[Project 326] has a cheap thermal camera that plugs into a smart phone. Sure they are handy, but what if you could hack one into a microscope with a resolution measured in microns? It is easier than you might think and you can see how in the video below.
Of course, microscopes need lenses, but glass doesn’t usually pass IR very well. This calls for lenses made of exotic material like germanium. One germanium lens gives some magnification. However, using a 3D printed holder, three lenses are in play, and the results are impressive.
The resolution is good enough to see the turns of wire in an incandescent light bulb. A decapsulated power transistor was interesting to view, too. Imaging heat at that much resolution gives you a lot of information. At the end, he teases that using first surface mirrors, he may show how to build an IR telescope as well.
Presumably, this will work with just about any IR camera if you adapt the lens holder. The unit in the video is a UNI-T UTi-260M. So when he says he spent about $35 on the build, that’s not including the $400 or so camera module.
IR imaging can pull off some amazing tricks, like looking inside an IC. If the thermal camera used in the video isn’t to your liking, there are plenty of others out there.
No need for devices under test to warm up until the false color view occasionally provides higher contrast.
With periodic pulsing and some post-processing, spatial resolution and contrast can be improved (lock-in thermography):
https://www.tek.com/en/blog/speeding-up-failure-analysis-using-lock-in-thermography
I’ve been wanting to try the lock in approach, I know the cell phone thermal cameras have a frame rate about 25Hz so could just sample lots of frames and cope with the jitter.
Does anyone know of any versions of these cheaper thermal cameras with a trigger input?
I liked this write up in case you didn’t see it here six months ago. https://hackaday.com/2024/10/28/lock-in-thermography-on-a-cheap-ir-camera/
Use a camera, or phone, stand to get video of the thermal image. Dont need to deal w/ jitters.
“microns” what like 800 to 900? Not entirely sure how much use a thermal camera with a macro lense outside of “gee whiz”
I have found the P2 Pro macro lens to be very useful in identifying e.g. overheating SMD resistors.
I used that camera with macro lens to identify a bad LED (of 250 in parallel).
https://hackaday.io/project/196782/instructions
The UNI-T Uti260M camera module is still expensive in my book (if you don’t have funding from a consulting project), but its retail cost is closer to $200-ish, not $400 (at least in the US). Electronics are generally more expensive in the UK and EU due to high VAT (value-added taxes) that consumers have to pay, which is why Europeans love to combine travel to the US with shopping sprees for expensive camera equipment, etc. What I don’t know is how modest, direct-from-China online purchase are taxed in VAT countries.
Actually, shopping abroad is now considered illegal in most EU nations unless you apply for a vat extemption. I wouldn’t do it unless your looking for some happy fun time in prison shower with Mr. Bear and his 10 inch friend.
And paying vat and possibly a customs fee makes it all legal again. Some larger sellers outside of Europe even have those parts automated so the customer doesn’t get multiple invoices.
Can you provide the legal support for your comment?
If you give an honest customs form with everything you bring in, they usually let you keep the items if you pay the tariffs, taxes, or VAT that is due for importation. This is how stuff is imported period. People import through customs, pay whatever, then sell to consumer.
Great video, definitely looking forward to seeing the telescope video. One question: Where the glass slide almost completely blocked the thermal imaging the glass around the lightbulb did not. Any idea why this is the case?
Halogen bulbs are often made of quartz glass (almost pure SiO2) which is transparent to lot of the infrared and ultraviolet range.
‘Posh’ says halogen lamp which implies quartz, which is reasonably transparent to IR.
We evaluated this approach in the lab about a decade ago and for simple uses (industrial detection of material failure in our case) it worked pretty well and was very cheap even then (at the low-cost end we were using GridEye 8 x 8 sensors so the resolution would come from the optics rather than the sensor).
Several lenses add to the magnification of course , but the most annoying thing was getting the assemblies focused since the index of refraction is different and the lenses are opaque.
Lepton series FLIR sensors and boards are available for “not-too-much” these days , as well as secondhand handhelds, so it could be better now. USB phone modules available from the usual sources seem likely to work well though the data would have to be parsed since they’re usually bundled with a proprietary app.
There’s python scripts (and I believe ffmpeg) can directly parse the video feeds from those chinese usb thermals.