JPH065684B2 - Bonding device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a system for bonding wires or dies such as light emitting diodes and transistors using an image recognition device.

[Conventional technology and its problems]

When manufacturing a semiconductor device, die bonding for fixing a semiconductor chip (die) onto a predetermined substrate and Al wire or Au between a predetermined position of the semiconductor chip and a predetermined position of the lead electrode.
Wire bonding is performed by connecting wires such as wires. In these bonding, generally, the position where the semiconductor chip should be fixed or the position where the wire should be connected is automatically determined by image recognition, and an automatic bonding device (hereinafter referred to as an automatic bonder) for bonding the semiconductor chip or the wire to the position is automatically determined. Done using. Typical operating elements of this type of automatic bonder are as follows.

The lead frames are taken out one by one from the input side magazine containing a large number of unbonded lead frames and sent to a predetermined position.

The lead frame is temporarily fixed, image recognition is performed, and the coordinates (position) for semiconductor chip or wire bonding are determined.

Bonding is performed at the determined position.

Send out the lead frame after bonding,
Store in output magazine.

FIG. 8 is a conceptual diagram showing a form in which a plurality of automatic bonders (1a) (1b) ... (1n) are arranged and mass-produced. The worker (2) who manages multiple automatic bonders (1a) to (1n) issues a call signal (lamp lighting, buzzer call, etc.) that tells that the worker's hand is needed due to a trouble. Go to the stopped automatic bonder, perform necessary processing, and return to the operating state. An example of a situation in which the automatic bonders (1a) to (1n) are stopped and the coping work performed by the worker (2) at that time is as follows.

(B) When the image cannot be recognized, the image is judged by visually observing the CRT screen, the bonding position is determined, and the bonding instruction is given.

(B) When the wire is broken, the wire is pulled out and a ball is formed at the tip of the wire (in the case of ball bonding).

(C) If the input / output magazine is requested to be replaced, the magazine is replaced.

(D) In the case of a trouble such as a defective feed, deal with each situation.

The rate of occurrence of (a) to (d) is overwhelmingly high when the automatic bonder (1) stops due to image recognition in (a).
Generally, the order is (b), (c) and (d). Still, I
In many silicon devices such as C and power transistors, the number of times of stopping the automatic bonder (1) with (a) as the lead is within an allowable range, and the management form of FIG. 8 is often adopted.

However, this does not apply when the occurrence rate of (a) is high, such as wire bonding to an LED (light emitting diode) chip. The reason why (a) is high in the wire bonding to the LED chip is as follows.

(a) A tilted die-bonded LED chip produces a dark chip image, which makes image recognition difficult. Most of the chips of the silicon device are plate-shaped chips whose plane size is sufficiently larger than the thickness of the chip, so that the problem of tilt does not substantially occur. However, the LED chip has a very small plane size, for example, Since it is a dice-shaped micro chip with a side of 0.25 mm, it is easy to tilt.

(b) The chip image may have black stripes on the surface of the chip due to epitaxial growth, which may make image recognition difficult.

If the number of times the automatic bonders (1a) to (1n) are stopped is large, the worker (2)
Is a heavy burden, and most automatic bonders
Automatic bonders (1a) to (1n), such as the worst time zone where (1a) to (1n) are stopped waiting for the operator (2) to take measures
Operation rate will decrease. For this reason, it is necessary to reduce the number of automatic bonders (1a) to (1n) managed by one worker and deal with it.

In order to solve such a problem, the management form of FIG. 9 was devised. In this form, the worker (2) is resident in the control center (3), and when the image recognition of (a) with a high occurrence rate is not possible, the worker (2) conventionally has the corresponding automatic bonders (1a) to (1n). The control center
Performed by remote control from (3). Therefore, compared with the case of FIG. 8, one operator can manage the automatic bonders (1a) to (1
As the number of n) increases, the operating rates of the automatic bonders (1a) to (1n) improve.

However, regarding the work of (b) to (d), the worker
There is no change in that (2) must move to the corresponding automatic bonders (1a) to (1n). Although the number of coping tasks in (b) to (d) is smaller than that in (a),
There are quite a few occasions. Therefore, the worker (2) has n units of automatic bonders (1a) to (1n) and one control center (3).
It is necessary to carry out the coping work at +1 place, and the place where the coping work is performed is increased by one place. It does not mean that the number of coping tasks will be reduced. As a result, the number of automatic bonders (1a) to (1n) that can be managed by one worker (2) cannot be increased significantly.

