JP2012086478A - Liquid discharge failure detector, ink jet recorder, and liquid discharge failure detecting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately determine liquid discharge failure.SOLUTION: The liquid discharge failure detector 1 irradiates a light beam 104 from 102 on a flight path of a liquid droplet 107 discharged from a plurality of nozzles 101, receives forward scattering light 109 generated when the light beam 104 collides with the flying liquid droplet 107 at a light receiving part 106, and acquires waveform data corresponding to the change of light intensity of the forward scattering light 109. A discharge control part 111 discharges the liquid droplet 107 from the nozzles 101 which determine the failure, at such an interval that one liquid droplet enters within the beam diameter of the light beam 104. A data processing part 112 stores the waveform data acquired from the discharged liquid droplet 107, and counts feature amounts of the waveform data acquired from the plurality of liquid droplets 107. A failure determination part 113 determines the liquid discharge failure based on the counted feature amounts.

Description

  The present invention relates to a liquid discharge failure detection device, an ink jet recording apparatus, and a liquid discharge failure detection method.

  An inkjet recording apparatus includes an inkjet head of each color that ejects minute ink droplets (droplets) from minute nozzles, and ejects droplets while moving the inkjet head relative to a recording medium such as paper. An image is formed on a recording medium. In order to form a high-resolution image, it is necessary to reduce the size of the droplets by reducing the size of the nozzles. There is a problem in that an ejection failure (liquid ejection failure) occurs, dot missing or the like occurs in the image and the image quality deteriorates.

  In order to solve this problem, the ink jet recording apparatus is provided with a liquid discharge failure detection device that detects a liquid discharge failure. For example, an output voltage obtained by receiving light (shielded light) shielded by irradiating a droplet discharged from a nozzle with laser light emitted from a light emitting element such as a laser diode by a light receiving element such as a photodiode. And a reference voltage value are known, and a liquid ejection failure detection device that determines whether or not a droplet has been ejected normally is known.

  In the liquid discharge failure detection (shielded light method) using the above-described shielding light, in order to increase the S / N ratio of the output voltage (light receiving signal) obtained by the light receiving element, the liquid droplets are ejected at intervals of a plurality of drops within the laser diameter. The S / N ratio is improved by putting a plurality of droplets within the laser diameter (Patent Documents 1, 5, and 6). In the shielded light method, the output voltage obtained by the light receiving element varies depending on the difference in transparency depending on the color of the ink. Therefore, in order to obtain an equivalent output voltage regardless of the color of the ink, There is one that changes the number of droplets entering and changes the volume of the droplets (Patent Document 7). In addition, as a method for discriminating liquid ejection failure, a method of discriminating the output voltage obtained by the light receiving element by putting it in a comparator (Patent Document 4), an average waveform of a plurality of good waveforms and a waveform to be inspected are compared. In addition to a light detection method such as a shielded light method, there is a method of determining a liquid ejection failure based on a waveform of a current flowing in a piezoelectric element inside the head (Patent Document 2).

Japanese Patent No. 4227395 JP 2008-254239 A JP 2002-11871 A Japanese Patent No. 3820830 Japanese Patent No. 4074920 Japanese Patent No. 3876700 JP 2000-233520 A

  However, the conventional liquid discharge failure detection apparatus described above may not be able to detect a liquid discharge failure with sufficient accuracy, which may result in an erroneous determination. For example, in the shielding light method, in order to increase the S / N ratio of the light reception signal, it is necessary to put a plurality of droplets within the laser diameter, but with the number of droplets that is within the laser diameter (approx. 5). , The S / N ratio may not be sufficiently obtained. In addition, a method other than the light detection method needs to incorporate a dedicated circuit for measuring the current waveform flowing in the piezoelectric element in the head, resulting in high cost.

  The present invention has been made in view of the above, and an object of the present invention is to provide a liquid discharge failure detection device, an ink jet recording apparatus, and a liquid discharge failure detection method capable of accurately determining a liquid discharge failure. To do.

  In order to solve the above-described problems and achieve the object, the liquid ejection failure detection device of the present invention irradiates a flight path of droplets ejected from a plurality of nozzles with a light beam from a light emitting element, and the light beam An acquisition unit that receives scattered light generated by colliding with the flying droplets by a light receiving element and acquires waveform data according to a change in light intensity of the scattered light, and performs defect determination among the plurality of nozzles An ejection unit that ejects the droplets at intervals of one droplet within the beam diameter of the light beam from a nozzle; and a storage unit that stores the waveform data acquired from the droplets ejected by the ejection unit; And tally means for tallying the feature values of the waveform data acquired from the plurality of droplets, and detecting means for detecting a liquid ejection failure based on the tallyed feature quantities.

