Using Phase Change Properties of Pure Water

to Determine the Time Constant of a K-type Thermocouple

Jay Huber, Aliya Kusumo, Rachel Williams, Matthew Nelson
Dept. of Mechanical Engineering, University of California at Berkeley
(Date: November 24, 2020)

Abstract Our lab positions the thermocouple in pools

of ice water and boiling water to measure
Thermocouples are used in a wide variety of significant sensor data points at different phases
cases as sensors for measuring temperature. In of pure water. The controlled parameters include
order to properly use data from this sensor, it is the volume of water subjected to temperature
important to understand the sensitivity of the change, phase change temperatures, sheathed
thermocouple. This lab focused on taking thermocouple properties and dimensions, and
advantage of the solid to liquid and liquid to DAQ circuitry. An AD8495 voltage amplifier
vapor phase processes of pure water to use as and ESP32 microcontroller were incorporated to
known calibration points for the thermocouple. magnify the mV signal and collect temperature
Data was amplified and recorded as the values during the experiment.
thermocouple was moved from its steady state The model for the thermocouple depends on
position in ice to a pot of boiling water until it its capacity to convert internal temperature
again reached a steady state. From this, the time changes into voltages. Each thermocouple uses a
constant was determined and compared against linear correlation between voltage and
its theoretical calculation. The measured time temperature difference to convert the signals
constant values were consistent across the four received by the thermocouple into usable
data sets and fell within the expected theoretical voltage data, known as the Seebeck Effect
range. Equation (Eq 1). ΔV is the voltage difference
between the chromel & alumel leads of the
1. Introduction thermocouple, α is the Seebeck coefficient that
Accurate temperature measurements are an scales the correlation depending on the type of
invaluable resource in both commercial and thermocouple (a K Type thermocouple has α =
residential applications. Whether it comes to 41[µV/℃]), TM is the measured temperature (in
monitoring the common household stove or the this case a boiling ~1.7L pot of water at 100℃
sterilization of biotechnical machinery, to simulate a temperature step response), and
thermocouples of varying degrees are TRef is the reference temperature of the
incorporated to provide quality measurement environment the system is in (we used an ice
accuracy in high thermal load environments. In bath as our reference temperature to begin the
this lab, we aim to characterize the associated sensor at ~0℃).
time constant of the 2M K Type Temperature
∆𝑉 = α(𝑇𝑀 − 𝑇𝑟𝑒𝑓) (1)
Sensing Line TES1310 TM902C Thermocouple
and its corresponding uncertainty. Using the The temperature measurements taken from
time constants, our objective is to derive an the thermocouple were analyzed to determine
analytical model for our thermocouple and the time constants and steady state temperatures
compare measured data to theoretical time of the experiments. These measurements were
constants for K Type thermocouples, which will then used to verify the validity of the analytical
allow us to tabulate corresponding uncertainties model predicted for the K Type sensor.
and gauge the quality of our sensor.