It is also possible to improve the image recognition rate of the automatic bonders (1a) to (1n) and reduce the number of occurrences of (a) itself. However, in reality, if the image recognition rate is improved, mass productivity does not immediately improve. This is the first
The description will be given based on the examples of the automatic bonder A having a low image recognition rate and the automatic bonder B having a high image recognition rate in the table.

In addition, in this Table 1, only the work for dealing with (a) is focused. This is an example when the image recognition rate is at the lowest level, and the time is a value per chip.

As shown in Table 1, the automatic bonder A has a total of 1
The wire bonding time per chip is 1.60 seconds. However, since the automatic bonder B performs high-grade image recognition, the image recognition time is 0.35 seconds, which is longer than that of the automatic bonder A (0.2 seconds).
The wire bonding time per chip is 1.55 seconds,
Almost the same as the automatic bonder A. In the automatic bonder B, the number of coping work for the worker is reduced due to the improvement of the image recognition rate, and the number of automatic bonders that can be managed by one worker can be increased, so that the labor cost can be reduced. However, although the bonding processing amount (the number of products) per unit time of the automatic bonder B is almost unchanged, the apparatus price of the automatic bonder B is naturally higher than that of the automatic bonder A. In order to increase the bonding processing amount per unit time,
If the speed of the pixel calculation processing in the image recognition portion is increased to suppress the increase in the image recognition time, the device price of the automatic bonder B becomes higher. Therefore, it cannot be generally said that the automatic bonder B is better than the automatic bonder A in terms of mass productivity in which the labor cost, the amount of bonding process per unit time of the automatic bonder, and the apparatus price of the automatic bonder are comprehensively taken into consideration.

Therefore, an object of the present invention is to provide a bonding apparatus in which the number of automatic bonders that can be managed by one operator is increased, the bonding processing amount of the automatic bonder is kept at a high level, and the increase in the apparatus price is minimized. To do.

[Means for Solving the Problems] To achieve the above object, the present invention provides a bonding tool (8) for bonding a die or a wire, and an imaging for imaging an image recognition area for bonding. Device (7) and monochromatic binarized image data connected to the image pickup device (7) and expressing the density of each pixel in two gradations based on an analog output signal obtained from the image pickup device (7) Is connected to the bonder image reading device (9) that forms the image and stores it in the memory, and the bonder image reading device (9) and the bonding tool (8). The position to be bonded to the target image recognition is automatically detected, the die or wire is automatically bonded to the detected position, and the position to be bonded is not detected. Connected to a plurality of bonders (1a to 1n) each including a bonder computer that sends out a signal indicating this when in Noh, and to the imaging device (7) of the plurality of bonders (1a to 1n). The plurality of image signal selection / acquisition switches (16a-16n) and the plurality of bonder (1a-1n) imaging devices through the plurality of image signal selection / acquisition switches (16a-16n) (7) is selectively connected to form multi-valued image data or color image data in which the density of each pixel is expressed in three gradations or more based on the analog output signal obtained from the image pickup device (7). And a single image capturing device (12) for support stored in a memory, and bonding by image recognition including subjecting the multi-valued image data or color image data to image calculation processing. One support computer (13) for automatically detecting the position and notifying the detected position to the bonder, and a bonder of the plurality of bonders (1a-1n) for selectively forming a data transmission path A plurality of computer interconnection switches (19) respectively provided between the computer (10) and the one support computer (13).
a to 19n) and a signal indicating whether or not the bonder computer (10) of the plurality of bonders (1a to 1n) is in a state where the position to be bonded cannot be detected, the support computer cyclically. (13) to send to the plurality of computer interconnection switches (19a-1)
9n) is controlled, and the image connected to the bonder indicating that the position to be bonded cannot be detected when a signal indicating that the position to be bonded cannot be detected is generated from the bonder computer (10). By controlling the signal selection and capturing switch and the interconnection switch to be in the ON state, the signal obtained from the image pickup device of the bonder whose position cannot be detected is converted into the support image reading device (1).
2) and control means for sending to the bonder whose position cannot be detected, which shows the position to be bonded detected by the supporting computer (13) based on the data of the supporting image reading device (12). The present invention relates to a bonding apparatus including:

[Advantageous Effects of the Invention] The present invention has the following advantageous effects.

