurality of test results obtained in the operating state to the calculation unit 240. In addition, the details of the test performed by the test device 100 will be described later.

[0049] The learning unit 230 uses the plurality of test results obtained in the normal state to determine the change in the normal state. Here, the change in the normal state may be an average of the changes in the test results of the plurality of jigs 148 in the normal state. In addition, the learning unit 230 uses the plurality of test results obtained in the normal state to learn a covariance matrix, which is used to calculate the Mahalanobis distance. This part will be described later. The learning unit 230 supplies the change in the normal state and the covariance matrix to the determination unit 250.

[0050] The calculation unit 240 uses a plurality of test results obtained in the operation state to calculate the variation of the test results for each jig 148. The calculation unit 240 supplies the variation of the test results for each jig 148 to the determination unit 250.

[0051] The decision unit 250 determines the maintenance timing of the jig 148 based on the change in the test result of each jig 148. At this time, the decision unit 250 may compare the change in the test result with the change in the normal state to calculate the difference, and determine the maintenance timing of the jig 148 based on the difference. The decision unit 250 supplies the maintenance timing of the jig 148 to the output unit 260.

[0052] The output unit 260 outputs the maintenance timing of the jig 148 determined by the determination unit 250. At this time, the output unit 260 may display and output the maintenance timing on a monitor, etc., or may output voice through a speaker, etc., or may output data to various memory components, etc.

[0053] Figure 2A detailed example of the probe card 140 of the present embodiment is shown. The upper part of the figure shows a side view of the probe card 140. In addition, the lower part of the figure shows a back view of the probe card 140. A plurality of probes 146 are arranged corresponding to each of a plurality of electrode pads in some of the plurality of measured components 30 formed on the measurement object 20. Furthermore, the plurality of probes 146 corresponding to each of the plurality of electrode pads in one measured component 30 are grouped (probe units) in a roughly square or rectangular shape, for example, to form a jig 148. The figure shows the following as an example: the probe card 140 has four jigs 148a~d in two rows in the X direction and two rows in the Y direction, and the jigs 148a~d are arranged in a square shape. Thus, the test device 100 can simultaneously test (simultaneously test) several (four in this embodiment) components 30 to be measured by contacting the probe card 140 with the measurement object 20. In addition, the number or arrangement of the jigs 148 of the probe card 140 is not limited to such an embodiment. The probe card 140 may have more or less than four jigs 148. Furthermore, the probe card 140 may also have a plurality of jigs 148 arranged in a shape different from a square, such as a straight line.

[0054] Figure 3 An example of bringing the probe card 140 of the present embodiment into contact with the measurement object 20 is shown. As shown in this figure, a plurality of components to be measured 30 are formed on the measurement object 20. Here, when a specific component to be measured 30 is to be distinguished for description, the component to be measured 30 (x, y) is represented using the coordinate x in the X direction and the coordinate y in the Y direction.

[0055] The test device 100, for example, uses Figure 2The test apparatus 100 uses the probe card 140 shown in the figure to test up to four components 30 to be measured at the same time. That is, the test apparatus 100 uses the probe tester 150 to align the positions of the probe card 140 and the measurement object 20, and makes the probe card 140 touch the measurement object 20 (the first time). In addition, the test apparatus 100 simultaneously performs the following tests: testing the component 30 (3, 1) to be measured via the fixture 148a, testing the component 30 (4, 1) to be measured via the fixture 148b, testing the component 30 (3, 2) to be measured via the fixture 148c, and testing the component 30 (4, 2) to be measured via the fixture 148d. In this way, the test apparatus 100 obtains the test result a1 of the component 30 (3, 1), the test result b1 of the component 30 (4, 1), the test result c1 of the component 30 (3, 2), and the test result d1 of the component 30 (4, 2). However, the component 30 (3, 1) to be measured is not actually formed on the measurement object 20. Therefore, the test device 100 obtains "no result" as the test result for the test result a1 of the component 30 (3, 1) to be measured. In this way, the test device 100 can obtain "no result" as the test result for the test result in the area where the component 30 to be measured does not exist.

[0056] Next, the test device 100 changes the relative position between the probe card 140 and the measurement object 20 by the probe tester 150, and makes the probe card 140 touch the measurement object 20 again (second time). In addition, the test device 100 simultaneously performs the following tests: testing the measured component 30 (5, 1) via the jig 148a, testing the measured component 30 (6, 1) via the jig 148b, testing the measured component 30 (5, 2) via the jig 148c, and testing the measured component 30 (6, 2) via the jig 148d. In this way, the test device 100 obtains the test result a2 of the measured component 30 (5, 1), the test result b2 of the measured component 30 (6, 1), the test result c2 of the measured component 30 (5, 2), and the test result d2 of the measured component 30 (6, 2).

