One of the key elements in successfully utilizing a textile color matching and quality control system is the accurate and repeatable measurement of the samples being evaluated. Poor technique in sample measurement will greatly diminish the accuracy of the formulas produced by the shade matching software and will provide misleading results from the quality control software. Mistakes in either area can lead to considerable waste of time and money. This paper will focus on the techniques required to ensure repeatable sample measurement.

An understanding of spectrophotometers and how they work is useful in determining the appropriate conditions for sample measurement. Several types of spectrophotometers are available and most can be configured to meet certain measurement requirements as determined by the type of material being measured.

Geometry

The term "geometry" refers to the placement of a sample relative to the light source and measuring lens in a spectrophotometer. The most common geometries found in modern instruments are diffuse/8 and 45/0. Each instrument has a particular area where it is most useful.

In a diffuse/8 spectrophotometer, the sample is illuminated by use of an integrating sphere as indicated in Figure 1. Light that enters the sphere from the light source reflects off the coated interior surface of the sphere and strikes the sample from all angles. The reflected light is then measured at an angle of eight degrees relative to a perpendicular line drawn from the sample. This type of spectrophotometer tends to minimize the influence of surface irregularities on the light reflected from the sample, allowing measurement of reflectance due to color rather than surface variations. This is especially helpful in laboratory shade matching and in production dyeing where it is preferable to measure the difference in color of two samples with little consideration for differences due to construction or surface irregularities.

The 45/0 spectrophotometer depicted in Figure 2 makes use of one or more light sources to illuminate the sample at an angle of 45° with a lens placed at 0° to measure the amount of light reflected from the sample surface. This type of instrument is said to be more sensitive to surface irregularities than the diffuse/8 instrument and measures "appearance" as well as color. For this reason, 45/0 instruments are often used in quality control applications where differences in surface texture and finish are important.

Either instrument may be used for measuring textile materials as long as correct measurement techniques are used. Before a spectrophotometer is purchased, a decision must be made as to whether the color of the material or its appearance is to be measured.

Because of the differences between the readings produced on these types of instruments, reflectance data cannot be transferred between two systems unless instruments with the same geometry are used. There will still be some difference in readings produced on two instruments with the same geometry due to manufacturing variance, but the color difference between two samples measured on two instruments will be similar.

Light Sources

Spectrophotometers make use of one of two light sources - either a tungsten filament bulb which is similar to a common projector bulb, or a xenon flash bulb which is similar to the flash bulb of a camera. Certain guidelines should be followed depending upon the type of bulb being used.

Tungsten filament bulbs burn continuously and generate a considerable amount of heat. For this reason, temperature and light sensitive samples should not be placed at the instrument port until immediately prior to measurement of the sample. Extended exposure of some samples to the heat and light from the tungsten filament bulb will cause a dramatic change in the color of the sample.

Xenon flash bulbs do not generate heat so there is no concern for the sample heating previously mentioned. These bulbs are, however, very rich in ultraviolet energy. This ultraviolet energy will excite any fluorescing chemicals or dyes present in the samples and will lead to inaccurate match predictions. Filters are usually available to minimize the effect of ultraviolet energy for shade matching or to calibrate the amount of UV energy for accurate whiteness calculations. If the instrument being used does not contain a UV filter for calibrating ultraviolet energy, standards must be remeasured each time a batch is to be evaluated for whiteness. Stored standards can be used for evaluating whiteness only if they were measured on an instrument equipped with a UV filter. If a filter is not present, use of a small aperture on the instrument will also reduce errors due to fluorescence.

Viewing Area

Spectrophotometers typically come equipped with a range of aperture sizes to allow measurement of both small and large samples. It is always preferable to use the largest aperture size possible to minimize the influence of unlevel dyeings, but smaller ports may be used as necessary for measuring even the smallest of samples. Lab and production standards should be prepared with the intention of using the largest area view available on the spectrophotometer. Samples measured with small apertures will require additional reads to insure minimal measurement error. It is often impossible to measure some customer standards, especially when only threads or small clippings are provided for shade matching.

Specular Gloss

Many materials used as standards for shade matching appear glossy when viewed from a particular angle. These materials include shiny paint chips, plastic panels, and magazine pages among others. When these types of materials are measured for new shade formulation or when samples are measured behind a glass plate, it is important to exclude the glossy reflectance from the sample measurement. The glossy reflectance – or specular gloss – is automatically removed when using a 45/0 instrument and can be removed from a diffuse/8 instrument by use of a specular gloss port. The gloss port is located at an eight degree angle opposite the lens port and opens automatically when selected during the calibration routine on most software programs. The specular gloss can be safely included when comparing two identical materials but may be excluded when comparing two materials with significantly different gloss levels as long as the same measurement technique is used.

