Professional Documents
Culture Documents
I. I NTRODUCTION
N TEMPERATURE (T ) measurements, different solutions
I of sensors with the operating principle based on specific
physical phenomena are used. For example, in parametric
sensors with an electrical output signal, the influence of
temperature on the resistance of a conductor, of a semicon-
ductor, and on the voltage or the current of an n-p-junction
is used. In a group of parametric sensors, resistive temper-
ature detectors (RTD) are important, especially those made
of platinum (Pt). The T -to-resistance (Rs ) processing charac-
teristics of platinum RTDs and their classes of accuracy and
measurement ranges corresponding to them are defined by a
standard [1].
An RTD is connected with a temperature transducer (TT)
in a two-, three-, or four-wire way [2], [3]. In most cases,
TT output requires a digital signal whose value corre-
sponds with the measured value T . In TT input circuits, the divider [2]–[11]. Another solution is resistance-to-time (Rs /t)
RTD-resistance-to-voltage (Rs /V ) processing is applied by processing [12]–[14]. In a TT with Rs /V processing, an
means of an unbalanced Wheatstone bridge or a voltage amplifier (A), an analog-to-digital (A/D) converter, and a
Manuscript received March 1, 2018; revised July 2, 2018; accepted microprocessor (μP) are used. A TT with Rs /t processing
July 4, 2018. The Associate Editor coordinating the review process was comprises a voltage comparator, a counter, and a microproces-
Dr. Huang-Chen Lee. (Corresponding author: Sergiusz Sienkowski.) sor. In order to ensure high accuracy of the measurement of the
The authors are with the Institute of Metrology, Electronics, and Computer
Science, University of Zielona Góra, 65-246 Zielona Góra, Poland temperature T , appropriate hardware and software solutions
(e-mail: W.Miczulski@imei.uz.zgora.pl; M.Krajewski@imei.uz.zgora.pl; are applied to minimize processing errors of the measuring
S.Sienkowski@imei.uz.zgora.pl). signal. Table I presents various solutions in this regard. Some
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. of these solutions deal with intelligent temperature transducers
Digital Object Identifier 10.1109/TIM.2018.2857901 (ITTs). The existing ITT solutions do not include autocali-
0018-9456 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
Fig. 3. Error of temperature measurement (T ) for an ITT with no Fig. 4. Uncertainty intervals of temperature measurement (UT ) for a 99%
autocalibration procedure and with the influence of the ambient temperature confidence level as a function of the measured temperature as condition (3)
(at values Ta = 35 °C and Ta = 75 °C) on the values of the bridge resistors is not fulfilled due to the resistance scatter l_St j,S (0.01, 0.03, and 0.05 ).
(R2 , R3 , and R4 ), changes in the supply voltage V by −0.5% and in the gain
G by −0.1%, the amplifier imbalance voltage V0 = −80 mV.
resistances Rl_S and Rl_St j , only in the case of nonfulfillment results for such a solution for two values of the coefficient
of condition (3). TCR R St j (2.5 and 5 ppm/°C) and for the assumed value of
In the case of applying an autocalibration procedure, the ambient temperature of the standard resistors Ta(R St j ) =
the uncertainty of temperature measurement also depends on 75 °C determined with the error T a(R St j ) = ±0.5 °C and
the manufacturing accuracy of the standard resistors R St j T a(R St j ) = ±1 °C are presented in Fig. 7. The proposed
which influences the accuracy of determining the standard solution reduces the error T tenfold in relation to the case
temperatures TSt j . The simulation research results presented when the ambient temperature of the standard resistors is not
in Fig. 5 show unequivocally that achieving small values measured. The values of the error T (Fig. 7) are close to the
of the uncertainties UT requires applying precision resistors. values of the errors for nominal conditions (Fig. 2).
Ensuring the uncertainty of the measurement of T at a value Fig. 8 presents the results of simulation research on deter-
level of 0.03 °C requires applying standard resistors with mining the uncertainty intervals of temperature measurement
accuracies not lower than 0.002%. by means of the MC method for the assumed numbers of bits
Changes in the ambient temperature Ta(R St j ) of the standard of the A/D converter. The research indicates a necessity to
resistors R R St j will considerably affect the error of tempera- employ A/D converters with a minimum resolution of 18 bits
ture measurement. Fig. 6 shows the results of the simulation so that the uncertainties of the measurement of T on account
research describing the influence of the ambient temperature of this are not greater than 0.02 °C.
