JP2011101908A - Cutting device and cutting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a cutting method capable of grasping the cutting state of a workpiece with high accuracy in real time. <P>SOLUTION: The cutting method cuts the workpiece 10. The cutting method includes a process for generating reference data based on a first output signal from an acceleration sensor 25 mounted in a spindle 20 when performing a test cut for the workpiece 10, a process for generating actual cut data based on a second output signal from the acceleration sensor 25 when performing an actual cut for the workpiece 10, a process for deciding whether or not the actual cut data are deviated from the range of the reference data, and a process for outputting an alarm when the actual cut data are deviated from the range of the reference data. The reference data and the actual cut data are generated by performing waveform processing using a high-pass filter for each of the first output signal and the second output signal from the acceleration sensor and doubly integrating the signals obtained by the waveform processing. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a cutting apparatus and a cutting method for cutting a workpiece.

  2. Description of the Related Art Conventionally, a cutting apparatus provided with a dicing blade has been used to cut a workpiece sealed with a resin by mounting a semiconductor element on a lead frame. In recent years, various materials have been used as a resin sealing member of a workpiece, and for example, it is necessary to cut a fragile material or a composite material. For this reason, in order to ensure the quality of the workpiece after cutting, it is important to evaluate the cutting surface (cutting evaluation) for each type of workpiece (resin sealing member).

  Conventionally, in the cutting evaluation of such a workpiece, a test cut is performed when the first workpiece is cut, and the cutting surface is evaluated by image processing using a camera or visual observation with a microscope by an operator. When it is determined that the workpiece is non-defective in the test cut, the distance that can be cut with the dicing blade (that is, the number that can be processed) is estimated from experience based on the state of the cutting surface.

  On the other hand, Patent Document 1 discloses an automatic blade replacement system in which a vibration meter is provided on an outer cylinder of a spindle. This automatic blade replacement system is configured to rotate the spindle after replacing the blade assembly, detect the vibration state with a vibrometer, and issue an alarm when the vibration exceeds a predetermined level.

Japanese Patent Laid-Open No. 10-340867

  However, as in the past, in the evaluation method in which a test cut is performed at the time of the first workpiece cutting and the cutting surface is evaluated by image processing using a camera or visual observation with a microscope by an operator, the wear state of the dicing blade is changed. It is difficult to respond quickly (in real time).

  Patent Document 1 discloses a configuration in which a vibration is detected by using a vibrometer and a warning is issued. Specifically, any method is used to issue a warning based on the spindle vibration state. Neither is it disclosed, nor is it disclosed a configuration for detecting a vibration state with high accuracy.

  Therefore, the present invention provides a cutting method and a cutting apparatus capable of grasping a cutting state of a workpiece with high accuracy in real time.

  A cutting method according to an aspect of the present invention is a cutting method for cutting a workpiece, and when performing a test cut of the workpiece, reference data is obtained based on a first output signal from an acceleration sensor attached to a spindle. A step of generating actual cut data based on a second output signal from the acceleration sensor when the actual cut of the workpiece is performed, and whether the actual cut data is out of a range of the reference data And a step of outputting an alarm when the actual cut data is out of a range of the reference data, and the reference data and the actual cut data are obtained from the acceleration sensor. Waveform processing using a high-pass filter is performed on each of the first output signal and the second output signal, and the signal obtained by the waveform processing is double-integrated. Produced by.

  A cutting apparatus according to another aspect of the present invention is a cutting apparatus that cuts a workpiece, and includes a blade that cuts the workpiece, a spindle that includes a motor that rotates the blade, and an acceleration that measures acceleration of the spindle. A sensor and a control unit that controls to output an alarm based on an output signal from the acceleration sensor, and the control unit outputs a first output from the acceleration sensor when performing a test cut of the workpiece. Reference data is generated based on the signal, and actual cut data is generated based on the second output signal from the acceleration sensor when the workpiece is actually cut, and the actual cut data is out of the range of the reference data. And control to output the alarm when the actual cut data is out of the range of the reference data. And the actual cut data is obtained by subjecting each of the first output signal and the second output signal from the acceleration sensor to waveform processing using a high-pass filter, and double-doubling the signal obtained by the waveform processing. Generated by integrating.

  Other objects and features of the present invention are illustrated in the following examples.

