The resistance of the thermistor depends on the temperature. If the temperature is increased, the resistance of the thermistor would decrease, thus decreasing the amount of potential difference across the thermistor. To do this I needed to find a way of regulating the temperature. Heating water and putting the thermistor into this was out of the question as the water could interfere with the circuit, so I chose ethanol Therefore if the potential difference across the thermistor is measured at different temperatures, it would become possible to measure the temperature using the potential difference across the thermistor.
INITIAL EXPERIMENTS AND CALCULATIONS
The first step I took was to assess if the thermistor I was supplied with did decrease the potential difference split across it as the temperature increases. I did this by placing the thermistor next to a heat source. Once I established that my thermistor worked, I measured the potential difference (3.56 volts) and the current in order to work out the resistance of the thermistor at room temperature. I then found a fixed resistor of about the same resistance (2.2kΩ) to try to split the potential difference across both of the components roughly equally at room temperature. I then checked this by putting a voltmeter across the thermistor, which read 1.76 volts.
After doing the calculations above, I then followed through the plan for my experiment to see if it worked. I found some problems in my experiment, which needed to be eradicated to improve the reliability of the results. My findings are explained in the conflicts section below.
CONFLICTS
One problem that I did encounter occurred during the hating and cooling of the ethanol. The thermometer reading the temperature of the ethanol was resting on the bottom of the small beaker. The temperature at the bottom of the beaker seemed to be different than just under the surface, where my thermistor was situated. This difference was further increased under cooling, where the bottom of the beaker measured zero degrees but when the thermometer was bought up to just under the surface it was actually ten degrees Celsius. Because of this I clamped the thermometer so the bulb was just below the surface of the ethanol, where the thermistor was in the ethanol. This ensured I measured the temperature where the thermistor was in the ethanol.
The other problem I encountered was that the voltmeter seemed slow to respond to any change in temperature. This wasn’t so much of a problem at higher temperatures as it took a long time for the ethanol to heat up anyway, but when the ethanol was first put into the hot water the temperature increased at a rapid rate, and a slow response time was going to be a problem. My solution to this was to first start with mildly warm water to heat the ethanol slowly, and adding more boiling water as I needed it, but only just enough to keep the temperature rising slowly. This ensured the voltmeter had time to respond to the change in temperature and made sure my results were much more accurate.
RESULTS
Experiment 1
Experiment 2
Experiment 3
Average Results
A GRAPH TO SHOW THE RELATIONSHIP BETWEEN TEMPERATURE AND ONE DIVIDED BY THE POTENTIAL DIFFERENCE ACROSS THE THERMISTOR
CONCLUSION
From the results I gained, I was first able to construct average values of the potential differences for each temperature I measured. This then enabled me to draw a graph of temperature against potential difference (supplied with my report). This graph is my calibration graph, and I can use this to find the temperature of the surroundings of the thermistor for any given reading on the voltmeter.
One notable problem with the calibration graph is that the relationship between the temperature and the potential difference across the thermistor is non-linear. This is a problem, as a movement of 0.1 volts on the voltmeter is not going to mean the same temperature change for whatever temperature the thermistor is measuring. This means my sensor has different resolutions as the temperature increases, meaning the thermistor is more sensitive at some temperatures than at others.
An example of this is between 10°C and 15°C. The difference between the potential differences across the thermistor at these temperatures is 0.6 volts. As the voltmeter used for my sensor can measure up to a hundredth of a volt, the resolution between these two temperatures is:
Resolution = 5°C ≈ 0.08°C
60
This means the smallest change the voltmeter can detect between these two temperatures is 0.08°C. Comparing this to the resolution between 55°C and 60°C, which is 1.25°C, it is clear to see that my sensor is much more sensitive at lower temperatures.
To try to overcome this lack of sensitivity at lower temperatures, I decided to increase the overall potential difference across the circuit to around 5 volts (5.02 was the actual value). My plan was to try to increase the number of volts across the thermistor so the resolution of the sensor would increase overall, which would be especially useful at the higher temperatures. These were my results:
Using this table, the resolution between 55°C and 60°C has actually changed very little. The smallest change my sensor can read is still only 1°C between these temperatures. However, the sensitivity at lower temperatures has increased slightly, allowing my sensor to measure a change of 0.07°C. In theory, increasing the voltage further would have increased the sensitivity of my sensor, but this would not be possible for long with the thermistor I was supplied.
The other way I investigated increasing the sensitivity at higher temperatures was changing the resistance of my fixed resistor to ensure the thermistor had a larger share of the potential difference at room temperature. However, this did not work to my plan either. I tried to set up a 3:1 ratio by replacing my 2.2kΩ fixed resistor with a 500Ω resistor. The problem with this was that although more potential difference was across my thermistor to start with, there was also a considerable amount more potential difference across the thermistor at the higher temperatures. This meant between the temperatures 10°C to 60°C I was a restricted to 1.5 volts of change (2.92 volts at 10°C, 1.4 volts at 60°C), where as in my original experiment I had a whole 2 volts.
One other problem with my calibration graph is that it is hard to read of a temperature below 10°C due to the curvature of the graph. To try to combat this, I plotted the graph of temperature against the potential difference to the power of minus one (supplied with the report). From the graph, you can see that there is some degree of proportionality. This means that any temperatures not covered in my calibration curve can still be calculated by simply putting the relative potential difference to the power of minus one and reading of the temperature.
The other important meaning of this graph is that the temperature is found to be inversely proportional to the potential difference across the thermistor e.g.
Temperature α 1
V
This would allow me to calculate an equation to put any given output into in order to work out the temperature directly if necessary. This equation could replace the need for any sort of calibration graph, but although this would be preferable, I am unsure whether this equation would be exact as the graph is not perfectly linear.
One factor that my sensor does have a slight problem with is response time. The voltmeter display doesn’t react quickly to quick changes in temperature, so my sensor would have to be used to measure a slow rise in temperature. This did lead to a small inaccuracy in my results, as the difference between the potential differences across the thermistor between 20° and 25°C is actually bigger than the gap between 15°C and 20°C. This is because this is the area in which the temperature of the ethanol increased at the fastest rate and the voltmeter reading struggled to keep up with the temperature rise, despite my attempts to heat the ethanol slowly. However, the calibration graph allowed me to discount this small area by leaving this point out whilst drawing the curve, but it is still clear that my sensor would be most effective at measuring small temperature increases over time.
The good qualities my sensor has involved an excellent sensitivity at lower temperatures, and could measure very small changes. The calibration curve has been constructed using three experiments averaged out, so any noise or random fluctuations which may have affected my sensor have been effectively eliminated, so the temperature could be calculated quite accurately.
Overall, my sensor is very good. Only the presence of a voltmeter with a much quicker response time could improve its performance at lower temperatures, at which my sensor is very sensitive. It could be used at higher temperatures but would only be used to measure large changes such as 1°C.