Ta(R St j ) at 35 °C and 75 °C for two assumed values, equal to Another problem in ITTs is noise. Fig. 9 shows an example
2.5 and 5 ppm/°C, of the coefficient TCR R St j of the standard simulation result illustrating the effect of standard deviation
resistors on the temperature measurement error. The presented σnoise of Gaussian noise added to the bridge output voltage on
research results indicate a need for applying precision standard temperature measurement uncertainty. The presented results
resistors of very small TCR R St j values. refer to the measurement case T = 180 °C for which the
In order to reduce the error value T caused by the changes highest uncertainty values were obtained.
in the standard resistors ambient temperature (Ta(R St j )), From Fig. 9, it follows that to obtain an uncertainty of a
an extension to the autocalibration procedure was proposed single T measurement below 0.02 °C, the noise must be less
involving the use of the additional temperature measurement than 2 μV. Reducing the impact of noise and electromagnetic
Ta(R St j ) of an independent sensor with its digital output interference on the final measurement result can be obtained
connected directly to the microprocessor circuit. The research by [21] as follows:
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
A. Simulation Tests
The evaluation of the designed measurement circuit was
conducted by means of the MC method. The determination
of the uncertainty of the temperature T measurement was
realized in the MathCAD program, according to the following
Fig. 9. Effect of σnoise on the uncertainty of a single temperature measure- assumptions.
ment for a 99% confidence level, for T = 180 °C, measured by the ITT in
nominal conditions. 1) The number of random experiments M was 106 .
2) Sources of uncertainty were defined as measure-
ment input quantities with values: δ R St j (i ), TCR St j (i ),
1) averaging the results of multiple measurements; δ R2 (i ), δ R3 (i ), δ R4 (i ), TCR S (i ), δV (i ), δ1 (i ), 1 (i ),
2) shielding and grounding the connections, especially the T a(R St j )(i ), T a (i ), and l_St j,S (i ), generated ran-
RTD connection with the bridge; domly (for i = 1 . . . M) according to their uniform
3) using analog filters; probability distribution.
4) appropriate selection of elements, e.g., low-noise resis- 3) The output quantity was defined as an error of tem-
tors, used in the ITT processing channel. perature measurement whose values were determined
The test results shown in Figs. 8 and 9 imply a need to from (8).
use an 18-bit A/D converter and a low-noise level in the ITT. 4) Based on (1) and (2), mathematical models were
Such requirements are met by popular sigma–delta converters, designed of the bridge voltage V1_S (i ) (for temperature
which can contain a programmable amplifier in their structure. sensor) and V1_St j (i ) (for standard resistors), whose
values depend on δ R St j (i ), TCR St j (i ), δ R2 (i ), δ R3 (i ),
III. E VALUATION OF T EMPERATURE δ R4 (i ), TCR(i ), δV (i ), δ1 (i ), 1 (i ), T a(R St j )(i ),
M EASUREMENT ACCURACY T a (i ), and l_St j,S (i ), and are described by the depen-
The evaluation of the temperature measurement accuracy dencies
within a range of 0 °C–200 °C by means of the ITT was
V1 S (i ) = k S (i )V ∗ (i )(1 + δ1 (i )) + ∗1 (i ) (9)
conducted for the designed measurement circuit realizing an
∗
autocalibration procedure taking into account the research V1_St j (i ) = k St j (i )V (i )(1 + δ1 (i )) + ∗1 (i ) (10)
results presented in Section II. In the designed circuit (Fig. 1), where
the amplifier (A) and the A/D converter were replaced with
R S + Rl_S (i ) R 2 (i )
an Agilent 3458A multimeter, and the μP was replaced with k S (i ) = −
a PC. The evaluation of temperature measurement was made R S + Rl_S (i ) + R4 (i ) R2 (i ) + R3 (i )
for the following measurement conditions. (11)
R St j (i ) + Rl_St j (i )
1) In the bridge circuit, RWE 0411 standard metal resistors k St j (i ) =
(R St0 –R St4 ) with a coefficient TCR St j = 5 ppm/°C and R St j (i ) + Rl_St j (i ) + R4 (i )
values of 1020.435, 1199.436, 1379.225, 1579.465, and R2 (i )
− (12)
1759.106 , measured by an Agilent 3458A multimeter R2 (i ) + R3 (i )
with the accuracy δ R St j = 0.002% were applied. R2 (i ) = R2 (1 + δ R2 (i ))(1 + T C RT a (i )T a (i ))
2) In the bridge circuit, metal resistors with the nominal (13)
values: R2 = R3 = R4 = 1000 in the first of the
R St j (i ) = R St j (1 + δ R St j (i ))
analyzed cases, and R2 = 910 and R3 = R4 =
1000 in the second, with the tolerance δ R2 = δ R3 = × (1 + T C R R St j (i )T a(R St j )(i )) (14)
δ R4 = 0.5% and TCR = 50 ppm/°C were applied. Rl_St j (i ) = Rl_St j + l_St j (i ) (15)
3) The bridge was powered by the voltage V = 2 V from an Rl_S (i ) = Rl_S + l_S (i ) (16)
SQ11 calibrator with the measured short-time instability
V ∗ (i ) = V (1 + δV∗ (i )) (17)
δV = ±10 ppm.
