EEVblog Electronics Community Forum
EEVblog => EEVblog Specific => Topic started by: EEVblog on December 22, 2012, 11:57:25 pm
-
EEVblog #402 - Flir E60 IR Thermal Camera (https://www.youtube.com/watch?v=8Oxky0qyOuw#ws)
Dave.
-
Why cant they just make a 4K resolution thermal camera with a lens mount supporting a decent f1.4 anamorphic lens to maximize detail when outputting content to 16:9 displays, and possibly even a cinema mode where the raw video is recorded with an S-Log profile to allow for 13 stops of dynamic range in order to get the most out of the video in post production?
-
I think the reason most thermal cameras have limited resolution is because it's very difficult to make the individual sensor pixels small.
-
Why cant they just make a 4K resolution thermal camera with a lens mount supporting a decent f1.4 anamorphic lens to maximize detail when outputting content to 16:9 displays, and possibly even a cinema mode where the raw video is recorded with an S-Log profile to allow for 13 stops of dynamic range in order to get the most out of the video in post production?
They can. It's called the James Webb Space Telescope.
-
Why cant they just make a 4K resolution thermal camera with a lens mount supporting a decent f1.4 anamorphic lens to maximize detail when outputting content to 16:9 displays, and possibly even a cinema mode where the raw video is recorded with an S-Log profile to allow for 13 stops of dynamic range in order to get the most out of the video in post production?
Because you need the lens elements to pass IR very efficiently, something conventional lens materials don't do. Also I feel this is sarcastic.
-
Can you estimate how the 12000 AUD get distributed in the BOM?
I guess most goes to the IR sensor... but what other expensive components are there?
-
Probably 2/3 is the sensor and the associated lens, along with the mount and the isothermal block for them. Then the rest is split between display, case, battery and battery management and the rest of the electronics. Probably the single biggest coat after the IR block is the rest of the case, as the moulds to make them are not cheap in something that complex, and there are most likely not many units made to drop the amortisation cost of the dies to an insignificant amount. Consider car body moulds generally are used to make 100k plus units before they are refurbished. Similar cost of mould, but the production run is likely to be only 10k units of the IR camera per year, not 500k like the car. VW here in SA had to do a major refurbishing of the VW Golf tooling, as it was 30 years old, and was well worn, about 10 years ago. Those have made millions of vehicles.
-
Why cant they just make a 4K resolution thermal camera with a lens mount supporting a decent f1.4 anamorphic lens to maximize detail when outputting content to 16:9 displays, and possibly even a cinema mode where the raw video is recorded with an S-Log profile to allow for 13 stops of dynamic range in order to get the most out of the video in post production?
Because you need the lens elements to pass IR very efficiently, something conventional lens materials don't do. Also I feel this is sarcastic.
The lens are made of Germanium.
Dave.
-
The problem is the market for TI cameras is small, and there aren't many manufacturers, so you have the old chicken/egg situation of small marked due to high price and high price due to small volumes. Given sufficient volume I'm sure he sensor price could come down significantly - I don't know enough about the intrinsic costs of making germanium optics to know how scaleable they are. I think the key there would be to get the sensor size down to allow the use of smaller optics
Silicon can also be used as a lens material, but not sure how useable it is at imaging, as opposed to detection type applications.
e.g. this Omron 4x4 sensor (http://www.omron.com/media/press/2012/06/e0627.html)uses a silicon lens.
Devices like the Melexis 16x4 sensor (http://www.melexis.com/Infrared-Thermometer-Sensors/Infrared-Thermometer-Sensors/FIRray16X4-Far-InfraRed-Array-776.aspx) could shake up the bottom end of the market, especially with projects like this kickstarter for a $175 phone add-on (http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory?ref=live)
Which provide enough functionality for things like electrcal faultfinding and building heat loss, which are probably the main potential volume markets.