  ADVANTAGE OF THE INVENTION According to this invention, the cutting device and cutting method which can grasp | ascertain the cutting condition of a workpiece | work in real time with high precision can be provided.

It is a schematic block diagram of the cutting device in a present Example. It is a flow of the cutting method in a present Example. It is a flow of the test cut process of the cutting method in a present Example. It is an example of the acceleration waveform data acquired with the cutting method in a present Example. It is an example of the FFT conversion result of the acceleration waveform data in a present Example. It is an example of the double integral data in a present Example. It is an example of the double integral data in a present Example. It is an example of the defect determination performed with the cutting method in a present Example. It is a flow of the actual cutting process of the cutting method in a present Example.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the description of each figure, overlapping description is omitted.

  First, the cutting apparatus in a present Example is demonstrated with reference to FIG. FIG. 1 is a schematic configuration diagram of a cutting apparatus 100 in the present embodiment. In FIG. 1, reference numeral 10 denotes a work (cutting object) in which a plurality of semiconductor elements (IC chips) are mounted on a lead frame and resin-sealed. As will be described later, the workpiece 10 is cut (cut) by the cutting device 100 into individual pieces. In this embodiment, the workpiece 10 is cut in the X-axis direction and the Y-axis direction to produce 24 semiconductor packages (individualized workpieces). As such a semiconductor package, for example, a collective sealing BGA (ball grid array) package can be cited.

  However, the present embodiment is not limited to this, and the cutting apparatus 100 can be separated into semiconductor packages other than the above number. The cutting device 100 is used not only for cutting the workpiece 10 into individual pieces, but also for cutting the workpiece 10 (making a cut at a predetermined portion of the workpiece 10). Furthermore, the cutting apparatus 100 is not limited to a batch-sealed BGA (ball grid array) package for the workpiece 10 provided with semiconductor elements, and for example, a plurality of elements such as an LED package provided with light-emitting elements are provided. A package that is collectively sealed, or WLP (Wafer Level Package) that is obtained by directly resin-sealing a semiconductor device in a wafer state can be widely applied.

  In the cutting apparatus 100, 20 is a spindle. The spindle 20 includes a spindle motor 21 (driving means), and is configured to be rotatable by the driving means. Reference numeral 22 denotes a dicing blade (cutting blade: hereinafter simply referred to as “blade”). The blade 22 is attached to the tip of the spindle 20. The blade 22 rotates with the rotation of the spindle 20 and can cut the workpiece 10 placed on the table 23. The spindle 20 is attached to the apparatus frame with a support portion provided close to the blade 22. The spindle 20 is configured to be movable in the X-axis direction (left-right direction in FIG. 1) by a positioning motor (not shown), and can change the cutting position of the workpiece 10 in the X-axis direction.

  The table 23 is configured to be capable of being moved in the Y-axis direction (vertical direction in FIG. 1) by being driven by a table motor 24, and further configured to be rotatable in the XY plane. For this reason, the cutting direction of the workpiece 10 placed on the table 23 can be arbitrarily set. The rotational position by the table 23 is controlled based on position information obtained from an imaging device (not shown). The imaging device is configured to be able to capture an image of the cutting position of the workpiece 10 and configured to be output to the display unit 60 via the control unit 40.

  Reference numeral 25 denotes an acceleration sensor. The acceleration sensor 25 is attached to the spindle 20 and measures spindle vibration (acceleration) generated during cutting of the workpiece 10. The vibration (acceleration) measured by the acceleration sensor 25 includes vibration caused by resistance during cutting of the workpiece, in addition to vibration caused by rotation of the motors 21 and 24. By providing the spindle 20 with the acceleration sensor 25 as in the cutting apparatus 100 of the present embodiment, for example, it is possible to detect the deterioration of the blade 22 during workpiece cutting and the initial abnormality of the blade 22 when the blade 22 is replaced. it can. For this reason, it is possible to prevent occurrence of cutting defects by the blade 22 and provide a high-quality semiconductor package.

