Latest from Issues and Ideas

Awcnz62 | Dreamstime
Tadeusz Ibrom | Dreamstime
Dmytro Hrushchenko | Dreamstime
Assn. for Advancing Automation
Robert Kneschke | Dreamstime
Spaxia | Dreamstime

Two-Dimensional Codes Leave Their Mark

April 30, 2004
Reliability of reading two-dimensional matrix codes by vision scanners has been the initial focus of this new identification technology. Cast surfaces pose some particular problems, but actually imprinting readable two-dimensional codes quickly and cost-e

In 1988, the first truly two-dimensional (2D) bar code was introduced by Intermec Corporation, when they introduced Code 49 for use in small packaging applications in the healthcare industry. Since that time, a number of other codes have been developed or have been redesigned to meet the need for placing a database in as little space as possible. One 2D code in use today, the Data Matrix code from Acuity CiMatrix, may allow manufacturers to include additional information in a compact way that’s impossible with bar codes.

A Data Matrix symbol can store between one and 500 characters. The symbol is also scalable between 1-mil square and 14-inch square, theoretically permitting 500 million characters to the inch. Practically, density application is limited by the resolution of the printing and reading technology used. The data on this code is a compilation based on a square matrix broken down into a series of smaller squares, or cells, that can be square, hexagonal, or circular in shape. Date is encoded by the position of the light and dark areas. Compared to one-dimensional bar codes, the 2D code is really a database, containing much more information than a bar code can provide.

The codes can be applied to surfaces using a variety of printing technologies including ink jet, dot matrix, lithography, lasers, chemical etching, and now by dot-peen marking. As well as being scaleable, in terms of physical size, the Data Matrix codes can be rotated up to 360° without affecting readability.

The initial focus of this technology was creating the image recognition systems to effectively and reliably read the 2D matrix codes on a variety of surfaces as well as certain subsequent manufacturing processes like bending, cleaning and painting. Other limitations may include space availability, issues of contrast, reader requirements and cost. However, with vision scanning, the technology is now available to read most types of imprinted codes.

Direct Part Mark Verification for Data Matrix Codes

The Association of Automatic Identification and Data Capture Technologies (AIM) Symbology specification calls out the Data Matrix verification methodology as the measure of; symbol contrast, print growth, axial non-uniformity and unused error correction. The mark quality is only as good as the lowest grade given in the previously cited categories.

With parts that are directly marked, this AIM method misses several key measures and will almost always report a final score based on symbol contrast, which for most direct marked parts will be meaningless, as it is not something that can be controlled. For example, dot-peened marks on a metallic surface seldom add much contrast. If companies rely exclusively on AIM parameters they may get bag readings on good marks, or vice versa.

With ink jet, dot-peening and laser marking (the three most often used marking methods), the two most important aspects to measure are the “normalized” center offset and size offset values. These measures are made of all the “ON” cells (those actually marked vs. the spaces left unmarked) within the borders of the code. The scores are normalized because with a Data Matrix code, if all cells are shifted the same amount — or are smaller or larger by the same amount — there is nothing to adjust. This is because most readers will deal with the scale change and no marker maintenance is necessary. However, a difference in cell position within the coder indicates service is needed. The same is true for size offset where “ON” cells of a matrix would be the same size and a variation in size could indicate pressure problems on a dot-peen marker.

One of the marking processes that is expanding the use of 2D matrix codes on actual parts is dot-scribe marking. Columbia Marking Tools’ line of high-speed Unigrav markers can produce clear and precise Data Matrix codes by dot-scribe marking. A recent test conducted by the HTE Vision Lab concerning the reliability of dot-peen marks on cast aluminum surfaces revealed the machine to be highly reliable in producing consistently readable marks on rough surfaces. Tests were conducted in accordance with NASA STD-6002 and NASA HDBK-6003.

The Test

The HTE vision lab successfully read the dot-scribe 2D Data Matrix codes on an automotive cast aluminum part with both fixed station and portable handheld Mix readers. The purpose of the tests was to supply automotive manufacturers with relevant information about the marking and reading of dot-scribe marks on cast surfaces, ultimately aiding in the production of quality marks that can be read with both fixed station and handheld readers. without difficulty. For example, the dot-peen 2D code produced by Columbia’s Richter Unigrav, marking system successfully read 10 out of 10 attempts on a rough aluminum casting with the hand-held reader.

Mark Analysis

The 2D code produced was of high quality, even though it was tested with a mark that had slightly “under printed” cells. Ideally the cells should be touching. Two approaches could be taken to improve the mark. The data cells could be spaced closer together, creating a smaller data matrix within the current cell size. The second option would be to increase the cell size so the cells are just touching, maintaining the current data matrix size. Cells can be made larger by a feature that simply increases either the marking pressure or the gap between the stylus and the part.

On cast surfaces, larger cell sizes tend to read better. The NASA Technical Standard NASA-STD-6002A, 5.1.9 states:

“Cast surfaces present a unique symbol decoding challenge because the surface irregularities (pits) create shadows that can be misinterpreted by the decoding software as dark data cells. Consequently, individual data cells in the symbol must be larger than the surface irregularities so that the decoding software can differentiate between the two features.”

There appears to be sufficient room on the marking pad boss on the aluminum part to produce a larger data matrix so the individual data cells are touching.

The Importance of Proper Cell Spacing for Readability

The nominal shape of a dot-scribed data cell is circular. Dot spacing is the horizontal or vertical distance between center points of contiguous cells. Dot spacing should be equal to the cell size or a printed Data Matrix symbol, (Fig. 6). The dot diameter should be 80-percent of the dot spacing to allow for a dot-size tolerance of ±10 percent.

In addition, dot spacing should be adjusted to ensure that the individual data cells touch or are in close proximity to each other. Gaps that exceed 50-percent of the dot size will adversely affect symbol readability. As a result, data cells should not overlap.

Cell Size Factors

The decoding of machine-readable codes on most castings is difficult because of surface roughness. The rough surfaces of most common castings make it difficult for Direct Part Mark readers to distinguish the difference between light and dark cells. 2D optical readers use the discrimination of light and dark cells to determine 1s and 0s to decode information contained in the marks.

On rough surfaces, cell sizes should be made as large as possible to overcome the confusion of light reflections (Fig. 8) that are returned to cameras from irregular surfaces. Overall symbol sizes should not exceed the available field of view of the cameras used to decode the mark. NASA marking standards dictate that many of the available marking processes not be used on cast surface parts that exceed defined roughness levels. The NASA standard states that cell size must increase in direct proportion to the average surface roughness of cast parts.

As a rule of thumb, the rougher the surface, the larger the cell size needs to be to insure proper decoding, especially with handheld readers that do not have the light adjustment capabilities of fixed station cameras.

For part surfaces with varying roughness, dot-scribe marking of Data Matrix codes appears to be the most effective, but after marking is necessary to confirm the mark quality is met.

Andy Habedank is an application engineering manager with Columbia Marking Tools Inc.

Len Dorazio is a project engineer with HTE Vision Lab.