Actions
Forum Thread Details
  • State Verified Answer
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 285 subscribers
  • Views 6952 views
  • Users 0 members are here
Related

Inaccurate Readings from Thermal Imaging Camera on Copper Heatsink-Anyone Else Had this Happen?

kkazem

I have a Fluke Ti25 Thermal Imaging Camera that I used on a 1,200 watt PFC Controller to measure the power FET and power diode temps. The Fluke camera has an adjustment for emissivity, which I set to copper (I don't recall the value). I chose this as my heatsink was a copper bar. Using both my finger (with the power off, of course) and a thermocouple, I'm confident the actual temperature was about 85 Celsius, but the Fluke IR camera said it was 15 Celsius. Clearly, this reading was wrong as my lab is in Southern California and it was done in late summer. The temperature never gets below 70 F (21.1 C) and this heat sink was hot to the touch. 

The literature says it may be due to reflected energy instead of emitted energy. So I changed the angle of the camera with respect to the heatsink and it didn't make much difference. The literature also says to put black tape on the surface or to paint it black, but both of those would be problematic on my heat sink as it would interfere with convection cooling off of the copper bar heat sink. The devices were TO-247 (one FET and one SiC Diode) on a 4" x 1" x 0.19" thick copper bar. The devices were mounted with a 6-32 screw and nut and tightened appropriately. 

Has anyone else had similar problems and if so, do you have any tips on how to correct it to obtain more accurate temperature values from the IR camera?

image

Parents
  • 0

    An infrared camera measures the long-wave infrared light emitted by a hot object and does not measure temperature directly. It converts the measured amount of infrared to a temperature by assuming a value for the emissivity of the object (typically, approximately 0.85 but this can be set in some higher-end cameras). The emissivity is a measure of the ratio of energy emitted from the material's surface to that of a perfect black-body radiator.

    Depending on the type of surface, the emissivity value can vary dramatically - some materials like shiny metals are very low emissivity. Worse still, they may have a tendency to reflect IR energy more than they emit, so you may be getting a false reading from a reflection of a hot component nearby.

    Wikipedia has a list of common emissivity values - https://en.wikipedia.org/wiki/Emissivity#Emissivities_of_common_surfaces - for example, Aluminium foil has 0.03 while paper has about 0.89.

    The most accurate measurements will be had from a contact measurement system - e.g. by thermocouple, RTD probe. Some cameras will have an external thermocouple input for this reason. If a little bit of error is acceptable, painting or sticking on some matte finish painters tape is often recommended as an alternative but as you identified is a potential problem as it will change the behaviour of the system. Adjusting the in-camera emissivity value to match the material you are measuring for should provide more accurate results, but at such low emissivity values, it is more likely to result in a very noisy reading. Perhaps roughening the surface of the heatsink would help.

    Another thing to be worried about is the area of the heatsink. The spot-distance ratio for the camera and minimum focal length need to be respected - if the heatsink is not filling the spot's field of view, then the measured temperature could be influenced by surrounding components. Likewise, below the minimum focal length, it can be hard to align the camera to the heatsink at all. Depending on the unit you are using, the central "spot" temperature reading may actually come from another sensor entirely (e.g. with the FLIR TG267 that I have) which means that at close range, you have a vertical parallax error to account for, although I think the Fluke doesn't work in the same way.

Reply
  • 0

    An infrared camera measures the long-wave infrared light emitted by a hot object and does not measure temperature directly. It converts the measured amount of infrared to a temperature by assuming a value for the emissivity of the object (typically, approximately 0.85 but this can be set in some higher-end cameras). The emissivity is a measure of the ratio of energy emitted from the material's surface to that of a perfect black-body radiator.

    Depending on the type of surface, the emissivity value can vary dramatically - some materials like shiny metals are very low emissivity. Worse still, they may have a tendency to reflect IR energy more than they emit, so you may be getting a false reading from a reflection of a hot component nearby.

    Wikipedia has a list of common emissivity values - https://en.wikipedia.org/wiki/Emissivity#Emissivities_of_common_surfaces - for example, Aluminium foil has 0.03 while paper has about 0.89.

    The most accurate measurements will be had from a contact measurement system - e.g. by thermocouple, RTD probe. Some cameras will have an external thermocouple input for this reason. If a little bit of error is acceptable, painting or sticking on some matte finish painters tape is often recommended as an alternative but as you identified is a potential problem as it will change the behaviour of the system. Adjusting the in-camera emissivity value to match the material you are measuring for should provide more accurate results, but at such low emissivity values, it is more likely to result in a very noisy reading. Perhaps roughening the surface of the heatsink would help.

    Another thing to be worried about is the area of the heatsink. The spot-distance ratio for the camera and minimum focal length need to be respected - if the heatsink is not filling the spot's field of view, then the measured temperature could be influenced by surrounding components. Likewise, below the minimum focal length, it can be hard to align the camera to the heatsink at all. Depending on the unit you are using, the central "spot" temperature reading may actually come from another sensor entirely (e.g. with the FLIR TG267 that I have) which means that at close range, you have a vertical parallax error to account for, although I think the Fluke doesn't work in the same way.

Children
No Data