[0057] Similarly, the test apparatus 100 changes the relative position between the probe card 140 and the measurement object 20 by the probe tester 150, and brings the probe card 140 down to contact the measurement object 20 again (the third time). Furthermore, the test apparatus 100 simultaneously performs the following tests: testing the component 30 (7, 1) to be measured via the jig 148a, testing the component 30 (8, 1) to be measured via the jig 148b, testing the component 30 (7, 2) to be measured via the jig 148c, and testing the component 30 (8, 2) to be measured via the jig 148d. Thus, the test apparatus 100 obtains the test result a3 of the component 30 (7, 1), the test result b3 of the component 30 (8, 1), the test result c3 of the component 30 (7, 2), and the test result d3 of the component 30 (8, 2).

[0058] The test device 100 tests different plural components 30 to be measured in sequence through plural jigs 148 by repeating such position alignment and touchdown multiple times. Furthermore, the test device 100 obtains plural test results obtained from the test of the plural components 30 to be measured, and stores them in association with information indicating which jig each of the plural test results was tested through. The test device 100 can output the plural test results stored in this way to the maintenance device 200 via wired or wireless means. The maintenance device 200 of this embodiment analyzes the plural test results obtained in this way and determines the maintenance timing of the jig 148.

[0059] Figure 4 An example of a process for the maintenance device 200 of the present embodiment to determine the maintenance timing of the fixture 148 is shown. The maintenance device 200, in step 410, obtains a plurality of test results in a normal state. Here, the plurality of test results in a normal state may be, for example, a plurality of test results obtained by sequentially testing a plurality of different components to be measured 30 through a plurality of normal fixtures 148. As an example, the test device 100 obtains the following test results: a plurality of test results obtained by sequentially testing all components to be measured 30 formed on a measurement object 20 (for example, a wafer) using a probe card 140 that has been cleaned or has not been used, and through a plurality of fixtures 148 in the probe card 140. Here, a measurement object 20 may be, for example, the initial (first) wafer when mass production of a new type of wafer is to begin. Furthermore, when the test apparatus 100 acquires a plurality of test results in a normal state, the probe card 140 may be cleaned again during the test of the one measurement object 20. The input unit 210 receives the input of the plurality of test results in a normal state acquired in this way and supplies the input to the acquisition unit 220.

[0060] That is, the acquisition unit 220 acquires the following test results respectively: a plurality of test results a1 to a32 obtained by sequentially performing tests on a normal (cleaned or unused) fixture 148a, a plurality of test results b1 to b32 obtained by sequentially performing tests on a normal fixture 148b, a plurality of test results c1 to c32 obtained by sequentially performing tests on a normal fixture 148c, and a plurality of test results d1 to d32 obtained by sequentially performing tests on a normal fixture 148d.

[0061] In addition, here, each of the plurality of test results, as an example, may indicate whether the test performed on the measured component 30 is passed, failed, or there is no test result. Generally speaking, the test device 100 performs tests on each of the measured components 30 for a plurality of test items corresponding to a plurality of categories, such as an open circuit and short circuit test, a DC test, a functional test, a frequency modulation (FM) test, and a USB test. Therefore, each of the plurality of test results, for example, may indicate whether the plurality of test items performed on each of the measured components 30 are all passed (P: overall passed), at least one item fails (F: overall failed), or there is no test result (N / A).

[0062] As an example, the acquisition unit 220 acquires the test results such as "a1 = N / A, a2 = P, a3 = P, a4 = F, a5 = N / A, a6 = P, ..., a32 = P" as a plurality of test results obtained by testing with the normal jig 148a. Similarly, the acquisition unit 220 acquires a plurality of test results b1 to b32 obtained by testing with the normal jig 148b, a plurality of test results c1 to c32 obtained by testing with the normal jig 148c, and a plurality of test results d1 to d32 obtained by testing with the normal jig 148d. The acquisition unit 220 supplies the plurality of test results in the normal state to the learning unit 230.

[0063] In step 420, the test device 100 determines the change in the normal state. Here, the change in the normal state can be the average of the change in the test results of the plurality of jigs 148 in the normal state. For example, the learning unit 230 analyzes the plurality of test results in a predetermined period among the plurality of test results obtained in the normal state while sequentially changing the starting point of the period, thereby determining the change in the normal state. As an example, the learning unit 230 uses the plurality of test results in the 20 touchdown periods as the plurality of test results in the predetermined period. In this case, the learning unit 230, for example, calculates the average overall failure rate of the plurality of test results a1 to a20, b1 to b20, c1 to c20, and d1 to d20 obtained by the 1st to 20th touchdowns at the starting point position m=1. That is, the learning unit 230 calculates the overall failure rate u1, which indicates how many overall failures there are among the plurality of test results a1 to a20, b1 to b20, c1 to c20, and d1 to d20 (excluding "no test results").

