Propeller-based Thermal Imager

Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
edited 2013-01-19 - 22:16:43 in Propeller 1
I've been working on a possible new product for the Propeller: a 16x4-pixel thermal imager. Here's a photo of my hand-wired prototype plugged into a Prop BOE:

attachment.php?attachmentid=95347&d=1346870385

I wrote a Perl script to display the raw image data on my PC. Here's a video that demonstrates the unit capturing the heat from my hand and also from a hot soldering iron:

[video=vimeo;48896841]

In the video, I've expanded each pixel to a 3x3 array and interpolated the eight outer subpixels to give a smoother image. Although the resulting 48 x 12 image doesn't contain any more information than the 16x4 image, it does look nicer (i.e. less jagged).

Anyway, so far, so good. Correcting the raw image data for gain, offset, die temperature, etc., and converting it to actual pixel-by-pixel temperatures is the next step. The math to do so is a bit hairy, though, involving fourth powers and roots. This may be one case where I succumb to using floating-point math. Also, I want to be able to display the image directly to video and VGA, rather than depending on a PC. Being able to overlay the image on a video feed would be nice, too, although that would have to be done in grayscale, not color.

Stay tuned for further developments!

-Phil
522 x 382 - 42K
«13

Comments

  • RobotWorkshopRobotWorkshop Posts: 2,301
    edited 2012-09-05 - 11:58:28
    Looks pretty cool Phil. I think this sensor would be great for robotics and would help with people following, etc. The only other similar sensor I have seen is this one:

    http://www.robot-electronics.co.uk/acatalog/Thermal_Array_Sensor.html

    It looks like the one you have will have a better resolution than that one. I can think of a couple robot projects I've done where this sensor would help.

    Robert
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-05 - 12:17:40
    Wow, this sounds really interesting!

    I hope you develop it - yes, definitely get it on the shelf!

    Any ideas yet about how sensitive it can be? What's the wavelength range?
  • LeonLeon Posts: 7,620
    edited 2012-09-05 - 12:48:12
    I worked on the prototype of this thermal imaging system:

    http://www.irisys.co.uk/thermal-imaging/

    It used a 16x16 IR array with 256 levels, and a germanium lens.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-05 - 14:22:20
    Any ideas yet about how sensitive it can be? What's the wavelength range?
    The only data I have so far is the temperature range reportable by each pixel for objects in its field of view, which is -50°C to +300°C.

    -Phil
  • lardomlardom Posts: 1,585
    edited 2012-09-05 - 14:57:40
    Can you tell us what you're using as a sensor?
  • Duane DegnDuane Degn Posts: 10,157
    edited 2012-09-05 - 19:44:55
    lardom wrote: »
    Can you tell us what you're using as a sensor?

    I'm going to bet it's a MLX90620.

    Here it is at DigiKey. I don't know why I'm surprised to see it cost as much as it does at $64.45.

    Edit: BTW, this looks super cool.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-05 - 20:39:53
    Yes, Duane is right: it's an MLX90620. And, yes, they are a bit pricey, but way less expensive than larger-array thermal imagers. I had actually contemplated building one with surface-mount thermistors, at one point, using a largish parabolic mirror for forming an image on the PCB. I'm glad I waited! :) The Melexis sensor is a much more elegant solution.

    Anyway, my initial plan is to design a board with a TO-239 socket to accommodate the sensor. That way, people can choose between the 40-degree model and the 60-degree model -- or anything else Melexis might come up with.

    -Phil
  • lardomlardom Posts: 1,585
    edited 2012-09-05 - 21:37:51
    $65 doesn't seem like a lot. I wanted to buy a flir in '07. It was $7k or $8k then.
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-05 - 21:45:21
    I think $65 is just for the sensor by itself, not all the work that Phil must do and the parts that go into making it interface with the world.

    Nevertheless, it looks like an awesome project.

    I love this FAQ on page 37 of the data sheet:

    I consider using MLX90620AAA to measure temperature within car compartment, but I am embarrassed about the Sun light that may hit the module. Is it a significant issue?
  • lardomlardom Posts: 1,585
    edited 2012-09-05 - 22:19:23
    @ElectricAye, Duane Degn posted a link to Digikey that showed the sensor was $64.45 which I think is a great price.
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-06 - 06:58:08
    lardom wrote: »
    @ElectricAye, Duane Degn posted a link to Digikey that showed the sensor was $64.45 which I think is a great price.

    Yes, I was just pointing out that the effort and intelligence it takes to get that sensor to work with the Propeller, etc. is far from trivial. Phil needs to be compensated along with Parallax, and even though there might be discounts involved in buying a multitude of sensors, Phil needs to keep Browser happy in fancy shrimp and raccoon furs.

    attachment.php?attachmentid=92366&d=1336604042
  • lardomlardom Posts: 1,585
    edited 2012-09-06 - 07:39:58
    @ElectricAye :smile: Agreed!
  • graffixgraffix Posts: 388
    edited 2012-09-06 - 09:34:47
    Very nice Phil, I knew it could be done at a much cheaper price than the open market has them. This could make ya lots of money for your efforts. IMO. I think the cheapest one I saw was like $1200? Yeah most were 7-8k on up. One could be rented or leased for like $120 a month I think. Can you add features for a home inspection application? record/playback feature? Just some thoughts, Congrats
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-06 - 11:34:57
    Will this thing have a lens? I know IR lenses are kinda pricey. DIY own out of blocks of salt?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-06 - 11:41:49
    The module has a built-in lens.

