Two-stage segment linearization as part of the thermocouple measurement chain

• Published in: Measurement and control (London) Vol. 54; no. 1-2; pp. 141 - 151
• Main Authors: Živanović, Dragan, Simić, Milan
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01.01.2021; Sage Publications Ltd; SAGE Publishing
• Subjects: Approximation; Chains; Error analysis; Linearization; Lookup tables; Mathematical analysis; Microcontrollers; Nonlinearity; Reduction; Segments; Software; Thermocouples; Transducers; Transfer functions; virtual instrumentation; approximation error reduction; Two-stage linear approximation; thermocouple transfer curve
• ISSN: 0020-2940; 2051-8730
• DOI: 10.1177/0020294020986833

An implementation of a two-stage piece-wise linearization method for reduction of the thermocouple approximation error is presented in the paper. First, the whole thermocouple measurement chain of a transducer is described, and possible error is analysed to define the required level of accuracy for linearization of the transfer characteristics. Evaluation of linearization functions and analysis of approximation errors are performed by the virtual instrumentation software package LabVIEW. The method is appropriate for thermocouples and other sensors where nonlinearity varies a lot over the range of input values. The basic principle of this method is to first transform the abscissa of the transfer function by a linear segment look-up table in such a way that significantly nonlinear parts of the input range are expanded before a standard piece-wise linearization. In this way, applying equal-segment linearization two times has a similar effect to non-equal-segment linearization. For a given examples of the thermocouple transfer functions, the suggested method provides significantly better reduction of the approximation error, than the standard segment linearization, with equal memory consumption for look-up tables. The simple software implementation of this two-stage linearization method allows it to be applied in low calculation power microcontroller measurement transducers, as a replacement of the standard piece-wise linear approximation method.