# A Crude Thermometer

1. Feb 15, 2005

### richnfg

Speaking to my physics teacher the other day, he said that you can use a thermistor as a crude thermometer...He didn't really explain it very well and I just got completely confused. :grumpy:

Can anyone explain it to me? Thanks!

2. Feb 15, 2005

### Gokul43201

Staff Emeritus
Why crude ?

Thermistors are used to build excellent thermometers. A thermistor is simply a resistor whose resistivity is a nearly linear function of temperature over a significant range. You also have NTC thermistors (usually made of nickel, cobalt or manganese oxides) which have a Negative Temperature Coefficient. Thermometers built from thermistors may have more than one thermistor in them (containing both NTC and PTC thermistors), which enables them to have a greater linear range with high sensitivity. I think the common material used for making PTC thermistors is still BaTiO3.

3. Feb 15, 2005

### richnfg

My teacher just kept saying that it would be a crude thermometer, I guess its because we have really cheap thermistors! How do you actually use a thermistor as a thermometer and get readings and stuff?

Thanks

EDIT: btw, I knew about the NTC and PTC thermistors :) We only have NTC at our school.

Last edited: Feb 15, 2005
4. Feb 15, 2005

### pervect

Staff Emeritus
You use a thermistor as a thermometer by measuring its resistance. You measure it's resistance by (for instance) forcing a known current to flow through the thermistor, and measuring the voltage across it.

5. Feb 15, 2005

### richnfg

so you're measuring the temperature in terms of resistance?

6. Feb 15, 2005

### Gokul43201

Staff Emeritus
Yes, since the resistance is a function of the temperature, you can determine the temperature by measuring the resistance and looking up a calibration chart to tell you the temperature at which it would have that resistance.

Usually however, you would buy a module that included an ADC, a microprocessor and a display. So, the module measures the voltage across the thermistor (as pervect explained above); converts this into a digital signal, and calculates the temperature (from an equation known as the Steinhart-Hart formula) which it then displays.

With a single NTC Thermistor, you can get accuracies of about +/-0.1C as opposed to a typical junction sensor, which rarely does much better than +/- 1C. So, that way, they are hardly crude. Often the greatest source of error may be self-heating effects. You have to put a current through the thermistor to measure the voltage. This current causes joule heating and hence must be kept to a minimum, lest the self-heat distort the measurement.

I think there are (relatively new?) devices called Composite Thermistors (which use both NTC and PTC thermistors?) which have even better performance. Standard temperature sensors, however, are usually just NTC.

7. Feb 16, 2005

### richnfg

Ok, that sounds all good. I guess I will recieve a calibration chart with the thermistor when the school buys them (well, at least I hope I do!).

I understand all of this, its seems pretty simple (except for that formula, but I guess that might be above my level judging by what it is when I looked for it on google :rofl:).

For my physics project I am testing the linearity of a thermistor, my teacher said that could relate to using it as a thermometer (you would need the linearity in there somewhere)...He tried explaining but it went over my head (hes not good at explaning ideas at all! ).

Thanks for helping so much, I see why you got the helper award!

8. Feb 16, 2005

### Integral

Staff Emeritus
You will not need any fancy electronics for your experiment. A battery, a resistor and a millivoltmeter should do the trick. You will want to create your own calibration table by recording the voltage across your resistor at 2 known temperatures (Ice/water slurry for 0C, boiling water for 100C) then assume the linearity. Linearity means that a change in temperature causes a proportional change in voltage.

Suppose that at 0C you read 1V and at 100C you read 6V if you assume that the change is linear it means that you have $\frac {100C} {5V} = \frac {20C} {1V}$. Knowing this if you read a voltage of 2V you would read the temperature as 40C.

These of course are all made up numbers to demonstrate the concept of linearity, they will not correspond to actual measurements.

EDIT:
OOPS! For the correct conversion of the voltage to temperature you must first subtract your 0 voltage point so your temperature will be found by

$$T_r = (V_r - V_0) K$$

Where $$K= \frac { \Delta T} {\Delta V}$$
and
$$\Delta T = T_{max} - T_{min}$$ or Change in Temperature

$$\Delta V =V_{T_{max}} - V_{T_{min}}$$ or Change in Voltage

In the above example
$$\Delta T = T_{max} - T_{min}= 100C -0C =100C$$

$$\Delta V =V_{T_{max}} - V_{T_{min}}= 6V - 1V= 5V$$

$$K= \frac { \Delta T} {\Delta V}= \frac {100C} {5V} = 20 \frac V C$$

Now we if we measure a voltage of 2V the temperature is found by

$$T_r = (V_r - V_0) K = (2V -1V) 20 \frac C V= 20C$$

Some points of interest, the relationship:
$$T_r = (V_r - V_0) K$$

Contains 2 variables T and V and a constant K,and one variable is a constant multiple of the other. This is the definition of a linear relationship.

Pay attention to the use of units in the constant K

$$K= \frac { \Delta T} {\Delta V}$$

They are $\frac C V$ when you multiply by a Voltage the units of your reading (V) cancel the units in the denominator (V) leaving only C units of temperature. Learn to use the units of the quantities in your problems to guide your operations. If you find yourself adding quantities with different units, there is a good chance you are doing something wrong. Or if your final answer does not have the correct units (after doing all of the algebra on the units alone) you have made a mistake.

Last edited: Feb 16, 2005
9. Feb 16, 2005

### rayjohn01

Calibration

I agree that calibration is required , the reason being that thermistors are not that accurate they have two tolerances
a) the resistance at a given temperature and
b) the slope of dr/dT
1 degree in the 0 to 100 is obviously 1 % -- they are not that accurate unless preselected by binning.
You can always look up a Manufacturers catalogue such as Dale they include some of this data.

10. Feb 17, 2005

### richnfg

Wow, thanks for the replies. :) I will probably use the equation and it will be some help thanks.
I'm actually testing the linearity of a NTC thermistor by taking different readings (resistance at different temperatures) and then like plotting a graph to check (its obviously not linear). I'm then going to put a resistor in parallel with the thermistor as it supposedly helps to linearize it (but not fully).
I have to apply my experiment to a real life situation and thats where my teacher about the thermometer. I understand how everything works now, thanks to you guys :), but still don't see if I would need linear thermistor for that (maybe a formula that works out the temperature for a linear resistor, but not actually assuming its linear with minimum and maximum points?)

I don't know, all a bit confused.

If anyone thinks of an easier real life example where linear thermistors would be needed , please tell me. :P

11. Feb 17, 2005

### richnfg

I guess its all going to go a bit quiet now

Why did I choose this project...

12. Mar 7, 2005

### richnfg

Ok, sorry about the bringing up the same topic again but I have another question which is related and didn't think it would be worth making a new thread for.

How come NTC thermistors have non-linear resistance? If I'm going to predict this, what would be my reasoning behind it? I know they are non linear, but I dont know why