  In the present embodiment, the acceleration sensor 25 is a three-axis acceleration sensor that measures acceleration in the three-axis directions of the XYZ axes, but is not limited thereto. At the time of cutting the workpiece 10, the change in vibration caused by the resistance during cutting is relatively larger in the cutting direction (Y-axis direction), in other words, the resistance in the feed direction than the resistance in the other axial direction. Since it is easy to detect, for example, an acceleration sensor 25 that measures only acceleration in the Y-axis direction may be used. In addition, the acceleration sensor 25 is preferably arranged apart from the support portion that supports the tip portion to which the blade 22 is attached, in order to amplify the amplitude of vibration received by the spindle 20 during workpiece cutting and effectively detect it. . For example, the acceleration sensor 25 of the present embodiment is provided at an end portion on the opposite side to the tip portion to which the blade 22 is attached. However, the present invention is not limited to this, and the acceleration sensor 25 may be provided in the vicinity of the blade 22 as long as the vibration of the spindle 20 can be detected.

  The acceleration (output signal from the acceleration sensor 25) detected by the acceleration sensor 25 is input to the A / D converter 30. The A / D converter 30 converts the output signal of the acceleration sensor 25, which is an analog signal, into a digital signal for performing various processes in the subsequent stage. Specifically, the A / D converter 30 obtains a digital signal by sampling an analog signal from the acceleration sensor 25 at regular intervals. The digital signal output from the A / D converter 30 is input to the control unit 40.

  The controller 40 performs various signal waveform processes on the input digital signal as will be described later. The control unit 40 performs feedback control based on the detection value (output signal) of the acceleration sensor 25. When the actual cut data (displacement amount of the spindle 20) obtained by the double integration of the acceleration of the spindle 20 deviates from the reference data Ds (see FIG. 7) obtained at the time of the test cut. The alarm unit 70 is controlled to output an alarm (when a predetermined threshold value is exceeded). For example, the control unit 40 can also control the spindle 20 to stop rotating when the acceleration detected by the acceleration sensor 25 exceeds the predetermined threshold. However, the present embodiment is not limited to this, and when a predetermined threshold value is exceeded, for example, the motor rotation speed of the spindle 20 is lowered to control to reduce the resistance at the time of workpiece cutting, and the cutting Generation of defects due to interruption may be prevented.

  The storage unit 50 is configured by, for example, a semiconductor memory or a hard disk. The storage unit 50 exchanges information with the control unit 40 while the control unit 40 performs various signal waveform processing. For example, as will be described later, generated data such as reference data Ds and actual cut data is sequentially stored. Moreover, since the memory | storage part 50 is comprised so that the detection value of the acceleration sensor 25 may be recorded as a log, the investigation after workpiece | work cutting is attained. In addition, numerical analysis such as FFT analysis or statistical analysis can be performed on the information stored in the storage unit 50.

  The display unit 60 displays information on triaxial acceleration (vibration) during workpiece cutting. Further, the display unit 60 includes, for example, each cutting condition information such as a cutting state of the workpiece 10, a cutting position of the workpiece 10, a rotation speed of the spindle 20, a load current, a cutting speed, and a cutting acceleration, and a signal processing result to be described later. Are displayed in batch on a multi-screen. By adopting such a display unit 60, it is possible to grasp the cutting state of the workpiece 10 and the like in real time, and to easily grasp the change point during the workpiece cutting.

  The alarm unit 70 is configured to issue an alarm when the actual cut data deviates from the reference data Ds. As the threshold value for operating the alarm unit 70, for example, the same value as the threshold value for stopping the spindle 20 can be used. In this case, the control unit 40 operates the alarm unit 70 and stops the motor of the spindle 20 at the same time. However, the present embodiment is not limited to this, and the threshold for operating the alarm unit 70 may be set lower than the threshold for stopping the spindle 20. In this case, after operating the alarm unit 70, the control unit 40 stops the spindle 20 only when the displacement amount of the spindle 20 further increases. Further, the control unit 40 is configured to cause an operator (operator) to determine when to stop the spindle 20 only by operating the alarm unit 70 when a predetermined threshold value is exceeded. Also good. In this case, the operator determines whether or not to continue the rotation operation of the spindle 20.

  Next, with reference to FIG. 2 thru | or FIG. 9, the cutting method in a present Example is demonstrated. In addition, the cutting method demonstrated in a present Example is performed based on the instruction | command of the control part 40 of the cutting apparatus 100 except for the one part process performed by an operator. FIG. 2 is a flow of a cutting method (cutting) in the present embodiment. First, in the cutting process of the present embodiment, a test cut is performed in step S101. The test cutting process is performed before starting the cutting of the first workpiece 10. FIG. 3 is a flow of the test cut process (step S101) in the present embodiment.

