Evaluation of a Modified ADC-Based Thermometry Bridge

Abstract

This article presents the modification and testing of an ADC-based thermometry bridge. The instrument under investigation is an Anton Paar MKT 50 Millikelvin Thermometer (developed at the IFE, TU-Graz) based on a precision analog-to-digital converter (ADC). During preliminary testing, it was found that the MKT 50 performs better than its declared uncertainty (1 mK equal to 1 ppm when using a 100 Ω PRT) and is comparable to thermometry resistance ratio bridges typically used in secondary thermometry laboratories (with typical uncertainties from 0.1 mK to 1 mK). The modifications to the original bridge were undertaken by the development team of the MKT 50 at the Graz University of Technology, Austria. Measurements and evaluation of the modified instruments were performed at the MIRS/UL-FE/LMK. For the MKT 50 to be used in thermometry laboratories as a reference unit, measuring parameters of the instrument had to be changed. During the first modification, the upper limit of the instrument range was decreased from 400 Ω to 133 Ω, this is a preferred range for standard platinum resistance thermometers (SPRTs). This also meant an increase in the measuring current from 0.5 mA to the more frequently used 1 mA. A modification of the programmable ADC control unit increased the resolution from 24 bit to 27 bit. By adding a switch, the use of an external standard resistor was enabled. After this stage of the modification, the first tests on the instrument were performed. The second stage was aimed at the removal of noise sources. The instrument was prepared in such a way that it only used two input channels, one connected to the SPRT and the other to the standard resistor. Also, the components of the ADC were upgraded to further reduce noise. The elimination of one input channel sped up measurements, making the PC software capable of taking several readings in a shorter time period. All tests were performed in laboratory conditions, where precision AC and DC resistance ratio bridges are typically used. The non-linearity was assessed by the use of an automated resistance bridge calibrator (RBC, Model RBC100), while the noise value was determined both from the standard deviation of RBC measurements as well as from comparison measurements of two standard resistors. All tests were repeated several times to assure confidence in the results. With its lowered range of 133 Ω and an increased resolution of 27 bit, the instrument non-linearity, its value below 2 μΩ, was comparable to primary resistance ratio bridges such as the ASL F900 or MI 6015T. However, the noise of the instrument remained relatively high at 4 μΩ. Since the modified MKT-50 is a much faster instrument than AC and DC bridges, averaging was used for true comparison. Measurements done with the modified MKT 50 were also averaged every 15 s (the time a classic resistance ratio bridge takes for one measurement). When measurements with the MKT-50 were averaged, the noise of measurements would be comparable to primary resistance ratio bridges.