EVB90620 Datasheet, PDF (36/39 Pages) Melexis Microelectronic Systems – High precision non-contact temperature measurements

MLX90620
16x4 IR array
15 FAQ
When I measure aluminum and plastic parts settled at the same conditions I get significant
rrors on aluminum. Why?
Different materials have different emissivity. A typical value for aluminum (roughly polished) is 0.18
and for plastics values of 0.84…0.95 are typical. IR thermometers use the radiation flux between the sensitive
element in the sensor and the object of interest, given by the equation
q ₁. ₁. T₁₄ . .A₁.F_{a b 2} . T₂₄ . .A₂ ,
Where:
1 ^and 2 ^{are the emissivities of the two objects,}
1 ^{is the absorptivity of the sensor (in this case),}
is the Stefan-Boltzmann constant,
A₁ and A₂ are the surface areas involved in the radiation heat transfer,
F_a-b is the shape factor,
T₁ and T₂ are known temperature of the sensor die (measured with specially integrated and calibrated
element) and the object temperature that we need.
Note that these are all in Kelvin, heat exchange knows only physics.
When a body with low emissivity (such as aluminum) is involved in this heat transfer, the portion of the
radiation incident to the sensor element that really comes from the object of interest decreases – and the
reflected environmental IR emissions take place. (This is all for bodies with zero transparency in the IR band.)
The IR thermometer is calibrated to stay within specified accuracy – but it has no way to separate the
incoming IR radiation into real object and reflected environmental part. Therefore, measuring objects with low
emissivity is a very sophisticated issue and infra-red measurements of such materials are a specialized field.
What can be done to solve that problem? Look at paintings – for example, oil paints are likely to have
emissivity of 0.85…0.95 – but keep in mind that the stability of the paint emissivity has inevitable impact on
measurements.
It is also a good point to keep in mind that not everything that looks black is “black” also for IR. For example,
even heavily oxidized aluminum has still emissivity as low as 0.30.
How high is enough? Not an easy question – but, in all cases the closer you need to get to the real object
temperature the higher the needed emissivity will be, of course.
With the real life emissivity values the environmental IR comes into play via the reflectivity of the object (the
sum of Emissivity, Reflectivity and Absorptivity gives 1.00 for any material). The larger the difference between
environmental and object temperature is at given reflectivity (with an opaque for IR material reflectivity equals
1.00 minus emissivity) the bigger errors it produces.
After I put the MLX90620 in the dashboard I start getting errors larger than specified in spite
that the module was working properly before that. Why?
Any object present in the FOV of the module provides IR signal. It is actually possible to introduce
error in the measurements if the module is attached to the dashboard with an opening that enters the FOV. In
that case portion of the dashboard opening will introduce IR signal in conjunction with constraining the
effective FOV and thus compromising specified accuracy. Relevant opening that takes in account the FOV is
a must for accurate measurements. Note that the basic FOV specification takes 50% of IR signal as threshold
(in order to define the area, where the measurements are relevant), while the entire FOV at lower level is
capable of introducing lateral IR signal under many conditions.
When a hot (cold) air stream hits my MLX90620 some error adds to the measured temperature
I read. What is it?
IR sensors are inherently sensitive to difference in temperatures between the sensitive element and
everything incident to that element. As a matter of fact, this element is not the sensor package, but the sensor
die inside. Therefore, a thermal gradient over the sensor package will inevitably result in additional IR flux
3001090620
Rev 4.0
Page 36 of 40
Datasheet IR16x4
19 September 2012