4) The accuracy of the voltage V1 measurement by means where R S is the temperature sensor resistance calculated
of an Agilent 3458A multimeter is V 1 = δ1 V1 + 1 = based on (7), R3 (i ) and R4 (i ) are similarly calculated
14 × 10−6 V1 + 0.3 μV within a 1 V range. according to (13), and ∗ signifies that the values of
5) The ambient temperature of the standard resistors a particular quantity were independently generated for
(Ta(R St j )) and of the remaining part of the circuit (Ta ) each connected standard resistor and the Pt1000 sensor
is 25 °C with the accuracy T a(R St j ) = T a = 1 °C. in a bridge configuration.
6) In the bridge circuit, a compensating loop and a unit of 5) The calculations of the temperature TMC (i ) were con-
reed relay switches (SU) were applied. Next, based on ducted based on (4) and (5) substituting the values of
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
TABLE II
C OMPARISON OF T AND T _meas VALUES O BTAINED AS A R ESULT
OF S IMULATION AND E XPERIMENTAL T ESTS FOR
THE T WO A DOPTED R2 VALUES
1) a fourth-degree polynomial describing the processing [16] S. Pradhan and S. Sen, “An improved lead compensation technique
characteristic of the ITT measurement channel; for three-wire resistance temperature detectors,” IEEE Trans. Instrum.
Meas., vol. 48, no. 5, pp. 903–905, Oct. 1999.
2) precision standard resistors with a 0.002% accuracy and [17] J. Rivera, M. Carrillo, M. Chacón, G. Herrera, and G. Bojorquez, “Self-
TCR = 5 ppm/°C together with a sensor for measuring calibration and optimal response in intelligent sensors design based on
their ambient temperature with an accuracy not worse artificial neural networks,” Sensors, vol. 7, no. 8, pp. 1509–1529, 2007.
[18] D. R. White, K. Jones, J. M. Williams, and I. E. Ramsey, “A simple
than 1 °C; resistance network for calibrating resistance bridges,” IEEE Trans.
3) line resistance fulfilling the condition Rl_S = Rl_St j with Instrum. Meas., vol. 46, no. 5, pp. 1068–1074, Oct. 1997.
a scatter not greater than 0.01 ; [19] J. R. Smith et al., “Method and apparatus for auto-calibration of signal
conditioning electronics,” U.S. Patent 4 642 636, Nov. 16, 1983.
4) an A/D converter with a resolution not worse than 18 bit. [20] W. Miczulski, “Method for measuring nonelectric quantities with
Reducing the given temperature range while the degree of bridged resistive sensor,” Polish Patent 229 338, Jul. 31, 2018.
[21] Resistive Bridge Basics, Part One, Tutorial 3426, Maxim Integr.
the polynomial describing the ITT processing characteristic Products, Inc., San Jose, CA, USA, 2004.
remains the same will lead to the reduction of systematic [22] Evaluation of Measurement Data—Supplement 1 to the ‘Guide to the
errors. Applying other measurement range values requires expression of uncertainty in measurement’—Propagation of Distribu-
tions Using a Monte Carlo Method, document JCGM 101:2008, 2008.
selecting an appropriate polynomial degree depending on the [23] M. Krajewski, “Constructing an uncertainty budget for voltage RMS
assumed temperature measurement uncertainty. measurement with a sampling voltmeter,” Metrologia, vol. 55, no. 1,
In terms of the achieved accuracy of the temperature T pp. 95–105, 2018.
measurement, the ITT solution proposed in this paper is
similar to the solutions presented in [13] and [14] (Table I)
where TT input circuits are based on Rs /t processing.
The obtained very small values of temperature measurement Wiesław Miczulski received the M.Sc. degree in
errors indicate the possibility of the developed ITT to cooper- electrical engineering from the College of Engineer-
ate with precise IEC Class AA platinum RTDs [1]. ing, Zielona Góra, Poland, in 1980, the Ph.D. degree
in electrical metrology from the University of Tech-
nology, Wrocław, Poland, in 1989, and the D.Sc.
R EFERENCES degree in electrical metrology from the University
of Zielona Góra, Zielona Góra, in 2002.
