The Role of Inspection Systems in Card Manufacturing Defect Control

Introduction

With the automotive industry, general retailers, banks, and institutions increasing the use of plastic cards, manufacturing of card products is increasing at an accelerated rate. Special features such as magnetic stripes, embedded microchips, and many types of security features only make for a complex method of assuring the quality such markets demand. Printed images and/or graphics already have their share of inconsistencies caused by process fluctuations such as temperature, pressure, and color control - and these now highly desired "strategically positioned" functional components have vision inspection systems competing for the challenge of maintaining quality assurance.

The roles of vision inspection systems

Vision inspection systems play many roles in various inspection environments - from two-dimensional associative assessment of graphical images and/or components - to three-dimensional comparative measuring of objects. As for the plastic card industry, vision inspection systems perform print image and surface anomaly verification utilizing four cameras: two cameras for image processing (for print/graphical movement detection, magstripe, holograms, and other surface components); and two cameras for surface anomalies or scratch measurement and identification - one of each type of camera for each side of the card.

Currently there are three providers of high speed card vision inspection systems (in excess of 333 cards/minute or 20,000 cards/hour): Louda Systems, Spartanics, and Mühlbauer (also responsible for the Focus Cardscan System), all consisting of two main assemblies; that of a mechanical card handling system and the other, an image processing system.

Card inspection system providers - quite instrumental in providing for high-speed vision inspection for an industry whose products are being manipulated by its market to provide card visual appeal, financial agility, personal identification, and the highest degree in security requirements - have secured an interest highly sorted in an industry challenged with prompt and viable solutions to market demands i.e.: three dimensional (3D) holographics, smart chips, multiple magnetic stripes, complex graphics, lenticular and transparent plastics, and many others. Such demands have forced card manufacturers to extend further the art of printing in order to compensate, include, and allow for further processes to take place, yet retain the requirements of compliance with the "form, fit, and function" of a desired product.

With Visa, MasterCard, American Express, and other secured card clients requiring tight tolerances of component positions, the normal wear and tear of hot stamp, punch, silkscreen, and other support equipment demands close monitoring and inspection of production output in the assurance of a stabilized and controlled process. Unfortunately, and when processing large quantities of cards, typical manual processing will allow for undesired movement in components, text, and graphics. Even when using measuring tools, tolerances are typically smaller than can be immediately identified with the unaided eye. Product color and tone - probably the most important measured attribute in the card industry - are a feature of human perception and not an objective phenomenon unwavering by any means other than the clients' acceptance.

Auto-vision inspection

Auto-vision inspection systems as provided by Louda Systems, Mülhbauer and Spartanics, contain technologically advanced subsystems capable of required inspection accuracy and color measurement. The collective use of optics and software of each of the systems provides excellent analysis of captured images as long as each image presented to the vision system is visually introduced with exactly the same lighting and physical position in order to differentiate between the initially trained image (golden card) and newly introduced images (cards under inspection) containing potential differences in color, cosmetic irregularities, component misplacement, and surface anomalies.

Since each type of inspection system introduces cards to the image processing subsystem via some mechanical handling means i.e.: rails, conveyor belts, rotating drums, or suction cups (all assisted by vacuum in securing the card during transport), this involves the crossing of these two technologies - that of first generation mechanics and third generation electronics. Unless communicating the cards' physical position and making x, y, and delta positional adjustments in compensation for product skewing (caused by minute backlash in gearing, belt stretch or any other external disturbances to the cards momentum) - this marriage of technologies demands a synchronization which can only be met by image processing software providing for such listed compensations - including lighting error correction. Under ideal conditions (a well tuned and optically calibrated system), such software corrections translate to accurate measuring of the card product.

Unlike many types of high speed robotic driven vision systems (primarily used in electronics manufacturing), using close-loop servo systems compensating for undesired product movements by calculating present and desired positions, thereafter making servo driven adjustments - Louda, Muhlbauer, and Spartanics vision systems compensate using complex algorithms from image processing software which use factory set or adjustable (virtual movement) tolerances for the various selected inspection areas on the card.

Defect detection

To determine and train a vision inspection system as to what is an acceptable card is in itself a task quite illusive, as manufactured card products are forever dynamic in their processing. Even when process results are anticipated and calculated during initial product design, such projections only prevent gross anomalies from occurring, and it is the variability of processes at initial introductions, during, and after various process elements that introduce defects and/or anomalies upon the card(s).

