Instructables
Picture of Thermal Camera
Have you ever had the desire to see what can not be seen? I never had the ability to see in thermal infrared, and it is rather cool to be able to see in the thermal infrared part of the electromagnetic spectrum. Unfortunately, most thermal (FLIR) cameras are very expensive. In 2011 cameras frequently cost several thousand dollars. This project was created to develop a cheaper solution at a fraction of the cost.

The primary reason for the cost difference is the thermal infrared sensor. Instead of having a 2 dimensional array of sensors, a single sensor is moved in a raster scan pattern. Moving this single sensor takes time. So while you have fewer costly sensors, it takes longer to take a picture. This means the camera can not take images of moving objects, and the objects have to be very still during the scanning.

The original instructions can be found on my web site http://www.centralnexus.com/thermal/. These instructions require some soldering ability. It can take most of your weekend to assemble this camera once you have the parts. If you're good with soldering you may be able to assemble it faster.
 
Remove these adsRemove these ads by Signing Up

Step 1: Get The Parts

Picture of Get The Parts
The following are the parts used in creating this camera. The links to the web sites that carry these parts can be found on my web site. It's been a while since these parts were ordered, and you may need to find other web sites with similar parts if a part is no longer available. It's highly recommended that you do not substitute the MLX90614ESF-DCI sensor because it's the most accurate with the narrowest field of view.

    Arduino Project Enclosure
    2x Servo - Small
    Pan/Tilt Bracket
    Arduino Uno or an Arduino variant
    Logic Level Converter needed when using a 5 volt version of the Arduino, which is the most common version.
    Break Away Headers - Straight
    Break Away Male Headers - Right Angle
    Microsoft LifeCam VX-700
    Protoshield PCB for Arduino. Almost any Arduino prototype shield will do.
    MLX90614ESF-DCI. This model has the best accuracy and narrowest field of view.
    About a foot of ribbon cable. Almost any wire will do. At least 5 wires are needed to connect the sensor to the Arduino and laser.
    .100 KK Housings & Terminals 6 Cir. (2.54mm) for the ribbon cable.
    PCB board for holding the laser and sensor.
    A switch to turn the laser on or off.
    A laser for aiming the sensor and aligning the pictures.
    Small nuts and bolts (#2-56 x 1/4") to attach the servos.

The following additional parts and tools are used in creating this camera.

    Soldering iron
    Solder
    Wire cutters
    A standard drill or a Dremel drill for creating holes in the case
PureE made it!16 days ago

you have to switch to the flir lepton module, check it out here

http://www.pureengineering.com/projects/lepton

I made a small breakout board for it.

IMG_20140801_165701~2.jpg
Black_Diamond (author)  PureE16 days ago

There's 2 problems with the Flir Lepton. The bare module seems to be more expensive than my entire camera, and it's difficult to buy just one module. Once you get up to that price, you might as well consider buying a Flir One instead. That recently released camera probably uses the Flir Lepton as the sensor, but it works out of the box for a little more money. The downside to the Flir One is that all the photos get the annoying Flir logo. An advantage to those Flir modules and cameras are that they create a photo much quicker but at the cost of your wallet. So the trade off still exists. The trade off gap just isn't so large as it use to be.

I'm sure my comments will be out of date in a few years as this technology progresses.

Also you can now buy single lepton modules from this site:

https://groupgets.com/campaigns/27-flir-lepton-the...

You are right that it is more expensive, but you do get a better sensor. And the relative cost once everything is put together is not much more. Especially when compared to the 2-10k FLIR systems.
Also the lepton sensor outputs the raw data, there is no Flir logo or anything like that.
Also from a size point of view the lepton is much smaller so you can apply it to more applications.

I think if you put the lepton on your pan/tilt setup you could get some really high resolution photos from it as well.
Just wanted to make you aware of the module.
turbonegro1 year ago
Very nice
I absolutely love thermal cams
Lots of people dont seem to grasp how useful they can be
Wish i had the time to build something like this
The soldering would not pose a problem i do that for a living
Considered a kit ?
I cant wait for the day when cellphones will have this ;-)
I'll just leave this here... =) http://www.muoptics.com/

These geeks at mu optics never delivered. They took the money and now seem to have disappreared

A Dream come true (or rather just spawning)
THX for the link x 1000
Will look into it
hfb turbonegro1 year ago
If you want a kit, you can go to the original cheap, thermal cam site where it all started.
http://www.cheap-thermocam.net/
Max is a great guy and has been very helpful when I made my own V1 about a year ago. Here's how I did mine.
http://amazingrcvideos.blogspot.com/2012/08/my-own-thermal-camera.html
Black_Diamond (author)  hfb1 year ago
I didn't notice that he updated his design recently. He still has the v1 software that we both worked on. He has a couple of pictures of my home in his thermal image gallery.
Yeah, the design looks better than before and more portable. I actually ended up backing a Kickstarter project, IR-Blue, that used the 16x4 array that connects via Bluetooth to a smartphone and it works fairly well. It's not the higher res images you get from this thermal camera we've made but it's essentially instantaneous and useful.
Black_Diamond (author)  mark4291 year ago
It looks like their camera is getting into the affordable range with $400 MSRP. It's still more expensive than this one, but I'm sure it's also faster.
Black_Diamond (author)  turbonegro1 year ago
I'm not entrepreneurial. So I haven't created a kit. Some people have asked me to make them one too, and they offered to pay for parts and labor. I declined the work though. I was just doing this for fun and not for money. I'm more of a software person than a hardware soldering person.
Sorry that was in response to turbo since they were looking for a kit/product to buy. I in no way meant to diminish your project which I find way more impressive. I'm still trying to figure out what to do with my uno... I think your project is awesome! I took a risk and backed the muoptics imager on indiegogo for a song compared to the retail price, I think it was $150 or so... You still need a phone to use it so most of he heavy lifting is done on your phone... Sorry for the rant Awesome work! - mark429
hilukasz1 year ago
very cool!
Have you looked into a thermopile array? They cost a bit more, but will greatly improve scan time.
More sensors mean a lot more cost and more complexity in aiming and wiring. You can buy the parts for this camera for about $150 + tax & shipping. I wanted to keep this under a $200 budget. 4 of these sensors would cost over $200 alone, and it would blow up my budget. I can't use generic thermopiles because I also need a narrow field of view with an easy way to accurately query the temperature. It's all about trade-offs. :-)
It looks like the MlX90614ESF-DCI has a collimator attached, correct? It seems unfair that they would charge almost 4x the price for such a simple modification. Looking at the datasheet it appears that it also uses an inferior sensor as well (compared to the AAA model). The settling time and noise are much worse. It might be worth it to just get the bare sensor and make my own collimator.
If you can find a cheap, portable, easily available and accurate alternative, I'd be interested in seeing the design. The settling time is modifiable in the sensor. The timing is an area that I'd like to tweak in the future. I need to slow the servos down near high temperature changes, but I can go faster near similar temperatures. The noise may be due to the field of view that it has to be sensitive to. There is less area to sample with a narrow field of view.
Very interesting.... I already know exactly how I shall spin this into an entirely new application.
kmitchell51 year ago
That's a super instructable. I always wanted one to use on our home. Thanks for sharing your knowledge.