[1] Industrial Platinum Resistance Thermometers and Platinum Tempera- He is currently a Professor with the Institute
ture Sensors, 2nd ed., document IEC 60751, 2008. of Metrology, Electronics, and Computer Science,
[2] W. Kester, “Bridge circuits,” in Practical Design Techniques for Sensor University of Zielona Góra. He was a Supervisor
Signal Conditioning. Norwood, MA, USA: Analog Devices, 1999, and a Team Leader in many research projects for the
pp. 2.1–2.18. industry (power engineering, telecommunication, information technology). His
current research interests include electrical measurements, methods, and the
[3] “Citeria for temperature sensor selection of T/C and RTD sensor types.
application of artificial intelligence in instruments and measurement systems.
The basics of temperature measurement using RTDs,” Acromag, Inc.,
Wixom, MI, USA, White Paper 8500-917-A11E000, Part 2/3, May 2011,
pp. 1–22.
[4] R. Radetić, M. Pavlov-Kagadejev, and N. Milivojević, “The analog
linearization of PT100 working characteristic,” Serbian J. Elect. Eng.,
vol. 12, no. 3, pp. 345–357, 2015. Mariusz Krajewski received the M.Sc. degree in
[5] B. Trump, “Analog linearization of resistance temperature detectors,” digital measurement systems from the Technical
Analog Appl. J., vol. 4Q, pp. 21–24, 2011. University of Zielona Góra, Zielona Góra, Poland, in
[6] M. Nadi, C. Margo, M. Kouider, J. Prado, and D. Kourtiche, “Embedded 2001, and the Ph.D. degree in electrical engineering
system design and implementation of standard auto-calibrated measure- from the Faculty of Electrical Engineering, Com-
ment chain,” Int. J. Smart Sens. Intell. Syst., vol. 1, no. 1, pp. 21–33, puter Science, and Telecommunications, University
2008. of Zielona Góra, in 2009.
[7] M. Looney, “RTD interfacing and linearization using an He is currently an Assistant Professor with the
ADuC706x microcontroller,” Analog Devices, Norwood, MA, USA, Institute of Metrology, Electronics, and Computer
Appl. Note AN 0970, 2009, pp. 1–16. Science, University of Zielona Góra. His current
[8] K. V. Santhosh and B. K. Roy, “An adaptive calibration circuit for research interests include issues related to high-
RTD using optimized ANN,” in Proc. 7th Int. Conf. Intell. Syst. precision voltage measurements, digital signal processing, and determination
Control (ISCO), Coimbatore, India, Jan. 2013, pp. 49–54. of measurement uncertainty.
[9] A. Ryan and W. Collin, 3-Wire RTD Acquisition System Accurately
Measures Temperature From −200 °C–850 °C, Texas Instrum., Dallas,
TX, USA, 2013, pp. 1–38.
[10] J. S. Coursey, K. C. Hasson, and G. H. Owen, “Systems and methods for
auto-calibration of resistive temperature sensors,” U.S. Patent 8 337 082,
Dec. 25, 2012. Sergiusz Sienkowski received the M.Sc. degrees
[11] RTD Embedded Technologies, Inc. (2014). 12-Bit Analog I/O dataMod- in industrial mathematics and industrial computer
ule, User’s Manual. [Online]. Available: http://www.rtd.com systems from the Technical University of Zielona
[12] G. C. M. Meijer and X. Li, “Smart sensor interface electronics,” in Proc. Góra, Zielona Góra, Poland, in 2001, and from
23rd Int. Conf. Microelectron., vol. 1, May 2002, pp. 67–74. the University of Zielona Góra, Zielona Góra, in
[13] F. Reverter, J. Jordana, M. Gasulla, and R. Pallàs-Areny, “Accuracy and 2003, and the Ph.D. degree in electrical engineering
resolution of direct resistive sensor-to-microcontroller interfaces,” Sens. from the Faculty of Electrical Engineering, Com-
Actuators A, Phys., vol. 121, no. 1, pp. 78–87, 2005. puter Science, and Telecommunications, University
[14] P. R. Nagarajan, B. George, and V. J. Kumar, “Improved single-element of Zielona Góra, in 2011.
resistive sensor-to-microcontroller interface,” IEEE Trans. Instrum. He is currently an Assistant Professor with the
Meas., vol. 66, no. 10, pp. 2736–2744, Oct. 2017. Institute of Metrology, Electronics, and Computer
[15] T. K. Maiti, “A novel lead-wire-resistance compensation technique using Science, University of Zielona Góra. His current research interests include
two-wire resistance temperature detector,” IEEE Sensors J., vol. 6, no. 6, signal processing, processing of measured data, measurement systems, and
pp. 1454–1458, Dec. 2006. evaluation of measurement uncertainty.