[0064] The learning unit 230 repeats the calculation while sequentially changing the starting point of the period. That is, the learning unit 230 then calculates the overall failure rate u2 of the plurality of test results a2 to a21, b2 to b21, c2 to c21, and d2 to d21 obtained by the second to 21st touchdowns by the same means as described above at the starting point position m=2. Similarly, the learning unit 230 calculates the overall failure rate u3 of the plurality of test results a3 to a22, b3 to b22, c3 to c22, and d3 to d22 obtained by the third to 22nd touchdowns at the starting point position m=3. In this way, the learning unit 230 determines, for example, the change u (u1, u2, u3, ..., un) in the normal state by sequentially changing the starting point of the period. The learning unit 230 supplies the determined change u in the normal state to the determination unit 250.

[0065] In step 430, the maintenance device 200 calculates a covariance matrix. For example, the learning unit 230 uses a plurality of test results obtained in a normal state to learn a covariance matrix for calculating the Mahalanobis distance. Generally speaking, the Mahalanobis distance relative to a group of values ​​represented by a multivariate vector x is defined as follows, where the mean vector of the multivariate vector x is u and the covariance matrix is ​​Σ. More specifically, the Mahalanobis distance quantitatively represents how much a certain data deviates from the mean, and represents the distance from the data group taking into account the degree of dispersion of the data in various directions. Therefore, when the multivariate vector x is consistent with the mean vector u, the Mahalanobis distance is 0, and when the multivariate vector x is not consistent with the mean vector u, the Mahalanobis distance is greater than 0.

[0066] [Mathematical formula 1]

[0067]

[0068] The learning unit 230 calculates the covariances and learns a matrix in which the covariances are arranged, namely the covariance matrix Σ, which is the average value of the product of the average deviation value between each variable. The learning unit 230 supplies the learned covariance matrix Σ to the decision unit 250.

[0069] In step 440, the maintenance device 200 obtains a plurality of test results in the operating state for each fixture. For example, the acquisition unit 220 obtains the following test results for each fixture 148: a plurality of test results when different plurality of measured components 30 are tested sequentially by the plurality of fixtures 148 in the operating state. That is, the acquisition unit 220 obtains the following test results: a plurality of test results a1 to a32 obtained by sequentially testing via the fixture 148a, a plurality of test results b1 to b32 obtained by sequentially testing via the fixture 148b, a plurality of test results c1 to c32 obtained by sequentially testing via the fixture 148c, and a plurality of test results d1 to d32 obtained by sequentially testing via the fixture 148d in the operating state. Here, each of the plurality of test results may indicate any of the following, similarly to the plurality of test results in the normal state: all the plurality of test items performed on each of the measured components 30 passed (overall passed), at least one item failed (overall failed), or there was no test result. In the operating state, the acquisition unit 220 sequentially acquires the plurality of test results and supplies them to the calculation unit 240.

[0070] In step 450, the maintenance device 200 calculates the test results in the operating state for each fixture 148. For example, the calculation unit 240 uses the multiple test results obtained in the operating state to calculate the change of the test results for each fixture 148. At this time, the calculation unit 240 can analyze the multiple test results in a predetermined period while sequentially changing the starting point of the period and calculate the change of the test results in the same way as the learning unit 230. In this way, the maintenance device 200 calculates the change of the test results with good sensitivity because it calculates the change of the test results based on the test results in a certain period instead of the accumulation of the test results. In addition, the calculation unit 240 can calculate the change of the failure rate in the multiple test results at this time. Here, the failure rate can be an overall failure rate indicating that at least one of the multiple test items has failed. Thus, the maintenance device 200 calculates the fluctuation of the test results using a relatively simple index such as whether the test is completely passed or completely failed, and thus the calculation burden of the fluctuation of the test results can be reduced.

[0071] As an example, the calculation unit 240 uses the plurality of test results derived during the 20 touchdowns as the plurality of test results in the predetermined period. In this case, the calculation unit 240 calculates the overall failure rate of the plurality of test results a1 to a20 obtained by the 1st to 20th touchdowns through the jig 148a at the starting position m=1. For example, the calculation unit 240 calculates the overall failure rate xa1, which indicates how many overall failures there are in the plurality of test results a1 to a20 (excluding "no test results").

[0072] The calculation unit 240 repeats the calculation while sequentially changing the starting point of the period. That is, the calculation unit 240 then calculates the overall failure rate xa2 of the plurality of test results a2 to a21 obtained through the jig 148a by the second to 21st touchdowns at the starting point position m=2 by the same method as described above. Similarly, the calculation unit 240 calculates the overall failure rate xa3 of the plurality of test results a3 to a22 obtained through the jig 148a by the third to 22nd touchdowns at the starting point position m=3. In this way, the calculation unit 240 calculates the changes xa(xa1, xa2, xa3, ..., xan) of the test results of the jig 148a by sequentially changing the starting point of the period.