  In addition, the liquid discharge failure detection method of the present invention includes a step of discharging the droplets from a nozzle that performs defect determination among the plurality of nozzles, and a light emitting element on a flight path of droplets discharged from the plurality of nozzles. Irradiating a light beam, receiving scattered light generated by colliding with the droplet flying by the light beam with a light receiving element, obtaining waveform data according to a change in light intensity of the scattered light, and the ejection A step of storing the waveform data acquired from the droplets discharged by the means; a step of counting feature amounts of the waveform data acquired from the plurality of droplets; and a liquid discharge based on the totaled feature amounts And a step of detecting a defect.

  According to the present invention, it is possible to accurately determine a liquid discharge failure.

FIG. 1 is an explanatory diagram showing a configuration example around the main part of the liquid ejection failure detection device according to the present embodiment. FIG. 2 is a diagram illustrating a configuration example of the light receiving unit and the data processing unit. FIG. 3 is a flowchart illustrating an example of a liquid ejection failure detection process. FIG. 4 is a graph showing an example of waveform data. FIG. 5A is a conceptual diagram for explaining the acquisition of the feature amount of the waveform data. FIG. 5B is a conceptual diagram for explaining the acquisition of the feature amount of the waveform data. FIG. 5C is a conceptual diagram for explaining the acquisition of the feature amount of the waveform data. FIG. 5-4 is a conceptual diagram for explaining the acquisition of the feature amount of the waveform data. FIG. 6 is a conceptual diagram illustrating the discharge of droplets. FIG. 7 is a conceptual diagram showing an example of waveform data. FIG. 8 is a flowchart illustrating an example of a liquid ejection failure detection process according to the modification. FIG. 9 is a conceptual diagram showing an example of an FFT spectrum obtained from waveform data. FIG. 10 is a conceptual diagram showing an example of an FFT spectrum obtained from waveform data. FIG. 11 is a conceptual diagram showing the configuration of an example of an ink jet recording apparatus provided with a liquid ejection defect detection device from the front. FIG. 12 is a view of a part of the ink jet recording apparatus observed obliquely from above.

  Exemplary embodiments of a liquid ejection failure detection apparatus, an ink jet recording apparatus, and a liquid ejection failure detection method according to the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 1 is an explanatory diagram illustrating a configuration example around a main part of a liquid ejection failure detection device according to an embodiment. As shown in FIG. 1, the liquid ejection failure detection apparatus 1 has a plurality of nozzles 101, and performs recording on a recording medium such as paper by ejecting ink droplets (droplets) from each nozzle 101 according to an image signal. This is a device for detecting a liquid ejection failure of the head 100, and is provided in an ink jet recording apparatus (details will be described later).

  Specifically, the liquid discharge failure detection apparatus 1 discharges a droplet 107 from a nozzle 101 that performs failure determination among a plurality of nozzles 101, and travels from the light emitting element 102 such as a laser diode to the flight path of the droplet 107. A light beam 104 (laser light) is irradiated through a collimating lens 103 that corrects to parallel light. That is, in the course of the flight path of the droplet 107, the light beam optical axes 105 of the light beams 104 intersect so as to be substantially orthogonal. At this time, the light beam 104 collides with the flying droplet 107 to generate scattered light 108 and forward scattered light 109.

  The light receiving unit 106 receives forward scattered light 109 generated at an angle θ when the light beam 104 collides with the droplet 107 at a position deviating from the beam diameter of the light beam optical axis 105. FIG. 2 is a diagram illustrating a configuration example of the light receiving unit 106 and the data processing unit 112.

  As shown in FIG. 2, when a liquid discharge failure is detected, among the plurality of nozzles 101 of n1 to nN, droplets 107 (nX) are detected at predetermined intervals from the nozzle 101 (nX) that detects the liquid discharge failure. d1 to dK) are discharged. The discharged droplet 107 intersects with the light beam 104 when it has traveled a distance H from the nozzle 101. By the intersection of the light beam 104 and the droplet 107, the above-described scattered light 108 and forward scattered light 109 are generated. The generated forward scattered light 109 is received by the light receiving unit 106 disposed below the light beam 104. The distance H is the height from the bottom surface of the recording head 100 where the nozzle 101 is provided to the light beam optical axis 105, and is specifically about 3 mm to 7 mm.

  The light receiving unit 106 includes a light receiving element 1061, a current / voltage conversion unit 1062, a high pass filter unit 1063, an amplification unit 1064, and an A / D conversion unit 1065. The light receiving element 1061 is a photodiode or the like that outputs a current corresponding to the received light intensity. The current-voltage converter 1062 is a circuit that converts an input current into a voltage, and converts the current output from the light receiving element 1061 into a voltage and outputs the voltage. The high-pass filter unit 1063 removes and reduces the DC component (noise component) of the voltage output from the current-voltage conversion unit 1062. The amplifying unit 1064 amplifies the voltage output from the high pass filter unit 1063. The A / D conversion unit 1065 digitally converts the voltage (analog) amplified by the amplification unit 1064 and outputs it to the data processing unit 112.