However until there is a mass-market for higher-res sensors they will never be cheap. I wonder what this 1kx768 TI camera costs.... (http://www.thermoteknix.com/brochures/night_vision/microcam_HD_high_def_thermal_imager.pdf)
-
Melexis 16x4 sensor could shake up the bottom end of the market, especially with projects like this kickstarter for a $175 phone add-on (http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory?ref=live)
I wonder if you could get better resolution out of one of those 16x4 sensors if you rotated it 90deg and mounted it to a motor that scanned a scene twice a second.
Maybe you could up the res to 30x16 or something.
The array has a framerate of up to 512Hz, so it might work.
-
Melexis 16x4 sensor could shake up the bottom end of the market, especially with projects like this kickstarter for a $175 phone add-on (http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory?ref=live)
I wonder if you could get better resolution out of one of those 16x4 sensors if you rotated it 90deg and mounted it to a motor that scanned a scene twice a second.
Maybe you could up the res to 30x16 or something.
The array has a framerate of up to 512Hz, so it might work.
I think the thermal mass of the sensor means the effective update rate would be somewhat less, but maybe still useable in a a slow-scan mode.
-
Why cant they just make a 4K resolution thermal camera with a lens mount supporting a decent f1.4 anamorphic lens to maximize detail when outputting content to 16:9 displays, and possibly even a cinema mode where the raw video is recorded with an S-Log profile to allow for 13 stops of dynamic range in order to get the most out of the video in post production?
Because you need the lens elements to pass IR very efficiently, something conventional lens materials don't do. Also I feel this is sarcastic.
The lens are made of Germanium.
Dave.
Indeed, which means you can't just go out and grab a panavision anamorphic cine lens and slap that on the mount.
-
I think the thermal mass of the sensor means the effective update rate would be somewhat less, but maybe still useable in a a slow-scan mode.
Seeing how easy IR is reflected you might get away with a rotating mirror like in a laserprinter (I'm sure lots of you have taken apart).
You could even scan multiple vertical lines by slightly slanting the mirrors, if I remember correctly most lasers have 6 or 8 mirrors on the motor.
If the sensor can do 512 images per second you could get 10FPS out of it with a resolution of 80*40 isn't it?
Hmm, doesn't sound all that much after all.
-
Probably 2/3 is the sensor and the associated lens, along with the mount and the isothermal block for them. Then the rest is split between display, case, battery and battery management and the rest of the electronics. Probably the single biggest coat after the IR block is the rest of the case, as the moulds to make them are not cheap in something that complex, and there are most likely not many units made to drop the amortisation cost of the dies to an insignificant amount. Consider car body moulds generally are used to make 100k plus units before they are refurbished. Similar cost of mould, but the production run is likely to be only 10k units of the IR camera per year, not 500k like the car. VW here in SA had to do a major refurbishing of the VW Golf tooling, as it was 30 years old, and was well worn, about 10 years ago. Those have made millions of vehicles.
I bought an "old" used Flir B2 thermal camera a year ago.
it is "only" 122 x 80 pixels sensor and it costed new more than 8000 euros.
the battery was weak and a new one is 150 euros that's not too much
but I asked for another lens you can have 3 different types : each one costs 2000 euros !!!
so the sensor costs a lot but the lens I think it costs as much as the sensor.
actually you can get a "night vision" camera on german high end cars mercedes or bmw
the option costs 2500 euros, and it is in the end a nice thermal camera
they sell more, so the price goes down.
-
I think the thermal mass of the sensor means the effective update rate would be somewhat less, but maybe still useable in a a slow-scan mode.
Seeing how easy IR is reflected you might get away with a rotating mirror like in a laserprinter (I'm sure lots of you have taken apart).
You could even scan multiple vertical lines by slightly slanting the mirrors, if I remember correctly most lasers have 6 or 8 mirrors on the motor.
If the sensor can do 512 images per second you could get 10FPS out of it with a resolution of 80*40 isn't it?
Hmm, doesn't sound all that much after all.