1
2. Theory can become saturated vapor, and steam will rise
off of the pot.
2.1 Heat Transfer Therefore, when a substance is observed
Heat transfer occurs through three modes, changing phase at a known pressure, the exact
which are conduction, convection, and radiation. temperature can be easily determined. At
Conduction is heat transfer through a material as atmospheric pressure (101.325 kPa), Tsat for pure
energy is transferred from more energetic water is known to be 99.97 °C [3]. Likewise, at
particles to less energetic ones [1]. Convection is 101.325 kPa, water solidifies at 0.01 °C [3].
the transfer of heat by macroscopic motion of
fluid [1]. Radiation is energy emitted by some 2.3 Thermocouples
matter due to its temperature [1]. All three heat Design and Function
transfer mechanisms need to be considered to do Thermocouples are devices that measure
a comprehensive analysis, although one or two temperature differences. They are made of two
mechanisms dominate in most real-world heat wires of dissimilar metals, which are welded
transfer cases. These three modes can be together at one end (referred to as the hot
expressed with Eqs (2) to (5), as follows: junction). The other end of the thermocouple is
referred to as the cold junction (see Figure 1 for
𝑄˙𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 = 𝑘
𝐴
𝐿
(𝑇 − 𝑇∞) (2) a thermocouple circuit diagram). When there is a
𝑄˙𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛 = ℎ𝐴𝑠(𝑇 − 𝑇∞) (3) difference in temperature between the hot and
cold junctions, a very tiny electromotive force is
𝑄˙𝑟𝑎𝑑𝑖𝑎𝑡𝑖𝑜𝑛 = εσ(π𝐷𝐿)(𝑇 − 𝑇∞)
4 4
(4) generated [4]. Therefore, the hot junction
between the wires is used as the testing tip. The
Σ 𝑄˙ = 𝑄˙𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 + 𝑄˙𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛 + 𝑄˙𝑟𝑎𝑑𝑖𝑎𝑡𝑖𝑜𝑛 millivolt-order voltage produced by a
(5) temperature differential is amplified by an
attached corresponding ANSI-type amplifier.
Thermocouples are impacted primarily by The thermocouple’s ANSI-type and
convection. Conduction and radiation heat losses amplified voltage output can be looked up in a
from thermocouples are not significant because thermocouple reference table to find the
the wire is insulated and does not reach a high corresponding temperature. Many types of
temperature (see Eq. 4), and conduction is not a thermocouples exist and are classified based on
significant concern because of its reliance on the their metal composition, operating temperature
cross sectional area, which is small here (see Eq. range, and durability. These experiments use a
2). The primary focus of these experiments is the K-type ANSI-code thermocouple, which must
time constant, which is different depending on specifically contain a nickel-chromium or
the type of fluid measured (gas or liquid), due to nickel-aluminum conductor and provide a
the convection coefficient (see Eq. 3) [2]. standardized error of ± 2.2 °C (or ± 0.75 %,
whichever is greater) [2]. Based on ANSI
2.2 Phase Changes standards for K-type thermocouples, expected
Pure substances have three distinct phases voltage output values for 0 and 100 °C are 0.000
(solid, liquid, and gas), which are classified at and 4.096 mV, respectively [5]. Amplifier
specific states, each defined by two independent, manufacturers use these specifications in
intensive properties. At any given pressure, there creation of their corresponding-typed devices.
is a saturation temperature at which the
substance will change state [3]. Two phases can
coexist at equilibrium at a saturated state [3].
When a body is evenly rising in temperature,
such as a pot of water boiling on a stove, the
whole volume will rise to the saturation
temperature. At this point the saturated water

2
Figure 1. Thermocouple circuit diagram
Figure 2. Temperature progression over time of
Thermocouple and Amplifier thermocouple
The amplifier used for these procedures is
an Adafruit 8495 Analog Output K-Type The time constant of a thermocouple is
Thermocouple Amplifier. The device influenced by the metal alloy used (determined
documentation gives an equation to interpret by the type of thermocouple), the dimensions of
readings from the amplifier (output voltage to the wire, if the lead is exposed, and type of
input temperature conversion), listed as Eq (6). media being measured (liquid or gas). Our
A voltage range of 3.3 V allows a temperature thermocouple has an unsheathed lead, indicating
range of -250°C to +410°C input to the a fast response time, Tr, of 0.6 seconds, listed by
thermocouple. This range is well suited for the the manufacturer. The theoretical τtheory = ⅕Tr =
temperatures in the procedures here, which 0.12s [6]. τtheory is achieved at 63.2% as shown in
range from 0°C to 100°C. Figure 2. Our experimental data will be
measured against this τ to gauge the sensitivity
𝑉𝑜𝑢𝑡 − 1.25
𝑇= (6) and quality of our sensors.
0.005

3. Methods
Time Constants
One parameter for describing thermocouples 3.1 Equipment
is the time constant (τ), defined as the time 1. Stovetop
required for the sensor to reach 63.2% after an 2. Oven mitt
instantaneous temperature change. The time to 3. Pot full of water
reach the 63.2% of total temperature change will 4. Bowl of ice
be how the time constant is experimentally 5. Smartsails 2M K Type Temperature
found. This value can also be measured using Sensing Line (TES1310 TM902C)
Eqs (7) - (8). The progression of these equations 6. Analog Output K-Type Thermocouple
demonstrates the relationship between τ and Amplifier (AD8495 Breakout)
temperature with respect to time as certain 7. Microcontroller
values are reached. 8. Breadboard

3.2 Setup
−𝑡

𝑇 = 𝑇0 − (𝑇𝑓 − 𝑇0)(𝑒 τ ) (7)