  Thereby, the data processing unit 112 obtains a voltage value (waveform data) corresponding to the change in the light intensity of the forward scattered light 109. In the drawing, the light receiving unit 106 is disposed below the light beam 104, and a light receiving element 1061 is disposed as close to the center of the light beam optical axis 105 as possible at a position where the light beam diameter and the light receiving surface of the light receiving unit 106 do not overlap. By doing so, efficient detection becomes possible. In addition, the light receiving unit 106 receives the scattered light 108, particularly forward scattered light 109 having a strong light intensity in the scattered light 108, so that the light intensity can be measured as a voltage value. A voltage value having a sufficient S / N ratio can be obtained if one droplet 107 falls within the diameter, so that the behavior of the droplet 107 for each droplet can be understood as a waveform of the voltage value. In addition, since there is no difference in voltage value output depending on the ink color, it is not necessary to change the ejection conditions for each ink color.

  The data processing unit 112 ejects a plurality of droplets 107 and stores waveform data for each droplet 107 in a memory 1121 as a storage unit. The feature amount (details will be described later) of the stored waveform data is used as a totaling unit. The calculation unit 1122 counts. In the liquid discharge failure detection device 1, it is determined whether or not the droplet 107 has been normally discharged based on the collected values.

  As a configuration for controlling the above-described operation, the liquid ejection defect detection device 1 includes a control unit 110, a nozzle driving unit 120, and a light emission driving unit 130. The control unit 110 is composed of a microcomputer system having a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a timer, and the like. Are sequentially executed to centrally control the operation of the liquid ejection failure detection apparatus 1. The control unit 110 realizes functions as the ejection control unit 111, the data processing unit 112, and the defect determination unit 113 by the above-described CPU sequentially executing programs.

  The discharge control unit 111 outputs a control signal to the nozzle drive unit 120 that discharges the droplet 107 from the nozzle 101 and the light emission drive unit 130 that emits light from the light emitting element 102, and thereby the liquid discharge failure of the recording head 100 is detected. The ejection of the droplet 107 and the light emission of the light emitting element 102 at the time of detection are controlled. Specifically, when the detection of the liquid discharge failure of the recording head 100 is started, the light emitting element 102 starts to emit light, and the discharge of the droplet 107 from the nozzle 101 that detects the liquid discharge failure is started. The ejection interval of the droplets 107 by the ejection control unit 111 is an interval that allows one droplet 107 to enter the beam diameter of the light beam 104. Design items such as the speed and the beam diameter of the ejected droplet 107 Is a value (for example, 500 Hz) set in advance at the time of factory shipment.

  The data processing unit 112 performs data processing for discharging a plurality of droplets 107 and summing up feature amounts of waveform data for each droplet 107. Specifically, the work area secured in the RAM of the control unit 110 is the memory 1121, and the CPU performs the functions as the calculation unit 1122 by sequentially executing the program, thereby performing the above-described data processing. The failure determination unit 113 detects a liquid discharge failure by determining whether or not the droplet 107 has been normally discharged based on the values counted by the data processing unit 112.

  Here, details of the liquid discharge failure detection processing by the data processing unit 112 and the failure determination unit 113 will be described. FIG. 3 is a flowchart illustrating an example of a liquid ejection failure detection process. As shown in FIG. 3, in this embodiment, liquid ejection failure is based on the Mahalanobis distance MD obtained by summing up the feature values of waveform data obtained from a plurality of droplets 107 using the MT (Mahalanobis Taguchi) method. Judgment is made.

  Prior to the liquid discharge failure detection process, the waveform data detected by the light receiving unit 106 when the plurality of droplets 107 are discharged from the nozzle 101 (nX) that detects the liquid discharge failure is stored in the memory 1121. Is stored. FIG. 4 is a graph showing an example of waveform data. As shown in FIG. 4, when, for example, three droplets 107 (d1 to d3) are ejected from the nozzle 101, the memory 1121 stores waveform data including peaks corresponding to d1 to d3. Specifically, the peak of the voltage vd1 at time td1 corresponding to the droplet 107 of d1, the peak of the voltage vd2 at time td2 corresponding to the droplet 107 of d2, and the voltage vd3 at time td3 corresponding to the droplet 107 of d3. Waveform data with a peak is stored.