(A) Since the supporting image capturing device (12) and the supporting computer (13) that handle multi-valued image data or color image data having three or more gradations are provided, the image recognition rate can be improved.

(B) Since one support image capturing device (12) and one support computer (13) are provided for a plurality of bonders (1a) to (1n) and these are shared, the cost is increased. Can be suppressed.

(C) The upper support image capturing device (12) and the support computer (13) are the lower bonders (1a) to (1n).
Since it is used only when the position cannot be detected, the image recognition time on average does not differ much from the case of only the lower bonders (1a) to (1n).

(D) Computer interconnection switch (19a)
Since (19n) is cyclically controlled, it is possible to automatically monitor the plurality of bonders (1a) to (1n) and quickly respond to a position detection failure.

[Embodiment] Next, an LED wire bonding system according to an embodiment of the present invention will be described with reference to FIGS.

First, the state of LED wire bonding will be described with reference to FIG. In FIG. 3, (20) is a dice-shaped LED chip having a substantially positive surface shape, (21) is a lead frame, and (22) and (23) are leads in the lead frame (21). The chip (20) is fixed to the dish-shaped recess (24) provided at one end of the lead (22) with a silver paste (25).
The periphery (26) of the bottom of the dish-shaped recess (24) is not covered with the silver paste (25). (27) is an Au wire, one end of which is ball-bonded to the center of a minute circular electrode (28) on the chip surface (the Au tip is made into a ball shape, and this ball-shaped section is subjected to thermocompression bonding while applying ultrasonic energy. The other end is nail head bonded to the end face of the lead (23). In the following, the present system will be described by taking wire bonding to the electrode (28) on the chip surface as an example.

The system for bonding shown in FIG. 3 is configured as shown in FIGS. 1 and 2. FIG. 2 is a conceptual diagram of a system according to the present invention shown in correspondence with the conventional system of FIG. In this system, high speed, versatility,
Multiple (8 in this example) automatic bonders (1a) (1b) to (1n) that are relatively low-grade image recognition functions with a focus on low cost
Image recognition support device (4) having a higher-grade image recognition function than the automatic bonders (1a) to (1n)
And a switching device (5) for selectively connecting the plurality of automatic bonders (1a) to (1n) and one supporting device (4).

FIG. 1 shows the system conceptually shown in FIG. 2 in more detail. Each of the automatic bonders (1a) to (1n) is a camera for imaging the bonding stage (6) and the LED chip (20) mounted on the lead frame (21) on the stage (6).
(7), a bonding tool (8) that can move in the X, Y, and Z axis directions integrally with the camera (7), an image reading device (9), a computer (10), and a controller (not shown) for each unit Including etc.

The camera (7) as an image pickup device is a general one which sends out an analog image signal in a television signal format. camera
The image reading device (9) connected to (7) includes an A / D converter and a frame memory, and outputs an analog image signal corresponding to the light and darkness of a subject (chip) obtained from the camera (7).
The / D converter converts the image data into binary image data and writes the image data in the frame memory. The frame memory has an address corresponding to the pixel of the camera (7),
Image data is written here. Note that another memory such as a field memory may be used instead of the frame memory.

The computer (10) is associated with the image reading device (9), the camera (7), the bonding tool (8), the bonding stage (6), etc., and written in the frame memory of the image reading device (9). An image calculation is performed by the pattern matching method based on the binarized image data to determine the bonding position. The computer (10) also controls and manages the automatic bonder (1a) including the movement of the camera (7), the bonding tool (8), the bonding stage (6) and the check and notification of the operating state.

In addition, in this example, since a plurality of (30) LEDs are mounted on the single lead frame (21), a plurality of (5
Each LED bonding position (coordinates) is continuously determined and stored in a memory, and then data is read from the memory to wire bond a plurality of LEDs continuously. Automatic bonder (1
a) to (1n) are configured.

The internal structure of the second to nth automatic bonders (1b) to (1n) is not shown, but the first automatic bonder (1a) is not shown.
Is configured exactly the same as. Each automatic bonder (1
The items a) to (1n) correspond to the automatic bonder A having a low image recognition function shown in Table 1.