[0073] The calculation unit 240 calculates the variation xb (xb1, xb2, xb3, ..., xbn) of the test result for the jig 148b, the variation xc (xc1, xc2, xc3, ..., xcn) of the test result for the jig 148c, and the variation xd (xd1, xd2, xd3, ..., xdn) of the test result for the jig 148d by the same means. In this way, the calculation unit 240 calculates the variation x of the test result for each jig 148 using the plurality of test results in the operating state. The calculation unit 240 supplies the calculated variation x (xa, xb, xc, xd) of the test result for each jig 148 to the determination unit 250.

[0074] In step 460, the maintenance device 200 calculates the difference from the variation u in the normal state. For example, the decision unit 250 may compare the variation x of the test result of each jig 148 with the variation u in the normal state to calculate the difference, and determine the maintenance timing of the jig 148 based on the difference. Here, the difference may be the Mahalanobis distance.

[0075] That is, the decision unit 250 substitutes the variation x(xa, xb, xc, xd) of the test results of each fixture 148 calculated in step 450 as the multivariate vector x in (Formula 1), substitutes the variation u(u1, u2, u3, ..., un) in the normal state calculated in step 420 as the average vector u, and uses the inverse matrix of the covariance matrix Σ learned in step 430 to calculate the Mahalanobis distance.

[0076] In step 470, the maintenance device 200 determines whether the difference calculated in step 460 exceeds a threshold value. For example, the decision unit 250 determines whether the Mahalanobis distance calculated in step 460 exceeds a predetermined threshold value. When it is determined that the Mahalanobis distance does not exceed the threshold value, the decision unit 250 determines that it is not necessary to maintain the jig 148 and advances the process to step 490. On the other hand, when it is determined that the Mahalanobis distance exceeds the threshold value, the decision unit 250 determines that it is necessary to maintain the jig 148 and advances the process to step 480. At this time, the decision unit 250 may determine the timing of maintaining the jig 148 after the number of times the Mahalanobis distance exceeds the threshold value reaches a predetermined number of times. In this way, the decision unit 250 determines the timing of maintaining the jig 148 based on the variation x of the test results of each jig 148. Here, the decision unit 250 may decide the cleaning period of the jig 148 as the maintenance period. However, the present invention is not limited to this type. The decision unit 250 may decide a maintenance period different from the cleaning period, such as the replacement period of the jig 148, as the maintenance period.

[0077] In step 480, the maintenance device 200 performs maintenance on the jig 148. As an example, the maintenance device 200 may clean the jig 148 as the maintenance of the jig 148. Once the maintenance device 200 performs the maintenance on the jig 148, the process proceeds to step 490.

[0078] In step 490, the maintenance device 200 determines whether the test of the measurement object 20 is completed. If it is determined that the test of the measurement object 20 is completed, the maintenance device 200 ends the processing. On the other hand, if it is determined that the test of the measurement object 20 is not completed, the maintenance device 200 returns the processing to step 440 and repeats the analysis while sequentially changing the starting point of the period.

[0079] Figure 5The figure shows the change of the test result of each jig 138 and the Mahalanobis distance calculated by the maintenance device 200 of the present embodiment. In this figure, the horizontal axis represents the starting point position m. In addition, in the upper part of this figure, the vertical axis represents the overall failure rate [%]. Therefore, the upper part of this figure shows the change of the test result (overall failure rate) of each jig 148 when the starting point position m is changed sequentially. In addition, in the lower part of this figure, the vertical axis represents the Mahalanobis distance. Therefore, the lower part of this figure shows the Mahalanobis distance calculated based on the change of the test result of each jig when the starting point position m is changed sequentially. As shown in this figure, the maintenance device 200 of this embodiment calculates the change of the test result for each of the plurality of jigs 148. That is, the maintenance device 200 will calculate the change of four test results when there are four jigs 148. However, the maintenance device 200 calculates the Mahalanobis distance based on the variation of the test results of each of the plurality of jigs 148, and determines the maintenance timing of the jig 148 based on the Mahalanobis distance. In this way, the maintenance device 200 determines the maintenance timing of the jig 148 based on only one index such as the Mahalanobis distance, so the maintenance timing of the jig 148 can be determined relatively simply.

[0080] Generally speaking, defective components to be measured 30 will occur evenly in the measurement object 20. In addition, the plurality of jigs 148 in the probe card 140 will be suddenly contaminated with a small probability. Therefore, when the plurality of jigs 148 test different plurality of components to be measured in sequence, the probability that the plurality of test results will fail at the same time is very small. The maintenance device 200 of this embodiment uses such actual conditions to determine the maintenance timing of the jig 148 based on the changes in the test results of each jig 148. Generally speaking, the cleaning of the jig 148 is carried out by lowering the jig onto a file and performing a needle measurement stroke. At this time, the front ends of the plurality of probes 146 in the jig 148 will gradually wear out. Therefore, the life of the probe card 140 is roughly determined by the number of times the cleaning is performed. According to the previous test device, the jig 148 is cleaned at a cycle predetermined by empirical rules or preliminary tests. However, the regular maintenance of the jig 148 may be excessive or insufficient for testing the measured component 30. For example, when the maintenance of the jig 148 is excessive, the life of the probe card 140 will be shortened faster than actually required. On the other hand, when the maintenance of the jig 148 is insufficient, the yield will be reduced as the contact resistance of the probe 146 increases. In this way, there is a trade-off between the life of the probe card 140 and the yield. In response to this, according to the maintenance device 200 of the present embodiment, because the maintenance timing of the jig 148 can be determined according to demand based on the changes in the test results of each jig 148, the timing of maintaining the jig 148 can be optimized. Thereby, according to the maintenance device 200 of the present embodiment, the reduction in yield can be suppressed and the life of the probe card 140 can be extended.