  In the test cut process, first, in step S201, acceleration waveform data of a test cut for cutting the test workpiece 10 under the same cutting conditions as the product workpiece 10 is acquired. Using the acceleration sensor 25 of the cutting apparatus 100, the acceleration generated in the spindle 20 is measured for each of the XYZ axes, and the waveform data (first output signal from the acceleration sensor 25) is acquired. In this case, the operation of moving the blade 22 to the cutting start position of the dicing line in the workpiece 10 and the operation of cutting the workpiece 10 with the blade 22 in the dicing line are repeated. Thereby, a series of waveform data at the time of performing a series of cutting in a plurality of dicing lines similar to the actual cutting process is acquired. The acquired waveform data is subjected to fast Fourier transform (FFT transform), and the frequency components contained in the waveform data can be visualized.

FIG. 4 is an example of acceleration waveform data. In FIG. 4, the vertical axis represents the acceleration (m / s ² ) of the spindle 20, and the horizontal axis represents time (Sec). Further, a section C in the figure corresponds to a period during which the workpiece 10 is being cut (the same applies to FIGS. 6 to 8). This section C can be set as appropriate based on the image captured by the imaging device, the cutting sound, the cutting operation program, and the like. In an acceleration waveform that is not subjected to signal waveform processing as will be described later, the section C has a waveform like white noise that is not numerically different from other sections. For this reason, it is impossible to detect the peak of the displacement amount that changes in accordance with the change in the cutting resistance, and it is impossible to grasp the cutting state of the workpiece.

  FIG. 5 shows the FFT conversion result of the acceleration waveform data at the time of the test cut. In FIG. 5, the vertical axis represents the spectral intensity, and the horizontal axis represents the frequency. As shown in FIG. 5, the acceleration waveform data includes many frequency components of 100 Hz or less and frequency components of around 500 Hz. In the present embodiment, a waveform when the spindle motor 21 is driven at a rotation speed of 30000 rpm, that is, a frequency of 500 Hz is targeted. For this reason, many frequency components around 500 Hz corresponding to the motor rotation speed of the spindle 20 are generated.

  On the other hand, in FIG. 5, the spectrum is thickly distributed below 50 Hz. This corresponds to a frequency component (rotational frequency of the table motor 24) generated by the movement of the table 23 on which the workpiece 10 is placed during cutting. Specifically, the maximum number of times of the table motor 24 when the table 23 on which the workpiece 10 is placed moves is 3000 rpm, that is, operates at a maximum frequency of 50 Hz. When the workpiece 10 is actually cut, the moving speed and the cutting speed are set to optimum values according to the machining conditions with the maximum frequency as an upper limit. The vibration caused by the movement of the table 23 (that is, the rotation of the table motor 24) is not the vibration that occurs when the workpiece 10 is cut and the displacement (amplitude) changes according to the cutting resistance. In order to detect only the vibration generated by the cutting of the workpiece 10, the low-frequency component having a large amplitude is preferentially removed, and thus the frequency appearing according to the actual moving speed of 50 Hz or less is removed. Is preferred.

  Therefore, in this embodiment, signal waveform processing using a high-pass filter (filter circuit) is performed in step S202. In this embodiment, the signal waveform processing is performed by changing the frequency setting of the high-pass filter by the digital signal processing of the control unit 40. The high-pass filter is a filter circuit that removes low-frequency components and passes only high-frequency components. In this embodiment, in order to remove the frequency component corresponding to the rotation of the table motor 24 at 50 Hz or less, the signal waveform of the test cut is subjected to waveform processing using a high-pass filter that passes a frequency higher than this frequency component.

  By performing the waveform processing of the acceleration waveform data using such a high-pass filter, it is possible to effectively detect only the vibration generated with the cutting of the workpiece 10 if the frequency setting of the high-pass filter is appropriate. It is possible to grasp the cutting state of the workpiece (the worn state of the blade 22) with high accuracy. In this embodiment, a high-pass filter is used as the filter circuit. However, when it is necessary to remove, for example, a high-frequency component in accordance with a frequency component to be removed, a band-pass filter that passes only a predetermined frequency range may be used. it can.

