National Calibration Facility for Retroreflective Traffic Control Materials (2005)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CHAPTER 2 SOURCE CHARACTERISTICS The source was designed to meet or exceed the requirements listed in the documentary standards. The source uses an Abbe projection system to image a uniformly illuminated field aperture at the specimen location through a source aperture of defined angular extent. Two systems for achieving a uniformly illuminated field aperture were investigated – an imaged 100 W tungsten strip lamp and a pumped integrating sphere. STRIP LAMP PROJECTION SYSTEM The first system is a 100 W tungsten strip lamp imaged onto the field aperture. This is the system used in the NIST retroreflectance facility in the 1980s. Figure 3 shows a schematic of the projection system. An aspheric lens with a diameter of 60 mm and a focal length of 39 mm images the 2 mm x 18 mm tungsten strip on a 5 mm field aperture. The 5 mm field aperture limits the transmitted light such that the image is from a 0.83 mm diameter circle of the tungsten strip. One of a variety of a chromatic projection lenses is chosen to image the field aperture at the retroreflective device after passing through the source aperture. If the image of the field aperture is larger than the size of the retroreflective device, then the device is the aperture stop of the source and the source aperture is the field stop. However, if the image of the field aperture is smaller than the size of the retroreflective device, for example if a smaller field aperture is used, then the image is the field stop and the source aperture is the aperture stop. Table 1 lists the lenses and image sizes for various distances to the retroreflector device. The strip lamp system is the source that will be used on a routine basis for calibrations and research activities. A summary of the requirements determined from standards and the characterization of the strip lamp projection system is presented in Table 2. Stability and Correlated Color Temperature The stability of the source is primarily dependent on the current control of the lamp. The lamp is powered by a constant current power supply that is analog controlled from the output of two 12-bit digital-to-analog controllers. One analog output is divided by 4096 and then added to the second analog output to achieve a 22-bit signal. Two bits of signal are lost in the addition. The 22-bit signal controls the current supply with a resolution of 0.004 mA. The digital-to- 13

analog controller is programmed based on the voltage drop across a calibrated shunt resistor. Over the course of a year setting the current has an uncertainty of 0.013 % (k=2), as calculated from Table 3. This table follows the procedure used in the “Guide to the Expression of Uncertainty in Measurement” (GUM) published by the International Organization for Standardization (ISO).(1) This Guide establishes general rules for evaluating and expressing uncertainty in measurement. Listed in the table are the dependent quantities for the final result, the symbol for these quantities, the value and the standard uncertainty, the unit of the quantity, the type of evaluation method, which is Type A for statistically based and Type B for all others, the degrees of freedom, which for a Type A measurement is typically the number of repetitions and the sensitivity coefficient, which is the partial derivative of the model or equation with respect to the quantity. The sensitivity coefficient can be calculated or determined experimentally. The uncertainty contribution is the standard uncertainty times the sensitivity coefficient. The combined standard uncertainty is the sum of the squares of the uncertainty contributions. Therefore, the last line in the table gives the resultant, the unit and the combined uncertainty. Typically, uncertainties are expressed as an expanded uncertainty with a coverage factor, k. A coverage factor of 2 is similar to a confidence interval of 95.45 % if the degrees of freedom are large. Using this feedback loop the current will vary over 24 h by less than 0.008 % (k=2), as calculated from Table 4. The difference is the calibration of the voltmeter for a day versus a year. The luminance of a tungsten lamp is approximately proportional to the current raised to the power of 7. Therefore, the luminance of the tungsten lamp will vary less than ± 0.056 % (k=2) over a day from the current setting. Figure 4 shows that once the lamp has stabilized, the intensity fluctuation is less than ± 0.025 %, which is within the expected uncertainty. The correlated color temperature (CCT) is determined by a spectrometer that is calibrated against the NIST photometric color temperature standards using a pressed PTFE plaque. The PTFE plaque is placed in front of the projection system and viewed by the spectrometer, and the current is set so that the light has a 2856 K ± 10 K (k=2) spectral distribution. The CCT of the strip lamp changes 1 K for every 7 mA at the operating current for Illuminant A. Since the current is controlled to ± 0.008 % (k=2) or ± 2 mA over a 24 h period, the CCT should not change by more than ± 0.3 K. Figure 5 shows that the CCT only changes within the ± 1 K resolution of the spectrometer used to monitor the CCT. The final uncertainty budget 14