[0081] In addition, the above description only shows the case where the maintenance device 200 determines the maintenance of the jig 148 according to the demand, but it is not limited to such a type. The maintenance device 200, for example, can also regularly maintain the jig 148 as in the previous method, and further maintain the jig 148 according to the demand. That is, the maintenance device 200 can also maintain the jig 148 by combining regular maintenance and maintenance according to the demand.

[0082] Furthermore, the example shown in the above description is that the maintenance device 200 implements the determination of the change in the normal state and the learning of the covariance matrix based on the test results of the initial wafer when mass production of new wafer varieties is to begin. However, it is not limited to such a type. For example, the maintenance device 200 can also implement the determination of the change in the normal state and the learning of the covariance matrix based on the test results of the initial wafer during the learning period, and then in the operating state, if the test results of the measurement object 20 with a higher yield than the initial wafer are obtained, the change in the normal state and the covariance matrix are updated based on the test results.

[0083] Furthermore, the example shown in the above description is that the maintenance device 200 performs the determination of the change in the normal state and the learning of the covariance matrix based on the test results of one measurement object 20 (one wafer), but it is not limited to such a type. The maintenance device 200 can also perform the determination of the change in the normal state and the learning of the covariance matrix based on the statistics (e.g., average) of the test results of a plurality of measurement objects 20.

[0084] In the above description, the maintenance device 200 learns the covariance matrix and uses the Mahalanobis distance as the difference between the variation of the test result of each jig 148 and the variation in the normal state, but the present invention is not limited to this type. For example, the maintenance device 200 may input a plurality of test results in the normal state as teaching data into a learning model such as a neural network, and determine the maintenance timing of the jig 148 based on the difference between the variation of the test result of each jig 148 and the variation in the output of the learning model.

[0085] Furthermore, the maintenance device 200 may use the Euclidean distance instead of the Mahalanobis distance as the difference. In this case, the determination unit 250 may, for example, calculate the difference by subtracting the change u in the normal state from the change x in the test result for each of the components and taking the absolute value, thereby calculating the sum of these values ​​as the difference.

[0086] In the above description, the case where the Mahalanobis distance is used is shown as an example in which the decision unit 250 calculates the difference by machine learning. However, the decision unit 250 may calculate the difference by other machine learning algorithms.

[0087] Figure 6 A maintenance device 200 according to a modified example of the present embodiment is shown. Figure 6 In the Figure 1 The same reference numerals are given to components having the same functions and configurations, and descriptions other than the differences will be omitted below. The maintenance device 200 of this variation includes a model generation unit 610 instead of the learning unit 230 .

[0088] The model generation unit 610 generates a model by machine learning, and the model detects the deviation from the change u in the normal state based on the change x of the test result. As an example, the model generation unit 610 can generate a model by an unsupervised anomaly detection algorithm. For example, the model generation unit 610 uses a plurality of test results obtained during the learning period as learning data to calculate the characteristics of the statistical distribution of the change u in the normal state. In addition, the model generation unit 610 generates a model based on the calculated statistical distribution characteristics, so that when the change x of the test result is input into the model, the model can detect the deviation from the change u in the normal state. In this way, the model generation unit 610 generates a model by unsupervised learning, so that the annotation work used to create teaching data can be omitted, and the model can be generated more simply. However, it is not limited to such an implementation. The model generation unit 610 can also generate a model by various learning algorithms such as a supervised learning algorithm, a semi-supervised learning algorithm, and a reinforcement learning algorithm. The model generation unit 610 supplies the generated model to the determination unit 250 .

[0089] The determination unit 250 calculates the difference based on the output of the model supplied from the model generation unit 610. For example, the determination unit 250 inputs the variation x of the test result of each jig 148 calculated by the calculation unit 240 into the model generated by the model generation unit 610, and calculates the output of the model as the difference.

[0090] In this way, the maintenance device 200 of this modification calculates the difference based on the output of the model generated by machine learning. Thus, according to the maintenance device 200 of this modification, it is possible to realize the determination of the maintenance timing of the jig 148 by various machine learning algorithms.

