JPH01140737A - Wire bonding apparatus - Google Patents

Abstract

PURPOSE:To execute a wire bonding operation while an automatic inspection is being carried out without lowering a production capacity by installing the following: a pattern recognition part; an image memory which can memorize an image of a pressure-bonded part after a bonding operation in connection with a wire number and the pressure-bonded part. CONSTITUTION:A pattern recognition part and a bonding control part are equipped with separate central processing units (CPU's) 18, 20; in the pattern recognition part, an image memory 12 used to temporarily memorize an image of an object to be inspected is constituted. Accordingly, immediately after a wire bonding operation or after the completion of a whole bonding operation for one IC, an image of a bonding part is stored in the memory 12 in conjunction with the bonding control part; an inspection is carried out during a next wire bonding operation or during the bonding operation of the IC and during a transfer duration of the IC and in parallel with an operation of a wire bonding mechanism 6. By this setup, the effective bonding operation due to a parallel operation can be achieved.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to wire bonding of ICs, etc., and in particular, uses a pattern recognition method or the like immediately after wire bonding to perform automatic visual inspection while reducing assembly throughput. The present invention relates to a bonding method that is effective for wire bonding without wire bonding.

[Conventional technology]

Conventionally, a bonding apparatus that performs an automatic visual inspection immediately after wire bonding has a wire bonding mechanism, its control circuit, an imaging system for detecting the position of the bonding part, and a recognition system, as described in Japanese Patent Laid-Open No. 57-134944. It consists of a circuit, an image processing circuit that performs the inspection, and a central control circuit that controls all of these, and the method is such that bonding and inspection are performed sequentially.

[Problem that the invention seeks to solve]

It has been argued that the above conventional technology can perform inspection immediately after wire bonding, but according to this device, for bonding, the position of the pellet is first detected, and then the wire bonding is performed, and each bonding area is then inspected. This is the sequence for performing bonding inspection. Therefore, compared to a wire bonding apparatus that does not perform bonding inspection, it is thought that the throughput is reduced by the increased number of inspection steps. In other words, compared to an apparatus that does not perform an inspection, the time required for inspection is added, and no consideration has been taken to reduce this loss and not reduce assembly throughput. Therefore, there was a problem in that the production capacity was significantly reduced and overall effects could not be obtained.

SUMMARY OF THE INVENTION An object of the present invention is to provide a wire bonding device that can perform wire bonding while performing automatic inspection without reducing the production capacity of the bonding device, even when automatically inspecting the bonding part immediately after bonding. .

[Means for solving the problem] □The above purpose is to provide a pattern recognition section that detects the pad position by pattern recognition for wire bonding, and a bonding control section that performs bonding to the position determined by the pattern recognition section. The two have separate central processing units (CPUs), and the pattern recognition section also configures an image memory to temporarily store images of the inspection target, so that it can be used immediately after wire bonding or after all bonding for one IC. After completion, the image of the bonding section is stored in the memory in conjunction with the bonding control section, and the mechanical section operation of the wire bonding is performed during the first wire bonding operation time, during the next IC bonding operation time, during the IC transportation time, etc. This is achieved by testing in parallel.

[Effect]

The control of the wire bonding mechanism, the pattern recognition process, and the image recording process are operated independently by separate CPUs, and automatic inspection by the pattern recognition unit is possible during wire bonding operation, making it possible to perform effective bonding operations in parallel. It becomes possible. In addition, since both operations are synchronized by the central processing unit (CPU) through communication, the operation of the XY child table bonding mechanism for image capture and the image data recording operation of the pattern recognition unit may malfunction. You can definitely do it without any trouble.

〔Example〕

Hereinafter, one embodiment of the present invention will be described with reference to FIGS. 1, 2, 3, and 4.

