SWIR Images - InGaAs Linescan Camera

[Andy Perrin]

Stefano, that's actually making an incorrect deduction -- microbolometers have a different operating principle than semiconductor detectors, so cannot be directly compared. Microbolometers have a suspended IR-absorbing material that is "floating" over the substrate to thermally isolate it:

The circuitry in the substrate records the temperature from a thermister.

 

This is compared to the usual silicon or InGaAs sensors which generate images by some kind of semiconductor junction which generates a current like a solar cell when light is absorbed and causes an electron to jump from valence band to conduction band.

 

Credit to https://en.wikipedia.org/wiki/Microbolometer for the picture.

 

 

[Stefano]

Yes, that's quite amazing. It's true, they are two completely different technologies, but I wonder why your SWIR sensor needs cooling. As I understand, uncooled thermal cameras have lower image quality than cooled ones because the sensor can pick up its own blackbody radiation, but this doesn't happen at SWIR wavelengths. I am sure the topic is much more complex, but I don't think SWIR photodiodes need cooling in general. Does your MWIR camera use a cooled sensor?

[Andy Perrin]

I think the MWIR one does have cooling? Not 100% sure on that. 

 

Quote

e lower image quality than cooled ones because the sensor can pick up its own blackbody radiation, but this doesn't happen at SWIR wavelengths.

Incorrect! It even happens in visible light. Blackbody radiation is the same conceptually as the main source of thermal noise in electronic circuits (Johnson-Nyquist noise). It leads to "amp glow" in ordinary cameras when the temperature is too high, from taking many photos in a short time for example. It comes from electronic fluctuations in the circuitry, not just external blackbody radiation.

 

Anyway, as for why my Triwave needs such a lot of cooling, I think it comes down to the signal being very small, so the noise must be reduced a lot to become smaller than the signal. Again, I think my Triwave was a prototype and probably this was an issue they fixed eventually.

[Stefano]

Thanks.

[Daniel Csati]

Regarding the -80C Triwave I found some further good reading. Apparently the high dark current is intrinsic property of Ge-Si direct bond. -80C was always necessary for this to work. 

See the discussion in comments:

2010

2015

[Andy Perrin]

Wow, well that does shed some light on why they went under. 

[Daniel Csati]

Here is a little insight how the recording works with my camera:

 

[lukaszgryglicki]

Great video - thanks for sharing - seems like you had a lot of fun with it :)

 

[Daniel Csati]

6 minutes ago, lukaszgryglicki said:

Great video - thanks for sharing - seems like you had a lot of fun with it :)

 

Haha thanks, sure, a LOT of fun xD

[colinbm]

Thanks for showing us the video, you have put in a lot of work to get this far.
Can you see at night time with moon light ?

[Andy Perrin]

My cost for the Triwave was very similar - it was $3000. It’s analog but the frame grabber was cheap ($0 for the one I used originally because I already owned it for watching over-the-air TV…). The software for the Triwave is clearly not for consumers. It is similar to what you have, it sends serial commands. It does have sliders for gain and exposure though so I don't have to do it numerically. 
 

I have seen InGaAs 320x240 area cameras for around $5000 on ebay. That might be another option for people. 

[Daniel Csati]

2 hours ago, colinbm said:

Can you see at night time with moon light ?

Hi Colin, I tried without moon (moon behind mountain) up to 11ms exposure time and maximum digital gain but the image was just noice. I'll give it a go some other time with longer exposure and maybe lower sensor temperature. I really expected to see something but I'm not giving up just yet :)

[Andy Perrin]

I have so much light pollution here in Boston that I don’t think I would be able to distinguish nightglow from the rest of it. 

[Daniel Csati]

Another SWIR, this time a video. The magic of line scan cameras!! Enjoy

 

 

[Andy Perrin]

You know, they tell me that cars are more compact in Europe, but I had no idea! 

[Daniel Csati]

I admit, the view is not ideal from my window but next time I will record it from the sidewalk :)

The camera can't handle this heat wave anyway :)

[Daniel Csati]

16 minutes ago, Andy Perrin said:

You know, they tell me that cars are more compact in Europe, but I had no idea! 

Oh so you really mean it. :) There is a lot of compact cars, that's true. When I was once in the US, it didn't seem that much different to me - but I remember definitely more trucks than here

[Andy Perrin]

No, it's (supposedly) a true fact but I was joking about the foreshortening effect caused by the camera scan rate. (In the time it takes to scan one vertical line of pixels, the car moves several pixels horizontally, that is the cause of the effect, not your angle.)

 

A lot of American appliances etc. tend to be physically larger than European ones, according to friends. I have not been there so all this is second-hand knowledge.

[lukaszgryglicki]

I definitely see the difference between average cars in US and EU. I'm usually 2-3 times/year in US.

 

[dabateman]

Looks like a video game to me.

Did the frog make it to the other side?

[colinbm]

Thanks Daniel, I hope you get a result next time you try with the moon.
The second video is funny.

[Daniel Csati]

SWIR Portrait with 1575nm bandpass filter. I tried to do this inside using an incandescent bulb and scattered sunlight but it needed a very long exposure time. That explains then why I didn't see anything in the night without light. 

 

 

My girlfriend has blond hair, interesting that it doesn't look different at all in SWIR

 

[colinbm]

These look good.
The Moon reflects the Sun's light, but not SWIR wavelengths ?

[Daniel Csati]

2 minutes ago, colinbm said:

The Moon reflects the Sun's light, but not SWIR wavelengths ?

Andy's moon photos are proof that it reflects SWIR too, but the amount of light is just not enough. I could probably see something if I set the exposure time to 1s but an image with 1840x lines would take then 1840s. :/ 

[colinbm]

Yes, I see the problem now, thanks.