The support device (4) includes an image capturing device (12) and a computer (13).
Including and This support device (4) is equipped with n automatic bonders (1a).
.About. (1n) are provided one by one, and are selectively connected to the automatic bonders (1a) to (1n) by the switching device (5).

The switching device (5) uses image signal lines (14a) (14b) ... (14n) for transmitting the output of the camera (7) of each automatic bonder (1a) to (1n) as a common image signal acquisition line ( 15) has switches (16a) (16b) ... (16n) for selectively connecting to the signal line (1) connected to each automatic bonder side computer (10).
Switch (19a) (19b) ... (19n) for selectively connecting 7a) (17b) ... (17n) and the signal line (18) connected to the assisting device side computer (13) Have. Each automatic bonder side computer (10) and support device side computer (13)
And communicate with each other.

The image capturing device (12) of the support device (4) is an automatic bonder (1a) ~
The analog image signal from the camera (7), which has become unrecognizable in (1n), is taken in through the switching device (5) and converted into digital image data of 256 gradations by the A / D converter. Then, it is configured to write to the frame memory. Compared with the image reading device (9) that performs A / D conversion into two gradations of automatic bonders (1a) to (1n), the image capturing device of the support device (4)
Since (12) is A / D converted into 256 gradations, it is possible to obtain image data that can be finely identified.

The assisting device side computer (13) is configured to execute a three-step image recognition process as shown in principle in FIGS. 5, 6 and 7. That is, the image data of 256 gradations is read from the frame memory of the image capturing device (12),
Based on this data, high-grade image recognition calculation processing is performed,
The result is sent to the unrecognizable automatic bonders (1a) to (1n) by the signal line (18). In this example, the switches (16a) to (16n) and (19a) to the switching device (5)
(19n) is controlled by the computer (13) on the support device side, the computer (13) uses the line (5a) to switch.
It is connected to (5). In addition, the support device side computer
The item (13) is also configured to check the operation status of the automatic bonders (1a) to (1n) and totalize them.

(Operation) Multiple automatic bonders (1a) to (1n) automatically detect the center coordinates of the electrode (28) to which the Au wire (27) should be wire-bonded, based on the image recognition and processing functions of each. Then, wire bonding is performed at that position. Automatic bonder (1a) ~ (1
When the above condition (a) does not occur in n), the support device (4) patrols (checks and counts the operating state) the automatic bonders (1a) to (1n). That is, the switching device (5) cyclically connects the signal lines (17a) to (17n) to the signal line (18) in response to a command from the support device (4). This operation is easily achieved by forming the switches (19a) to (19n) in a multiplexer configuration. When the automatic bonder side computer (10) and the assisting device side computer (13) are cyclically connected via the signal line (18), the automatic bonder side computer (1
The image recognition state display signal (a signal indicating impossible or possible) obtained in 0) is given to the assisting device side computer (13).
Since the switches (19a) to (19n) can be switched at extremely high speed, the assisting device side computer (13) has multiple automatic bonders.
This means that (1a) to (1n) are always monitored.

In this patrol, when the support device (4) detects that the first automatic bonder (1a) has fallen into the situation (a) (the image cannot be recognized) described above, based on a command from the support device (4). The image signal line (14a) of the automatic bonder (1a) whose patrol was interrupted by the operation of the switching device (5) and the image was unrecognizable is connected to the line (15) via the switch (16a), and the camera ( The analog image signal obtained from 7) is input to the image capturing device (12), A / D converted into 256 gradations, and written in the frame memory. Support device side computer
(13) performs image recognition based on the image data of 256 gradations in the frame memory by a higher-level algorithm than the automatic bonder computer (10). As a result, target image recognition is possible except in exceptional cases.
The center coordinates of the electrode (28) to be wire-bonded to the line (27) are determined.

When the first automatic bonder (1a) cannot be recognized, the switch (19a) is turned on, and the data indicating the bonding position obtained from the computer (13) is the automatic bonder (1a).
Computer (10). In addition, line (17a)
~ (17n) and (18) are used in time division for both patrol and position data transmission. The automatic bonder-side computer (10) writes the data determined by the support-apparatus-side computer (13) in the internal memory, and executes the intended automatic bonding based on this.

When the data transfer from the assisting device side computer (13) to the automatic bonder side computer (10) ends, the assisting image recognition function of the assisting device (4) ends, so the patrol operation is restarted.