FIG. 1 shows the configuration of a wire bonding apparatus according to the present invention. In FIG. 1, the target ICI supplied to the bonding stage 4 by the transport device 2 is first moved while moving the field of view by driving the XY child table 7 on which the imaging system 5 is mounted, in order to inspect the position of the bonding part. The positions of two predetermined representative points within the pellet are detected by the pattern recognition unit 32, and the bonding pad position within the pellet is calculated. Further, since the bonding position of the external lead is maintained accurately only by mechanical positioning, it can be positioned without detecting the position. The bonding operation in which bonding coordinates are determined in this way and then bonding is performed between each pad and lead using the bonding tool 6 and the XY child table 7 is exactly the same as in the prior art. However, the present invention is characterized by a repetitive operation from the end of bonding of one IC to the end of bonding of the next IC, and the characteristic operation is as follows.
The operation will be explained from FIG. 2 and FIG. 4 according to the operation flow shown in FIG. In the flow of FIG. 2, in operation 1, the target ICI shown in FIG. 1 is transferred from the loader 2 to the bonding table 4.
At the same time, the bonded IC is sent to the unloader 3 and a series of bonding operations described below are performed.

In operation 2, the IC 4 supplied to the bonding table 4 is subjected to detection of the pad position on the pellet for wire bonding.

As is already well known, the pad position detection of the pellet is performed by moving the TV camera 5 inside the pellet using the XY child table 7 shown in FIG. The image data 7 is supplied to the pattern recognition device 32, and representative positions are detected from the image data to recognize each pad position. Then, in operation 3, based on the pad position data detected in operation 2 that is sent from the pattern recognition device 32 to the bonder control unit via the communication signal 19, the bonder control central processing unit 20 and the bonding tool Wire bonding is performed by operating the XY child table and the bonding tool 6 with a command value of 5.26 from the table driving device n.

At this time, the position of the external lead of the IC can be determined by means of obtaining position data in advance only by mechanical positioning, or by position detection using pattern recognition similar to pellet position detection.

Next, operation 3 and subsequent steps will be explained. In the conventional method, the second
As shown by the dotted line in the flowchart of the figure, the process returns to operation 1, but in the present invention, operations 4 and 8 are performed in order to inspect the bonding crimped portion after bonding. In addition, if this inspection is simply performed after bonding is completed, the time for each inspection will be accumulated, and the overall throughput will be less efficient by the amount of inspection time. The ratio of bonding to throughput is minimized. That is, in operation 4, based on the pad position and lead position that have already been detected, the image of the bonding and crimping part, which is the inspection part, is used to connect the XY child table (not shown in FIG. 1) to the control central processing unit 20 and the table drive circuit. The image is stored in the image memory 12 of the pattern recognition device by the TV camera 5 driven by the command value. and,
When operation 4 is completed, operation for wire bonding is performed as operation 5. Action 6. Proceed with action 7. Here, action 5 is the same action as action 1 already explained, and action 6 = action 2.
．． Operation 7=Operation 3. Operation 8 operates in parallel with operations 5 to 7, and sequentially reads image data from the image memory 12 for each image to be inspected, and performs the inspection while being inspected by the image processing area ji2i 16 and the pattern recognition central processing unit 18. At the same time as storing the results, it determines whether the product is good or bad and outputs the results. By adopting such an operation method, it is possible to inspect the crimped shape immediately after bonding compared to the conventional method, and highly reliable wire bonding can be realized in a short time, which is only the time required to capture an image of the inspection target part.

Next, the system configuration functions for realizing operations 4 to 8 will be explained. After the bonding in operation 3 is completed, the image to be inspected in operation 4 is captured in the bonding areas indicated by the diagonal lines (■ to ■), as shown in the plan view of the state after bonding seen from directly above in Figure 3. capture and store image data. Normally, all points can be inspected for bonding, but here we will explain the bonding inspection for 8 points and 4 wires as an example.

Due to the accuracy and capacity of the imaging system 5 and pattern recognition unit 32, it is impossible to capture all the pellets and leads shown in FIG. 3 in one screen due to the inspection performance. is enlarged at a predetermined magnification, and only the image shown in the diagonal frame is recorded. Therefore, the imaging system 5 shown in FIG. 1 drives the XY table correction and
Position to ~■. This positioning is performed based on the coordinate position for wire bonding, the XY child table, and the drive signal 26 from the bonding tool drive circuit based on the command value 21 from the bonding control central processing unit 20. Let it work. Image input is performed by converting the video signal 7 from the imaging system 5 into the digital signal 9 in the human/D conversion circuit 8 of the pattern recognition unit 32 and converting it into the image memory 1.
Record in 2. This image recording is performed by a command value 13 from the pattern recognition central processing unit 18.