[0091] Furthermore, as described above, when the calculated difference exceeds a predetermined threshold value, the decision unit 250 decides to perform maintenance on the fixture 148. At this time, the predetermined threshold value may be a fixed value pre-set by the operator based on a rule of thumb, but is not limited to this embodiment. As an example, the predetermined threshold value may also be determined by machine learning. For example, the maintenance device 200 may also learn the threshold value used to determine the timing of maintenance of the fixture 148 by machine learning. At this time, the maintenance device 200 obtains the performance data of the maintenance of the fixture 148, such as data indicating at which timing (time point) what kind of cleaning was performed. In addition, the maintenance device 200 uses the data of the difference calculated by the decision unit 250 and the performance data as learning data to learn the relationship between the difference and the maintenance performance. Thus, the maintenance device 200 determines the threshold value by machine learning, and the threshold value is used to determine the extent of the difference when the maintenance of the jig 148 is to be performed, that is, to determine the maintenance timing. Thus, according to the maintenance device 200, the threshold value used to determine the maintenance timing can be set to an optimal value based on actual performance without human effort.

[0092] Furthermore, in the example shown in the above description, the failure rate used as the test result is an overall failure rate indicating the failure of at least one item (also called a bin or category) among a plurality of test items, but it is not limited to such an implementation. The failure rate may also be a partial failure rate indicating the failure of a specific item among a plurality of test items. That is, the maintenance device 200 may also use a partial failure rate as the test result, which indicates that a specific item has failed, regardless of whether other items among the plurality of test items have failed. Here, such a specific item may be one item or a plurality of items. When there are a plurality of specific items, the maintenance device 200 may use the logical sum of the test results in each item as the partial failure rate. In this way, according to the maintenance device 200, either the overall failure rate or the partial failure rate can be used as the test result, and it is possible to respond flexibly to various situations.

[0093] Furthermore, in the example shown in the above description, the determination unit 250 determines the cleaning period of the jig 148 as a maintenance opportunity. However, in addition to this, the determination unit 250 can also further determine the cleaning content of the jig 148. There are various types of cleaning of the jig 148, such as cleaning with a brush, cleaning with a polishing sheet, cleaning with a cleaning wafer, etc. Furthermore, in addition to the cleaning that can be completed in the device as described above, there is also cleaning that is applied to the outside of the device, such as removing the probe card and cleaning it with other devices. The determination unit 250 can determine the most suitable cleaning content from such a plurality of cleaning methods.

[0094] At this time, the decision unit 250 can determine the cleaning content of the jig 148 based on the change in the test results of each of the multiple test items. For example, when the difference in the change of the failure rate in the open-circuit short-circuit test exceeds a predetermined threshold value, the reason is that the possibility of the needle tip being dirty is high, so the decision unit 250 can decide to implement cleaning using a polishing sheet. Similarly, when the difference in the change of the failure rate in the leakage test (a part of the DC test) exceeds a predetermined threshold value, the reason is that the possibility of a short circuit between the pins is high, so the decision unit 250 can decide to implement cleaning using a brush. In this way, the maintenance device 200 can determine the most suitable cleaning content according to which test item determines that the jig 148 should be cleaned.

[0095] Furthermore, in the example shown in the above description, when obtaining multiple test results in the normal state, the first (first) wafer at the beginning of mass production of the new variety of wafers is used as the measurement object to obtain multiple test results. However, the finished product level of the wafer itself is different in each wafer, and each wafer may have inherent characteristics. Such inherent characteristics usually do not have much deviation within the same batch, but the deviation between batches may become significant. Therefore, the maintenance device 200 can also obtain multiple test results in the normal state by using the head wafer of the batch as the measurement object through the normal fixture 148 after each batch of the measurement object is switched. That is, the maintenance device 200 can update the change u in the normal state that will become the difference reference after each batch is switched. In this way, the maintenance device 200 can distinguish whether the reason for the increase in the difference is on the fixture 148 side or the wafer side.

[0096] Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where a block may represent (1) a stage of a process for performing an operation or (2) a section of a device responsible for performing an operation. Specific stages and sections may be implemented by dedicated circuits, programmable circuits provided together with computer-readable instructions stored on a computer-readable medium, and / or processors provided together with computer-readable instructions stored on a computer-readable medium. Dedicated circuits may include digital and / or analog hardware circuits, and may also include integrated circuits (ICs) and / or discrete circuits. Programmable circuits may include reconfigurable hardware circuits, which may include AND logic gates, OR logic gates, XOR logic gates, NAND logic gates, NOR logic gates, and other logical operations, flip-flops, caches, field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and other memory elements.

[0097] Computer-readable media may include any physical component capable of storing instructions that can be executed by appropriate components. As a result, a computer-readable medium having instructions stored therein will have a product and contain instructions that can be executed in order to generate means for performing the operations specified in the flowchart or block diagram. Examples of computer-readable media include: electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, etc. More specific examples of computer-readable media include: floppy disks (registered trademarks), magnetic disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disks (CD-ROM), digital versatile disks (DVD), blue-ray (RTM) disks, MS memory cards, integrated circuit cards, etc.