  In this embodiment, the waveform data of three drops (K = 3) ejected from the normal nozzle 101 is used to obtain unit space information necessary for determining a liquid ejection failure based on the Mahalanobis distance MD. Are acquired as samples for 100 waveforms (n = 100). Here, in order to consider variation for each nozzle 101, samples may be acquired using different nozzles 101.

  First, the unit space information acquisition flow for determining a liquid discharge failure based on the Mahalanobis distance MD will be described. As shown in FIG. 3, when the unit space information acquisition flow is started, the data processing unit 112 sequentially extracts feature amounts for 100 waveforms acquired as samples of the normal nozzle 101 from the i-th. = 1 is set (S10).

  Next, the calculation unit 1122 receives input of waveform data stored in the memory 1121 from the i-th sample (S11), and removes a noise component by a digital filter (S12). Next, the computing unit 1122 acquires a waveform feature amount from the waveform data from which the noise component has been removed (S13).

  The waveform feature amount may be any feature amount that produces a difference between the waveform of the normal nozzle 101 and the waveform of the nozzle 101 having a liquid ejection defect, such as the standard deviation of the voltage value at the peak of the waveform. It's okay. In the present embodiment, a plurality of reference voltages are set as threshold values, and the amount of change and the amount of existence calculated by the intersection of the threshold value and the waveform are set as feature amounts.

  5A to 5D are conceptual diagrams for explaining the acquisition of the feature amount of the waveform data. As shown in FIGS. 5A to 5D, in this embodiment, the intersections with the waveforms are obtained using the voltages Vt1 to Vt4 as threshold values. The amount of change represents the number of intersections with waveforms in each of the voltages Vt1 to Vt4. Specifically, in the case of FIGS. 5-1 to 5-3, since the number of intersections between the voltages Vt1 to Vt3 and the waveform is 6, the amount of change in the voltages Vt1 to Vt3 is 6. In the case of FIG. 5-4, since the number of intersections of the voltage Vt4 and the waveform is 2, the amount of change in the voltage Vt4 is 2.

  The abundance represents the total area from the rising edge to the falling edge at the intersection with the waveform at each of the voltages Vt1 to Vt4. For example, in the case of FIG. 5-1, at the intersection of the voltage Vt1 and the waveform, the sum of the areas from the rising edge to the falling edge (the area of the waveform passing through the voltages vd1, vd2, and vd3 from the intersection (the shaded area in FIG. 5A). Since the area)) is 22, the abundance at the voltage Vt1 is 22. Similarly, the abundance at the voltage Vt2 is 14 (area of the hatched portion in FIG. 5-2), the abundance at the voltage Vt3 is 10 (area of the shaded portion of FIG. 5-3), and the abundance at the voltage Vt4 is 2 (FIG. 5). 5-4). This area may be obtained by simply adding the differences from the threshold voltage to the peak voltage. Therefore, in S13, eight feature amounts of the change amount and the existence amount in the voltages Vt1 to Vt4 are acquired. The eight feature amounts are variable (1) to variable (8), respectively.

  Next, the calculation unit 1122 increments i (i = i + 1) (S14), and determines whether i is equal to or greater than the number of samples (n) (S15). If the number is not more than the number of samples (S15: NO), the process returns to S11 to process the next sample. If the number is greater than or equal to the number of samples (S15: YES), the process of S11 to S14 is finished and the process proceeds to S16.

  In S16, the calculation unit 1122 assigns the variables (1) to (8) in the samples 1 to n to the unit space data table. The following (Table 1) is an example of the unit space data table.

  In the present embodiment, since the number of variables is 8, k = 8. The variables (1) to (8) of each sample are stored in the cell of the corresponding sample number. In each variable, the average value is calculated by the calculation unit 1122 using the equation (1). The calculated average value is stored in a cell for each variable.

  In each variable, the standard deviation is calculated by the calculation unit 1122 using the equation (2). The calculated standard deviation is stored in a cell for each variable.

  Next, the calculation unit 1122 normalizes the unit space data table exemplified in (Table 1) (S17). This normalization is performed using equation (3).

  (Table 2) is an example of a unit space data table after normalization.

  Next, the calculation unit 1122 obtains a correlation matrix R using the normalized unit space data table exemplified in (Table 2) and Expression (4).

  The obtained correlation matrix R is as shown in Equation (5).

  Since the unit space information for determining the liquid ejection failure based on the Mahalanobis distance MD is an inverse matrix of the correlation matrix R, the calculation unit 1122 acquires the inverse matrix of the correlation matrix R (S18), and the unit End the spatial information acquisition flow.