Instrument Diagnostics

Prior to measuring samples on the spectrophotometer, perform diagnostic tests to check the accuracy of the measurements. These tests should include a drift test to check for read to read repeatability, a diagnostic tile test to insure long term repeatability, and a standardization -- or calibration -- with a white tile. Only the white tile calibration is required daily while the other diagnostic tests can be performed on a weekly basis. Any poor test results should be resolved prior to measuring any samples. Long term stability of the instrument is critical due to the fact that standards and dye primaries are often used for many years.

As previously stated, the ability to measure a sample repeatably is essential for successful use of a color formulation or quality control program. Before any permanent samples are measured and stored into the computer database, a repeatable measurement technique must be established and observed. Samples should always be measured multiple times with the largest area view available on the spectrophotometer being used as long as the samples are large enough to completely cover the viewing area. Sample conditioning should also be considered because variations in temperature and moisture content can contribute to variations in measurement data.

Sample Thickness

Two to four layers will be sufficient for most knitted and woven materials to achieve an opaque sample for presentation to the instrument. If the material is not opaque, light will pass through the sample and reflect off the backing material or sample holder and produce misleading reflectance data. Lightweight and translucent materials will often require so many layers to become opaque that the material is forced into the interior of the instrument when measuring, causing inaccurate reflectance measurement (Figure 3). For

these types of materials, repeatable results can be obtained by measuring only a few layers of material backed with a white ceramic tile similar to the instrument’s calibration tile. The portion of the reflectance due to the color of the backing can be ignored when comparing two samples if the backing is the same for both. When measuring a sample for new shade formulation, however, the reflectance due to the backing will lead to errors in the predicted dye formula.

Sample Positioning

Sample rotation and repositioning will reduce measurement variability due to fabric construction, directionality of yarns, and unlevel dyeings. A common practice in sample measurement is to place the sample at the instrument port and simply rotate the sample for four or more measurements. This technique is quick, but it will not account for variations due to unlevel dyeing and should be avoided. A better technique is to remove the sample from the instrument and refold or reposition it before additional readings. Care should always be taken to avoid any areas of the sample that are contaminated by dirt, finger prints, creases, dye blotches, or other substances.

Developing a Repeatable Technique

An optimum measurement technique has been established when a sample can be measured, removed from the instrument, and remeasured with a variation of less than 0.15 DE (CMC) units. Higher variation will decrease the confidence level in the quality of the stored data and lead to less accurate match predictions.

A simple technique for determining the correct number of reads to use is to first produce an average reading for a sample by measuring it eight times -- being sure to rotate and reposition the sample after each read -- and saving the average. This should produce the most repeatable read even though it is not practical for day-to-day operations. Remove the sample and then measure it again using the same technique -- eight reads with rotation and repositioning. The color difference between these two averages should be very low. Remove the sample and then measure it again, but this time use only seven reads with rotation and repositioning. Repeat the process using six reads, five reads, four reads, three reads, and finally two reads.

After obtaining color difference data between each test and the original sample measured eight times, identify the point at which the DE(CMC) exceeds the desired limit of 0.15. As an example, if the DE(CMC) of the four read sample is 0.08 and the DE(CMC) of the three read sample is 0.21, samples should be read four times to ensure a variation of less than 0.15 DE(CMC). When the correct number of reads has been determined, measure the sample at least four more times using the required number of reads to confirm that all reads are less than 0.15 DE(CMC). If any of the measurements are greater than 0.15, the technique must be altered either by modifying sample placement or by taking additional reads. It may seem too time consuming to measure a sample three or more times, but the time taken in the beginning to ensure accurate measurements will translate into better results in the end. The measurement speed of modern spectrophotometers will reduce the time required to make additional reads to only a few seconds.

Measurement repeatability is especially critical as it affects computer pass/fail programs in use in the quality control and final inspection areas. If a sample is measured and the DE(CMC) is calculated as 0.80 but the measurement variability is 0.30, then the true reading can average from 0.50 to 1.10. This may mean the difference between a rating of pass or fail if the pass/fail tolerance is less than 1.10.