Elements such as ultraviolet light (electromagnetic radiation) used to cure inks, convection heat transfer via heated fluids supporting card laminator processes, and conductive thermal transfer during hot stamping operations are just one type (heat transfer) of the myriad of elements that induce distortion, poor overlay to ink adhesion, and exaggerated impressions to plastic cards - not to elaborate on the many other type of machines that further process these cards or the implications of using fine microscopic powders for the control of static electricity and their secondary effect(s), emulating and misdiagnosed as process induced printing defects upon cards when poorly controlled.

Other areas of concern that affect the acceptance of a final product are holographic stamps, special foils, microchips, and the continuous mishandling of cards during and in route to the various processes.

Vision inspection systems are designed with programmable software tools to capture and train themselves (some systems do not have self training capability) to view a card initially considered acceptable by their human programmers and thereafter made to compare the captured and trained card to all remaining batches of cards assigned for inspection. This in itself infers that the final "inspected" product is of the quality level determined by its programmer. This being so permits one to assume that the programmer is well informed of client supplied specifications, internally developed manufacturing specifications, and as a minimum, well versed on industry related International Organization for Standards (ISO) specifications for the card type being programmed. Without such background, programming of these cards would be left to subjectivity comparable to that of manually inspected cards as practiced by organizations not equipped with vision inspection capability.

Image processing units within inspection systems provide programmers many "object oriented" software tools for use in capturing and analyzing areas of concern on the card, and depending on the programmer's philosophy of what "quality" is, depends on whether he/she programs the system to identify previously sampled defects or to inspect the card as a whole, then ensuring that quality was built into the processing of the card.

Those who desire to inspect out undesired defects may, in the course of inspection, allow unsuspecting fluctuations in color, image mis-registration, shifts in component positioning, and numerous other anomalies not identified during previewing or batch sampling of cards. Even when the quantity of these undesirable defects is of low percentages, it can give the impression that 100 percent client satisfaction is not on the top of the list.

Defect analysis

In order to define a defect, one must first understand what is an acceptable card - considering how fluid the card is during the myriad of manufacturing processes, yet still containing itself in a presentable manner to the client, should provide the best answer to this unforgiving question. Unfortunately such is not the case outside of the clients' initial proofing, thus leaving card manufacturers depending on their continuous process improvement efforts and stringent adherences to required specifications in order to maintain the confidence of the client after proofing has taken place.

Defects - displayed in many forms, shapes, and colors to the observer - provide only mathematical data to image processors. These processors allow interactive tools (software parameters or the manipulation of tolerances) to be used by programmers in order to give processors mathematical correlation of what would be considered allowable fluctuations in color, component movement, and other anomalies. Vision systems do not know what a good card is, and only the programmers' ability to understand and identify out-of-spec anomalies can provide the system for such separation of acceptable versus rejectable cards.

It cannot be mentioned enough that without the programmers' knowledge of required specifications, the analysis of the many types of fluctuations would (as mentioned earlier) remain to subjectivity, thus having numerous adjustments made to the program on an all too frequent basis. Here is where systems which contain self-training features or can adapt themselves against a series of pre-selected cards would provide for a more stable program and minimize frequent adjustments to the program, ultimately improving process throughput.

In analyzing cards, programmers rely on information given by the vision system in addition to that provided by product sampling plans in order to provide manual calculations to what should be acceptable product, thereafter manipulating parameters to segregate desirable from undesirable product. What is taken for granted is the way the system rejects cards for masked reasons not necessarily due to any card anomalies.


About Versatile Card Technology, Inc. As a globally recognized leader, and the largest American manufacturer of plastic cards, Versatile Card Technology produces cards for every application and offers complete fulfillment services and turnkey solutions. Facilities are located in Turkey, Germany, Mexico, Singapore, South America and India. The Visa/MasterCard state of the art facility in Chennai, India, has helped increase VCT's customer focus throughout Europe, Asia, Australia and Africa. VCT has increased its card capacity to produce over one billion cards per year. VCT is a certified manufacturer of Visa and MasterCard and produces a wide variety of products including, loyalty, membership, gift cards, SmartCards, access control and contactless cards.