When the disabled automatic bonder (1a) completes the predetermined bonding with the support of the support device (4), it returns to normal operation and starts the bonding by its own image recognition processing.

In some exceptional cases, the image cannot be recognized even by the support device (4). In this case, bonding is not performed there (pass bond), and the process proceeds to bonding to the next chip (20). The pass-bonded LED is excluded as a defective product in the subsequent characteristic inspection process of the product. Of course, when the image recognition of the support device (4) fails, the corresponding automatic bonders (1a) to (1n) issue a call signal and continue to stop, so that the worker (2) can also take countermeasures. Can be set.

When the above-mentioned situations (b) to (d) occur, the corresponding automatic bonders (1a) to (1n) generate a call signal and stop as in the conventional case, and the worker (2) must take corrective action. wait.

(Image Recognition Operation) Next, image recognition of the automatic bonders (1a) to (1n) and the support device (4) will be described. First, in the automatic bonders (1a) to (1n), an analog image signal taken in by a TV camera (7) using an image pickup tube is converted into binary image data by an image reading device (9). That is, pixels having a predetermined density (brightness) or higher are white (logic 1), and pixels having a density lower than the predetermined density are black (logic 0) to simplify the image information into two gradations. Then, the binarized image data is written in the frame memory, the computer (10) compares the image data with a preset binarized standard pattern, and the region in the binarized image data that matches the standard pattern is compared. To detect.
In this example, since the circular electrode (28) of the chip is brighter than the surroundings, a standard pattern including a white (logic 1) circle corresponding to the circular electrode (28) is set. When the coordinates of the area matching the standard pattern are detected, the coordinates of the bonding point preset for the standard pattern can be determined, and therefore the bonding is performed at these coordinates. In this example, the center point of the white (logic 1) circle is the bonding point. The number of pixels in the frame memory of the image reading device (9) is 120 × 120 = 14400, and the camera (7)
Corresponds to the half sampling of the pixel in the central part of the tube surface (about half of the tube surface). Of course, the pixel at the central portion of the tube surface of the camera (7) may be extracted in a non-sampling state.

By the above image recognition, wire bonding is performed for about 96% of the chips (20) even when the chip is low. In other words, from the above-mentioned problems (a) and (b) peculiar to the LEDs, when the number is large, the image recognition of the chip (20) becomes impossible (a state in which a region matching the standard pattern cannot be detected) at a rate of several%. In this case, the operator (2) instantaneously shifts to image recognition by the support device (4). The support device (4) executes three-step image recognition. FIG. 4 shows an example of an image of a region (29) which is the target of pixel calculation processing in the first stage, and an analog image signal taken by a TV camera is projected on a cathode ray tube (not shown). Is. Although the image actually has a slight density difference, in the drawing, the dark portion is shown by hatching and the whitish portion is shown by white. Reference numerals in FIG. 4 correspond to respective portions in FIG. In FIG. 4, black stripes (30) and (31) are formed on the chip (20) which may make the image impossible. Stripes (31)
If the electrode (28) is touched, the image cannot be recognized by the automatic bonders (1a) to (1n). A dark ring-shaped portion (32) appears around the circular electrode (28). This ring shaped part (32)
Occurs because the electrode (28) is a truncated cone-shaped protrusion and the inclined side surface becomes dark.