At this time, the field of view movement of the input point of the imaging system 5 and the image input operation are performed by synchronizing the central processing unit 18 of the pattern recognition section 32 and the central processing unit for bonding control using the communication signal 19. In other words, if the field of view movement and image capture by the This is because a misalignment will occur.

In addition, in the operation flow of FIG. 2, operation 6 and operation 8 for performing pattern recognition are processed in parallel, but in the configuration of this embodiment, there is only one pattern recognition device, so operation 6 and operation 8 are processed in parallel. It is necessary to give priority to either operation 8. In this case, unless priority is given to operation 6, it will be impossible to perform operations 7 and 8 in parallel, resulting in poor efficiency. Therefore, action 5. Action 6. In order to minimize the waiting time and increase the efficiency of the processing in operation 7, the pattern recognition central processing unit 18 and the bonding control central processing unit 20 need to operate in synchronization using the communication signal 19.

Image data is then recorded in a new memory area as the visual field moves. This image recording method is shown in Figure 3.
This will be explained using FIG. The Eirin signal 7 of the input image ■～■ is digitized by the A/D converter 8 and becomes the signal 9, which is an image memo! J 12 a to 12 n. Then, the pattern recognition central processing unit 18 receives the image human power request command, input field of view, bonding wire number, etc. from the bonding control central processing unit 20 as a communication signal 19.
and supplies the image input command 13 to the image memory control circuit 39. Image memory control circuit 3
In step 9, the image is recorded in the memory of the image memory 12 associated with the input visual field number in synchronization with the synchronization signal 11 from the image processing circuit 16, as shown in (1) to (2) in FIGS. 3 and 4.

When the image capture in operation 4 shown in FIG. 2 is completed, the first
The target ICI on the bonding stage shown in the figure is sent to the next station via transport system 3, and
From there, the product before bonding is supplied to the bonding stage 4. Then, operation 6 is IC pad position detection, and operation 7 is wire bonding. In parallel with these operations, in operation 8, an automatic inspection is performed based on the image of the bonding part input in operation 4, and the inspection results 34
, 35 are output. In operation 8, if pellet position detection and automatic inspection are performed in the same recognition processing circuit, operation 8 must be temporarily interrupted during operation 6.

However, this does not apply when using another system. Then, in synchronization with the end of operation 7, operation 8 is realized, and the process returns to operation 4, where wire bonding is performed sequentially.

Next, regarding the automatic inspection in operation 8, see Figures 5 and 6.
This will be explained using figures. FIG. 5 shows the bonding wire 40
This is a plan view showing an image of the crimped part of the pad 37, and shows a search area for inspection and a measurement location for determining the quality of the crimped part.

In FIG. 5, Nl is the first neck position because it is the base of the bonding wire 40 that begins to collapse when it is crimped, N2 is the second neck position because it is the end of the collapse, and Pc is the center position of the bonding pad 37. , Bc is the center position of the crimp part, and TL is N1
The most extreme position on the center line of the crimp part in the direction of the horse from Wl
, W will be explained below by defining the measurement point as the furthest point position of the collapse width in the direction perpendicular to the center line. The automatic inspection is performed according to the procedure of the flowchart shown in FIG. First, images of the crimped part and its surroundings recorded in the image memory 12 are selected. Next, the selected image is binarized by the binarization circuit 10 shown in FIG. The image data is supplied to the image processing circuit 16. At this time, the binarization threshold level for binarization can be determined in advance by teaching, or can be determined by analyzing data in the image memory. Next, from the binarized image of only the bonding crimping part, the approximate center of gravity of the crimping part of the bonding wire 40 is detected by the image processing circuit 16 and the central processing unit 18 for pattern recognition. This position of the center of gravity is determined by formula +11 from the pixels and coordinates of the crimping part.

