KR102455733B1 - Electronic device for manufacturing transistor outline package and method thereof - Google Patents

Abstract

In accordance with one embodiment of the present invention, a transistor outline package manufacturing apparatus may comprise: a lighting unit disposed toward at least one part of a TO package substrate to radiate a light source; a camera photographing an image of the at least one part of the TO package substrate; a deep learning processor performing machine learning based on a dataset, learning a TO package manufacturing process in a good or bad state, determining whether the TO package manufacturing process for the image photographed through the camera is good or bad, and, when the image of the at least one part of the TO package substrate photographed through the camera is inputted, determining whether a result of each TO package manufacturing process is good or bad, based on machine learning; a control unit performing a control operation to output the result of the TO package manufacturing process; and a communication unit configured to receive the dataset for each TO package manufacturing process in order to perform the machine learning by being connected with a wireless network through a wire or wirelessly. Therefore, the present invention is capable of increasing the accuracy of a transistor outline package manufacturing process.

Description

Transistor outline package manufacturing apparatus and manufacturing method thereof

The present invention relates to an apparatus for manufacturing a transistor outline (TO) package and a method for manufacturing the same, and more particularly, to improve the reliability of the manufacturing process by manufacturing a TO package by applying a deep learning technique it is to make

A TO package may consist of a TO header and a TO cap. The TO header forms the basis of the encapsulated component and provides power, where the TO cap can enable seamless optical signal transmission. It is important to form the TO package so that these two components are hermetically sealed to protect sensitive semiconductor components.

In the manufacturing process of the TO package, for each process, a person can visually determine whether the process is good or defective. On the other hand, the TO package is composed of a diode chip, a capacitor, or a wire connecting the components. As the components of the TO package have a size ranging from a centimeter to a millimeter, the human eye checks the accuracy of the process. There may be limitations in determining

An embodiment of the present invention learns images for various TO package manufacturing processes using machine learning techniques, acquires images of transistor outline packages through cameras disposed at various angles, and Based on the determination of good or bad for each process of the transistor outline package process, the purpose is to increase the accuracy of the process.

According to an embodiment of the present invention, an apparatus for manufacturing a transistor outline package includes a lighting unit disposed to face at least a portion of a TO package substrate and irradiating a light source, and a camera for photographing an image of at least a portion of the TO package substrate , by performing machine learning learning based on the dataset, learning the TO package manufacturing process in good or bad state, determining whether the TO package manufacturing process for the image taken through the camera is good or bad, and When an image of at least a portion of the TO package substrate is input through a camera, a deep learning processor that determines whether the result of each TO package manufacturing process is good or bad based on the machine learning, the result of the TO package manufacturing process a control unit controlling the output; and a communication unit configured to receive a data set for each TO package manufacturing process in order to perform the machine learning by being connected to the wireless network by wire or wireless.

According to an embodiment of the present invention, a method for manufacturing a transistor outline package includes a process of irradiating a light source to at least a portion of the TO package substrate through a lighting unit, and an image of at least a portion of the TO package substrate through a camera In order to perform machine learning, a process of receiving a dataset for each TO package manufacturing process through a wireless network, performing machine learning learning based on the received dataset, and performing good or bad state The process of learning the TO package manufacturing process of, when an image of at least a portion of the TO package substrate is input through the camera, determining whether the result of each TO package manufacturing process is good or bad based on the machine learning and outputting a result of the TO package manufacturing process.

According to an apparatus and method for manufacturing a transistor outline package according to an embodiment of the present invention, various images for each process of the transistor outline package are photographed, and the photographed image is a good or bad process by using a machine learning technique It may be determined whether the state is indicated, and accordingly, the accuracy of the transistor outline package manufacturing process may be increased.

1 is a block diagram illustrating an example of a configuration of a TO package manufacturing apparatus according to an embodiment of the present invention.
2 is a flowchart illustrating a method of manufacturing a transistor package according to an embodiment of the present invention.
3A and 3B are diagrams illustrating an example of a TO package to which a chip die is bonded according to an embodiment of the present invention.
4 is a diagram illustrating an example of a wire bonding process according to an embodiment of the present invention.
5 is a diagram illustrating an example of a cap welding process according to an embodiment of the present invention.
6 is a flowchart illustrating a method of manufacturing a transistor package package according to an embodiment of the present invention.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to a specific embodiment, it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. In describing each figure, like reference numerals have been used for like elements.

