A cheap infrared scanner can replace an expensive thermal-imaging camera
Images: David Schneider
TEMP WORK: Thermal portraits [top left] are fun, but the device also finds energy leaks: Cool air produces the blue in this scan [bottom left]. A professional imager shows the effect in greater detail [bottom right].
A recent energy audit at my place of worship indicated that the insulation in the walls and ceiling was pretty poor. That was obvious enough from casual inspection. But some subtle cracks revealed themselves only through the use of high technology—thermal imagery, which showed the locations of several air leaks in vivid detail.
Seeing those results, I was so impressed that I went off to the Web to check out thermal-imaging cameras, only to discover that they typically run several thousand dollars. Ouch. This type of imagery shouldn't be confused with run-of-the-mill infrared photography, which uses wavelengths only slightly longer than 0.75 micrometers or so. Thermal-imaging cameras sense much longer wavelengths, typically 8 µm or more.
The difference is critical: Unless you're taking a picture of something heated to the point of being almost red hot, infrared photography, like normal photography, requires that the object be illuminated. And all you see in the image is how much of the incident infrared light is reflected. Thermal imaging, on the other hand, senses the infrared radiation given off by everyday objects of modest temperature. So it can work stealthily in total darkness, which is why the military, which cares less about cost than I do, often uses it to sense enemy movements.
But my Web search also turned up a US $200 device capable of creating thermal images. Its resolution is lower than that of commercial thermal-imaging cameras, and each image takes much longer to obtain. Those seemed reasonable trade-offs. Of course, I'd have to build it myself.
The Cheap Thermocam is the brainchild of two 18-year-old students, Max Ritter and Mark Kohl, from Mindelheim, Germany. The project earned them an award in the 2010 Jugend Forscht, the largest science and technology competition for young people in Europe. Ritter has a website where he documents what is needed to build the device, including sources for virtually all the required components.
The strategy for reducing cost here is to avoid having to buy an imaging array of any sort. Instead, the thermocam uses what you might think of as a single-pixel infrared sensor: Melexis's MLX90614‑DCI (
$67 $52 from Future Electronics. Note: Buy only the DCI version, which has high sensitivity and a very narrow field of view). That sensor is attached to a simple pan-and-tilt mechanism, which does a line-by-line scan to produce an image. A laser pointer (also mounted on the pan-and-tilt) and a webcam affixed to the main enclosure allow the thermocam to generate a matching picture of the area scanned.
The project also includes an Arduino microcontroller, which handles low-level communication with the Melexis sensor and generates the necessary signals for the servomotors in the pan-and-tilt mechanism. A program written in Java does most of the computational heavy lifting. I ran it on a PC under Windows 7 and had no problem.
On his website, Ritter indicates that he will soon be selling a custom printed circuit board to replace the generic Arduino and enclosures to make construction that much easier. I simply purchased an Arduino Uno board ($30) along with a nice powder-coated metal enclosure for it (PRT-10033, $30 from SparkFun Electronics).
I departed from Ritter's design in other minor ways, too. I bought what I thought would be more robust servos (Hitec HS425BB, two for $13 each) and a pan-and-tilt mechanism (DDT500H, $25), all from Servo City. I got a laser module with digital control (COM-08654, $19) along with a plastic mount for it (COM-08674, $5), both from SparkFun, instead of the $8 module that Ritter had suggested, mostly because feedback on the SparkFun website indicated that the power wires on the cheaper module have a tendency to work loose.
Assembly took just a few hours, mostly spent modifying the metal enclosure so that it would hold the pan servo firmly in place and I could mount the whole thing securely to a tripod. The Java software worked straightaway, although the Arduino code took a bit of tweaking to get the thing to pan in the correct direction with my particular servos.
The one thing that put me on edge, though, was a secondary Arduino program that Ritter includes with his code distribution. This code issues a stern warning up front: "This program will change the EEPROM settings of your MLX90614‑DCI sensor to work best with the Cheap Thermocam. Please make sure you only use this with the DCI version, otherwise you will destroy your sensor." The word destroy certainly got my attention. And Melexis's documentation confirmed that it is perfectly possible to change certain settings that shouldn't be altered. I didn't want to run this program until I had figured out exactly what it did and until I had recorded the factory EEPROM settings on the sensor so that I could restore them.
Meanwhile, I was scanning various corners of my own home to check out what the Cheap Thermocam could do. Quite a lot, really. It could, for example, show where the weather stripping was leaking air on one side of my front door. Sure, my hand could sense the ingress of cold air there, too, but the thermal images also showed other, more subtle areas of heat loss, such as along baseboards or at electrical outlets.
To get a better sense of how valuable the Cheap Thermocam would be for tracking down such things, I contacted Mark Bashista, owner of Home Performance NC, a home-energy auditing business based in Pittsboro, N.C. He was kind enough to let me tag along on one of his energy audits, where we compared the Cheap Thermocam's images with ones from his $8000 FLIR B60 thermal-imaging camera.
"I was impressed by the contrasts [the Cheap Thermocam] showed, and the resolution isn't horrible," says Bashista. "But to have to use a tripod and to wait 3 minutes makes it impractical for a home-energy auditor." Bashista suggests that the unit might be more valuable just using the Java software's live-temperature function, which reports in real time the temperature of a single spot.
Bashista wasn't overly concerned about the limited resolution, because the images he obtains are often most useful not for discovering flaws, he says, but for communicating results to his clients. "It helps to explain what's going on."
Energy audits aside, the Cheap Thermocam has considerable novelty value, particularly if you use it to scan people. Forget Kirlian photography and energy auras. Now you've got a gizmo that can measure the actual energy people emit. You'll of course need to switch off the laser module so as not to risk eye damage, but it shouldn't be too hard to get things framed up without it. That the subject has to hold still for a few minutes during the scan demands a little patience, but I suspect it will only add to the sense of metaphysical intrigue.
Oh, and as for that second Arduino program? As it turns out, my concerns were unjustified: The new settings adjust the internal digital filters to reduce the sensor's time constant (which is initially set to about half a second) by a factor of 10 or so. The resulting images don't appear noticeably noisier and are much less blurred—just the ticket for a high-quality energy portrait.