  Next, a flow for determining a liquid discharge failure based on the Mahalanobis distance MD will be described. When the determination flow is started, the calculation unit 1122 receives input of waveform data (see FIG. 4) of K droplets ejected from the nozzle 101 (nX) that detects a liquid ejection failure from the memory 1121 (S20). Next, the computing unit 1122 removes the noise component by the digital filter (S21), and acquires the feature amount of the waveform from the waveform data from which the noise component has been removed (S22).

  The acquisition of the feature amount of the waveform is performed in the same manner as in S13 described above, and the change amount and the existence amount in the voltages Vt1 to Vt4 are acquired as the feature amount. The acquired change amount and existence amount are assigned to y1 to yk in Expression (6). Specifically, the acquired variables (1) to (8) are assigned to y1 to y8.

  Next, the calculation unit 1122 performs normalization using Equation (3) using the average value and standard deviation value of the unit space data table exemplified in (Table 1) and the data assigned to Equation (6) (S23). ). However, yj = xij. Equation (7) shows the result of normalization.

  Next, the computing unit 1122 calculates the Mahalanobis distance MD using the equation (8) from the result shown in the equation (7) and the unit space information (inverse matrix of the correlation matrix R) previously obtained in S18 (S24). ).

  In S24, the Mahalanobis distance is similarly obtained for the n samples for which the unit space information is generated. The average of the Mahalanobis distance of n samples is 1. That is, since it can be determined as a waveform away from the average waveform from the n-sample Mahalanobis distance, that is, a waveform of liquid discharge failure, the good / failure determination is performed using the n-sample Mahalanobis distance as a threshold for detecting liquid discharge failure. It becomes possible. In addition, it is preferable to set the threshold value determination method within a range in which a plurality of defective samples are acquired, the Mahalanobis distance concerning the defective samples is acquired, and the Mahalanobis distance of the previously acquired n samples can be completely separated. Within this range, it may be set small when it is desired to reduce overdetection for determining good as defective, and it may be set large when it is desired to reduce erroneous detection when determining defective as good. In order to obtain sufficient detection performance, the number of ejected droplets K = 100, the variable k = 52 (the number of reference voltages is 26, the amount of change: 26, the amount of existence: 26), and the number of unit space samples n = about 1000 It is good to do.

  Next, the failure determination unit 113 compares the Mahalanobis distance MD calculated in S24 with a threshold value based on the n-sample Mahalanobis distance, and determines whether the Mahalanobis distance MD is less than the threshold value (S25). As described above, if it is less than the threshold value (S25: YES), it is judged good because it is close to the average waveform without liquid ejection failure (S26), and the process is terminated. If it is equal to or greater than the threshold (S25: NO), it is determined that there is a liquid ejection failure because it is far from the average waveform (S27), and the process ends.

  As described above, in the liquid discharge failure detection apparatus 1, the light beam 104 is irradiated from the light emitting element 102 onto the flight path of the droplet 107 discharged from the plurality of nozzles 101, and the light beam 104 collides with the flying droplet 107. The forward scattered light 109 thus generated is received by the light receiving unit 106, and waveform data corresponding to a change in the light intensity of the forward scattered light 109 is acquired. The ejection control unit 111 ejects the droplets 107 from the nozzle 101 that performs the defect determination at an interval at which one droplet falls within the beam diameter of the light beam 104. The data processing unit 112 stores the waveform data acquired from the ejected droplets 107 and totals the feature amounts of the waveform data acquired from the plurality of droplets 107. The defect determination unit 113 determines a liquid ejection defect based on the total feature amount.

  Therefore, in the shielded light method, in order to increase the S / N ratio of the light reception signal, it is necessary to put a plurality of droplets within the laser diameter. Therefore, waveform data having a sufficient S / N ratio can be acquired with one drop. In addition, since the liquid ejection failure is determined by aggregating the feature values of the acquired waveform data, liquid ejection is possible without incurring costs such as incorporating a dedicated circuit for measuring the current waveform flowing in the piezoelectric element inside the head. Defects can be determined with high accuracy.

<Modification>
Next, a modified example of the liquid discharge failure detection process described above will be described. In the modification, as pre-processing for acquiring the waveform feature amount, processing for selecting (extracting) only a specific waveform, correction of the amplitude of the waveform, and the like are performed.

  FIG. 6 is a conceptual diagram illustrating the discharge of droplets. As shown in FIG. 6, when droplets are ejected from the nX nozzles, a minute droplet d1a (hereinafter referred to as satellite) follows the main droplet d1 regardless of whether the liquid ejection is defective or normal. Discharged. The satellite d1a can be made slower than the droplet d1 by increasing the distance H (see FIG. 2) to about 7 mm. Therefore, since the satellite d1a slowly passes through the beam diameter, the satellite component appears as a low-frequency component in the waveform data. Therefore, when the distance H is increased, the satellite component is removed by the high-pass filter unit 1063. However, when the distance H is reduced (about 3 mm), for example, to prevent over-determination, the satellite component cannot be completely removed by the high-pass filter unit 1063.