[0098] Computer-readable instructions include: assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microprogram code, firmware instructions, state setting data, or object-oriented programming languages ​​such as Smalltalk, JAVA (registered trademark), C++, and previous procedural programming languages ​​such as "C" programming language or equivalent programming languages; it may also include any source code or object code written in any combination of one or more programming languages.

[0099] Computer readable instructions, for a processor or programmable circuit of a general purpose computer, a special purpose computer or other programmable data processing device, can be provided via a local or local area network (LAN), a wide area network (WAN) such as the Internet, and executed in order to generate means for performing the operations specified in the flowchart or block diagram. Examples of processors include: computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc.

[0100] Figure 7 An example of a computer 2200 that can implement the whole or part of the multiple aspects of the present invention is shown. The program installed in the computer 2200 can make the computer 2200 function as an operation associated with the device of the embodiment of the present invention, or one or more sections of the device, or can execute the operation or one or more sections of the device, and / or can make the computer 2200 execute the process of the present embodiment or the stage of the process. Such a program can be executed by the CPU 2212 to make the computer 2200 perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described in this specification.

[0101] The computer 2200 according to the present embodiment includes a CPU 2212, a RAM 2214, a graphic controller 2216, and a display device 2218, and these components are connected to each other through a host controller 2210. The computer 2200 also includes input / output units such as a communication interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, and these components are connected to the host controller 2210 via an input / output controller 2220. The computer also includes conventional input / output units such as a ROM 2230 and a keyboard 2242, and these components are connected to the input / output controller 2220 via an input / output chip 2240.

[0102] The CPU 2212 operates according to the programs stored in the ROM 2230 and the RAM 2214 to control each unit. The graphic controller 2216 obtains the image data provided in the frame buffer in the RAM 2214 or generated therein by the CPU 2212, and displays the image data on the display device 2218.

[0103] The communication interface 2222 communicates with other electronic components via a network. The hard disk drive 2224 stores programs and data to be used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0104] The ROM 2230 stores therein a boot program to be executed by the computer 2200 at startup, and / or a program dependent on the hardware of the computer 2200. The input / output chip 2240 can also connect various input / output units to the input / output controller 2220 via a parallel port, a serial port, a keyboard port, a mouse port, and the like.

[0105] The program is provided via a computer-readable medium such as a DVD-ROM 2201 or an IC card. The program is read from the computer-readable medium and installed in a hard disk drive 2224, a RAM 2214, or a ROM 2230, which can also be an example of a computer-readable medium, and is executed by the CPU 2212. The information processing written in the program is read by the computer 2200 to achieve cooperation between the program and the various types of hardware resources described above. The device or method can be constructed by realizing the operation or processing of information by using the computer 2200.

[0106] For example, in the case where communication is performed between the computer 2200 and an external component, the CPU 2212 may execute a communication program read into the RAM 2214, and based on the processing written in the communication program, issue a communication processing instruction to the communication interface 2222. The communication interface 2222, under the control of the CPU 2212, reads the transmission data stored in the transmission buffer processing area and transmits the read transmission data to the network, or writes the reception data received from the network into the reception buffer processing area, etc., wherein the transmission buffer processing area and the reception buffer processing area are provided on a storage medium such as the RAM 2214, the hard disk drive 2224, the DVD-ROM 2201, or the IC card.

[0107] Furthermore, the CPU 2212 can read the entirety or a necessary portion of a file or database stored in an external storage medium such as the hard disk drive 2224, the DVD-ROM drive 2226 (DVD-ROM 2201), or an IC card into the RAM 2214, and perform various types of processing on the data on the RAM 2214. The CPU 2212 then writes the processed data back to the external storage medium.

[0108] Various types of information such as various types of programs, data, tables, and databases can be stored in the storage medium and subjected to information processing. The CPU 2212 can perform various types of processing described in various places of the present disclosure on the data read from the RAM 2214, and write the results back to the RAM 2214. The above processing includes various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc. specified by the instruction sequence of the program. In addition, the CPU 2212 can search for information in files, databases, etc. in the storage medium. For example, in the case where a plurality of entries are stored in the storage medium, and the entries have attribute values ​​of the first attribute that are associated with the attribute value of the second attribute, respectively, the CPU 2212 can search for an entry that matches the attribute value condition of the specified first attribute from the plurality of entries, and read the attribute value of the second attribute stored in the entry, thereby obtaining the attribute value of the second attribute that is associated with the first attribute that meets the predetermined condition.

[0109] The programs or software modules described above may be stored in a computer-readable medium on or near the computer 2200. In addition, a storage medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet may be used as a computer-readable medium, thereby providing the program to the computer 2200 via the network.

[0110] The present invention has been described above using the embodiments, but the technical scope of the present invention is not limited to the scope described in the aforementioned embodiments. It is clear to those skilled in the art that various changes or improvements may be made to the aforementioned embodiments. It is clear from the description of the scope of the claims that the embodiments obtained by adding the changes or improvements will also be included in the technical scope of the present invention.