Terms such as first, second, A, and B may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component. and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being “connected” or “connected” to another component, it may be directly connected or connected to the other component, but it is understood that other components may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that no other element is present in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. It should be understood that terms such as “comprise” or “have” in the present application do not preclude the possibility of addition or existence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification in advance. .

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

In addition, each configuration, step, process, process or method included in each embodiment of the present invention may be shared within a range that does not technically contradict each other.

1 is a block diagram illustrating an example of a configuration of a TO package manufacturing apparatus according to an embodiment of the present invention.

Referring to FIG. 1 , the TO package manufacturing apparatus 100 may include a control unit 110 , a deep learning processor 111 , a lighting unit 120 , a camera 130 , and a communication unit 140 .

According to an embodiment of the present invention, the control unit 110 may perform the overall operation of the TO package manufacturing apparatus 100 . The control unit 110 is functionally connected to the deep learning processor 111 so that the deep learning processor 111 performs machine learning to learn data models for various TO package manufacturing processes, and based on the learned data model, It can be controlled to determine whether the result of each TO package manufacturing process is good or bad.

According to an embodiment of the present invention, when the deep learning processor 111 determines whether the result of the TO package manufacturing process is good or bad, the control unit 110 of the TO package manufacturing apparatus 100 is at least one output device (not shown) (eg, a speaker, a vibrator, or a display) may be controlled to output the result of the determined manufacturing process. For example, the output device may be connected to the inside or outside of the TO package manufacturing apparatus 100 to convert data representing the result of the manufacturing process into various forms such as sound, vibration, or image, and output the data to the outside.

According to an embodiment of the present invention, the deep learning processor 111 may determine the result of each process by using a data-based method as a technique of machine learning. For example, the data-based method learns a model through a dataset having numerous images and labels (eg, dogs, cats, etc.), and in this case, learning the model is called machine learning. The machine learning model trained in this way receives a new image and predicts (prediction/test) the label of the image. In other words, a data-based method of making a model for classifying images is called machine learning, and a machine learning model has a process of learning an image from a dataset and a process of predicting a new image. The dataset for training the machine learning model can be directly collected from the web or the like.

According to an embodiment of the present invention, the deep learning processor 111 of the TO package manufacturing apparatus 100 can build a machine learning model (eg, convolutional neural networks: CNN) using public datasets such as ImageNet and CIFAR. and the user can directly input various data sets. For example, the CIFAR-10 dataset consists of 10 classes, 50000 training sets, 10000 test sets, and so on. The deep learning processor 111 may build a model for determining whether each process is good or bad using the dataset for each process.

According to an embodiment of the present invention, the lighting unit 120 of the TO package manufacturing apparatus 100 is a TO package substrate to determine a bonding state between the substrate 300 and the wire when the wire is bonded to the TO package substrate 300 . It may be arranged to face a part of 300 . In addition, of course, the lighting unit 120 may be arranged in various directions toward the TO package substrate 300 in order to accurately grasp various other process states.

According to an embodiment of the present invention, the camera 130 of the TO package manufacturing apparatus 100 may take an image for determining whether various processes performed in the TO package substrate 300 are in a good or bad state. For example, the camera 130 is arranged to photograph the wire bonded to the TO package substrate 300 to photograph the TO package substrate 300 and the wire bonded to the TO package substrate 300, The TO package substrate 300 may be arranged to capture an image of a specific portion by photographing an entire image or a partial image.

According to an embodiment of the present invention, the communication unit 140 of the TO package manufacturing apparatus 100 may be connected to a machine learning/deep learning server through a wireless network to communicate various data for performing a machine learning operation. For example, when a data set for each TO package manufacturing process is received through the communication unit 140 , the deep learning processor 111 is photographed through the camera 130 based on the received data set. images can be analyzed.

2 is a flowchart illustrating a method of manufacturing a transistor package according to an embodiment of the present invention.