dependence on the CCT is more complex. The correction factor based on the CCT is the set CCT divided by the expected CCT. Since the expected CCT is 2856 K and the set CCT is 2856 K, the correction factor is 1. However, the expected CCT has no uncertainty where as the set CCT has an uncertainty of 10 K (k=2). The sensitivity coefficient used in the final uncertainty budget depends on the detector spectral response and the reflectance factor of the sample measured. The most time effective method of determining the sensitivity coefficient in this case is through simulation. The final results of the simulation are in Figure 6 and the calculation method is described below. The calculation is normalized to the values calculated at 2856 K. For example, since the illuminance measured at the sample is dependent on the spectral response of the detection system, ideally the spectral luminous efficiency function, Vλ, the voltage of the detector system has to be calculated with a given CCT spectral distribution and divided by the voltage of the detector system with the given CCT spectral distribution at 2856 K. The illuminance ratio is then divided by the ratio at the detector, which is now dependent on the reflectance factor of the sample directing the light to the detector. Equation 1 expresses this mathematically, ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ∫ ∫ ∫ ∫ = λλλλΦ λλλλΦ λλλΦ λλλΦ dsR dsR ds dsRNorm rsK rsCCT rK rCCT L 28562856 . (1) where ΦCCT(λ) is the spectral power distribution of the source at a given CCT, Φ2856K(λ) is the spectral power distribution of the source at 2856 K, sr(λ) is the spectral response of the detection system, Rs(λ) is the retroreflectance factor for a given sample, and the integration is over the visible region. The retroreflectance factor curves used in this calculation are measurements and are shown in Figure 7. The difference between the Vλ function and the spectral response of the NIST detection system is not visible in a figure and therefore is not shown here. The slopes of the curves in Figure 6 are the sensitivity coefficients. Table 5 summarizes the slope of the curves for the different materials and shows the relative expanded uncertainty for each material. The relative expanded uncertainty is the contribution that the CCT uncertainty will have on the final uncertainty budget for the measurement of the coefficient of retroreflected luminance, RL. 15

Source Aperture Uniformity and Size The uniformity of luminous flux at the apertures of the projection system is important because it affects the uncertainty budget in several ways. The uniformity of the illuminance at the source aperture determines the weighting of the observation angles that are integrated within the aperture area. The uniformity was measured by scanning a photometer with a 3 mm aperture horizontally and vertically perpendicular to the source aperture. The uniformity of the illuminance at the source aperture was determined to be within ± 3 % of the mean value and is shown in Figure 8. To determine the effect the source aperture illuminance uniformity has on the overall uncertainty budget, we developed a simulation based on the aperture synthesis procedure described in Appendix D of the document CIE 54.2 “Retroreflection: Definition and Measurement.”(2) The definition of RL uses the angles α, β1, β2, and ε, which are defined originating from points. The definition is understood as the limiting value for infinitesimally small source and observer apertures. The measurement using an aperture of reasonable size generally introduces approximations to this theoretical definition. For calibration purposes the aperture size is fixed. For example, a measurement may define the distance to be 15 m and the apertures to have an angular extent of 6’ from the sample. Therefore, the aperture should have a diameter of 26.18 mm. Any deviation from 26.18 mm causes uncertainty in the measurement. The source aperture made at NIST was measured at the Absolute Aperture Area Facility and has a diameter of 26.01216 mm with an uncertainty of 0.00087 mm (k=2). The aperture synthesis procedure uses a large number of RL measurements made using extremely small source and observer apertures. The reader can image a small area that is near the edge of the source aperture that is farthest from the detector aperture and vertically offset higher as shown in Figure 9. The light emerging from that small area goes to the sample and is retroreflected back to a small area near the edge of the detector aperture that is farthest from the source aperture and vertically offset lower. This path of light has a larger observation angle and a significantly different rotation angle than the path of light from the centers of the source and observer apertures. If the amount of light is much less or more at the edge of the aperture than the center, the average RL value determined for this aperture size will be different than that determined for a uniformly illuminated source aperture. By treating the small areas as vectors as described in CIE 54.2, the simulation was calculated rather easily. The simulation showed that 16

