IEEE WORKSHOP 2020 SENSORICA 2020 | May, 14th - 15th 2020, Mülheim an der Ruhr 8 Rejection of the Intrinsic Noise for Non-contact Temperature Measurement of an Object in a Closed Cavity D. Xu (1) (2) , J. Himmel (1) , D.Erni (2) , K.Thelen (1) (1) Measurement Engineering and Sensor Technology, University of Applied Sciences Ruhr West, 45479 Mülheim a.d. Ruhr, Germany E-Mail: dawei.xu@hs-ruhrwest.de, joerg.himmel@hs-ruhrwest.de, klaus.thelen@hs-ruhrwest.de (2) General and Theoretical Electrical Engineering (ATE), Faculty of Engineering, University of Duisburg-Essen, and CENIDE – Center for Nanointegration Duisburg Essen, 47048 Duisburg, Germany E-Mail: daniel.erni@uni-duisburg-essen.de Abstract – A novel and simple method to reject the intrinsic system noise within a non-contact absolute temperature measurement is presented. The absolute radio thermometry usually uses Dicke’s radiometer with a known reference temperature source, which has a fixed impedance [1] [2]. This can be applied to the radiometer, if a constant impedance of the antenna exists [3]. However, it is not suitable for the measurement of an object with variable impedance [4]. The impedance of the antenna, including the nearfield coupled object, might change its physical conditions. Moreover, the intrinsic noise of the measurement system depends on it strongly. Using the stochastic method based on a least mean squares algorithm [5] allows a measurement and calculation of the complex noise parameters. As an extension, this method is used to calibrate the intrinsic system temperature and compensate it for the measurement. An improvement of measurements accuracy and a simplification of the measurement procedure is achieved. Theory – The noise source of an amplifier is modelled by a current and a voltage source, which are partly correlated. The noise characterization of an amplifier can be parametrized with three coefficients, the variances of the noise current source , the noise voltage source and the complex correlation coefficient , corresponding to four real numbers [5]. Fig. 1 shows the diagram of the noise model of the measuring amplifier with a coupled input impedance of the measured object, which has the temperature proportional to the variance of the available voltage . Fig.1: noise model of the measurement system The output voltage can be measured with a spectrum analyzer and calculated with equation (1). (1) where is the Boltzmann constant, the measurement bandwidth and the gain of the amplifier. For impedance matching we have to measure the input impedance of the amplifier and the coupled impedance using a network analyzer previously. As the three noise coefficients and the coupled impedance are independent of each other, the characterization of the noise coefficient is carried out by using a set of arbitrary but known input impedances at known temperatures. In order to obtain the four mentioned numbers a calibration with at least four arbitrary impedances needs to take place. In our case, we used 10 different 2 n E I é ù ë û 2 n E V é ù ë û [ ] [ ] n n n n Re{E I V *} Im{E I V *} × + × L Z L T 2 L L L E V 4kT f Re{Z } é ù = D ë û o V k f D G i Z L Z L Z { } { } [ ] { } { } [ ] { } ÷ ÷ ÷ ø ö ç ç ç è æ × × - × × - - ú û ù ê ë é × + ú û ù ê ë é + D + = * * 2 2 2 2 2 2 Im Im 2 Re Re 2 Re 4 n n L n n L L n n L L L i i o V I E Z V I E Z Z I E V E Z f kT Z Z Z G V