  Next, in step S203, double integrated waveform data is generated. FIG. 6 is an example of double integrated waveform data at the time of a test cut. In FIG. 6, the vertical axis represents the displacement (mm) of the spindle 20, and the horizontal axis represents time (Sec) (the same applies to FIGS. 7 and 8 described later). When the acceleration waveform data is double-integrated with time, the displacement amount of the spindle 20 is obtained. In the graph of the relationship between the amount of displacement and time (double integral waveform data) as shown in FIG. 6, the waveform processing is insufficient, and even in the section C, there is a large numerical difference from other sections. There will be no waveform. For this reason, in the graph shown in FIG. 6, it is difficult to detect the peak of the amount of displacement that changes according to the cutting resistance.

  Steps S202 and S203 are repeatedly executed until a peak due to cutting resistance is detected in the double integral waveform data at the time of the test cut (step S204). If no peak due to cutting resistance is detected, the process returns to step S202. At this time, signal waveform processing using a high-pass filter that passes only a higher frequency than the high-pass filter used in the previous signal waveform processing is performed. That is, a high-pass filter having a larger lower limit value of the pass frequency is used. Thus, by changing the frequency setting of the high-pass filter in step S202, it is possible to generate double integrated waveform data while increasing the passing frequency.

  Next, when a peak due to cutting force is detected in the double integral waveform data, the reference data Ds in step S205 is generated. FIG. 7 shows double integrated waveform data when a peak due to cutting resistance is detected. In this case, in the double integrated waveform data, the frequency component of the vibration generated by cutting is larger than the other frequency components, so the displacement amount in the section C is large. If a peak due to cutting resistance is detected in the double integrated waveform data, a step of generating reference data Ds is executed based on the double integrated waveform data in step S205. In this embodiment, the maximum value of the displacement amount in the section C of the double integral waveform data is set as the reference data Ds.

  When the generation of the reference data Ds is completed, the test cut process in step S101 ends (step S206). As described above, when the test cut of the workpiece 10 is performed, the reference data Ds is generated based on the first output signal from the acceleration sensor 25 attached to the spindle 20. More specifically, the reference data Ds is a signal waveform obtained by subjecting the first output signal from the acceleration sensor 25 to waveform processing using a high-pass filter and double integration of the signal obtained by this waveform processing. Based on.

  Next, in step S102 of FIG. 2, it is determined whether or not the cutting surface (cut surface) of the workpiece 10 by the test cut is smooth and within the specification range (whether or not it is a non-defective product) by visual observation with an operator's microscope. . This determination is performed from the viewpoint of enlarging the workpiece cut surface with a microscope and whether or not there is a chipping (chipping) having a size larger than a predetermined size on the cut surface. In this step, it is also possible to make a determination by calculating the amount of chipping at the edge of the cut workpiece 10 from the captured image captured by the imaging device by image processing. The control unit 40 is configured to calculate the total amount and maximum width of the chipping area by image processing on the captured image of the workpiece 10 after cutting, and compare these with a threshold value set in advance by the operator. May be.

  When it is determined that the cut surface is smooth (the workpiece is non-defective), non-defective product data registration is performed in step S103. That is, the reference data Ds acquired during the test cut, the frequency setting of the high-pass filter used for waveform processing when the reference data Ds is generated, and the processing condition information are registered as good product data (storage unit) 50). The registered non-defective product data is used in actual cutting described later. On the other hand, if it is determined that the cut surface is not smooth (the workpiece is a defective product), defective product data is registered in step S104. That is, the defective product reference data Df equivalent to the reference data Ds acquired during the test cut (see FIG. 8), the frequency setting of the high pass filter used for the waveform processing when the defective product reference data Df was generated, and Each piece of information of the processing condition information is registered as defective product data (stored in the storage unit 50). The defective product reference data Df registered together with the frequency setting of the high-pass filter used for the waveform processing when the defective product reference data Df is generated is also used in an actual cut described later. However, the processing condition information stored as defective product data is different from good product data in that it is not used in actual cutting.

