  17 April 2010
Author: Giorgos Lazaridis
Experimenting with Thermocouples

I am preparing a very interesting project with a soldering iron. For this project, i will need to measure the temperature of the solder tip. I could use a thermistor for this, yet, there will be a small problem... The thermistor will be tosted! Therefore, i will use something that can handle the soldering iron temperature. This "something" is called "a thermocouple". I have prepare a complete theory for "How Thermocouples Work". I strongly suggest you read this theory before, if you do not know how they work.

Before i start building my project, i had to run some experiments with thermocouples, as i was not very familiar with the subject. I got myself (thanks Kourt!) a pair of K-type thermocouple wire for this reason. The reason for these experiments is first of all to get the beginner's experience with thermocouples, and also to find the way that i will measure the temperature using a uController and the thermocouple junction.

Before i begin...

I have the table of the K-type thermocouple voltage to temperature, measured in mili-volts. You may need to use it yourself:

K-Type thermocouple voltage to temperature table

The first thermocouple experiment: Do they really work?

It may seem funny, but i always have this question when i first get in contact with new staff for me. I name this "the joy of discovering". This is how the first experiment goes: I connect a thermocouple pair to copper wires, and then to a multimeter with a 200 mV scale. Then i heat up the junction with a lighter to see if i get some readings:

The second thermocouple experiment: Does similar metals act like dissimilar metals? (Proving the first law of thermocouples)

According to the theory of thermocouples, this is out of the question. Yet, i had to see with my very own eyes:

The third thermocouple experiment: Proving the second law of thermocouples

In this experiment, i introduce a third dissimilar metal to the thermocouple pair like an extension. According to the theory of thermocouples, if the third metal has the same temperature across it's length, it will not affect the measurement of the thermocouple. So, i extended one cable of the thermocouple with a copper cable to see if the measurement is changed. Then, i changed the temperature of the new wire to see what happens:

The fourth thermocouple experiment: What happens if the temperature of the thermocouple to copper junction is changed?

This is extensively discussed in the theory of thermocouples. A temperature change in the thermocouple-to-copper junction will alter the readings, that is why all thermocouple pairs needs a hardware or software temperature compensation at this point. Look what happens:

The fifth thermocouple experiment: Measuring the boiling water and something... hotter!!!

The distiled water boils at 100oC. In this experiment, i measure the temperature of the boiling water with the thermocouple, and yet i get the result way below the 100oC. It proves that the measurement is absolutely depended from the thermocouple to copper junction temperature. Then, i have some more fun with something much hotter!

Another way to calculate the temperature from the voltage

In the theory of thermocouples, i explained how can someone solve a polynomial to calculate the temperature from the Seebeck voltage. Yet, this polynomial requires a lot of multiplications. For a PIC, this means a lot of time. Let's see something interesting. I made with open office spreadsheet a graph. The x axis had the voltage and the y axis had the temperatures from 0 to 1370oC, with a 10oC step. Surprisingly, this is what i got:  The open office spreadsheet to calculate the slope and offset for the line

In a first glance, this is a straight line. Well, actually it is not. The next step was to find out if this line could be useful to me. Using the open office spreadsheet again, i applied the least squares method to calculate the line. You can use my "least squares calculator". From there, i got the slope (a) and the offset (b) of the above line. Using these numbers and the line equation (line equation solver), i can calculate any Y (temperature) given an X (voltage).

What i actually did was this. For temperatures from 0 to 1300oC with a 10oC step, i calculated the temperature from the voltage using both the polynomial equation and the line equation. For each result, i calculated also the percentage error. The following screenshot shows this spreadsheet. I have mark with blue box the values that i am mostly interested at, for my coming project: The column with the red values, is the column with the error from the polynomial. The column with the green values has the error from the line equation. For the first 20 degrees, the error is unacceptable. As the temperature rises though, the error is decreased to very satisfying values. To tell you the truth, i calculated the slope and offset of the line for temperatures from 100 to 350oC, and that is why i get so good approach. For a wider range, i would not have these results.

And because a picture is like 1000 words, look at the following graph: The yellow line is the error from the line equation and the red is the error from the polynomial approach. Both error values are displayed in %. The temperature range (x axis) is form 0 to 1300oC. I think i will prefer this simply line equation (Y=aX + b)...

 Name  Email (shall not be published)  Website Notify me of new posts via emailWrite your comments below:BEFORE you post a comment:You are welcome to comment for corrections and suggestions on this page. But if you have questions please use the forum instead to post it. Thank you.  HOT in heaven!  NEW in heaven! New Theory: AC electric motor working principle