the relative standard uncertainty due to the NIST non-uniformity at the source aperture for an encapsulated bead sign material is 0.000012 %. The non-uniformity of the sample and the detector are not considered in this quantity, and will be included in their respective chapters. For prismatic material, the RL dependence on rotation angle was modeled as a sine wave that has three periods over 2π and oscillates from 0.8 % to 1.2 % of the RL at 0°. The relative standard uncertainty for a prismatic sign material is 0.0026 %, given the NIST source aperture illuminance uniformity. The initial angle values used in the simulation were α = 0.33°, β1 = -5°, β2 = 0°, and ε = 0°. When these values were changed, no large changes in the relative standard uncertainties were observed. NIST modeled some of the systems encountered in the initial survey of instruments available. One such instrument had a rather non-uniform source aperture illuminance because it produced an image of the lamp filament next to a reflected lamp filament image. The reflected filament image is typically 20 % less in flux than the filament image. To model this situation a continuous curve that peaks in the center and is 20 % less on one side was used. A plot of this curve is shown in Figure 10. The vertical axis in the plot is along the observation plane in this system. The correction for beaded material with this system is 0.016 % and for prismatic material the correction climbs to 0.20 %. If the center of the filament image was placed perpendicular to the observation plane, the correction for beaded material is 0.22 % and for prismatic material the correction drops to -0.20 %. The instrument examined positions the non- uniformity in the most favorable geometry. However, this may not be the case for all instruments. The aperture synthesis procedure was also used to model the uncertainty dependence on the size of the aperture. The relative RL values were calculated for apertures of 23 mm and 29 mm versus a 26 mm aperture for beaded sign material and prismatic sign material. The results are shown in Figure 11. As stated above, the diameter of the source aperture is 26.01216 mm; therefore, ∆d is -0.16779 mm. The correction factor calculated using the equations in Figure 11 is 0.99996 for beaded material and 0.99993 for prismatic material. Since the simulation is not based on detailed information for a particular material, the correction factor is included in the total uncertainty budget. The relative expanded uncertainty due to the size of the aperture is 0.0086 % (k=2) for beaded material and 0.014 % (k=2) for prismatic material. 17

Retroreflector Aperture Surface Uniformity and the Illuminance Measurement Document CIE 54.2 defines the retroreflector aperture surface as the area illuminated by the source and viewed by the detector given either by the retroreflector itself or by a diaphragm enclosing part of the retroreflector. Retroreflector aperture refers to the angular dimensions from the source point of reference to the retroreflector aperture surface. The uniformity of the illuminance at the retroreflector aperture surface has a strong dependence on the illuminance measurement method and on the light retroreflected to the detector in association with the uniformity of the coefficient of retroreflection for the device. The uniformity of the illuminance at the retroreflector aperture surface was determined to be within ± 1.8 % of the mean value and is shown in Figure 12. The uniformity was measured by scanning a photometer with a 5 mm aperture across the beam over a 40 cm range in both x and y perpendicular to the beam. The uniformity of the illuminance at the retroreflector aperture surface was scanned many times over several days after burning the lamp for days at a time. The structure of the uniformity did not change by more than 0.10 %. The absolute flux fluctuated less than 0.30 % on a day-to-day basis. Based on this data the illuminance for the given day can be measured by sampling the retroreflector aperture surface instead of performing a complete scan. To measure the illuminance the detector is positioned on the goniometer in the center of the retroreflector aperture surface. The center is measured and then the detector is moved in a circle 5 cm away from the center point where 8 additional measurements are made. The 9 voltages are averaged then adjusted by a correction factor based on the size of the sample to be measured. For example, if a 20 cm by 20 cm sample is measured, the average voltage is multiplied by 0.99585, because the average illuminance on the 20 cm by 20 cm area is less than the average illuminance of the area sampled. The relative expanded uncertainty for the correction factors and therefore the illuminance measurement is 0.212 % (k=2) based on the uniformity measurement, the sampling, and the expected change over time. The illuminance is based on the projected area of the sample; therefore, the illuminance needs to be re-calculated for every movement of the entrance angles. For the measurement of pavement marking material, an adjustable rectangular aperture in place of the 5 mm aperture in the projection system regulates the size of the illuminated area on the pavement marking material. The image formed is small enough that instead of sampling the 18