Samples to be measured can be produced in an assortment of forms, from fiber to yarn to sleeves. The type of sample produced is often dependent upon the equipment available or on the end-use of the yarn. Regardless of the form selected, an appropriate measurement technique should be developed as indicated above and the sample must represent the color of the batch as a whole. The contribution of the system operator to the measurement process cannot be overlooked and communication of proper techniques is critical.

Knitted Sleeves

Knitted sleeves provide the most repeatable measurements because of their uniform construction and size. The samples can be read very easily with the spectrophotometer’s largest view port. A typical measurement technique involves four measurements using four layers of material. Tests may show that as few as two reads are sufficient with fewer layers, but keep in mind that the deviation when repeating the sample measurement should be less than 0.15 DE(CMC). It is important to remember that simply rotating the sample without moving it to a new area on the sample will not produce the most accurate color assessment.

A common practice in the package dyeing industry is to include a small strip from a standard cone within the body of the knitted sleeve to be evaluated. Butt-knitting the standard and sample in this way will eliminate numerical shade variation that can result from comparing samples produced on different knitting machines or under different conditions. If the process of producing the knitted sleeves has been proven to be consistent, then it is acceptable to compare samples to stored standards produced using the same technique. Controlling knitting tensions may prove too difficult to eliminate the need to produce butt-knits. Several trials should be performed before a decision is made as to which route to follow.

Loose Fiber

Loose fiber is especially difficult to measure repeatably. A mass of fiber placed at the port of a spectrophotometer tends to protrude into the sphere in much the same way as too many layers of a sheer material. Not only does this introduce error into the reflectance measurement, but there is also the risk of loose fibers falling into the instrument and interfering with the measurement process. One technique to eliminate these errors is to place a piece of optically clear glass against the view port before measuring. It is critical that the instrument be configured to read in specular excluded mode to remove the glossy reflectance due to the glass. Other compensation factors may be required for shade formulation, but are not necessary for quality control if standards and batches are measured using the same piece of glass.

Repositioning the sample for multiple reads using only the instrument’s standard sample holder arm may provide measurements that are sufficiently repeatable. A better technique is to place an exact mass of fiber into a compression cell and apply a constant amount of pressure. This will eliminate errors due to gaps between fibers that exist under conditions of minimal pressure.

Yarn

The sample form that allows the quickest yet potentially most variable measurement is yarn. The most common method for measuring yarn is to obtain a small skein and simply place it against a small measurement port and rotate the sample for two or three measurements. The repeatability of this particular measurement method is questionable and must be confirmed as mentioned earlier. Individual yarns in the skein must be aligned to prevent the formation of shadows that the instrument would detect as depth of shade. The skein must also be thick enough to prevent light from passing through the strands and reflecting off the background which is typically the sample holder. An important fact to keep in mind is that skeins will produce lighter readings than sleeves knitted from the same yarn and these two materials should not be compared to each other for color difference calculations. Always compare yarn to yarn and sleeve to sleeve.

Yarns that are very bulky should be measured behind a glass plate to prevent the yarn from bulging into the spectrophotometer measurement area. Use the instrument’s specular excluded mode to remove the glossy reflectance from the glass surface. Loose pile goods such as carpet and towels can be measured in the same way to prevent the yarns from protruding into the instrument and to prevent fibers from falling into sphere instruments.

When matching shades for use in pile goods such as automotive or home upholstery, a skein of yarn is typically pulled through a cardboard tube and cut to allow evaluation of the tips of the yarn instead of the sides. These shades are typically evaluated visually because of the difficulty in presenting a uniform surface to the spectrophotometer for measurement. To successfully measure the color of a yarn pom, the density of the pom at the measurement plane must be controlled to prevent differences in depth of shade due to light trapped in open areas between the yarns. The pom must also be placed at the port in such a way as to prevent flattening of the pom which will expose the sides of the yarn to measurement. This can be very difficult without the use of a custom designed sample presentation apparatus.

Other useful techniques for yarn measurement include winding the yarn around a card or tab and using specially designed devices with springs that clamp the yarn securely to a plate. Yarn tension is a concern in either case and must be controlled from sample to sample to prevent measurement errors.

Package Measurement

Obtaining skeins and sleeves for shade analysis introduces additional time and effort into the process of determining the acceptability of a production dyeing. In addition, samples obtained from a small number of packages may not be representative of the color of the entire batch. A potential replacement for obtaining a sample from a few packages is to measure many packages using a portable spectrophotometer. The accuracy of this measurement technique must be confirmed relative to more traditional techniques of sample measurement involving skeins and sleeves.

