Look at the diagram above. It shows a cold start of a switched off camera.
This drift is a problem of the low cost Flir One. With my Flir E4 and Flir E40 this effect is significant lower.
Yes but it does not make sense not to calibrate for cold start. If the sensor reacts this way, it should be calibrated to work from the start and to show correct values. Hm, will do some testing, mines already shipped, not long to wait.
Here is Lepton 3 datasheet
http://www.file-upload.net/download-11989243/Lepton_Datasheet.pdf.html. There are some interesting points. For example, page 21, "Lepton 3 automatically prohibits the shutter from operating when it detects the temperature to be outside the range -10° C to +65° C. For example, if the camera is operating at a temperature of 70° C, no automatic FFC will be performed, and the camera will ignore any commanded FFC if the FFC mode is “automatic” or “manual.” ". Meaning if you want shutter control, there of course is another way - heat the thing up. Ok, do not listen to me, you can burn it. Just saying. Maybe E4 has heating element? Anyway, on page 57 Maximum Operating Temperature is -40 to +80C. Over that it shuts after some seconds has passed.
Datasheet: Flir Lepton 3 thermal sensitivity <50 mK =0,050 K.
This is a realistic value, but not so good like a upgraded E4. You see the difference ;-)
Yup, NETD is rated <50 mK (20 mK typical) according to datasheet at page 52.
But temperature is rated 0 K to >400 K at the same page 52. Hm, that means realy -273.15 C to >+126.85 C , meaning this sensor can actually sense that temperature. Ok, maybe over 120C it is not as accurate (or it just dies and sees same values for any higher?), but I believe Seek over 120C is not very acccurate too. You think Flir SDK let us see this whole range really?
Ok, I do not want to resurrect old discussions about the f. rate limit here, and firmware, etc, heck I do not even have the product in hands yet, well but being still careful, even though page 36 states that 26 frames are @8mhz, max @20mhz could be 60 or smth, and at page 42 you can see interesting statement "For each unique frame, two duplicates follow in the VoSPI stream", so this may as well be something different, but anyway what I am trying to say you know where the limit is (could be) implemented, so I wonder if temperature range is not limited there too.
Try Georgs Thermal Camera App. There is no temperature limit (but limited by SDK).
Definitely. I think I will buy it anyway, need to support the community