That still assumes the thermal response time of the sensor is fast enough. Remember a thermopile sensor is just bunch of small temperature sensors, which will have some thermal mass
-
I guess this is on its place :)
Emissivity:
http://www.ictinternational.com.au/brochures/everest/emissivity-theory.pdf (http://www.ictinternational.com.au/brochures/everest/emissivity-theory.pdf)
http://www.everestinterscience.com/info/emissivitytable.htm (http://www.everestinterscience.com/info/emissivitytable.htm)
(The same temperature can seem to be measured as a different temperature)
-
I wonder if you cool the sensor down, i.e. through a Peltier cooler, the thermal response would be better. Don't they do that for some sensors?
-
Only done for those that need low noise. The common sensors utilise thermal difference to make the output, cooling makes it lower noise, though to get an advantage that is better than the power consumed you need a 3 or more stage petlier unit that takes the sensor down to liquid nitrogen temperatures so that there is effectively nearly no noise compared to room temp. You really would use these if you want a ultra low noise signal like in a telescope. James Webb telescope cools the detector actively with such a cooler to get it cooler than 4k as that is what the background is. The biggest problem they have is stray heat entering the optical path.
-
FYI, the sensor these IR cameras use is called a Focal Plane Array Uncooled Microbolometer. http://en.wikipedia.org/wiki/Microbolometer (http://en.wikipedia.org/wiki/Microbolometer) and http://en.wikipedia.org/wiki/Bolometer (http://en.wikipedia.org/wiki/Bolometer).
-
To me the insane price seems to be one of two things.
They charge a ton of money to have a huge profit margin because they can get away with it due to lack of competition
Or the second reason (though highly unlikely), They charge enough to get many 10's of thousands % profit margin because the devices last long enough that they wont sell very often (eg in the mattress industry, they may charge $1000 for a mattress that cost about $30-40 to make)
I bet if someone took one apart and looked up all of the parts, they would get a BOM of less than $100
eg kinda like how smartphones sell for nearly $900 when they cost about $200 or less to make
http://www.isuppli.com/Teardowns/News/Pages/Many-iPhone-5-Components-Change-But-Most-Suppliers-Remain-the-Same-Teardown-Reveals.aspx (http://www.isuppli.com/Teardowns/News/Pages/Many-iPhone-5-Components-Change-But-Most-Suppliers-Remain-the-Same-Teardown-Reveals.aspx) (And those estimates don't even take into account volume discounts and other deals that the company may have made (since that info is not publicly available)
-
To me the insane price seems to be one of two things.
They charge a ton of money to have a huge profit margin because they can get away with it due to lack of competition
Or the second reason (though highly unlikely), They charge enough to get many 10's of thousands % profit margin because the devices last long enough that they wont sell very often (eg in the mattress industry, they may charge $1000 for a mattress that cost about $30-40 to make)
Or maybe BOM and R&D costs are high due to limited volume. You're not going to sell millions of these anytime soon.
I bet if someone took one apart and looked up all of the parts, they would get a BOM of less than $100
Tell me where to find just the sensor for $100, or just the lens.
-
How much are those 4x4 or 16x4 sensors?
I mean, even a 16x16 camera wouldn't be too bad, if you could put them in a matrix, even better.
-
Omron 4x4 GB£30
http://uk.farnell.com/omron-electronic-components/d6t-44l-06/sensor-thermal-mems-4x4/dp/2218000 (http://uk.farnell.com/omron-electronic-components/d6t-44l-06/sensor-thermal-mems-4x4/dp/2218000)
Panasonic 8x8 : US$39
http://www.digikey.com/product-detail/en/AMG8831/255-3509-1-ND/3461616 (http://www.digikey.com/product-detail/en/AMG8831/255-3509-1-ND/3461616)
But the bastards won't ship it to UK ( same with all Panasonic products from DK)
Melexis 16x4 US$65
http://www.futureelectronics.com/en/technologies/semiconductors/analog/sensors/temperature/Pages/5020660-MLX90620ESF-BAB-000-TU.aspx?IM=0 (http://www.futureelectronics.com/en/technologies/semiconductors/analog/sensors/temperature/Pages/5020660-MLX90620ESF-BAB-000-TU.aspx?IM=0)
Omron are promising a 16x16 sensor mid next year, which should be interesting.