Table one contains a compilation of data for a gray reactive dyed cotton shade gathered by measuring knitted sleeves and skeins on a bench-top diffuse/8 spectrophotometer and from dyed packages measured with a portable diffuse/8 spectrophotometer. The sleeves were folded to four layers and measured four times with rotation and repositioning for each reading. The skeins were also measured four times with rotation and repositioning. The packages were measured six times each with three reads on each end of the packages. The ends of the packages were measured because the curvature of the sides of the packages did not allow the portable instrument to be placed flush against the yarn. Packages with flat sides may be measured either on the ends or on the side, depending on which technique gives the most accurate results.

In this particular exercise, the first package was used as the standard and nine additional packages were compared to it to generate the color difference data. This provides data on package-to-package variability, but says nothing about the difference between the lot and the standard. True color differences between packages and the standard can be obtained with this technique only if a standard has already been measured in package form and is stored on the portable instrument prior to measuring the new lot. Without the standard, package measurements only provide data on variability within the lot itself, which is still useful data.

Examining the table gives us an idea of the repeatability of measurements between the three forms used. Measurement of knitted sleeves can be considered to be the most repeatable process and provides the truest indication of color difference between the ten packages. Theses sleeves were knit end-to-end on the same knitting machine, so tension variation does not contribute to color difference. The data is calculated using CMC equations. Because of the variability in the lightness experienced during the trial, the l:c ratio was set at 3:1 for skein and package measurements to minimize the influence of these variations on DE(CMC) and DL.

Looking first at the variations in color – or hue – as indicated in the DH column, we see that the largest difference between any pair of values for skein versus package is 0.06 and the average difference is 0.023. When comparing the package measurements versus the sleeve measurements, we see that the largest difference in hue is 0.11 and the average difference is 0.044. This indicates that variations in hue from one package to the next can be detected repeatably and accurately regardless of the form of measurement.

Examining the differences in brightness – or chroma – as indicated in the DC column of the table between the skein measurements and the sleeve measurements shows a maximum deviation of 0.10 and an average deviation of 0.065. For the package measurements the maximum deviation is 0.19 with an average deviation of 0.063. This data indicates that variations in chroma from sleeve to skein to package are also negligible.

The characteristic with the highest variability is lightness. The DL calculations were performed with an "l" value of 3.0 in the CMC formula to minimize the variations. The CMC formula allows this modification to increase the correlation between visual evaluation and measured data. For the skeins, the maximum deviation in DL was 0.28 and the average deviation was 0.104. For the packages, the maximum deviation was 0.46 with an average deviation of 0.196. These values indicate that lightness is the area to expect the most deviation from sleeve measurements. The l:c ratio can be changed further to increase the correlation of DL.

Similar results are seen when evaluating the DE(CMC) data. The maximum deviation in DE for the skein measurements relative to the sleeve measurements is 0.14 and the average is 0.065. For the package measurements, the maximum DE deviation is 0.28 and the average is 0.095. Eight of the nine packages had DE values less than 0.15 when compared to the sleeves, indicating excellent repeatability. This suggests that measurement of individual packages is a valid option for the purpose of detecting variation within the batch instead of knitting a sleeve of each package. Testing of a larger number of packages for a range of colors and yarn types would be necessary to confirm the results of this limited test. If the results do prove repeatable, then a program of measuring packages with a portable spectrophotometer will significantly reduce the time required to gather data on package variability which is a critical issue for some end-uses.

• All calculations are based on CMC formulas with l:c ratio 3.0:1.0.
• Package number 1 was used as the standard for all calculations.
• Sleeves and skeins were measured four times. Packages were measured six times.
  

If any degree of success is to be expected from spectrophotometers and computer formulation and quality control systems, repeatable techniques for measuring dyed samples must be established. Failure to establish a repeatable measurement technique will introduce a significant potential for error into all aspects of the formulation and quality control programs. If the old saying "garbage in - garbage out" applies to any process of data management, it applies especially well to the process of measuring dyed yarn.

A repeatable measurement technique includes specification of the number of layers of material to use, the positioning of samples, the number of measurements to make, instrument settings, and clear communication with the system operators. Failure to fully test and confirm the quality of a measurement technique will be a source of error for the life of the program.