Room temperature thermal imaging in SWIR

[GoodNUC]

Anyone ever see this before? This is room temp thermal imaging….in SWIR.

Last week I tested out a new lens on my Goldeye G033, an InGaAs camera , and was pushing its sensitivity to the max while thermally imaging some very hot water in a dark room.

As the water cooled, I upped the integration time, retaking dark frames along the way to compensate for the increase dark current.

Eventually it was imaging the water spitting distance from human body temperature so I decided to set the exposure to a value way higher than recommended and plopped myself in front of the camera in a pitch black room.

suddenly, something I’ve long wanted to see but haven’t seen anywhere, proved possible. Thermal images of ambient temperatures in shortwave.

This was a run of the mill spectral range InGaAs SWIR (0.94um-1.68um), not eSWIR or anything fancy, by the way. I’ve tested that with a monochrometer. And for this test I doubled checked that it wasn’t sensing longer wavelengths using an IR AR coated germanium window around 7mm thick. Couldn’t see a thing.

Closest I’ve seen to ambient thermal imaging in Swir used a 1-2.5um MCT cryocooled camera at -196C. That’s a lot longer wavelength than this topped out at. Also my camera was running at -14C.

What can you do at this short of a wavelength with thermal imaging? Well theoretically it’ll provide higher contrast than even MWIR, always, if you can get ample signal to noise. That’s just a consequence of spectrally integrating the Stephan Boltzmann law.

Camera: goldeye G033 TEC1, 640x512 15um pitch, -14C FPA with addition of a water cooling block I added.
Lens: SOLO50 F1.4 50mm from sensors unlimited

Since I definitively proved to myself swir cameras are room temp thermal cameras under the right conditions, I wanted to show that LWIR thermal cameras can “not” be room temp thermal cameras if they image mainly scattered light (in the same post). Something I rarely see as well, but is quite easy to show.

A 120C heating element did just the trick for this, hot enough to illuminate a scene with ample longwave, but not too much to cause surface heating from shorter wavelengths. A nearby fan helps with this since I still had to get real close to it for this to look nice.

The LWIR light overwhelms most of the thermal emission of an object revealing a completely invisible world that looks a lot more like SWIR aught to look at this temperature, rather than LWIR.

At 13000nm, water absorbs 99% of incident light in the first 25um or so, assuming no reflections. That’s roughly 200 times shorter than at even the strongest absorption band in swir (1450nm.) So in reflected LWIR, my entire eyeball’s surface is rendered a solid shade of black from its water content… wild.

LWIR images were captured with an HTI301 LWIR VOx 384x288 pix 17um pitch. “55mk” whatever that means in the context of the true noise specs of a camera. Nothing fancy, so feel free to give it a whirl on a higher resolution sensor and post!

[DaJMasta]

It's an interesting experiment with very low illumination imaging, but is it particularly meaningful to call it thermal imaging?  A blackbody radiator still radiates shorter wavelengths than the peak of the curve, so while the amount is negligible at shorter wavelengths when the temperature is low, with cooling of a sensor and long integration times, you can turn a negligible amount of photons into an image.  In the same vein, you could take a 50-60C heat source and use it as illumination in a completely dark space that with a cooled optical band camera and a long integration time, would give you an image of the scene.


In your camera, do you have pretty easy access to the back of the sensor itself, or are you cooling the outside chassis with the block?  Are you running it on an industrial chiller?  Have you considered just using a TEC and a heatsink/fan?

[GoodNUC]

Thank you, glad you found it interesting!

I think it is. It’s not any less meaningful by the definition of thermal imaging far as I can tell. It’s an image showing thermal emission of an object.

I typically think of thermal imaging as room temp thermal emission and it checks that box as well. If you could do that with an EMCCD in vis I’d have no problem calling that image a thermal image, but I would avoid calling any EMCCD a thermal camera.

Personally I find the whole distinction of there being a thermal camera that produces thermal images in the first place somewhat misleading. Seems more like marketing terms. Does an LWIR camera stop being a thermal camera if it images a scene too cold to emit light? What about if you can’t see thermal emission just scatter? No, but a NIR camera does? Kinda inconsistent but that’s just my opinion. It’s application specific.

There’s merit to letting consumers know what they’re buying. But at the same time, people might not know there’s more to see from the face value of that label alone if they are new to this. I imagine the people on EEV know better.

In this application it’s perhaps not practical for around the house imaging but is considerable if you’re doing long exposures in SWIR and assumed blackbody is a non factor at this temp, when it could be.

It’s just so cool to see some very ignored photons come to light in my opinion!

The water block is simply lassoed onto the side of the camera for now. I made a CAD model of a custom waterblock with 1/4-20s mounting threads that I debated CNC milling and swapping with the current side of the camera. Would look sleek, but isnt necessary, this method got the job done, and isn’t falling apart or cumbersome for the time being.

The water’s cooled through evaporative cooling via a swamp cooler for a mattress. Nice because it’s a passive means of insuring the body doesn’t get any cooler than the dew point, but at the same time is sub ambient. A lot cheaper than active electronics to do that, but there still might be a few more degrees of cooling I could get out of it with active cooling via TECs at the expense of added complexity and power draw. I want to run this off a battery remotely so power draw is a concern.

The sensor has a single stage TEC and a vacuum enclosure so I can get the FPA much colder than the body, I wouldn’t wanna open this guy up to change anything near the FPA considering it’s a $20,000 camera and space is at a premium in there.