Homepage » Allgemein »

Programming and ease are topical

Discussing the key driving factors for using AOI systems
Programming and ease are topical

AOI systems are indispensable in board assembly lines because of the extraordinary benefits to be gained in terms of efficiency and cost effectiveness, if properly used. There are systems now available for under E50,000, but retain many features comparable to machines in substantially higher price brackets. As the performance increases, the demands are also grow. So, what are the expectations and challenges, and how can they be met?

Paul Phillips, BSE UK/Goepel Electronic

The growing use and application of electronic assemblies represent a serious production challenge. Reasons for this include shorter product life cycles, the ubiquitous practice of just-in-time manufacture resulting in the need for instant changeovers on production lines, faster operating speeds and tremendous cost pressure. These pressures apply particularly intensively to contract electronics manufacturers as well as such of all high-mix/low-volume environments. AOI systems are increasingly being used to monitor the level of product quality, and to optimize processes using the immediate feedback they can provide.
Bearing this in mind we can draw up a list of key capabilities needed in an AOI system:
• Rapid test program generation with facilities for program optimization
• Simple user interface combined with intelligent test algorithm
• Low level of missed faults and false rejects
• The ability to rapidly accept alternative components
• High test efficiency and throughput
• Short amortization time
Efficiency relies on optimal data structure
Because of the complexity of electronic assemblies, a surprisingly large number of parameters must be defined to establish an effective optical inspection test program. This procedure can be aided by allowing default values to be specified for all non-visual related parameters. Nevertheless, it is essential to organize this data in a good practicable and manageable way, not least because of the need for simple and clear operation of the system. In fact, data management is arguably the single most important factor determining the usability of AOI systems.
Of particular significance is the data that describes the appearance of components and their solder joints. There are a number of formats which AOI systems can use: components stored as images, masks to define specific areas which contain valid test data, and data structures following intelligent rules which describe the components as models (such as neural network structures). The administration of this data in an optimal way within the system software is indispensable for the creation, debugging and fine-tuning of test programs without undue effort being required. Indeed, because programming effort is measurable, it can be used to directly evaluate the cost of ownership and efficiency of an AOI system.
During the evolution of optical inspection systems several methods of data handling have emerged. Some instruments use libraries with parameters for each type of component, using the same philosophy as in-circuit testers. The disadvantages of this approach are especially apparent with highend AOI systems which end up with many slightly different varieties of very similar components resulting in gigantic libraries which are difficult to maintain.
Following this, the number of systems without any libraries emerged. These take a board-oriented approach and sep-arately store the parameters for each component on a board. The problem here shows up when, for example, an electrically equivalent but visually different component is introduced to a PCB. Now, each single step of the AOI program, which features this component, must be individually re-programmed each time a component change is encountered. This re-sults in a great deal of unjustifiable (re-)-programming time especially for lines with frequent product and component changes.
Another type of data storage and program creation method uses one or more reference assemblies or so-called Golden Boards. In this case, assemblies, which represent an absolute „good“ pattern or a representative production profile, are used as the basis for comparison with the board and components to be tested. Apart from the difficulty in producing a board with 100% quality, real use of a Golden Board as a reference data base is impractical since components always have variations in their appearance caused by tolerances, process variations, placement angle and so on.
Finally, following careful analysis of these different types of data organization a manageable and practical library structure has triumphed. Goepel’s OptiCon family of AOI system creates test programs made up using libraries which can be expanded in a simple, yet effective way. The data structure is split into two categories: component parameters (placement tolerance, angle, etc.) and component models (their appearance). Only one instance of each model is required in the system, so component variations with another appearance just need to be added to the existing model. Having done this, inspection in multiple placement positions and with different programs can be executed using the modified library entry without further debugging. This library structure also allows the parameters (placement tolerances, etc.) for a single component or component type within a program to be optimized. This unique structure provides the basis for achieving the goals for AOI in production, particularly for rapid program generation and optimization.
Intelligent test algorithm
Another key factor in determining AOI system productivity is the algorithm used for the recognition process. It is clear that simple analytical processes in image processing such as pixel correlation, edge detection or brightness analysis are of limited use for the sophisticated needs of electronic production. There will be many situations whose demands will exceed thecapabilities of these low-level algorithms. Assemblers typically have very limited influence on the precise components or suppliers they will use in their processes. They need to constantly cope with a myriad of changes in, for example, component colors, cap sizes on chip components, PCBs and so on.
Under these circumstances, a system administrator can be faced with the laborious task of defining many inspection parameters in many component positions such that all possible eventualities are considered in the hope that, finally, a reliable verification can be a-chieved with an acceptable false-call rate. This is the main reason why the recognition algorithm needs true intelligence; to be able to recognize all component variations either in the inspection algorithm or in alternatives stored within the library model.