Referring to FIG. 2 , in step s210 , the TO package manufacturing apparatus 100 may perform a chip die bonding process and verify the bonded chip die.

In step s220, the TO package manufacturing apparatus 100 may perform a wire bonding process and verify the bonded wire.

In step s230 , the TO package manufacturing apparatus 100 may perform a cap welding process and a photodiode (PD) aligning process.

3A and 3B are diagrams illustrating an example of a TO package to which a chip die is bonded according to an embodiment of the present invention.

According to an embodiment of the present invention, in the chip die bonding process, the TO package manufacturing apparatus 100 mixes a predetermined amount (eg, 10 g) of the silver (Ag) component epoxy stored in a frozen state with the H77 epoxy at a preset ratio. Epoxy setting process can be performed.

According to an embodiment of the present invention, in the chip die bonding process, the TO package manufacturing apparatus 100 applies Ag epoxy using a syringe for epoxy, closely mounts TIA to a data line with a preset strength, and bakes in an oven for temporary curing TIA mounting step can be performed.

Referring to FIG. 3A , in the chip die bonding process, a plurality of capacitors 311-314 are mounted on a stem (or substrate) 300 and an Avalanche PhotoDiode (APD) chip 320 is mounted. may include steps. For example, in the TO package manufacturing apparatus 100, the APD chip 320 is photographed through the camera 130, and the APD chip 320 is well mounted based on the model trained in advance by the deep learning processor 111. It can be determined whether

Referring to FIG. 3B , the deep learning processor 111 learns an image in which the APD chip 320 is located in the center, and the camera 130 may be disposed to face the APD chip 320 in the air. . When the deep learning processor 111 determines that the APD chip 320 is located in the very center of the stem in the APD mounted image taken through the camera 130, it is determined that the APD chip 320 is mounted well and the next You can control the process to proceed.

4 is a diagram illustrating an example of a wire bonding process according to an embodiment of the present invention.

4, a wire bonder is provided to perform wire bonding, an air line is opened through the wire bonder to fix the sample to a jig, and a jig setting process for placing the wire bonder at a designated position; An equipment setting process of setting the power and buffer of the wire bonder may be performed.

According to an embodiment of the present invention, the wire bonder (not shown) may be connected to the TO package manufacturing apparatus 100 or included in the TO package manufacturing apparatus 100, wherein the wire bonder is the TO package manufacturing apparatus ( 100), the wire bonding process may be performed.

According to an embodiment of the present invention, the wire bonder performs the wire bonding process to bond the wire 401 between the APD chip and the stem, and the TO package manufacturing apparatus 100 has the bonded wire 401 good. You can check if it has been done. Since the length of the wire 401 in the wire bonding between the APD chip and the stem may affect the sensitivity of the TO package, the wire bonding can be performed with the shortest possible wire length that the APD chip and the stem can be connected to. have.

According to an embodiment of the present invention, the camera 130 in the TO package manufacturing apparatus 100 may be arranged to photograph a position where the wire is bonded. TO package manufacturing apparatus 100 controls the deep learning processor 111 to learn the wire-bonded image well, and the deep-learning processor 111 is wire-bonding photographed with the camera 130 based on the learned image. From the image, it can be judged whether the wire bonding between the stem and the APD chip is good.

According to an embodiment of the present invention, the cap welding process may include setting electrodes and equipment of a cap welder (eg, a passive cap welder) 500 , and inserting a lens cap to weld the cap. For example, in the setting step, the air pressure gauge of the cap welding machine 500 is set to a preset value (eg, 0.2 MPa), the power switch of welding power or work, etc. is turned on, and the charging voltage is set to a preset value (eg, , 150v) to set electrodes and equipment. In addition, in the setting step, the upper and lower electrodes of the TO package may be coupled. In the step of welding the cap by inserting the lens cap, the sample in which the wire bonding is completed by putting the lens cap under the electrode may be disposed.

According to an embodiment of the present invention, the cap welding machine 500 may be connected to the TO package manufacturing apparatus 100 or included in the TO package manufacturing apparatus 100 , wherein the cap welding machine 500 manufactures the TO package. A cap welding process may be performed under the control of the apparatus 100 .