In the first step of image recognition in the support device (4), a whitish quadrangle image having a size in a predetermined range is extracted from the image of FIG. 4 to obtain a rough position of the chip (20). To explain this in detail, the image data is read from the frame memory of the image capturing device (12) in which the analog image signal from the TV camera (7) is written as image data of 256 gradations (256 values), and the computer ( 2 in 13)
Convert to digitized image data. Then, pixel calculation processing is performed so as to perform square feature extraction based on the binarized image data. The square feature extraction is performed, for example, by ignoring white portions other than the square having a predetermined size and removing unnecessary black portions in the predetermined square. FIG. 5 shows the principle for determining the chip-corresponding square image (33) indicated by the dotted line and its center position in the first step. A logic "1" is obtained in the white part and a logic "0" is obtained in the black part corresponding to each pixel. As a result, a distribution pattern of logic "1" corresponding to the pattern of the chip (20) is obtained. Since the threshold value for obtaining the data in the first stage is different from the threshold value for reading the data into the frame memory on the automaticponder side, not only the electrode (28) but the entire upper surface of the chip (20). Is detected as white. In FIG. 5, the logical "1" is ideally distributed in a square, but the distribution is not necessarily such. Therefore, the sum of the data arranged in the Y-axis direction at each position on the X-axis is calculated as x ₁ = 0, x ₂ = 0, x ₃ =
4, x ₄ = 4, x ₅ = 4, x ₆ = 4, x ₇ = 0, x ₈ = 0, and this x ₁ , x ₂ , ... x ₈ is the width of the chip (20) The addition values A ₁ , A ₂ ... A ₈ of the number of pixels (4 pixels) corresponding to are sequentially obtained while moving to the right one pixel at a time. In this example, four sums are sequentially obtained, so A ₁ = x ₁ = 0, A ₂ = x ₁ + x ₂ = 0, A ₃ = x
₁ + x ₂ + x ₃ = 4, A ₄ = x ₁ + x ₂ + x ₃ + x ₄ = 8, A ₅ = x ₂ + x ₃ + x ₄ + x _{5
= 12, A 6 = x 3} + x 4 + x 5 + x 6 = 16, A 7 = x 4 + x 5 + x 6 + x 7 = 1
2, A ₈ = x ₅ + x ₆ + x ₇ + x ₈ = 8, and x ₆ on the x axis where the maximum value A ₆ is obtained is the right end of the square image (33) corresponding to the chip (20). I understand. Similarly at each position on the y axis, y ₁ , y
₂ , y ₃ ...... y ₈ is calculated, and B _{1 to} B ₈ are calculated so as to correspond to A _{1 to} A _8, and the upper end of the square image (33) is at the position of y ₇ on the y axis. decide. Since the size of the chip (20) (4 pixels in both the x-axis and y-axis directions) is known in advance, the position on the x-axis and the position on the y-axis of the quadrangle image (33) can be set to x ₃ to.
It can be determined as x ₆ , y _{4 to} y ₇ . Square image (3
The position of 3) can be determined based on x ₆ and y ₇ . In FIG. 5, in order to simplify the drawing, the number of pixels is 8 × 8 = 6.
However, in practice, the number of pixels is, for example, 18 × 19 = 342. The frame memory of the image capturing device (12) of the support device (4) is, for example, 128 × 128 =
It has 16384 pixels. This is 5 in camera (7)
This corresponds to 1/4 sampling of 12 × 512 pixels. The 18 × 19 pixels in the first stage are obtained by sampling the frame memory by 1/6. Since data is written in the frame memory with 256 gradations, binarized data is obtained with an appropriate threshold value, and this is used as the data of each pixel in FIG.

When the extraction of the rectangular image (33) is completed, the computer (13) determines whether or not the condition (which is set in advance) for this to be the chip (20) is satisfied. Move on to the second stage. If it is not satisfied, the binarization threshold value is changed and extraction is performed again. If the determination condition is still not satisfied, the image is not recognized.

In the second step, the pixel calculation processing target area is narrowed based on the result of the first step, and a more accurate position of the chip (20) is obtained. At this stage, as shown in FIG. 6, the areas of the horizontal band (35) and the vertical band (36) are set to be the target of the pixel calculation processing so as to be substantially aligned with the center (34) of the quadrangle image (33). Showing tip (20)
The position of each side of the square image (39) corresponding to is calculated. In this case, the gradation of the image data which is the target of the pixel calculation processing is set to 256 gradations so that it can correspond to a slight density difference. That is, the data of 256 gradations obtained from the frame memory of the image capturing device (12) is used as it is, and each pixel data of the horizontal band (35) and the vertical band (36) is processed according to the same principle as in FIG. Then, the position of each side of the square image (39) is accurately obtained.
In FIG. 6, the horizontal band (35) and the vertical band (36) are shown as 11 × 3 pixels for convenience of illustration, but in reality, the horizontal band (35) is 1 pixel.
8 × 5 pixels, the vertical band (36) is 5 × 14 pixels, and it is obtained by 1/4 sampling of the 128 × 128 pixel frame memory of the image capturing device (12).