  FIG. 7 is a conceptual diagram illustrating an example of waveform data, and more specifically, a conceptual diagram illustrating an example of waveform data including satellite components. As shown in FIG. 7, the waveforms of the satellites d1a, d2a, and d3a are included in the waveform data following the waveforms of the main droplets d1, d2, and d3. Specifically, a waveform having a voltage vd1a peak at time tw1a corresponding to satellite d1a, a waveform having a voltage vd2a peak at time tw2a corresponding to satellite d2a, and a peak of voltage vd3a at time tw3a corresponding to satellite d3a. Waveforms with are included. Therefore, when the amount of change and the amount of existence are acquired as the waveform feature values with the voltages Vt1 to Vt4, the components of the satellites d1a, d2a, and d3a are also included, which affects the detection accuracy.

  Therefore, the calculation unit 1122 selects only a predetermined waveform included in the waveform data, and acquires a feature amount from the selected waveform. Specifically, since the satellite component has a smaller output than the components of the main droplets d1, d2, and d3, the waveform data is divided at a time interval td at which the droplet is discharged, and the divided ranges r1 to r3 are divided. Only the waveform with the maximum peak is selected and extracted. As a result, the waveform feature amount can be acquired using only the components of the main droplets d1, d2, and d3.

  Further, the output voltage obtained by receiving the forward scattered light 109 generated by colliding with the droplet 107 varies depending on an error due to how the jig is attached or an error of the nozzle 101. Mainly important for detecting liquid discharge defects are the standard deviation of the peak value of the waveform and the component of the waveform width, so if the output voltage is stable above a certain level, the waveform width In most cases, a good judgment should be made unless there is an abnormality. However, if the output voltage is too low, there may be a case where it is desired to make it defective. Therefore, when the average of the peak values of the waveform is greater than or equal to a certain value, it is possible to prevent overdetection by correcting the amplitude of the waveform to a certain value.

  FIG. 8 is a flowchart illustrating an example of a liquid ejection failure detection process according to the modification. As shown in FIG. 8, before S13b and S22b performing the same processing as S13 and S22, the calculation unit 1122 corrects the amplitude of the waveform to a constant value (S13a and S22a).

  Specifically, in the case of the waveforms of the droplets d1, d2, and d3 illustrated in FIG. 7, the average Ave is obtained by Ave = (vd1 + vd2 + vd3) ÷ 3. Therefore, when correcting to a voltage value Vx as a constant value, W ′ [n] = W where the data string of the nth sample of the waveform is W [n] and the data string of the corrected waveform is W ′ [n]. By setting [n] × Vx ÷ Ave and multiplying the entire waveform by the correction value Vx ÷ Ave, the average value of the peak of the waveform can be corrected to be Vx.

  Further, as one of the feature quantities of the waveform, there is a method using the average of the peak value of the waveform. Some of the droplets 107 ejected from the nozzle 101 with poor liquid ejection are bent and ejected. If the droplet 107 thus bent and ejected flies in a place where the intensity distribution of the light beam 104 is weak, the output voltage at the time of receiving light decreases as a whole, and the average value decreases. Due to the decrease in the average value, a difference in feature amount appears between the waveform when there is no liquid discharge failure and the waveform when there is a liquid discharge failure. Further, since it is possible to estimate that the droplet 107 is bent and discharged by detecting a decrease in the average value, it is possible to detect a liquid discharge failure by comparing with a preset threshold value. is there.

  As one of the feature quantities of the waveform, there is a method using the standard deviation of the peak value of the waveform. Some of the droplets 107 ejected from the nozzle 101 with poor liquid ejection scatter and land. This variation results in a variation in the peak value of the waveform, so that the standard deviation is increased. Depending on the size of the standard deviation, a difference in feature amount appears between the waveform when there is no liquid discharge failure and the waveform when there is a liquid discharge failure. Further, since it can be estimated that the liquid droplets 107 are ejected in a dispersed manner by detecting an increase in the standard deviation, it is possible to detect a liquid ejection failure by comparing with a preset threshold value. is there.

  Further, as one of the feature quantities of the waveform, there is a method using a numerical value of the FFT (fast Fourier cosine / sine transform) spectrum shape of the entire waveform. Specifically, the calculation unit 1122 reads the waveform data stored in the memory 1121, performs FFT conversion, and calculates an FFT spectrum component for each frequency. As shown in FIG. 9, when an FFT spectrum of waveform data when droplets are ejected at time t intervals is obtained, a waveform having a peak every 1 / t appears. In addition, when there is no liquid discharge failure, the approximate curve s that connects the peaks by the least square method or the like becomes gentle.