    -Phil
  • ratronicratronic Posts: 1,451
    edited 2012-09-06 - 12:08:42
    If the Propeller firmware is written in Spin/Pasm instead of C I will get one.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-06 - 12:11:50
    ratronic wrote:
    If the Propeller firmware is written in Spin/Pasm instead of C I will get one.
    Since I'm writing the firmware, that's one thing you can count on! :)

    -Phil
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-10 - 13:05:34
    More info on the raw sensor itself. The video is somewhat dull, but I guess this guy doesn't get paid to tap dance.
  • Duane DegnDuane Degn Posts: 10,157
    edited 2012-09-10 - 13:44:57
    Phil,

    I've been playing around with the Propeller Backpack lately trying to create custom characters on the fly to generate an artificial horizon overlayed on my ELEV-8's video link.

    This got me thinking about having thermal information overlayed on a live image of the object being measured(with the thermal sensor). I personally think it would be cool if your sensor board would work with the Propeller Backpack this way. Possibly even having the sensor on a daughter board that plugs into the Backpack's 12 pin connector. (Though now that I type this, I realize the sensor may need to be piggybacked on the video camera used to create the image.)

    Anyway just some thoughts.

    Thanks for continuing to make such cool stuff. I'll probably buy whatever you come up with next (I'm always glad when I do).
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-10 - 14:32:59
    Duane,

    The sensor board is pretty much laid out. It will plug into the Backpack, Spinnret, and MoBo 12-pin connector. It will also have provision for plugging into a solderless breadboard, working with a pair of servo extension cables, and plugging into the mezzanine connector on the TSL1401-DB and PropCAM-DB for simultaneous visible and far-IR image captures. My thought is that it be a semi-kit, with all the SMD parts installed and the though-hole parts (connectors, TO-5 socket, and sensor) loose, so that users can configure it to suit their needs. It will work with both 3.3V and 5V systems, and on-board I2C pullups are jumper selectable, since some host systems already have pullups for SDA and SCL.

    I think that pretty much covers the bases. :)

    -Phil
  • ratronicratronic Posts: 1,451
    edited 2012-09-10 - 14:51:17
    I can imagine this being mounted to a servo and have your robot/etc. do a thermal scan of it's present position.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-10 - 15:04:38
    ratronic wrote:
    I can imagine this being mounted to a servo and have your robot/etc. do a thermal scan of it's present position.

    Yup, I've been thinking the same thing!

    -Phil
  • TJHainesTJHaines Posts: 2
    edited 2012-09-10 - 17:39:35
    Phil,
    I wanted to persue thermal imaging a while ago, but did not have the experience to do so. However I did find a company that has quite a few thermal imagers.
    It is my understanding that these sensors can ouput a couple different image formats over serial, spi, or i2c interface. The ones that caught my interest are the uncooled(thermal electric cooler-less) microbolometers. SOFRADIR- EC has an office in NJ with testing facilities, unlike ULIS which is based in France. I hope this is helpful!
    And sorry, as you may know this is my first post on this forum... well on any forum.

    SOFRADIR-EC website: http://www.sofradir-ec.com/
    ULIS-IR website: http://www.ulis-ir.com/
    SOFRADIR- EC uncooled microbolometers: http://www.sofradir-ec.com/products-uncooled.asp
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,694
    edited 2012-09-10 - 18:42:05
    TJ,

    Welcome to the forum! Are you affiliated with either of the companies you reference, by any chance?

    Interesting, but I did not see any pricing info. Any idea what those things cost ... with lenses?

    -Phil
  • TJHainesTJHaines Posts: 2
    edited 2012-09-11 - 06:30:36
    No, I am not affiliated. Just thought it would be something cool for a propeller project. Im pretty sure you can get a quote through the SOFRADIR-EC website. I wanted to try to get ahold of some of their imagers, but assumed that the only way to do so was to contact the company (which at that point I moved onto another project). Finding your original post about your thermal sensor project made me remember.
  • GarethGareth Posts: 278
    edited 2012-09-12 - 14:31:20
    That is one neat sensor you have cracked......
    Here is some Inspiration for you ...... as you mentioned overlays
    [video=youtube_share;pIb1scnD67o]
  • Bob Lawrence (VE1RLL)Bob Lawrence (VE1RLL) Posts: 1,638
    edited 2012-09-12 - 20:36:01
    Wow! great Phil. I'm in for one when their ready.. We'll have to start a new thread on unique ways to use it.

    re: A high speed I2C compatible digital interface
    Maybe you could try 128 of them on the I2C interface * $64.45 * = $ 8249.60 LOL
  • Mark_TMark_T Posts: 1,981
    edited 2012-09-16 - 05:32:29
    I presume people are aware of this thermal camera? http://www.cheap-thermocam.tk/
  • ElectricAyeElectricAye Posts: 4,561
    edited 2012-09-16 - 08:06:46
    Mark_T wrote: »
    I presume people are aware of this thermal camera? http://www.cheap-thermocam.tk/


    That's impressive. And just think: if the Ardweeny can provide results like that, just imagine what Phil is going to get his Prop-based system to do.
  • Bob Lawrence (VE1RLL)Bob Lawrence (VE1RLL) Posts: 1,638
    edited 2012-09-20 - 15:04:18
    @Phil Pilgrim (PhiPi)

    If it was in a water proof container could it work through a lens of some sort? i.e. under water inspection of an object.
Sign In or Register to comment.