  Subsequently, in step S105, it is determined whether or not non-defective product data is registered. If the non-defective product data is not registered, the process returns to step S101 and the above steps are repeated. In this case, in the second and subsequent steps S101, the frequency setting of the high-pass filter when the reference data Ds (or defective product reference data Df) is generated during the previous execution of this step may be read from the storage unit 50 and used. Is possible. For example, when conditions such that a peak due to cutting force is detected with the same high-pass filter frequency setting as the previous execution, the number of executions of waveform processing using the high-pass filter and generation processing of double integral waveform data should be reduced. The reference data Ds can be generated efficiently. Further, when machining condition information is stored as defective product data, the rotation speed, load current, cutting speed, cutting acceleration, and the like of the spindle 20 can be set based on the machining condition information. Conditions can be set efficiently.

  On the other hand, if there is registration of non-defective product data, in step S106, the registered non-defective product data is set as a reference value, and an allowable range (upper limit and lower limit) recognized as non-defective product is set. This reference value (allowable range) is stored in the storage unit 50.

  Next, an actual cutting process is performed in step S107. FIG. 9 is a flow of the actual cutting process (step S107) in the present embodiment. First, in step S301, it is determined whether or not the cutting process in the actual cutting process has been completed. When the cutting process is completed, the actual cutting process ends (step S308). When the cutting process is not completed, acceleration waveform data is acquired in step S302. In the actual cutting process, unlike the test cutting process, when the acquisition of acceleration waveform data is started in step S301, the processing of the next step is started.

The acceleration waveform data acquired here is subjected to waveform processing using a high-pass filter in step S303, and double integrated waveform data using double integration is generated in step S304. Steps S303 and S304 are the same as steps S202 and S203 in the test cut process (step S101) except for some settings. First, the acquisition period (length) of the acceleration waveform to be processed is different, and in the actual cutting process, the acceleration waveform data target in the period from the start of acquisition of acceleration waveform data to the time when signal waveform processing is performed is used. As the high-pass filter frequency setting, the high-pass filter frequency setting stored together with the reference data Ds (or defective product reference data Df) in steps S103 and S104 is used.
In this way, when actual cutting of the workpiece 10 is performed, actual cut data that is double integrated waveform data is generated based on the second output signal from the acceleration sensor 25. More specifically, the actual cut data is generated by performing waveform processing using a high-pass filter on the second output signal from the acceleration sensor 25 and performing double integration on the signal obtained by this waveform processing. Is done.

  Subsequently, in step S305, it is determined whether or not the actual cut data after the double integration generated in step S304 is out of the allowable range set in step S106. That is, it is determined whether or not the actual cut data is out of the range of the reference data Ds. When the actual cut data is within the allowable range, that is, smaller than the reference data Ds obtained in the test cut process, the process returns to step S301, and steps S301 to S304 are repeated. On the other hand, when the actual cut data is outside the allowable range, that is, when the actual cut data is greater than or equal to the reference data Ds obtained in the test cut process, an alarm is output in step S306 because the workpiece 10 may be non-defective (alarm output) processing). As described above, the cutting state of the workpiece 10 is grasped in real time by sequentially comparing with the reference data Ds while acquiring the actual cut data. At this time, the alarm unit 70 outputs an alarm based on a command from the control unit 40. The alarm is preferably output before the state grasped based on the acceleration waveform data of the spindle 20 becomes fatal, that is, in a state where there is a certain margin.

  FIG. 8 is an example of defect determination in the present embodiment. The waveform shown in FIG. 8 is cut data after double integration, and the waveform represented as a non-defective product has a lower value than the reference data Ds obtained in the test cut process. The positive and negative values of the reference data Ds are set, and the interval between the upper limit and the lower limit (arrows in the figure) corresponds to an allowable range for accepting a good product. If it is within this allowable range, it is determined as a non-defective product.

  On the other hand, the part exceeding the upper limit or the lower limit represented as a defective product (the part marked with a circle in FIG. 8) is out of the allowable range. This is because the sharpness of the blade 22 deteriorates due to the machining amount represented by the number of workpieces 10 and the machining distance, and the acceleration (vibration) of the spindle 20 increases. As a result, the acceleration (displacement amount) becomes an ideal value over time. This is because it deviates from the reference value. When such cut data is obtained, a portion exceeding the upper limit or lower limit is determined to be defective. Actually, it is desirable to determine that the product is defective when the set value exceeds the set number of times for noise removal.