According to an embodiment of the present invention, the TO package manufacturing apparatus 100 may control the PD aligning process to be performed through a PD aligner. For example, the TO package manufacturing apparatus 100 may be connected to a digital oscilloscope and a DC power supply. The DC power supply may be connected to a TO PD loading board, and a line from the TO PD loading board may be connected to an output stage 1 and an output stage 2 of the DC power supply. The first and second output stages of the DC power supply may be set to preset values (eg, the first output stage: 3.3V and the second output stage: 19.5V).

According to an embodiment of the present invention, the PD aligner (not shown) may be connected to the TO package manufacturing apparatus 100 or may be included in the TO package manufacturing apparatus 100 , wherein the PD aligner is the TO package manufacturing apparatus 100 . The PD alignment process may be performed under the control of the device 100 .

According to an embodiment of the present invention, the digital oscilloscope is connected through channel 1 (CH 1) of the digital oscilloscope when connected to the TO PD loading board, a source is connected to an input terminal, and a power meter is connected to an output terminal and the power can be adjusted to a set value (eg, -20dB) through an attenuator.

According to an embodiment of the present invention, in the PD aligner, a guide socket (receptacle) to which a jumper cord is connected is disposed on a jig (JIG) of the PD aligner, and the upper part of the TO PD is inserted into the socket in accordance with the pin direction. can get The PD aligner may include turning the Z-axis adjusting screw to bring the socket into close contact with the TO PD. Thereafter, the TO PD may be placed in the center of the socket by turning the adjusting screws of the X and Y axes.

According to an embodiment of the present invention, when a user presses a button connected to the first and second output stages of the DC power supply, the PD aligner may supply power to the TO PD. When the user adjusts the peak (Pk: peak) value of power by turning the adjusting screws on the X-axis, Y-axis and Z-axis, the PD aligner can check the point where the peak of the power is maximum as a point with good reception sensitivity. have.

5 is a diagram illustrating an example of a cap welding process according to an embodiment of the present invention.

Referring to FIG. 5 , in the cap welding process, the lens cap 510 is inserted into the cap welding machine 500 , the TO package 520 in which wire bonding is completed is placed, and then the user presses the welding operation button of the cap welding machine 500 . When pressed, the cap welding machine 500 may control the lens cap 510 and the TO package 520 to be welded.

6 is a flowchart illustrating a method of manufacturing a transistor package according to an embodiment of the present invention.

Referring to FIG. 6 , in step S601 , the TO package manufacturing apparatus 100 may irradiate a light source to at least a portion of the TO package substrate 300 through the lighting unit 120 .

In step S602 , the TO package manufacturing apparatus 100 may photograph an image of at least a portion of the TO package substrate 300 through the camera 130 .

In order to perform machine learning in step S603, the communication unit 140 of the TO package manufacturing apparatus 100 may receive a data set for each TO package manufacturing process from a machine learning/deep learning server through a wireless network.

In step S604 , the deep learning processor 111 of the TO package manufacturing apparatus 100 may learn a TO package manufacturing process in a good or bad state by performing machine learning learning based on the received data set.

When an image of at least a portion of the TO package substrate 300 is input in step S605 , the deep learning processor 111 may determine whether the result of each TO package manufacturing process is good or bad based on machine learning.

In step S606 , the controller 110 of the TO package manufacturing apparatus 100 determines whether the result of the manufacturing process is good or bad, so that the result of the TO package manufacturing process is at least one output device of the TO package manufacturing apparatus 100 . It can be controlled to be output through .

The embodiments of the present invention described above may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by a computer. Such recording media includes computer-readable media, and computer-readable media can be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Computer readable media also includes computer storage media, which include volatile and nonvolatile embodied in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data. , both removable and non-removable media.

