Print  |  Back

TECHNICAL ARTICLE

Originally Printed in the August 2006 Issue of SMT Magazine
 

QUALITY CONTROL SOFTWARE

The key to true quality control (QC) is prevention, rather than detection. Electronics-manufacturing operations demand quality management software that monitors and controls processes prior to and beyond inspection. This article explains the need for QC software beyond its traditional definition, and the product traceability it can provide.

Quality control is the foundation of the workmanship and industry standards that serve as guidelines for manufacturing and operational practices. As standards such as the ISO 9000 series have evolved into industry-specific versions, their goals ultimately translate to customer satisfaction. A quality control (QC) system must support this and offer proof of the results.

According to the American Society of Quality (ASQ), quality control is defined as “the observation techniques and activities used to fulfill requirements for quality. When applying these terms to circuit assembly, quality control covers the steps and measures to ensure the highest products quality throughout the manufacturing process. Ironically, the traditional thought of a QC software system is focused on the areas after these operations are performed. Today, most quality software systems limit their area of coverage to defect collection, repair automation, and even collation in a quantifiable form for analysis. This system may consist of multiple collection points, such as an automated optical inspection (AOI) machine linked to an integrated repair station, or multiple stations performing manual inspection and repair. For circuit board assembly, the common manual defect-collection method is form-based text and/or field-selection entry (operator selection of defect, defective location, and affected pin number). This method is not only time-consuming, but prone to entry errors. It requires the inspector to know specific connection numbers if pin-level detail is expected of them. A more refined manual collection approach uses an intelligent graphical rendering of the board derived from a computer-aided design (CAD) file. In this scenario, the inspector can capture defects visually by clicking the on-screen image of a suspect component, pin, or area on the board.

Repair and repair-loop automation is commonly supported, although the system may not be aware of the overall process route and cannot advise the technician where to direct the product after completion. For repair, a form-based method requires on-board inspection arrows for reference. Several actions may be taken, such as marking a given indictment as fixed, false positive, or closed if the technician is given QC privileges. With a visual image, a repair technician can retrieve a graphical depiction of defective areas and perform on-screen actions to address suspect items. All of these actions cannot exist as independent events within a QC system, and must have a correlation to each other and to the design, bill of materials (BOMs), and materials data.

Charting and analysis also vary. Many are derived from manual collation into Excel charts, while some systems offer real-time chart displays and out-of-control alarm events. These quality/repair systems often have captive data and do not extend coverage to where defect causes actually occurred.

The traditional quality-system concept is constrained with respect to a system that performs true quality control. Data-collection scope typically does not extend to other sources serving as evaluation points in the process. This includes AOI for solder paste, pre-reflow, post-reflow, post-wave inspection, and test equipment. Its scope also does not include process-variable data from production equipment and environmental-related sources, where events may trigger out-of-control events or reveal probable causes for issues found within the inspection area.

To support this spectrum of data sources, the quality system must also broaden its definition of collected data types. The most common type found in a traditional system - the indictment - is the cause of a failure, and is discovered by either an inspector or an AOI machine. However, test systems primarily develop symptom data, which must be diagnosed for its relationship to an indictment. A traditional system typically lacks knowledge of the test failure, and does not support a means to correlate a symptom to a defect-inspection code. A comprehensive system must understand all of these differences and relationships among particular quality incidents as they move through their life cycle.

Although the system described above is hampered by its narrow field of knowledge, it precludes the possibility of monitoring “defect charge-backs” to a particular process and therefore, any hope of proactive process control. A charge-back is the physical location or process that caused the defect. The inspection point is not responsible for the defect’s existence. With a system that correlates each indictment to the source of the cause, it not only provides the benefit of proactive process control, but also automates relevant defect per million opportunity (DPMO) calculations across each process step. This is only possible when the system broadens its coverage scope on the factory floor.

Key to Quality: Factory-wide Control
Beyond correlating symptoms and repair events to a given indictment, a QC system must have extensive knowledge of practically everything related to the product’s manufacture. Only when this expanded information set is mapped to quality records can a defect’s true root cause be discovered. One key set of information related to the product is the material. For instance, analysis of a component’s solderability issues may be narrowed down to a given material lot. Material scope is not necessarily limited to product content and may include consumables, feeders, tooling, and any other items required for assembly. With these, the system can relate problems, such as flaws in a given fixture. Events that occurred within an automated production machine may also be related to quality issues. Events that detail excessive dwell in a reflow zone or actual temperatures above/below set points may have a clear relationship to problems identified at test-and-inspection points. With quality systems that inherently contain CAD-level circuit board product intelligence, these event-defect relationships may be visually apparent when defect hot spots are displayed across the board. Without these types of manufacturing data relationships, the system remains an island limited in its ability to identify the true potential cause.

