What does a FLIR report when looking *through* hot air?

[IDEngineer]

We have some interesting, conflicting thermal data at a remote location. We don't have a FLIR at my location so I can't confirm my theory myself.

We created a couple of prototypes which include a thermistor to sense ambient air temperature and take action at a certain threshold. We calibrated it using a Fluke thermistor sitting right next to the one in the prototype. We figured "use the same type of sensor to get comparable data". All went well and we shipped the prototypes, which tested well at the remote location.

Except... our prototypes are triggering about 40F below the target threshold. The remote site sent images showing how they measured this. They created a test chamber and pumped in warm air, but instead of using an actual temperature sensor they poked in a FLIR camera pointed at our prototype's enclosure.

FLIR's report surface temperatures. As the ambient air temp increases, the enclosure will start to warm up... but will lag the ambient temperature. My theory is that the ambient AIR temperature is probably at the target, but the enclosure hasn't heat-soaked to that point yet due to its significant thermal mass.

What I don't know is how a FLIR would react to "seeing" a cooler target through warmer ambient air. Which temperature will it report? The typical use case for a FLIR (at least in my experience) is that the target is warmer than the intervening ambient atmosphere - in other words, you're looking at a warmer object through cooler air. In that case the object is the warmest thing in view and there's no question what you're measuring. But what happens if the intervening ambient is WARMER than the target? I haven't been able to find any answers nor even comments or discussion of this question online.

Thanks!

[bap2703]

Short answer:
Air is transparent in the wavelengths used by thermal infared systems.
So warm or cool air doesn't change anything.
-----------
More detailed answer:
In reality "air" isn't totally transparent: water vapour and CO2 emit/absorb depending on the temperature of the object behind, the concentration of that gas and the length the light has to go through.
Other gases also emit/absorb even more but in that case you would probably know what you are doing.

I guess you are just in the simple normal case right?


Since you mention a "test chamber" and the reference temperature measured by a thermal camera I would suspect a window issue: what kind of window is your remote team using?
It could be that the window is filtering out too much infrared radiation, hence their camera registers a lower temperature than it actually is inside the chamber.

[IDEngineer]

When I typed "they poked in a FLIR camera pointed at our prototype's enclosure" I meant that literally: They cut a hole and pointed the FLIR through it. No "window" material to affect the measurement.

OK, logically your answer makes sense. In the IR, given STP and typical humidity, the intervening air should not affect the FLIR's measurement of a remote surface. That removes this possibility.

More likely, then, that my basic presumption is valid: The air warms faster than the enclosure, and what the the FLIR is actually measuring is the enclosure temp while our thermistor is responding to ambient air. The "thermal delay" of the thermistor is estimated by its manufacturer to be ~6-7 seconds (lots of possible variables but I presume this is based on the mass, thermal conductivity, and surface area of its bead with some allowances for heat sinking through the leads). The "thermal delay" of the enclosure is likely to be substantially greater given its much larger mass, so our thermistor reacts faster to the ambient rise (and our circuitry responds) while the FLIR reports the lagging indicator of the enclosure temperature.

I'll try to get them to measure the actual AIR temperature with an appropriate sensor.

Thanks!

[Bud]

OK, logically your answer makes sense. In the IR, given STP and typical humidity, the intervening air should not affect the FLIR's measurement of a remote surface.

It is not exactly that. Ambient temperature causes a correction factor into the measurements if you want more accurate numbers.

[IDEngineer]

What we're doing doesn't require high accuracy. If we were +/-10F everyone would be delighted. But the target threshold is 180F and the FLIR is reporting 140F, so we're off by quite a bit and the remote location was startled by the discrepency. Since we confirmed proper operation here with another calibrated, thermistor-based sensor, we cannot accept that a 40F difference is just "normal tolerance". That's why I think we're getting two different values because the two different measurements are being taken with two different kinds of sensors, the first which measures ambient and the second which measures surfaces.

[Bill W]

It may even depend on what the enclosure is made of.  If it is thin polythene then the FLIR will look *through* that as well back into the outside world.

A bit of cardboard inside for the FLIR to look at might convince them that all is well, should keep close to the air temperature.

Bill

[coppercone2]

Might be an interesting idea to compare a hot air gun (dry air) to a steam kettle (water vapor mixed with air). Pure high pressure steam would be interesting to know about too (its actually dangerous since its invisible if there a high pressure steam leak, maybe this can help steam tunnel workers). Not too sure about how to generate it.

It would be also trivial to test other gases at various temperatures if desired, by using a solder sucking station as a blower, by connecting it to gas of choice rather then suction pump.

[IR_Geek]

Low hanging fruit first.       yes, definitely get an air temperature thermocouple.   That along with the surface temp thermocouple should give some more insight.

The air will add/subtract to the overall signal coming from the unit under test.  The key is calculating if it is significant or not.  With a +/- 10 I'd say no.  However, that will depend on your exact spectral band.   I'd assume you are using an uncooled longwave?    Uncooled LW are really bad about having poorly defined cutoff's

[IDEngineer]

yes, definitely get an air temperature thermocouple

Well, we used an air thermistor to calibrate our prototypes. We did so because that way we were measuring air temp, which is what was specified. And the reference thermistor would presumably measure air temp like our prototype's thermistor would. Trying to use similar devices to make similar measurements... almost like we did that on purpose. 

I believe the error comes from the remote site using a surface measuring tool (FLIR) instead of an air measuring tool (thermistor) to measure air temp.

[IR_Geek]

Yes your right, thermistor.    Just so I'm clear on the setup.   You have your prototype with a thermistor (inside or outside of prototype?) ,  which is then in a thermal chamber.  They are sticking an imaging FLIR camera to get the surface temp of the prototype.   How do they set the thermal chambers air temperature?  Or is this a Home Depot special ... several pieces of hard insulation made into a box with a hole for the FLIR and another with a fan heater  ... Done that a few times when our thermal chamber was occupied.

Yes, unless they have some calibrated surface temp vs. air temp lookup table, then I don't see how they could get the air temp.  OR ... they went up slow in temperature of many hours to heat soak the prototype.

Sorry for all the confusion, but a picture or diagram would be very helpful.  In all my years it is very apparent that everybody 'fills in the gaps' with our own assumptions  and people end up talking past each other.  Communication is usually the hardest part of any technical job.

[IDEngineer]

Just so I'm clear on the setup.   You have your prototype with a thermistor (inside or outside of prototype?) ,  which is then in a thermal chamber.  They are sticking an imaging FLIR camera to get the surface temp of the prototype.   How do they set the thermal chambers air temperature?  Or is this a Home Depot special ... several pieces of hard insulation made into a box with a hole for the FLIR and another with a fan heater  ... Done that a few times when our thermal chamber was occupied.

Bingo, that's what they described to us. A big box with a hot air gun and a hole for the FLIR to look through. Zero chance they allowed things to warm up slowly and heat-soak things, which is exactly why I think they are measuring surface temperature (a lagging indicator of ambient air) while we are measuring ambient air (which is what was specified).

[IR_Geek]

Good luck educating them.   Should be an interesting conversation.