Xi: X address of each pixel composing the image divided into regions as the bonding crimping section Yi: Y address of each pixel composing the image divided into regions as the bonding crimping section And, in the wire direction, also detected from the image of the crimping section However, since the positions of pads and leads are already known in wire bonding, it can be easily calculated from this data. Then, once the approximate center of gravity and direction of the crimp portion are known, the position of the neck portion Nl +Nt is next detected. In order to do so, we need to search the neck search range 4 from the direction and center of gravity position.
1 Set a, b, perform recognition processing based on the neck definition described above only in this range, and obtain Nl, N! Detect. The reason why the search range can be set in this manner is that the shape of the crimped portion is almost the same when it is normal. Then, based on N, -N, search ranges 42a and 42b are set for detecting the farthest point in the crimp width direction, and Wl, W,
Detect. In addition, in the wire direction, that is, the direction from N, to N, the distance from N to the farthest point in the wire direction is called the tail, and this farthest point is also detected based on the search range 43 based on N,,N,. and detect the TL. The crimp center position Bc can be detected from these measurement points. As described above, by measuring each point of the crimped part, the deviation amount (ΔX) of the crimped center is determined.

ΔY), crimp width W, and tail length T can be detected. Therefore, the quality of bonding can be determined from this value. In the above embodiment, as shown in the operation flow in FIG.
Although we have explained the method of capturing images after each individual wire bonding is completed, image capturing for inspection is carried out on the table on which the imaging system is placed so that the image capturing is carried out at the same time as the completion of each bonding. It is also possible to perform inspection by controlling the control and pattern recognition unit control, and the time efficiency is also good.

As described above, according to the embodiments of the present invention, by adding an image memory to the pattern recognition section of the wire bonder, the state after wire bonding can be recorded, and the central processing unit (CPU) for wire bonding control Equipped with a central processing unit for pattern recognition, automatic inspection can be performed during wire bonding operation, improving bonder function without reducing wire bonder production throughput and making pass/fail judgments possible.
It also has the effect of improving yield. In addition, for inspection purposes, the conventional technology of imaging and control systems for bonding pad position detection can be shared without adding a new recognition station and imaging system, which reduces the increase in manufacturing costs of wire bonder equipment due to additional functionality. It has the effect of requiring less.

〔Effect of the invention〕

According to the present invention, when a wire bonder performs automatic inspection after wire bonding, 1. Normally, the assembly time would be longer due to the addition of an inspection process, but the assembly time is increased in only a very short time required for image capture. This has the effect of improving yield by improving the reliability of the bonder itself and detecting defects early.

[Brief explanation of the drawing]

Fig. 1 is a configuration diagram of an embodiment of the present invention, Fig. 2 is an operation flowchart of the embodiment, Fig. 3 is a plan view showing the area after IC wire bonding and the image capture range for inspection, and Fig. 4 is a circuit configuration diagram for image recording, FIG. 5 is a plan view showing automatic measurement points in one embodiment of bonding and crimping part inspection, and FIG. 6 is an inspection flowchart of FIG. 5. 5... Imaging system 8... A/D conversion circuit 10... Z value conversion circuit 12... Image memory circuit 16... Image processing circuit 18... Central processing unit for pattern recognition 20... - Bonding control central processing unit n... bonding tool drive and table drive circuit. Agent: Patent Attorney Masao Ogawa ゛−′□゛1し−1＋＋＋＋−”

Claims (1)

[Claims] 1. In a wire bonder that automatically detects the bonding pad position of a pellet using pattern recognition and performs wire bonding, a pattern recognition section that detects the pad position and crimping after wire bonding to the pattern recognition section are provided. A wire bonding apparatus comprising an image memory for storing an image of a crimped part after bonding in association with a wire number and a crimped part in order to determine whether the condition of the part is good or bad. 2. When recording and inspecting images in the image memory, an image processing section that controls the image memory and a wire bonding control section are each provided with separate central processing units, and are operated synchronously to complete bonding. 2. The wire bonding apparatus according to claim 1, wherein the quality of the bonding crimping part is determined during wire bonding operations other than position detection for wire bonding by simultaneously recording images.