The process of evaluating and selecting a reading system is critical to the success of the overall direct park mark identification (DPMI) application. While many factors go into the selection of any new equipment, the capability of the reader to consistently read codes throughout the process is critical.

In DPMI applications, distortions to the code are quite common due to part composition, variations in part presentation, or variability caused by the manufacturing process. It is important to select a reader that can tolerate a wide range of distortions to the appearance of the code no matter what the cause.

A set of sample parts that represent the range of mark quality that a reader will need to handle should serve as the basis for a preliminary test of read rate. However, a more extensive pilot test is recommended so that more read rate statistics can be gathered and analyzed.

To consistently read codes and achieve the required read rates, the reader should tolerate changes in contrast, focus and degradation to the code without a need to change underlying parameter settings. With a fixed-mount reader, this is easily tested by starting with an optimal setup and adjusting the aperture to simulate change in contrast; focus to simulate change in depth of field; and position of the light to simulate background problems. A reader that can consistently read under these conditions will lower installation cost and minimize start-up issues.

In addition to read rate, it is important that the reader return a result quickly. In operations requiring handheld or presentation DPMI readers, accurate and fast decoding is also important. Although the main driver for the use of an automatic reader is to eliminate data errors, the readers cannot slow down the process.

Until recently, handheld readers were very slow in their decoding time of direct-marked parts, often resulting in a no read. This can lead to operator frustration, lack of use and wasted investment. To avoid that problem, engineers should look for handheld readers that provide laser-like scanning performance across the entire set of representative sample parts that are chosen. A "trigger to good read beep" that is consistently less than 1 second on all parts will appeal to end users.