Testing the Accuracy of the FLIR One

In my review of the FLIR One, I mentioned that this would be an excellent device for a physics lab. So, here is my first “lab”, hopefully there will be more posts like this. The FLIR One Tech Specs states the camera can detect temperature differences down to 0.1°C. However, detecting temperature differences and measuring temperatures are two very different things.

How Do You Measure Temperature?

This might seem like an easy thing to measure, but it isn’t. Here are three common methods for measuring temperature.

Thermal Expansion of a Liquid. When a liquid increases in temperature, it expands (true for solids and gases too). If you put this liquid in a tube, you can correlate the increase in volume with an increase in temperature. In the past, these thermometers were filled with mercury. However, mercury isn’t the best thing to have around if the thermometer breaks. Newer liquid thermometers have alcohol in them. How do you calibrate these thermometers? The typical method is to use at least two reference temperatures (freezing and boiling point of water) and then divide the scale into 100 pieces (for Celsius).

Electric Potential of Different Metals (Thermocouple). The basic idea is that two different metals can produce a change in electric potential when in contact with each other. This small voltage depends on the temperature of the two metals. That’s it. Just record the voltages for some known temperatures and use that to calibrate the device.

Measuring the Infrared Light. For any object, there are two sources of light. First, light (of any wavelength) could hit the object and then reflect off. This isn’t very useful for measuring temperature. Second, light is also emitted from objects. The wavelength and intensity of this light depend on the temperature of the object. We call this blackbody radiation (since a black body doesn’t reflect light). For most objects around you, the blackbody radiation is in the infrared region so that you can’t see it – but an IR camera can. By measuring this IR light, the camera can estimate the temperature.

Of course with all of these methods, there are problems. The thermocouple and liquid thermometer have to be in contact with the object to measure its temperature. This means that the location of the probe might be a different temperature than other parts of the object. Also, there is the calibration issue. In the end, measuring temperature isn’t as simple as you would like.

Comparing Temperatures

As a means for estimating the accuracy of the FLIR One, I am just going to set up a simple experiment. I will use a few different temperature measuring devices and compare the temperature values for water on a hot plate.

Image: Rhett Allain

I have three thermometers to measure the water temperature. There is a hand held digital thermometer, a digital thermometer with a probe on a wire (I assume both are thermocouples) and then a liquid thermometer. My plan was to measure the temperature from 0°C to 100°C (the stated range of the FLIR One), but those two end points didn’t give the best data. Also, the liquid thermometer only goes up to 50°C.

For the FLIR One measurements, the device needs to recalibrate to obtain a temperature measurement. For each temperature reading, I would recalibrate at least 5 times and get 5 different values for temperature. This way I could get both an average and standard deviation for the temperature.

Let’s just get to the data. Here is a plot of difference in temperature of different devices vs. a reference temp. For the reference temperature, I am going to use the readings from the digital thermometer with the remote probe. The liquid thermometer is probably the most accurate, but it only goes up to 50°C.

Ok, how about the average deviation from the reference temperature for each device?

Clearly there are some problems with this data. Here are a few of the things I can think of.

• It’s possible that the FLIR One is measuring reflected IR instead of IR from the water. I tried putting a box over the water to shield it and it didn’t seem to make a difference.


• Both the digital thermometers and the mercury thermometer take some time to reach an equilibrium temperature. The reading might not yet be from the equilibrium temp.


• The water isn’t all just at one temperature. I think the contact thermometers sort of average out any difference in water temperature by “touching” different parts. However, the FLIR one just looks at one spot.

Here is a quick video where you can see the temperature differences at the surface of the water.

You can see the problem of measuring temperature. However, notice that even if the FLIR One produces uncertainty in the temperature, you can still see relatively small temperature differences.

But what is the accuracy of the FLIR One? Based on the bar chart above, I can give the following conservative estimates for the accuracy.

• FLIR One = +/- 2°C


• Mercury thermometer = +/- 0.5°C


• Digital thermometer = +/- 1°C

That’s my best guess on the accuracy of the FLIR One.