[0111] It should be noted that the execution order of each process such as actions, steps, steps, and stages in the devices, systems, programs, and methods shown in the claims, specifications, and drawings can be implemented in any order unless it is specifically stated as "before", "before", etc., and the output of the pre-processing is not used in the post-processing. Even if the action flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, it does not mean that it must be implemented in this order.

[0112] 10: Measurement system

[0113] 20: Measurement object

[0114] 30: Component being measured

[0115] 100:Test device

[0116] 110: Tester body

[0117] 120: Test head

[0118] 130: Performance Board

[0119] 140:Probe Card

[0120] 142:Substrate

[0121] 144:Probe housing

[0122] 146: Probe

[0123] 148:Jig

[0124] 150: Needle test machine

[0125] 200:Maintenance device

[0126] 210: Input unit

[0127] 220: Acquisition

[0128] 230: Learning Department

[0129] 240: Calculation Department

[0130] 250: Decision Department

[0131] 260: Output unit

[0132] 610: Model generation unit

[0133] 2200: Computer

[0134] 2201:DVD-ROM

[0135] 2210:Host Controller

[0136] 2212:CPU

[0137] 2214:RAM

[0138] 2216: Graphics Controller

[0139] 2218: Display Devices

[0140] 2220: Input and output controller

[0141] 2222: Communication interface

[0142] 2224: Hard Drive

[0143] 2226:DVD-ROM drive

[0144] 2230:ROM

[0145] 2240: Input and output chip

[0146] 2242:Keyboard

Claims

1. A maintenance device comprising: An acquisition unit that acquires, for each of the plurality of fixtures, a plurality of test results when a plurality of different components to be measured are tested sequentially by each of the plurality of fixtures; a calculation unit that uses the plurality of test results to calculate a change in the plurality of test results for each of the jigs; and A determination unit determines a maintenance timing of the jig based on a change in the plurality of test results.

2. The maintenance device according to claim 1, wherein: The determination unit compares the variation of the plurality of test results with the variation in a normal state to calculate a difference, and determines the maintenance timing based on the difference.

3. The maintenance device according to claim 2, wherein: The variation in the normal state is an average of the variation in the test results of the plurality of jigs in the normal state.

4. The maintenance device according to claim 2 or 3, wherein: The determination unit determines to perform maintenance on the jig when the difference exceeds a predetermined threshold value.

5. The maintenance device according to claim 4, wherein: The aforementioned predetermined threshold value is determined by machine learning.

6. The maintenance device according to any one of claims 2 to 3, wherein: The determination unit calculates the difference by machine learning.

7. The maintenance device according to claim 6, wherein: The aforementioned difference is the Mahalanobis distance.

8. The maintenance device according to claim 7, wherein: The system further includes a learning unit that uses a plurality of test results obtained in a normal state to learn a covariance matrix for calculating the aforementioned Mahalanobis distance.

9. The maintenance device according to claim 6, wherein: A model generation unit is further provided, the model generation unit generating a model by machine learning, the model detecting a deviation from a change in a normal state based on a change in the plurality of test results; Furthermore, the determination unit calculates the difference based on the output of the model.

10. The maintenance device according to any one of claims 1 to 3, wherein: The calculation unit calculates a change in the plurality of test results by analyzing the plurality of test results in a predetermined period while sequentially changing a starting point of the period.

11. The maintenance device according to any one of claims 1 to 3, wherein: The calculation unit calculates a change in the failure rate among the plurality of test results.

12. The maintenance device according to claim 11, wherein: The aforementioned failure rate is an overall failure rate, which means that at least one item among a plurality of test items failed.

13. The maintenance device according to claim 11, wherein: The aforementioned failure rate is a partial failure rate, which indicates the failure of a specific item among a plurality of test items.

14. The maintenance device according to any one of claims 1 to 3, wherein: The determination unit determines a cleaning period of the jig as the maintenance opportunity.

15. The maintenance device according to claim 14, wherein: The determination unit further determines the cleaning content of the jig.

16. The maintenance device according to claim 15, wherein: The determination unit determines the cleaning content of the jig based on changes in the plurality of test results for each of the plurality of test items.

17. A maintenance method comprising the following steps: A plurality of test results when a plurality of different components to be measured are tested sequentially by each of the plurality of fixtures are obtained for each of the aforementioned fixtures; Using the plurality of test results, calculating the variation of the plurality of test results for each of the fixtures; and, The maintenance timing of the jig is determined based on the changes in the plurality of test results.

18. A storage medium storing a maintenance program, wherein the maintenance program is executed by a computer and causes the computer to function as the following means: An acquisition unit that acquires, for each of the plurality of fixtures, a plurality of test results when a plurality of different components to be measured are tested sequentially by each of the plurality of fixtures; a calculation unit that uses the plurality of test results to calculate a change in the plurality of test results for each of the jigs; and A determination unit determines a maintenance timing of the jig based on a change in the plurality of test results.

Images