The foregoing description of the present invention is for illustration, and those of ordinary skill in the art to which the present invention pertains can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and likewise components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

100: manufacturing device
110: control unit
111: deep learning processor
120: lighting unit
130: camera
140: communication department
300: TO package substrate
311 to 314: capacitor
320: APD chip
401: wire
500: cap welding machine
510: lens cap
520: TO package

Claims (6)

A transistor outline (TO) package manufacturing apparatus comprising:
Control to perform a TO package manufacturing process including a chip die bonding process, a wire bonding process, a cap welding and a PD (PhotoDiode) align process, and the result of the TO package manufacturing process is a control unit for controlling the output;
a lighting unit disposed to face at least a portion of the TO package substrate and irradiating a light source;
a camera for taking an image of at least a portion of the TO package substrate;
a communication unit connected to a machine learning/deep learning server through a wired network to receive a data set for each of the TO package manufacturing processes; and
By performing machine learning learning based on the received data set or the data set input by the user, the good or bad state of each TO package manufacturing process is learned, and the TO package is manufactured for the image captured by the camera. It is determined whether the process is good or bad, and when an image of at least a portion of the TO package substrate is input through the camera, it is determined whether the result of each TO package manufacturing process is good or bad based on the machine learning, , When it is determined that the result of the specific TO package manufacturing process is good, a deep learning processor that controls to proceed with the process of the next step;
The camera is arranged to photograph a position where the wire is bonded when the TO package substrate and the PD chip are bonded with a wire for the wire bonding process to photograph an image of the bonded wire,
The camera is arranged to photograph the PD chip in the air when the PD chip is mounted on the TO package substrate for the chip die bonding process to photograph the image of the PD chip mounted on the TO package substrate,
The deep learning processor learns a data model for a good wire-bonded image, and determines whether the bonding of the wire is good or bad in the image taken of the bonded wire based on the learned data model, Based on the image taken of the PD chip, it is determined whether the PD chip is located in the center of the TO package substrate to determine whether the mounting of the PD chip is good or bad,
The received dataset receives a plurality of images for each TO package manufacturing process from the machine learning / deep learning server, predicts the label of the image, and learns a model for classifying images based on the predicted label. has been collected,
The chip die bonding process includes an epoxy setting process of mixing the epoxy in a preset ratio, an epoxy application process of applying the epoxy using an epoxy syringe, and a TIA mounting process that temporarily passes by mounting the TIA on the data line with a preset strength. including,
In the wire bonding process, the length of the wire is a minimum wire length to which the TO package substrate and the PD chip can be connected, characterized in that wire bonding is performed. delete delete A method for manufacturing a transistor outline (TO) package, the method comprising:
performing a TO package manufacturing process including a chip die bonding process, a wire bonding process, cap welding, and a photodiode (PD) align process;
The process of irradiating a light source to at least a portion of the TO package substrate through the lighting unit;
photographing an image of at least a portion of the TO package substrate through a camera;
A process of receiving a dataset for each TO package manufacturing process by being connected to a machine learning/deep learning server through a wired network;
performing machine learning learning based on the received data set or a data set input by a user, and learning a TO package manufacturing process in good or bad state;
When an image of at least a portion of the TO package substrate is input through the camera, it is determined whether the result of each TO package manufacturing process is good or bad based on the machine learning, and the result of the specific TO package manufacturing process is good If it is determined that the process of controlling to proceed with the process of the next step; and
Including the process of outputting the result of the TO package manufacturing process,
The method is
The camera is arranged to photograph the bonded wire when the TO package substrate and the PD chip are bonded with a wire for the wire bonding process to take an image of the bonded wire, and the camera is the chip die bonding process When the PD chip is mounted on the TO package substrate for
Learning a data model for a well-bonded image, and determining whether the bonding of the wire is good or bad in the image of the bonded wire based on the learned data model, and photographing the PD chip Further comprising the step of determining whether the mounting of the PD chip is good or bad by determining whether the PD chip is located in the center of the TO package substrate based on the image,
The received dataset receives a plurality of images for each TO package manufacturing process from the machine learning / deep learning server, predicts the label of the image, and learns a model for classifying images based on the predicted label. has been collected,
The chip die bonding process includes an epoxy setting process of mixing the epoxy in a preset ratio, an epoxy application process of applying the epoxy using an epoxy syringe, and a TIA mounting process that temporarily passes by mounting the TIA on the data line with a preset strength. including,
In the wire bonding process, the method of manufacturing a transistor outline package, characterized in that the wire bonding is performed with a minimum wire length to which the TO package substrate and the PD chip can be connected.
delete delete

Images