This study aims to develop a practical radiation thermometry system of metals moving through in a high temperature furnace. In order to achieve this study, two problems; emissivity compensation of a target and elimination of background radiation noise filled in a furnace must be sorted out. We have successfully developed a method for simultaneous measurement of emissivity and temperature by use of polarized directional properties of the radiance from the target, in this case, stainless steel, and moreover a technique to eliminate background radiation noise using a pseudo blackbody installed in a furnace.

Laser polarimetry has been used for years to obtain normal spectral emissivity measurements on pulseheated materials. The method is based on the FRESNEL equations that describe reflection and refraction at an ideally smooth interface between two isotropic media. However, polarimetry is frequently used with surfaces that clearly deviate from this ideal condition. Questions arise with respect to the applicability of the simple FRESNEL equations to non-specularly reflecting surfaces. On the other hand, reflectometry utilizing integrating spheres provides a measurement of the hemispherical spectral reflectance for normal incidence, from which the normal spectral emissivity can be derived, regardless of surface texture. In a first effort to explore the limits of polarimetry in terms of surface roughness, room temperature measurements were performed on a number of samples using both an integrating sphere reflectometer and a laser polarimeter. In this paper, the two methods are briefly described and the results of the comparison are discussed.

The 1990 International Temperature Scale (ITS-90) substituted the platinum-platinum-10% rhodium thermocouple by the high temperature standard platinum resistance thermometer (HTSPRT) and the radiation thermometer, as an interpolating instrument in the 630°C to 1064°C range, due to the lower stability of the thermocouple. Although the uncertainty of reproducing the temperature scale became much lower, the cost of the required measuring equipments was raised. Aiming to offer the Brazilian Calibration Network accredited laboratories a lower cost temperature scale traceability alternative, at a smaller uncertainty than the standard type S thermocouple can provide, a 99,999% purity gold-platinum (AuPt) thermoucouple was exposed systematically to a high temperature environment (close to 1000°C) for more than 1500 hours, with its stability and homogeneity being evaluated with the aid of a silver fixed point cell. It was shown that a ± 25 mK (k=2) uncertainty can be achieved. This works details the methodology and the cares that have to be taken in order assure the reliability of the results.

Using integrated miniature fixed-point cells, a measuring uncertainty of < 1 K can be reached under operating condi tions in the superheated steam range of power plants by a periodic recalibration of the thermocouples, with operating times of > 20000 h. The fixed-point materials used for a temperature range from 500°C to 650°C are technically pure metals and binary alloys.

Resistance bridges, when used for temperature measurements with SPRTs, are able to achieve uncertainties better than 1 ppm. However, they have several shortcomings that prohibit their use in many applications. Among these are cost, size, slow speed, and limited range. An endeavor was made by the author to design a readout for resistance thermometers that achieves less than 1 ppm uncertainty in resistance ratio while overcoming some of the problems of resistance bridges. A new approach was taken with a design that uses the latest integrated-circuit analog-todigital converters. This allows the instrument to have lower cost, smaller size, the capability of increased speed, and additional features. Special effort was made to reduce errors caused by component drift, thermoelectric EMF, component offset, electrical noise, and nonlinearity. The new resistance thermometer readout was tested to identify and evaluate sources of measurement uncertainty. The combined uncertainty was calculated for resistance ratio and W(T90) measurements of an SPRT with self-heating corrections. Measurements made with the resistance thermometer readout were compared with measurements made with a resistance bridge. The results show that the standard uncertainty of the new resistance thermometer readout is about 0.34·10^-6 in measuring resistance ratio at 25 Ω/100 Ω and about 0.68·10^-6 in measuring W(T90) near the triple point of water.