I will set up the experiment as shown in fig. 1
I will fix the temperature sensor from a multimeter to the thermistor with insulations tape to hold them close together. This will limit the temperature variation between the two. I will then pour a small amount of hot (80°C) water into a beaker and place the instrument into the water. I am using a small amount of water because it will cool faster so I don’t have to wait as long for the results, also I am not being exactly precise with the amount of water because I think it wont affect the overall results and if I add ice to the water to speed up getting to the low temperatures it will be changing the volume of water anyway.
I will take measurements of the resistance every 5°C starting at 75°C and dropping to 25°C. These are suitable temperatures because a CPU core will never get as low as room temperature, normally being at least 5°C above it, and once it reaches 75°C there is a high chance it will crash or even short out. It’s at this temperature that the warranty on the CPU becomes void.
I will repeat the experiment 3 times to produce an accurate average and to avoid the chance of anomalous readings affecting the results.
Because of the properties of semiconductors like the ones used in thermistors I expect to see the resistance rise as the temperature falls.
Results:
See table fig.2 and graph fig.3
From my results I can see as expected the rise in resistance as the temperature falls. This does not have a straight line trend but is in a curve.
I ended up testing the thermistor 4 times instead of the stated 3 because I noticed more varied results in the first 3 tests than expected and I wanted to do an extra test to get a better average.
Circuit design:
Having collected the values I want for the resistance of the thermistor at different temperatures I can now plan my circuit for the fan controller.
I will use a potential divider circuit with a variable resistor and the thermistor as R1 and R2. I will use this to turn on a relay when the temperature is high enough and the relay will switch on the fan.
Here is my component list:
Thermistor
Rheostat
Power supply x2
Cables
Relay
80mm case fan
Multimeter - to help set up the resistance of the rheostat.
The relay switches at 3v and the power supplies will mimic the power supply in a computer for this prototype. This means I have the choice of 2 input voltages: 5v and 12v (as is standard for an ATX Power supply). For the potential divider I will use 5v because 12v is un-necessary but I will power the fan with 12v because that is its rated voltage.
The relay also has the ability to work in two ways – to break the circuit when powered or to make it. For this circuit I will use it to break.
I am using a rheostat instead of a fixed resistor because it gives me the opportunity to adjust the temperature at which the fan will come on. This is useful for using the design on different processors and in different situations as the safe temperature for different CPUs varies.
I will set up my circuit as shown in fig.4
The thermistor would be placed as close to the core of the CPU as possible to get the most accurate readings. For this investigation I will be setting it up to turn the fan on at 55°C which is a slightly high CPU temp.
Setting up the rheostat:
I need to work out what resistance to set the rheostat at for the fan to come on at 55°C. Because the relay switches at 3v there needs to be 2v over the rheostat at the same time:
Input voltage 5v – 3v = 2v
R1 is the rheostat and R2 is the thermistor, the voltage across each is V1 and V2 respectively.
As with all potential dividers:
V1 / V2 = R1 / R2
So for my circuit it will look like this:
2 / 3 = R1 / 278 – this is the resistance for the thermistor at 55°C
278 / 3 = 92.66
92.66 x 2 = 185.32
Therefore the resistance of R1 has to be set to 185 for the relay to switch at 55°C
Constructing, testing and conclusion:
I built my circuit as shown in the plan and used the multimeter to test the accuracy of the input voltage and the resistance of the rheostat. I then tested the circuit.
Because I didn’t have access to a computer in which to test the setup I had to mimic the heat and make sure it turned the fan on. To do this I turned once again to hot water. I re-attached the electronic thermometer to the thermistor and placed it in a very small amount of hot water. I only used a small amount this time because I wanted to expose the wires as little as possible to the water as they now had more power running through them. The water was so shallow only the head of the thermistor was submerged.
The fan came on when the thermistor was in the water as the rely was not powered enough to switch, then once it cooled down and passed 54°C the relay got enough power and switched, stopping the fan. I repeated this 4 times and each time the relay switches at 54-56°C. I think this is an acceptable margin of error and the circuit would work well when applied to a computer.
Evaluation:
Overall I think my investigation went well and was quite accurate.
For my preliminary experiment I was going to use the heat from a bulb to test the thermistor because I was worried about the water reducing the resistance of the thermistor as the current could pass through the water instead. However I reconsidered this when I decided to use the low powered ohm meter instead of a powered circuit. Also, although the bulb would give me greater flexibility with changing the temperature, there would be a greater error between the readings of the thermometer and the thermistor. This is because although they would be close together, the air and glass is not as good a conductor as water, so the temperature would not be as evenly spread and there may be hot spots etc. Water on the other hand conducts the heat around it evenly so there is less chance of error.
The accuracy of this was proved in my final test. The relay was switched within a degree of the target so I can be sure that my readings were correct. For this reason I do not see a reason to improve my testing method.
There is a question of safety when using electrical devices in water, but because the input voltage was 5v, and the power that was reaching the water wasn’t that high there was practically no chance of anything going wrong because it takes at least 12v for water to conduct electricity.
If I had more time I would firstly like to test the system in a real PC to see if it still works as well and to check that the fan has the desired effect on the computer temperature. In a real life situation the relay could turn on multiple fans for extra cooling power. Also I would like to make a numbered dial for the rheostat so that it can be adjusted to switch the fan on at a desired temperature without having to set up the rheostat with a ohm meter.
My circuit could also be applied to other applications that involve cooling such as turning on fans to cool rooms etc. It could also be reversed and used to turn on a heater if something gets too cold.
Fig.2
Fig.3