A problem with low-res sensors around 16x16 is that looking only at the IR image it can be hard to figure out what you're looking at (i.e. relating objects in the thermal image to real objects. The 16x4 strip inserted into the phone image on the IR_Blue helps a lot as you get a visual reference on both sides.
-
Hmm, patent expired on a thermal imaging sensor design? I heard FLIR's patent was due to expire in 2012-2013 some time
-
On the discussion of transducer arrays I can add the TPA81, an 8x1 thermopile array I have had experience with.
This sensor module can easily be panned with a servo motor, and in fact the distributor linked below has demo software for this purpose.
Making qualitative measurements to e.g. lock-on a heatsource is straightforward, but making quantitative measurements is very challenging.
http://www.robot-electronics.co.uk/acatalog/Thermal_Array_Sensor.html (http://www.robot-electronics.co.uk/acatalog/Thermal_Array_Sensor.html)
http://www.robot-electronics.co.uk/htm/tpa81tech.htm (http://www.robot-electronics.co.uk/htm/tpa81tech.htm)
Alex
-
Dave:
Something you missed, or didn't know is that it has a touch screen. Makes some actions a bit easier.
paul
-
Something you missed, or didn't know is that it has a touch screen. Makes some actions a bit easier.
Didn't occur to me at all to try and touch it!, and Charles didn't touch the screen when he quickly demoed it before he loaned it to me.
I hadn't read the manual or web site etc :-[
Dave.
-
If the lens is quite expensive but reflective objects reflect IR quite well, why not use a curved mirror instead of a lens? After all, good telescopes always use a concave mirror as the first stage of magnification.
-
If you want to use a symmetric mirror, than you have to position the sensor in front of that mirror, kind of like a catadioptric system (eg. telescope). This unit would have to be sealed to protect it from dust, so you would still end up with a fairly expensive plan parallel lens in front of the mirror. You might also run into issues if you try to design it for short focal lengths.
-
If you want to use a symmetric mirror, than you have to position the sensor in front of that mirror, kind of like a catadioptric system (eg. telescope). This unit would have to be sealed to protect it from dust, so you would still end up with a fairly expensive plan parallel lens in front of the mirror. You might also run into issues if you try to design it for short focal lengths.
..and if you seal it you need an IR transparent window.
-
My company bought the Flir E50 a couple of weeks ago to create themographic building reports. We bought the Extech MO297 with it, to measure the air humidity etc. The Flir E50 unfortunately doesn't have that capability. The brief pauses you experience when using the camera are caused by the calibration mechanism. You can hear a solenoid clicking when it covers the lens for a brief moment. See Mikes electric stuff: Thermal imaging camera teardown and repair Fireflir ff110 (https://www.youtube.com/watch?v=_65erGXqXZk#ws) where Mike does a tear down and repair of a IR head mount fire-fighter camera.
The Extech MO297 has a IR thermometer as well and I am really curious as how this works? Maybe you could do an article on that?
Simon
-
IR $399
http://www.ebay.com/itm/DT-9862-Infrared-Video-Thermometer-/230902710856?pt=LH_DefaultDomain_0&hash=item35c2dfc248 (http://www.ebay.com/itm/DT-9862-Infrared-Video-Thermometer-/230902710856?pt=LH_DefaultDomain_0&hash=item35c2dfc248)
:)
-
IR $399
Yeah, IR spot measurement, the video is visible light only >:D
-
Has anyone done a review of the new Fluke VT02 Visual IR Thermometer? It looks interesting.
-
IR $399
Yeah, IR spot measurement, the video is visible light only >:D
That's pretty damn useless for $399.