The OptiCon system uses an intuitive and highly intelligent test method. Prior to program creation, several images showing both good and bad examples of a component to create a component model. The model can accommodate a number of alternative different visual appearances (such as yellow and black tantalum capacitors) whilst also handling small variations in each. Then with a single mouse click the component appearance is stored in a classifier which takes into account various internal rules for recognition. The result is a small database (a few Kbytes) which encompasses all component variations and which is based on a neural network structure. An additional benefit of this approach is a quick inspection (just a few milliseconds per recognition process), along with a reduction in the volume of data to be accessed during verification.
The resulting classifier – a special sort of optical description of component types – is finally linked to the library entry and becomes available for inclusion in a program. At any time during program generation and debugging, the classifier can be modified or fine-tuned by adding other component images. This simple process can also be used during production whenever, perhaps because of a change of supplier, the appearance of a component changes. This method imparts significant advantages to subsequent stages of system programming and operation:
• The algorithmic process is easily manageable by choice of images to be input
• Debugging the program can be achieved by simply adding additional good and bad images to the classifier or by amending one quality threshold
• Only one inspection step is needed per chip component. This can be created manually with a mouse click or imported from the placement data
• Images from the repair station can be used to immediately improve models, accelerating process optimization, maximizing reliability and minimizing false calls
Rapid inspection program generation
The trend towards the use of libraries with default attributes and verification parameters is clearly the way forward. This allows very short generation times for both manual and automatic programming. So long as the basic library exists, only new components (not previously stored) need to be defined in addition to, perhaps, modifications to models where another physical appearance emerges on existing components.
The task of program generation thus becomes extremely simple. Only the type of component and its physical location on the board need to be defined. All of the many others parameters such as image recognition and tolerances for placement, twist, polarity, etc. are stored as default in the library. To accommodate specific needs, the parameters can be changed, either for an individual component or to new default values on a board-wide basis. Inspection steps can be inserted manually simply by placing the relevant library entry over the component as seen on the board by the system camera. Automated generation is accomplished by reading the component type and position directly from placement CAD data.
Another advance in verifying more complex components (such as plugs, connectors, ICs, etc.) is the possibility to link several inspection steps in a macro (analogous to a miniature program). Such a so-called inspection complex can be manually or automatically inserted into a program alongside the other steps. They are always available to verify similar parts located elsewhere on the board or, indeed, as part of another programs.
Effective inspection in production
In addition, test throughput is clearly a key issue if the AOI system is not to become the bottleneck in a production line. Also, the cost-effectiveness is enhanced where a system can be rapidly configured for different applications. This is particularly true when considering the location of a system within the production line(in-line or stand-alone, before and/or after soldering). AOI systems, which combine a modular architecture with a broad variety of options, al-low customer-specific configuration and have proven to be particularly effective in practice.
To meet these requirements, the OptiCon is available as the BasicLine (stand-alone system with manual loading) and the SpeedLine (in-line system for direct integration or as an island solution with magazine loader/unloader). Each variant is available in two throughput ratings for maximum cost-effectiveness.
Zusammenfassung
Nur dann lassen sich AOI-Systeme weitgehend optimal nutzen, wenn die Problematik der Programmierung gelöst ist. Weder darf der Tuning-Aufwand dafür ins Gigantische steigen, noch die Inspektionsabdeckung der Prüfobjekte zu gering sein. Mit einer Programmerzeugung, die sich zusätzlich noch der Auswertung von guten und schlechten Bildbeispielen bedient, lassen sich Zeit- und Kostenaufwand reduzieren.
Résumé
Une utilisation optimale des systèmes AOI est uniquement possible si le problème des programmes d’inspection est résolu. La mise au point nécessaire doit rester dans des proportions raisonnables et les inspections réalisables doivent être suffisantes. Une création de programme qui utilise de surcroît l’exploitation d’exemples d’images bonnes et mauvaises permet de réduire le temps et les coûts.
Sommario
I sistemi AOI si lasciano sfruttare in maniera ottimale solo se la problematica dei programmi di ispezione è stata risolta, cercando di non lasciar crescere in maniera sproporzionata le spese per l’equipaggiamento tecnico e allo stesso tempo di garantire una copertura di ispezione più che sufficiente. Con la generazione di un programma che si serve anche dell’elaborazione di immagini esemplificative di pezzi i.o. e di pezzi n.i.o. è possibile ridurre sia i tempi che i costi.
Current Issue
Titelbild EPP EUROPE Electronics Production and Test 11
Issue
11.2023
READ
Newsletter

Subscribe to our newsletter now

Webinars & Webcasts

First hand technical knowledge

Whitepapers

Find all current Whitepapers here

Videos

Find all current videos here


Industrie.de Infoservice
Vielen Dank für Ihre Bestellung!
Sie erhalten in Kürze eine Bestätigung per E-Mail.
Von Ihnen ausgesucht:
Weitere Informationen gewünscht?
Einfach neue Dokumente auswählen
und zuletzt Adresse eingeben.
Wie funktioniert der Industrie.de Infoservice?
Zur Hilfeseite »
Ihre Adresse:














Die Konradin Verlag Robert Kohlhammer GmbH erhebt, verarbeitet und nutzt die Daten, die der Nutzer bei der Registrierung zum Industrie.de Infoservice freiwillig zur Verfügung stellt, zum Zwecke der Erfüllung dieses Nutzungsverhältnisses. Der Nutzer erhält damit Zugang zu den Dokumenten des Industrie.de Infoservice.
AGB
datenschutz-online@konradin.de