Step Response of a Thermocouple

Author: Joshua Clark

Laboratory Partner: Kirk McKenzie

1. ABSTRACT

This paper describes the experimental determination of the time constant (𝜏) of a K-type

Thermocouple using an operational amplifier and a data acquisition device. The experimental apparatus

consisted of a beaker ¾ full of saturated water resting on a hot plate with a liquid in-glass thermometer

suspended above with the bottom of the thermometer submerged in the beaker of water. Utilizing an

operational amplifier and a voltage follower circuit, a thermocouple at ambient temperature was plunged

into a beaker of boiling water. At each plunge of the thermocouple, the time constant (𝜏) was calculated

describing the time required for the initial temperature of the thermocouple to change 63.2% of the way to

its final temperature. Of the three times this test was performed, the experimentally determined time

constants were 0.032, 0.054, 0.016 seconds. The average of the time constants is 𝜏 = 0.034 𝑠. This result

is consistent with stated values of K-type thermocouple time constants.

11. INTRODUCTION

A thermocouple is a temperature measurement device that utilizes two dissimilar metals joined

together at two points. Temperature measurement is attained using a thermocouple when a voltage is

developed between the two junctions in proportion to the temperature difference. This is also referred to as

the Seebeck effect, which describes the phenomenon in which a temperature different between two

dissimilar electrical conductors or semiconductors produces a voltage difference between the two
substances. Once this voltage is amplified and calibrated, it can be collected by a data acquisition device

and utilized to measure temperature.

In order to attain a temperature measurement using a thermocouple, a calibration is required. While

there are multiple ways to accurately calibrate a thermocouple, a simple method is presented here. By

recording the ambient and boiling water temperature using a liquid in-glass thermometer and comparing

these temperatures to the voltage reading of the thermocouple at ambient temperature as well as inside the

boiling water, a linear fit can be applied to the data points to generate a correlation. Applying this correlation

to the thermocouple output voltage results in a temperature reading in degrees Celsius.

By definition, a first order system has a delayed response to a change in input signal due to its storage

capacity. A thermocouple circuit can therefore be modeled as a first order system. Due to the limits of

today’s technology, an instantaneous output in reaction to an input is impossible to attain. Thus, a lag or

delay is created within the measurement. Fundamentally, the time constant represents the required time for

a system response to decay or stabilize at an equilibrium point given the initial value.

This paper describes the calculating of the time constant 𝜏 of a K-type thermocouple and is organized

as follows. The methods section begins with describing the experimental apparatus followed by the

determination of the thermocouples time constant. This is followed by the process in conducting the

experiment. The next section describes the results of the experiment using the plotted step response

of all three experimental tests. Finally, the discussion/conclusion section explains the correlation

between the experimental data and the theoretical curve by highlighting similarities differences

between the two.

EXPERIMENTAL APARATUS
The experimental apparatus consists of boiling water inside a beaker placed on a hot plate with a

liquid in-glass thermometer suspended on a vertical stand above the boiling water.

Figure 1. . A beaker filled with saturated water rests on a hot plate with a liquid in-glass thermometer suspended on a vertical
stand above. The voltage follower circuit and the data acquisition device are shown to the right of the beaker.

A K-type thermocouple was used in conjunction with an AD595AQ Monolithic Thermocouple

Amplifier with Cold Junction Compensation which it utilized to produce a high level voltage output

directly from the thermocouple. Since the AD595AQ has relatively low impedance, an LM358
operational amplifier is used to buffer the output from the AD595AQ to the NI-6001 USB data

acquisition device. A Simulink program was created inside MATLAB to read the input voltage from the

voltage follower circuit and convert these voltages to temperature readings in degrees Celsius. The

program was designed with a sample rate of 1000 Hz and utilizes a function block which applies our

calibration equation to the output voltage in order to get a reading in degrees Celsius.

Figure 2. Final thermocouple electrical circuit which shows the K-type thermocouple wired into the breadboard where the
AD595AQ Thermocouple Instrumentation Amplifier amplifies the voltage output, the LM350 Op-Amp buffers the output, the NI-
6001 USB DAQ records the output voltage, and the Simulink program plots the recorded data in MATLAB.
 EXPERIMENTAL DETERMINATION OF THE
THERMOCOUPLE TIME CONSTANT

To calculate the time constant of each experimental test, a temperature vs. time plot from before the

thermocouple is plunged into the boiling water to after was needed. The time constant is found by

calculating the time required for the temperature reading of the thermocouple to reach 63.2% of its final

steady-state value. The equation

𝜏 = 0.632(𝑌𝑠𝑠 − 𝑌0 ) (1)

is used to calculate 𝜏 where 𝑌𝑠𝑠 is the final temperature steady state value and 𝑌0 is the initial recorded

value. Plot tools in MATLAB allowed us to subtract the time at 𝑌0 right before the step response is

introduced from the newly found 𝑌𝑠𝑠 value to produce the time constant 𝜏.

EXPERIMENTAL PROTOCOL

First, the ambient air temperature must be measured using the liquid in-glass thermometer which we

recorded to be 27.0°C. The thermocouple output voltage is also measured at ambient temperature which

we recorded to be 0.24251 volts. The same process is then repeated with both the liquid in-glass

thermometer and the thermocouple. The LIGT inside the boiling water gave us a reading of 99°C and the

thermocouple had an output voltage inside the boiling water of 1.00297 volts. These recorded values were

then utilized to create a 2-point calibration in MATLAB and to generate a linear fit equation that modeled

the system. This equation was used in our Simulink function block so that our MATLAB program would

accurately output temperature in degrees Celsius from the thermocouple output voltage. Once this was

complete, the LIGT was removed for the remainder of the experiment.

The tests were conducted by starting with the thermocouple at ambient air temperature. Once the

Simulink program was started, the thermocouple was immediately plunged into the boiling water and held
there while the Simulink program recorded the data for the few remaining seconds. Once the program was

finished recording the data, the resulting plot appeared as an exponential curve. This experiment was

completed a total of three times and each set of data was compiled into a single plot where all three data

runs could be compared. The time constant was calculated from each individual run and then averaged to

get the average time constant for the thermocouple.

IV. RESULTS

As stated above, the calibration process generates a linear equation that represents the first order

system that is the thermocouple. The generated linear equation for the system is

𝑦 = 97.5688𝑥 + 1.1411 (2)

where x is the output voltage from the thermocouple circuit and 𝑦 is the temperature reading in degrees

Celsius. The plot below shows the dynamic response of all three runs. Figure 3 displays the average time

constant that was calculated from all three test runs.

From all three tests conducted in this experiment, the time constant proved consistent

with the theoretical time constants expected from the thermocouple used. The time constant

generated from this experiment was relatively low. Therefore, this device is a good choice when

one is attempting to record sudden changes in a first order system.

Slight variations in our experimental data could likely be contributed due to error in our

experiment. Significant errors could take place in the process of plunging the thermocouple into

the boiling water given the layer of steam the thermocouple must pass through in order to reach

the boiling water. This would increase the initial output reading of the thermocouple before it is

plunged into the water which would alter our calculated time constant. Error could also arise in

the thermocouple by being held too close to the hot plate effectively increasing the initial voltage

output of the thermocouple which would result in a quicker step response. Longer waiting

periods could have been utilized to ensure the thermocouple had returned to ambient temperature

before repeating the experiment. Other real-world variables that do not occur in mathematical

simulation could also be a factor for error in our calculations.

VI. REFERENCES

Figliola, R. S., and Donald E. Beasley. Theory and Design for Mechanical Measurements. Wiley, 2019.