A manufactured product in our industry is always built to a standard or a combination of standards specified by internal, industry, or customer sources. A factory-wide software system extends this coverage to automate as many steps as possible to mistake-proof multiple processes - from data preparation to shipment. Error prevention is the true definition of quality control and when practiced correctly, the role of inspection becomes the final assurance that the manufacturing process remains in control. Areas where error prevention is actively controlled by a system of this scope include:
Version control: Mistake-proofing in the manufacturing process beings in the factory office with work-instruction development and machine-program generation. It requires version control and safe delivery to the shop floor. For reliable quality control on the shop floor, all steps in the process, whether in production or inspection/test, must refer to the same set of revision-controlled information. A factory-wide software system manages these revisions and automates the revision-approval cycle. It also identifies a revision’s approval status and establishes links to production work orders. Operators should be limited in their choices for version options, and information access by work order (or a serial number already anchored to a work order) ensures the correct revision selection.

While products are built to a given revision, their assembly occurs in a volatile environment. Change is a constant; it is the nature of the high-mix/low-volume manufacturing footprint commonly found in North America and Europe. Change occurs faster than revisions can be handled appropriately. A QC software system should have not only awareness of deviations issued against an affected product, but should also control broadcast to the shop floor. The broadcast must be targeted to only the affected areas and products to the operators in an active manner. If not handled as such, an operator becomes immune to the flurry of notices not applicable to their process. They must remain relevant to the end-user to make applicable changes and prevent unnecessary errors from occurring and being identified later in the process.

Material control: Material-related issues are a key source of problems found in inspection. A factory-wide system - that is aware of where components are installed throughout the process and includes shop-floor-material control - ensures that the correct material is assigned to each workstation (and machine), and reduces the number of errors found later in the process. For setup-error prevention, the system offers operator guidance when establishing or changing over a work center. The system also prevents job execution until the setup of materials at each station is valid against a pre-determined bill of process (BOP). A work center may be an automatic placement machine where parts are assigned to specific pick locations, a manual work station where parts are assigned to that particular location, or a stencil-printing operation to confirm the correct paste, stencil, and blades are used. These actions are performed using bar-code scanning to expedite the setup process and eliminate key-entry errors. Using radio frequency identification (RFID) technology with tags on materials and tooling augments this operation and ensures error-free setup in a closed-loop manner. In either of these scenarios, the system restricts production from occurring until the material setup is completed successfully.

Route control: Route enforcement is critical to a QC system, and is virtually impossible to control without software. Even the most experienced operators can overlook an intermediate step for a given product not found within an inline process. A factory-wide system must have complete knowledge of the entire route and determine the next step for a product automatically based on status, trends, and alarms. A system offering this form of control ensures a correct assembly sequence and prevents products from continuing in a fallible process. The system enforces bar-code misreads, misroutes, work-order completion, statistical process control (SPC), out-of-control events, and defect-collection trends. For inline processes, the route is enforced with supplemental conveyor-mounted, logic-controller hardware using SMEMA communication. For operators, active on-screen notification prompts the user to direct the product to the correct route location, or describes the reason why a product may not enter the work center.

Pack-out control: When a product reaches final operation in the factory, pack out , a factory-wide QC system, assures its shippable readiness. It is capable of answering the question: Has the product successfully completed all of the necessary production, process, and inspection steps? A system that has direct knowledge of all these operations can confirm a product’s completion - rewarding the operator with a shipping label and producing a certificate of conformance per product requirements. If a product in the pack-out area has bypassed required steps or has unresolved defects, the quality software system serves as the final safety net to prevent a suspect product from shipping. A facility without a system offering this control faces the cost of product-return logistics, losing a reputation, and possibly losing a customer.

The By-product: Traceability
A QC system encompasses all knowledge of product life as built in a factory. As a result, the system contains rich information about a given serialized product (or work order for batch processes). The information is rich in respect to the factory operation’s breadth, and to a given work cell’s data depth. Product traceability extends from complete revision details, revision approvals, applied deviations, performed route steps, defect/repair history, measurement detail, machine events, measurements that occurred while the product was present, sub-assembly genealogy, and lot-level material detail to a given reference location. Traceability is the sum of the product, process, materials, and measurement data from the entire manufacturing life cycle. Availability of this comprehensive record of assembly should require no additional labor overhead. It is a reflection of all of steps that the system performs and controls.

Enterprise Quality Control Systems
To attain the level of control described, the QC software system must have product intelligence of the circuit board, route intelligence for the entire process, support both human and automated equipment data acquisitions, and exist on an architecture that permits reliable scalability within any enterprise. Product data intelligence of BOMs and design data is required because it is the foundation of all analysis when data received from the real-time process must be correlated into the product. It also supports more intuitive operator interaction with the system. The system must be involved in every process-point along the route. If not, the data set for analysis is truncated.
Data acquisition from machinery and operator stations must be done using standards, rather than proprietary methods, to ensure forward-compatibility and rapid incorporation of new data sources. The volume of transactional data emerging from such systems per minute demands an enterprise-capable design. Software tools and point solutions cannot handle such loads.

Conclusion
A common question heard throughout the industry is: We’re collecting all this data, but what do we do with it? Software systems for quality control must not only possess the ability to collect data from every step of the process, but also to understand the relationship of this data to design it, analyze it in this context, and react to it through process control. This holistic concept of full-factory information management has recently become possible without a heavily customized solution. Today, the industry can leverage information to improve quality - and as a by-product, yields total traceability.
 

Author Information:
Bob Miklosey, Vice President of Product Management
Aegis Industrial Software Corporation
220 Gibraltar Road, Suite 100
Horsham, PA 19044