In the third step, the center of the position of the chip (20), that is, the quadrangular image (39) obtained in the second step is brought into alignment with one another as shown in FIG.
The area of the horizontal band (37) consisting of the x-direction array of pixels and the vertical band (38) consisting of the y-direction array of 1 pixel is specified, pixel calculation processing is performed, and the position (center coordinates) of the circular electrode (28) is determined. Ask. At the start of the third stage pixel calculation processing, the horizontal band (37) and vertical band (3
The intersection (40) of 8) is at the center of the square image (39). here,
A dark ring-shaped area (32) around the circular electrode (28)
(See FIG. 4) to detect the position of the circular electrode (28). The gradation of the image data that is the target of pixel calculation is set to 25 so that it can correspond to the slight density difference on the surface of the chip (20).
There are 6 gradations. With respect to the calculation result, it is determined whether or not the preset condition for the electrode (28) is satisfied, and if not satisfied, it is treated as image unrecognizable. Note that, in FIG. 7, for convenience of illustration, the horizontal band (37) and the vertical band (38) are each 9 pixels, but actually, the horizontal band (37) is, for example, 30 pixels and the vertical band (38) is For example, the number of pixels is further increased, such as 25 pixels. Pixels in the horizontal band (37) and vertical band (38) in FIG.
Correspond with 1 and obtained in the non-sampling state.

The image data described so far are all monochrome (black and white). However, in the image recognition in the support device (4),
It may handle color image data. For example, red, green,
If the image data is separated into three primary colors of blue, even if the image data of each of the three primary colors is converted into binary image data corresponding to the density and handled, the same information as that of the mono-valued multi-valued image can be obtained. You can

The assisting device (4) is a condition for individual image recognition for each of the automatic bonders (1a) to (1n) (threshold value for binarization,
Chip size, electrode size, etc.) are set. Even when all the automatic bonders (1a) to (1n) are used to manufacture LEDs with the same specifications using the chips (20) with the same specifications, for example, the brightness of the lights of each automatic bonder (1a) to (1n) Is subtly different, and the threshold value is set for each automatic bonder (1a) to (1n). When the need for parallel production of multiple products and the use of simple chips (20) in units of automatic bonders (1a) to (1n), the chip size and electrode size are individually set accordingly.

Due to the three-step image recognition by the support device (4), the chip (20) that could not be recognized by the automatic bonders (1a) to (1n)
Of these, 90% or more can be image-recognized, and even when the total is low, about 99.6% of the chips (20) are automatically wire-bonded. As for the chip (20) whose image cannot be recognized when the number is large, about 0.4%, the pass bond or call signal is issued as described above. The image recognition and processing time of the support device (4) is an average of 0.7 seconds, and is a value suitable for managing the eight automatic bonders (1a) to (1n) under the above image recognition rate. Has become. The image recognition and processing time of the support device (4) depends on the automatic bonder (1
It is longer than the image recognition time of a) to (1n). This is contrary to the general relationship that higher systems are faster than lower systems. As described above, since the image recognition and processing device is not required to have high speed, the support device (4) operates an inexpensive general-purpose 16-bit personal computer under a program written in C language and assembler language. Pixel calculation processing.

In this system, even if the operator (2) performs a coping operation when the image of the support device (4) cannot be recognized (a method in which the operator (2) has a greater burden than the path bonding method), one person The worker (2) was able to manage the eight automatic bonders (1a) to (1n) smoothly. Moreover, one worker
When (3) manages three automatic bonders (1a) to (1n) in Fig. 8, or one operator (2) manages five automatic bonders (1a) to (1n). Compared to the case of managing in the form of FIG. 9, the automatic bonder (1a)-
The operation rate of (1n) improved by 25%. If the image recognition and processing time of the support device (4) is shortened by using a computer with a high calculation processing speed, the device price will rise, but an automatic bonder (1) that can be managed by one worker (2) 1a)
The number of ~ (1n) can be further increased.

The following Table 2 shows the operating status of this system (when the image recognition and processing time of the assisting device (4) is 0.7 seconds),
Corresponds to Table 1.

In addition, the numerical value in the parenthesis indicates the value when the assisting device (4) is operated by the path-bonding method when the image cannot be recognized.

As is clear from Table 2, this system has a high image recognition rate.
The high-quality image recognition and processing of 99.6% and the image recognition time of 0.23 seconds operate as if the automatic bonder for high-speed pixel calculation processing is used in the form of FIG. 8, and the bonding processing amount is at a high level. The price of the device is extremely low compared to the case where eight automatic bonders for high-grade image recognition processing and high-speed pixel calculation processing are used as described above. Therefore, it can be said that a system with extremely high mass productivity is constructed when the labor cost, the amount of bonding processing, and the apparatus price are comprehensively viewed.