  On the other hand, when there is a liquid ejection failure such as a liquid droplet splitting, as shown in FIG. 10, the approximate curve sa connecting the peaks by the least square method or the like swells. The liquid discharge failure can be detected by quantifying this swell and comparing it with a threshold value. Specifically, the calculation unit 1122 obtains an approximate curve from a point (peak) every 1 / t by the least square method or the like, and integrates the differential value of the curve over a certain interval, thereby quantifying the undulation. . For example, if the undulation is large, the integral value becomes large, and if the undulation is small, the integral value becomes small. Therefore, by detecting that the value obtained by quantifying the swell is larger than a predetermined threshold, it is possible to estimate the occurrence of a liquid ejection failure such as a liquid droplet splitting. In the case of detecting a liquid discharge failure from a value obtained by digitizing the shape of the FFT spectrum of the entire waveform, the droplet discharge interval is 500 Hz, the distance H is about 7 mm, and the order of the least squares method is about 6th. Preferably it is done.

  Further, in the case of detecting a liquid ejection failure from a value obtained by quantifying the shape of the FFT spectrum of the entire waveform described above, a low frequency component amount (for example, less than 1 kHz) in a value indicating the shape of the FFT spectrum as a feature amount. There is a method of using. This low frequency component amount is obtained by the calculation unit 1122 adding the components of the FFT spectrum that are less than 1 kHz. When the droplets 107 ejected from the nozzle 101 scatter and a plurality of minute droplets fly within the beam diameter of the light beam 104, the amount of the low frequency component described above increases. Therefore, by detecting that the low-frequency component amount is larger than the predetermined threshold value, it is possible to estimate the occurrence of a liquid discharge failure such as a droplet being scattered into a plurality of minute droplets.

  Here, an ink jet recording apparatus including the above-described liquid discharge failure detection apparatus 1 will be described with reference to FIGS. FIG. 11 is a conceptual diagram showing the configuration of an example of an inkjet recording apparatus 2 including the liquid ejection failure detection apparatus 1 from the front. FIG. 12 is a diagram in which a part of the inkjet recording apparatus 2 is observed obliquely from above.

  As shown in FIG. 11, a guide shaft 13 and a guide plate 14 are provided in parallel on the left and right side plates 11 and 12 of the casing 10 of the inkjet recording apparatus 2. The guide shaft 13 and the guide plate 14 are slidably penetrated through the carriage 15. An endless belt (not shown) is attached to the carriage 15. The endless belt is wound around a driving pulley and a driven pulley (not shown) provided on the left and right sides of the housing 10. And a driven pulley is driven and rotated with rotation of a drive pulley, and an endless belt is run. As a result, the carriage 15 is moved to the left and right as indicated by the arrows in FIG.

  In the carriage 15, inkjet heads 16 y, 16 c, 16 m, 16 b of four colors of yellow, cyan, magenta, and black (hereinafter referred to as representative of the head 16) are arranged in parallel in the moving direction of the carriage 15. . Each head 16 has a nozzle row in which a plurality of nozzles are linearly arranged on a downward nozzle surface. Although not shown, the linear nozzle row is provided in a direction orthogonal to the moving direction of the carriage 15.

  When the carriage 15 is in the rightmost home position as shown in FIG. 11, each head 16 faces the single recovery device 18 installed on the bottom plate 17 in the housing 10. The single recovery device 18 is a device that sucks out ink from a nozzle that has detected a liquid discharge failure by the liquid discharge failure detection device 1 and recovers the liquid discharge failure by the inkjet recording device 2 itself.

  The liquid ejection failure detection device 1 is disposed on the bottom plate 17 in the housing 10 adjacent to the single recovery device 18. The liquid discharge failure detection device 1 detects a liquid discharge failure of each nozzle of the head 16 by the detection process described above.

  A plate-like platen 22 is installed at a position adjacent to the liquid ejection failure detection device 1. On the back side of the platen 22, a paper feed base 24 that supplies the paper 23, which is a recording medium, onto the platen 22 is provided in an oblique manner. Although not shown, a paper feed roller for feeding the paper 23 on the paper feed table 24 onto the platen 22 is provided. Further, a transport roller 25 is provided for transporting the paper 23 on the platen 22 in the direction of the arrow and discharging it to the front side.

  On the bottom plate 17 in the housing 10, a driving device 26 is further installed at the left end. The driving device 26 drives a feed roller (not shown), a conveyance roller 25, and the like, and drives the above-described driving pulley to move the endless belt to move the carriage 15.