Mechanical Setup
1
𝑇𝑓 = 2
(𝑇𝑓 + 𝑇0) as T→ 𝑋(𝑇0) (8) The mechanical setup in this lab was to
…where X is the mean of T0 create solid-to-liquid and liquid-to-vapor phase
transformations of pure water in order to
establish calibration points for our thermocouple
(see Figures 2 and 3). There is no setup
necessary for the solid-to-liquid testing, but the
liquid-to-vapor testing requires boiling a pot of

3
water on the stove. Once the water is observed
to be boiling and producing steam, it is a
saturated liquid and is ready for testing.
3.3 Procedures
Safety Remarks
During testing, keep the electrical equipment
away from the water and steam. The
thermocouple cord is 2 meters, so this should not
be a problem. When putting your hand over the
boiling water or touching the pot, wear a
cooking mitt to prevent burns.
Procedure
Fill a pot of water, put it on the stove, and
start the stove. Once you notice that the water
Figure 3. Mechanical lab, solid-to-liquid phase starts to boil, you can start the first cold testing.
transition Start the data collection, then put the
thermocouple’s test prod into a bowl of ice
water (focus on ice), and hold it there for 15
seconds. Water is used with the ice to get clean
data (aka if there is significant noise). Heat
conductivity of liquid water is better than ice, so
ice water will work better here than solid ice
alone.
After the 15 seconds of ice water exposure is
up, using your mitt-covered hand lower the tip
of the thermocouple into the boiling water.
Figure 4. Mechanical lab setup, liquid-to-vapor Move the thermocouple probe from the ice to
phase transition the boiling water as quickly and as safely
possible. We desire as quick a temperature
Electrical Setup change as possible, because this will keep the
The thermocouple and amplifier used measured values as close as possible to the
here are a Smartsails K-type thermocouple and theoretically calculated value. It should be noted
an Adafruit 8495 thermocouple amplifier. The that moving through the steam may cause noise
𝑚𝑉
amplifier is designed to produce a 5 °𝐶 output in the data. Wait 15 seconds with the
thermocouple in the water, then remove it and
from a K-type thermocouple signal, with an stop data collection. After these steps are
accuracy of ±1°C around room temperature, finished, repeat once more (for a total of eight
±2°C for −25°C to +400°C. trials between the four thermocouples).

4. Results and Discussion

From the raw sensor data sets, we plotted
the step change temperature outputs resulting
from a change in voltage due to dramatic heating
and cooling of the thermocouple leads. The
graphs in Figures 6 - 9 (see Appendix) showcase
the linearity described in the Seebeck Equation.
We ran multiple tests with 4 different
thermocouples and tabulated the initial and final
temperature conditions of the boiling pot and the
Figure 5. Fritzing diagram of the testing circuit
ice bath. We recalibrated the data to homogenize
their stagnation point at 100℃, cropping the
range between (0, 100)℃. We then found