illuminance, the complete image is scanned and the voltages averaged. The relative expanded uncertainty of the illuminance measurement for pavement marking material is 0.06 % (k=2). The effect of uniformity of the retroreflective device would be insignificant if the illuminance at the retroreflector aperture surface were perfectly uniform. Since it is not, each retroreflective device needs to be examined for uniformity. The standard deviation of the illuminance for the NIST retroreflectometer across the retroreflector aperture surface for a sample 20 cm by 20 cm is ± 0.85 %. If the uniformity of the device were not sampled, the standard deviation of the illuminance would have to be included in the final uncertainty budget. To not correct for this means the final relative expanded uncertainty would be ± 1.7 % (k=2), based on given knowledge of the uniformity of the sample. To sample the uniformity of the retroreflective device, a small aperture flips into position immediately after the 5 mm aperture in the Abbe projector. This new field stop produces an illuminated area 3 cm in diameter at the retroreflector device. Through an automated process the goniometer moves the sample in a grid, 6 steps wide by 6 steps tall, adjusting the angles and the distance to make all 36 measurements as if the source was illuminating them in their centered original positions. Therefore, the observation, rotation, and entrance angles are adjusted slightly. Figure 13 shows a demonstration of the angles and distance changes required. The 36 RL measurements are mapped onto the illuminance uniformity of the projection source and a correction factor is determined. The correction factor is close to one and the relative expanded uncertainty is about ± 0.20 % (k=2). The magnitude for the correction factor and uncertainty depend on the magnitude of RL for the material and the change in RL across the device. The uniformity of RL for pavement marking samples is typically poor. The uniformity of the projection system is ± 1.8 % of the mean value from top to bottom. Changing the vertical width of the adjustable rectangular aperture assesses the uniformity of pavement marking materials. The uniformity of a typical marking sample may change over 40 % from top to bottom or along the length. Mapping the projection system illuminance uniformity onto the RL uniformity of such a sample produces a difference of 0.23 %. Since the correction factor is based on measurements that are weak in signal, it is not applied and the difference is added as a component into the overall uncertainty budget. 19

SPHERE PROJECTION SYSTEM The second system consists of a 5 cm diameter sphere made from Zenithpolymer, a proprietary material, with two 10 mm diameter entrance ports and an 8 mm diameter exit port. Zenithpolymer is a material of processed PTFE (polytetrafluoroethylene) that functions as a volume reflector. Typical reflectance is 98 % between 300 nm and 1700 nm. It is insoluble in water and its characteristics do not change from –200 ºC to 260 ºC. The light from four 410 W reflector lamps is coupled into the sphere; two lamps per entrance port. The exit port is imaged by an Abbe projection system similar to the system used in the strip lamp system. The sphere system provided similar characteristics to the strip lamp system with more operating complications. The most significant improvement is the uniformity at the retroreflector aperture surface as shown in Figure 14. The sphere source would reduce only one uncertainty component; therefore, it was determined that the additional effort of operating it is not worth the reduction in uncertainty. The real advantage of the sphere projection system is that any light can be coupled into the sphere without changing any of the projection optics. The high intensity discharge (HID) lamps that are available in cars have a very distinct spectral pattern. The retroreflectance of devices can be calculated if spectral coefficients of retroreflection are measured, but to experimentally verify the results, the sphere projection system is the only option. Additionally, for research purposes a portable tunable laser system can pump the sphere to create a monochromatic source to measure fluorescence bi-spectrally. In this configuration NIST can also determine if a laser beam can be used to measure retroreflection. A laser beam has coherence, which causes diffraction patterns under certain circumstances. By coupling the laser into the sphere through a fiber optic the coherence can be effectively removed. Comparing the results from a laser beam and the monochromatic sphere source can determine if retroreflective devices have issues with coherence. REFERENCES (1) “Guide to the Expression of Uncertainty in Measurement.” International Organization for Standardization, Geneva, Switzerland, First Edition, (1995) 101 pp. (2) “Retroreflection: Definition and Measurement.” Technical Report CIE 54.2-2001, Commission Internationale De L’Éclairage, Vienna, Austria (2001) 55 pp. 20