The image recognition time of the automatic bonders (1a) to (1n) or this system is the time from the start of image data acquisition to the automatic bonders (1a) to (1n) until the bonding point is determined. . The image recognition processing time of the support device (4) means that the support device (4) receives a signal indicating that image recognition is impossible from the automatic bonders (1a) to (1n), and after the image recognition step, the automatic bonder (4) It is the time until a command is issued to 1a) to (1n). In other words, when the image cannot be recognized by the automatic bonders (1a) to (1n),
It is the time when the assisting device (4) interrupts the automatic bonders (1a) to (1n).

[Modification]

The present invention is not limited to the above embodiment, but can be modified. For example, the image recognition function of the support device (4) uses the automatic bonders (1a) to (1
Any device may be used as long as it is capable of image recognition at a higher level than n). Further, the control of the switching device (5) may not be controlled by using the assisting device side computer (13) but may be controlled by a dedicated control circuit.

[Brief description of drawings]

1 is a block diagram showing a bonding system according to an embodiment of the present invention, FIG. 2 is a conceptual diagram of the system of FIG. 1, FIG. 3 is a perspective view showing an LED to be bonded by the system of FIG. 1, and FIG. FIG. 5 is a diagram illustrating an image of an object (LED) on a cathode ray tube, FIG. 5 is a diagram illustrating in principle the image recognition in the first stage in the assisting device of FIG. 1, and FIG. 6 is an assisting device. Of the second stage of image recognition in principle, FIG. 7 is a diagram of principle of image recognition in the third stage of the assisting device, and FIGS. 8 and 9 show conventional systems, respectively. It is a block diagram. (1a) to (1n) ... automatic bonder, (4) ... support device, (5) ... switching device, (7) ... camera, (9) ... image reading device, (10) ... computer, (12) ... image Capture Device, (13) ... Computer.

Claims (1)

[Claims] 1. A bonding tool (8) for bonding a die or a wire, an image pickup device (7) for picking up an image recognition area for bonding, and connected to the image pickup device (7), An image reading device for a bonder (9) for forming monochromatic binarized image data in which the density of each pixel is expressed in two gradations based on an analog output signal obtained from the image pickup device (7) and storing the image data in a memory. And the image reading device for the bonder (9) and the bonding tool (8)
Connected to, and automatically detecting a position to be bonded by image recognition of the binarized image data as a target of pixel calculation processing, and automatically bonding a die or a wire to the detected position. In addition, a plurality of bonders (1a to 1n) each including a bonder computer that sends a signal indicating this when the position to be bonded cannot be detected, and a plurality of bonders (1a to 1n). A plurality of image signal selection / acquisition switches (16a-16n) respectively connected to the image pickup device (7), and a plurality of image signal selection / acquisition switches (16a-1)
6n) is selectively connected to the image pickup device (7) of the plurality of bonders (1a to 1n), and the density of each pixel is set to the third floor based on the analog output signal obtained from the image pickup device (7). A single support image capturing device (12) for forming multi-valued image data or color image data expressed in a tone or higher and storing it in a memory; and a pixel arithmetic processing of the multi-valued image data or color image data. In order to selectively form a data transmission path with one support computer (13) for automatically detecting the position to be bonded by image recognition including the target of the above, and notifying the bonder of the detected position. A plurality of computer interconnectors provided between the bonder computer (10) of the plurality of bonders (1a-1n) and the one support computer (13), respectively. Switches (19a to 19n) and a signal indicating whether or not the bonder computer (10) of the plurality of bonders (1a to 1n) is in a state in which the position to be bonded cannot be detected cyclically. The plurality of computer interconnection switches (19a to 19n) are controlled so as to be sent to the bonding computer (13), and the bonder computer (10) generates a signal indicating the undetectable state of the bonding position. Sometimes the image signal selecting and capturing switch and the interconnection switch connected to the bonder, which indicates that the position to be bonded cannot be detected, is controlled to be in the ON state, and from the image pickup device of the bonder whose position cannot be detected. The obtained signal is sent to the supporting image capturing device (12), and based on the data of the supporting image capturing device (12). Bonding apparatus, characterized in that a control means for sending the bonder became the position undetectable indicating the position to be bonded detected by the computer (13) for the assistance are.