Another example concerns electronic equipment production. In a typical scenario, the following steps will take place:

 Raw printed circuit boards are produced and transferred to an assembly facility.

 In the assembly facility, PCBs are assembled from raw boards and electronic components, then sent to the customer.

 At the final assembly, PCBs are installed in the final product.

Each step holds multiple opportunities for mistakes, like using mismatched components or executing the wrong assembly and test procedures. Using a paper trail won't resolve some of these typical problems. For instance, using mismatched components during assembly may not be discovered until final tests, causing costly rework and yield reduction. Some problems may even go unnoticed to a customer site, which would result in customer dissatisfaction and potential recalls.

In most cases, the manufacturer of the final product will not have access to the quality control information from the previous stages, or will have only partial, incomplete data. Lack of sufficient data may cause problems, not only during production but later, during product maintenance, warranty repairs or if liability issues arise.

The complexity and the cost of fixing a problem grow significantly the longer the problem goes undiscovered. A potential manufacturing error that is noticed before it actually occurs may cost only $10, while the same error discovered during final tests may require a $1 million rework. Obviously, the cost-effective solution in this case is prevention.

The most direct way to ensure complete control of the production process and post-production product support is to use "cradle to grave" unit traceability. Full unit-level traceability is achieved by marking each individual part so that it can be identified reliably at any point between initial production and end of life. There are different ways of marking parts, and different ways of recording the information required to identify the part in each specific industry and production process. Both the identification method and the marking technique depend on many factors, including required volume of data, available space and type of production processes. Identification methods can range from labels to direct part marking. Part identification can be achieved using human readable code or encoded data, as in linear bar codes and 2-D codes, like Data Matrix.

Full unit traceability's many benefits include automated data collection throughout the process, shift of testing and quality control to the start of production, complete quality track record for future use in liability and other issues, compliance with ISO 9001 and other regulatory requirements, and automatic setup of manufacturing machinery.

Automated data collection eliminates human error because it identifies the part, lot, date, type of processing and any other recorded information. All the pertinent data can be stored in a database and used during production, scheduled maintenance or warranty repairs. Identifying each unit as it goes through production steps allows for automatic equipment setup, thus reducing human process errors.

Also, automatic equipment setup enables easy tracking of small or even single-unit lots, as well as just-in-time manufacturing. One example of capitalizing on such flexibility on the production floor is vehicle assembly based on incoming orders, with variability in model, color and interior controlled automatically.

Another advantage of part-level traceability is the verification of the production sequence. The damage caused to factory equipment due to wrong use of machinery can be prohibitively high. For instance, a furnace used in silicon wafer fabrication can be contaminated if a wafer is inserted into the wrong furnace. Not only will the wafer be destroyed, but a multimillion-dollar furnace may be out of commission for several days due to contamination. Such an occurrence can be prevented if the wafer is identified prior to each production step, and the next step is validated by the manufacturing control system.

Shifting quality assurance tracking to the early stages of production contributes to problem prevention and minimized downtime, and enables statistical process control.

Availability of complete data records greatly facilitates information archiving and retrieval in cases of potential recalls, corrective action and liability issues. Regulatory agencies require complete data records in most industries, and automatic data collection facilitates agency approvals.

Problems related to linear bar code identification include:

 Limited amount of information--A typical bar code can contain 10 to 20 alphanumeric characters, not enough in many cases to store all the necessary information. Insufficient amount of data may in effect negate the benefits of unit-level traceability.

 Required high contrast--Linear bar codes represent an analog data encoding. They are based on a high-contrast pattern of dark and light bars. Linear bar codes must be printed on labels to ensure readability.

 No data protection--If a linear bar code is damaged in any way (e.g., a label is scratched, soiled or ripped), the data encoded in the bar code will be lost. Such volatility works against the goal of complete and permanent part identification.

 Space requirements--Linear bar codes require a significant amount of space on the part. With the general trend toward miniaturization in many industries, this requirement often makes bar code use impossible. Such is the case with densely populated PCBs, medical devices and pharmaceutical containers.

Using labels for individual part identification presents its own set of problems:

 Labels may not survive harsh environments and processing--Label material, no matter how resilient, may not survive processes such as high temperature reflow in surface-mount PCB manufacturing, chemical solvent cleaning or sterilization of medical instruments.

 Labels peel off--Labels can't be guaranteed to remain intact during the entire life of the product. If the label peels off, the product information is lost.

 Cost of consumables and maintenance--Using labels adds to production costs--companies must pay for label stock and inventory control. The most advanced, damage-protected labels may be quite expensive.

 Labels may contribute to product failure--Depending on the label material and the adhesive, a label may even introduce additional problems.

A part-marking method that recently has gained wide acceptance in many industries is the 2-D digital code, like the Data Matrix invented by RVSI Acuity CiMatrix and maintained in the public domain by AIM USA.

A Data Matrix is a digital, machine-readable symbol that can encode more than 3,000 characters in a very small physical space. Figure 1 shows the words "Quality Digest" in a Data Matrix and a linear bar code (Code 128, also invented by RVSI Acuity CiMatrix). There is a dramatic difference in data capacity and required size.

The Data Matrix symbols contain built-in error correction that protects encoded data. Error correction allows the data to be retrieved in its entirety even if the mark is partially damaged or obstructed.

Unlike linear bar codes, the Data Matrix can be decoded with as little as 20-percent contrast. This allows the Data Matrix to be marked directly on the part surface using various techniques. Depending on the part material and the type of processing the part is subjected to, the symbol may be marked using such methods as ink jet printing, laser etching, chemical etching, dot peening, pin stamp and thermal transfer.

The symbol's small size permits the manufacturer to place the Data Matrix mark on virtually any product without compromising its quality and appearance. Direct part marking creates a permanent and reliable way to trace individual parts during manufacturing and throughout the product lifecycle.

Fixed-position readers typically are installed at various points in the production line. The readers may function independently or connect with the factorywide computer network. The decoded data from each part may be used to set up the equipment (e.g., a pick-and-place machine), validate processing or record production steps, and test results in a central database.

Quality control of the production process is not limited to automatic reading of direct part marks. There is also a need for operator-controlled, handheld part reading. Validation of rejects during production, final testing and warranty repairs are all examples of setups where a fixed position device is impractical or even impossible. In addition, manufacturing of large or odd-shaped parts, like jet engines or surgical instruments, must rely on handheld devices.

These goals can't be achieved in a cost-effective manner without individual unit-level traceability, which is best implemented with 2-D symbology. Using a Data Matrix on each manufactured part greatly facilitates global production control and data management, allowing the highest profits in the most efficient manufacturing environment.