  At the time of recording, the paper 23 is moved onto the platen 22 by being driven by the driving device 26 and positioned at a predetermined position. Further, the carriage 15 is moved and scanned on the paper 23, and droplets are ejected in order from the respective nozzles using the four-color heads 16y, 16c, 16m, and 16b while moving leftward. An image is recorded. After image recording, the carriage 15 is returned to the right and the paper 23 is conveyed by a predetermined amount in the direction of the arrow in FIG.

  Next, while the carriage 15 is moved leftward again, droplets are sequentially ejected from the respective nozzles using the four-color heads 16 y, 16 c, 16 m, and 16 b on the forward path, and an image is recorded on the paper 23. Similarly, after image recording, the carriage 15 is returned to the right and the sheet 23 is conveyed by a predetermined amount in the direction of the arrow in FIG. Thereafter, the same operation is repeated, and an image is recorded on one sheet of paper 23.

DESCRIPTION OF SYMBOLS 1 ... Liquid discharge defect detection apparatus, 2 ... Inkjet recording apparatus, 10 ... Housing, 11 ... Side plate, 12 ... Side plate, 13 ... Guide shaft, 14 ... Guide plate, 15 ... Carriage, 16 ... Head, 17 ... Bottom plate, 18 ... Single recovery device, 22 ... Platen, 23 ... Paper, 24 ... Paper feed, 25 ... Conveying roller, 26 ... Drive device, 100 ... Recording head, 101 ... Nozzle, 102 ... Light emitting element, 103 ... Collimating lens, 104 ... Light beam 105... Light beam optical axis 106. Light receiving unit 107 107 Droplet 108 108 Scattered light 109 Forward scattered light 110 Control unit 111 Discharge control unit 112 Data processing unit 113 Defect determination unit, 120 ... Nozzle drive unit, 130 ... Light emission drive unit, 1061 ... Light receiving element, 1062 ... Current-voltage conversion unit, 1063 ... High pass filter unit, 1064 ... Amplification unit, 065 ... A / D conversion unit, 1121 ... memory, 1122 ... arithmetic unit, H ... distance

Claims (12)

A light beam is emitted from a light emitting element onto a flight path of droplets ejected from a plurality of nozzles, and scattered light generated by the light beam colliding with the flying droplets is received by a light receiving element. Acquisition means for acquiring waveform data according to a change in intensity;
An ejection unit that ejects the droplets at intervals at which one droplet falls within a beam diameter of the light beam from a nozzle that performs defect determination among the plurality of nozzles;
Storage means for storing the waveform data acquired from the droplets discharged by the discharge means;
Aggregating means for aggregating feature quantities of the waveform data acquired from a plurality of the droplets;
Detection means for detecting a liquid ejection failure based on the total feature amount;
A liquid discharge defect detection device comprising: The aggregation means calculates a Mahalanobis distance by aggregating a plurality of types of feature amounts as the feature amounts of the waveform data,
The detection means detects the liquid ejection failure by comparing the calculated Mahalanobis distance with a preset threshold;
The liquid discharge defect detection device according to claim 1. The counting means extracts a waveform of a predetermined droplet included in the waveform data, and totals the feature values of the extracted waveform;
The liquid discharge defect detection device according to claim 1 or 2. The aggregating means aggregating feature values of the corrected waveform data in which the average of the amplitude of the waveform data is corrected to be constant;
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device. The feature amount is a value obtained by aggregating at least one of a change amount at a break position obtained by dividing the waveform data by a predetermined reference value, and a waveform existing amount at the break position;
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device. The feature amount is an average of peak values of the waveform data;
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device. The feature amount is a standard deviation of a peak value of the waveform data;
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device. The feature amount is a value indicating the shape of the FFT spectrum of the entire waveform applied to the waveform data,
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device. The feature amount is a low-frequency component amount in a value indicating the shape of the FFT spectrum;
The liquid discharge defect detection device according to claim 8. The detecting means detects a liquid ejection failure by comparing the total feature amount and a preset threshold value;
The liquid discharge failure detection device according to claim 1, wherein the liquid discharge failure detection device is a liquid discharge failure detection device.   An ink jet recording apparatus comprising the liquid ejection defect detection device according to claim 1. A step of discharging the droplets from a nozzle that performs defect determination among the plurality of nozzles;
A light beam is emitted from a light emitting element onto a flight path of droplets ejected from a plurality of nozzles, and scattered light generated by the light beam colliding with the flying droplets is received by a light receiving element. A step of acquiring waveform data corresponding to a change in intensity;
Storing the waveform data acquired from the droplets ejected by the ejection means;
Summing up the feature values of the waveform data acquired from a plurality of the droplets;
A step of detecting a liquid ejection failure based on the total feature amount;
A liquid discharge defect detection method comprising:

Images