4
respective uncertainties of initial and final all four of the thermocouples tested, thereby
temperature values within a 95% confidence validating the reliability of this setup to find the
interval, shown in Table 1. Using Equation 9 we time constant of the thermocouple. The
calculated the uncertainties for each test. The overlapping ranges also support our assumption
uncertainty values were defined as bias error, to neglect radiation and conduction.
implying a systemic imperfection in the sensor The error within our t value, and
calibration that was consistent throughout our consequently 𝜏, can be attributed to error in our
repeated measurements. The error attributed an DAQ setup, varying transition time between the
average fluctuation of ±2.2℃ in our experiment, ice bath and boiling pot, and the theoretical
which deviates our values from their theoretical models reliance on an instantaneous temperature
magnitudes. change (which cannot be practically achieved).
2 2 Furthermore, the combined data for transient
𝑈 = 𝑈 𝑆
+𝑈 𝑅
(9) temperature could be improved with a higher
where sampling frequency that followed the parameters
US = bias error from thermocouple = 2.2 ℃ of the Nyquist Sampling Theorem [7]. Our DAQ
UR = random error = 2*stdev system did not have the bandwidth to accurately
collect data at higher frequencies, forcing us to
Table 1. Initial (ice) and Final (boiling water) contend with 20 Hz and a time tracker delay of
temperatures at 95% confidence 50ms. This, unfortunately, only provided us with
Test Initial Temp [℃] Final Temp [℃] 7 data points from the beginning of the transient
1 1.5 ± 2.9 100.0 ± 3.4 stage to Tf , for each experiment.
In order to improve our data in repeated
2 0.8 ± 2.7 100.0 ± 3.0
experiments, we would need to improve the
3 1.0 ± 3.2 100.0 ± 2.9 sampling rates of our ADC and increase the
4 1.1 ± 2.7 100.0 ± 3.1 resolution frequency to mitigate any associated
noise.
The transient stage indicates the time taken
to transfer the thermocouple from the ice bath to Table 2. Final temperature of half-life and time
the boiling pot. Because the temperature change constants found at 95% confidence
occurs so quickly, the time constant was found Test T = 63.2% (Tf - T0) [℃] 𝜏 [s]
with respect to the time required to reach 63.2% 1 63.8 ± 2.2 0.17 ± 0.03
of the total temperature change.
2 63.5 ± 2.1 0.13 ± 0.03
We cropped the temperature and time values
within an upper and lower transient temperature 3 63.6 ± 2.3 0.19 ± 0.03
range, then found the time it took to reach 63.2% 4 63.6 ± 2.0 0.16 ± 0.03
of the total temperature change to find the time
constants of our data, seen in Table 2. There was Acknowledgements
variation in output due to the process of We wish to acknowledge the ME103 Fall
calculating the time measurement, t, to reach 2020 teaching team for providing the sample
63.2% of the final temperature. We linearly template and contents which could be
interpolated t using values higher and lower than reproduced with permission.
our 63.2% of temperature change TM, where TM
= (TF - T0)*.632 + T0. Although this Table 3. Team member contributions to the lab
interpolation does not directly align with the
Name of author Contributions
exponential behavior of the transient
temperature range, it remains effective due to the Huber, Jay Theory, Methods
small time difference between T0 and Tf.
The theoretical time constant was Kusumo, Aliya Results and Discussion
determined to be 0.12 ± 0.03 s in the Theory Nelson, Matthew Intro, Results and Discussion
section. The range for this time constant
Williams, Rachel Abstract
overlaps with the ranges of time constants from

5
References 7. Keim, Robert. “The Nyquist–Shannon
Theorem: Understanding Sampled Systems -
1. Incropera FP, DeWitt DP, Bergman TL, Technical Articles.” All About Circuits,
Lavine AS. Fundamentals of Heat and Mass 2020,
Transfer. 6th ed. Hoboken: John Wiley & www.allaboutcircuits.com/technical-articles/
Sons; 2007. nyquist-shannon-theorem-understanding-sa
2. REOTEMP Instrument Corporation. Type K mpled-systems/.
Thermocouple [Internet]. San Diego, CA;
2011 [cited 2020 Nov 13]. Available from:
https://www.thermocoupleinfo.com/type-k-t
hermocouple.htm
3. Cengel YA, Boles MA. Thermodynamics:
An Engineering Approach. 8th ed. New
York: McGraw-Hill; 2015.
4. Pyromation Incorporated. Thermocouple
Theory [Internet]. Fort Wayne, IN; 2009
[cited 2020 Nov 13]. Available from:
https://www.pyromation.com/Downloads/D
oc/Training_TC_Theory.pdf
5. Omega Engineering Incorporated. Type K
Thermocouples [Internet]. Norwalk, CT;
2019 [cited 2020 Nov 13]. Available from:
https://www.omega.com/en-us/resources/k-t
ype-thermocouples
6. Barani, Jan. “Difference between Sensor
Response Time and Sensor Time Constant τ
(Tau) 63.2%.” BARANI DESIGN
Technologies, BARANI DESIGN
Technologies, 7 May 2019,
www.baranidesign.com/faq-articles/2019/5/
6/difference-between-sensor-response-time-
and-sensor-time-constant-tau.

8.

6
Appendix
Appendix A: Plots, figures, and tables

Figure 6. Voltage change over time for thermocouple 1

Figure 7. Voltage change over time for thermocouple 2

7
 Figure 8. Voltage change over time for thermocouple 3

Figure 9. Voltage change over time for thermocouple 4

8
Appendix B: MATLAB Codes