  Next, in step S307 in FIG. 9, it is determined whether or not to stop the rotation operation of the spindle 20. The reference for stopping the spindle 20 is preferably set wider than the reference for outputting an alarm (the allowable range in step S305). For example, as shown in FIG. 8, this standard can set defective product reference data Df that clearly shows that the cut surface of the workpiece 10 is a non-smooth defective product. If the spindle 20 is not stopped in step S307, the process returns to step S301. On the other hand, when the spindle 20 is stopped, the actual cutting process (step S107) ends (step S308). After the spindle 20 is stopped, dressing of the blade 22 or replacement of the blade 22 is performed. Instead of stopping the spindle 20 in step S307, the rotation speed of the spindle 20 may be reduced to reduce the resistance during cutting.

  Subsequently, in step S108 of FIG. 2, it is determined whether or not the machining of all the workpieces 10 to be cut has been completed. If there is an unprocessed workpiece, the process returns to step S107, and if all the workpieces have been processed, the cutting process ends (step S109).

  According to the present embodiment, it is possible to provide a cutting apparatus and a cutting method capable of grasping a cutting state of a workpiece in real time with high accuracy.

  In addition, since it is possible to specify the dicing line corresponding to the portion that becomes the defective product exceeding the upper limit or lower limit of the allowable range in the actual cut data and the position in the dicing line, Even when defective products coexist, it is possible to easily identify a portion that seems to be a defective product.

  The embodiment of the present invention has been specifically described above. However, the present invention is not limited to the matters described as the above-described embodiments, and can be appropriately changed without departing from the technical idea of the present invention.

  For example, it is also possible to employ a method of performing signal waveform processing without performing all of a series of cuttings on one workpiece 10 in the test cutting process. In this case, an acceleration waveform including at least one set of an operation for moving the blade 22 and a cutting operation in one dicing line may be used as a signal waveform processing target. Thereby, the amount of displacement can be compared at the time of cutting and other times, and the reference data Ds can be generated. Therefore, a plurality of test cuts can be performed with one workpiece 10, and the reference data Ds can be generated efficiently.

  Further, a twin spindle configuration in which the blades 22 of the two spindles 20 are provided to face each other and the workpiece 10 can be simultaneously cut may be employed. In this case, by providing the acceleration sensor 25 on each spindle 20, the wear state of each blade 22 can be grasped individually and replaced at an appropriate time. It is also possible to employ a configuration in which the acceleration sensor 25 is provided only on one spindle 20 and the cutting state of the workpiece 10 is grasped based on the detected acceleration.

10 Workpiece 20 Spindle 21 Spindle motor 22 Dicing blade (blade)
23 Table 24 Table motor 25 Acceleration sensor 30 A / D converter 40 Control unit 50 Storage unit 60 Display unit 70 Alarm unit

Claims (4)

A cutting method for cutting a workpiece,
Generating reference data based on a first output signal from an acceleration sensor attached to a spindle when performing a test cut of the workpiece;
A step of generating actual cut data based on a second output signal from the acceleration sensor when performing actual cutting of the workpiece;
Determining whether the actual cut data is out of the range of the reference data;
A step of outputting an alarm when the actual cut data is out of the range of the reference data,
The reference data and the actual cut data are obtained by performing waveform processing using a high-pass filter on each of the first output signal and the second output signal from the acceleration sensor, and obtaining the signal. A cutting method characterized by being generated by double integration.   The cutting method according to claim 1, further comprising a step of stopping the rotation operation of the spindle after the step of outputting the alarm. A cutting device for cutting a workpiece,
A blade for cutting the workpiece;
A spindle provided with a motor for rotating the blade;
An acceleration sensor for measuring the acceleration of the spindle;
A control unit that controls to output an alarm based on an output signal from the acceleration sensor,
The control unit generates reference data based on a first output signal from the acceleration sensor when performing a test cut of the workpiece,
When actual cutting of the workpiece is performed, actual cutting data is generated based on the second output signal from the acceleration sensor,
Determining whether the actual cut data is out of the range of the reference data;
Control to output the alarm when the actual cut data is out of the range of the reference data,
The reference data and the actual cut data are obtained by performing waveform processing using a high-pass filter on each of the first output signal and the second output signal from the acceleration sensor, and obtaining the signal. A cutting device characterized by being generated by double integration.   The cutting apparatus according to claim 3, wherein the control unit stops the rotation operation of the spindle after outputting the alarm.