Seek Thermal: Hardware modifications to improve image quality

[frenky]

Tnx for your input. You gave me a lot to think about and to try out.

[frenky]

Today I had some time to try the effect of Peltier element strapped to the back of the seek.
I polished the back of the seek to expose bare metal (magnesium alloy) for better contact. And I also opened it and stuffed 3 layer of thermal pads between the pcb and the housing.
All with the idea of cooling down the sensor and the shutter.

Peltier is TEC1-12706 (12V 60W). I was running it at 0.5A 2V (1W).

Seek + peltier + big heatsink:


Image of room corner (5*C temp diff) without peltier:


Image with peltier on:


As you can see the difference is huge. No wonder that old TI all had some sort of cooling system...

Just for fun 20 images stacked with ImageMagick "convert *.png -evaluate-sequence median OUT.png":


I'm pretty happy with the result. 

[tomas123]

great or better: cool

which temperature do you reach with the Peltier element?

[frenky]

According to this document http://www.hebeiltd.com.cn/peltier.datasheet/TEC1-12706.pdf I could push it to -41*C. 
Well I did not measure the temperature during the test but I did found out that even if I went to 1A with current the image did not get better (I think it was even worse).
I'll repeat the test soon... 

[Fraser]

There is an issue with this type of cooling....... condensation. You need to be careful about condensation on the SEEK camera's PCB.

Well done for doing this test though. As you say, cooling lowers noise in many electronics systems that deal with very small signal levels. The Agema Thermacam 400 runs the detector at -70C.

Fraser

[frenky]

I was thinking about condensation before experiment. My plan was not to go below 10*C, because that is dew point in my room (40% humidity, 25*C).



But when I started getting nice images I kind of forgot about it. 

I guess condensation is one of the reasons that some actively cooled sensors are in vacuum?

Hopefully condensation did not occur on sensor of my seek I'll make sure that I don't go below 10*C in the future tests...

I'll also try to find the highest (actively cooled) temperature that still improves seeks images...

[joe-c]

the Stacked image looks nearly the E4+ Images.
Really cool 

the "evaluate-sequence median" means Average right?

[frenky]

Yes kind of. If you have values 0,0,9,9,12 the average is 30/5=6. But median is 9 which is much better for noise reduction.

[frenky]

It seems that I may have destroyed my seek... 

The last thing I did was test the fit of custom made copper radiator that wold go on the back of the pcb...
Today I turned it on and the screen is solid orange with 300*C in the middle. I have no idea what happened to it...



I am really disappointed that I didn't get to try out my new setup with Peltier module and custom heatsink that fits pcb perfectly... and moved internal shutter in front of lens...

[frenky]

I'm one happy hippo. 


So what was the problem...
During my tests I have removed the internal shutter and I was shure that this will just distort output images a little as seen in Mike's video.
But during the power on the seek takes basic calibration frame with shutter covered. And without shutter this frame had really high values and software then pushed all real values to the 300*C.
Now I covered the lens and the shutter hole during the startup and the Seek works just fine. Image is distorted because there is no shutter to calibrate it abut it works!

I'm so happy.

[Fraser]

Excellent news Frenky 

Fraser

[IwuzBornanerd]

Has anyone who has cracked one of these Seek cameras open tried to find a way to reduce the random noise in the image via hardware changes such as better bypassing, rounding clock signals or anything like that?  There are noise patterns also, but lots of seemingly random crap, and maybe some "rolling through" noise.

[frenky]

On my to-do list was removal of R1 resistor and injecting 5V from 4xAA batteries to input of U1 regulator which supplies 2.8V to sensor.
This should remove most of the powerline noise...



EDIT: Perhaps you coluld even bypass U1 with two NiMh accus charged to 1.4V so you would supply 2.8V directly to sensor...

[joe-c]

This should remove most of the powerline noise...

it sounds really interesting. it would be great if there is a hardware way to reduce the noise.

good luck 

[frenky]

There is. Just strap a peltier module to the back of the seek and cool it to 15*C...
https://www.eevblog.com/forum/thermal-imaging/seek-thermal-hardware-modification-to-improve-image-quality/msg946675/#msg946675

[joe-c]

There is. Just strap a peltier module to the back of the seek and cool it to 15*C...

yes, i saw this images some weeks ago.

Maybe i should assemble the Board on a small Tec, and build a housing with a thermal Window and than use a vacuum pump to remove as much air as possible from inside...
i should go to bed, i am tired and have crazy ideas 

[frenky]

Good news (at least for me)... 

I had some spare time yesterday and I decided to try and fix my broken SeekThemal module.
A few freeze spray bursts and it came back to life. 
It seems that one of the solder joints on flash ic has cracked so I re-flowed four joints and now it's good as new.

So now I can go back to hacking the hell out of it because I don't care if I destroy it since I already have much better module in hand (i3 T.E.+).

[Fraser]

Great news 🙂👍

[joe-c]

gratz frenky.

I have seen, that the Seek hold the ADC Offsets as long as the device is Powered.
so if you change after warm-up to raw mode, you can use this mode next day if the Seek was just powered over the external Hub, while the PC is down.

If the TE works with the Planck formula, the Seek will be the next 

[Uho]

I did a small upgrade of seek thermal reveal . Added a telephoto lens and eyecup.
https://youtu.be/rt4EHTFfKVw

[meanie2]

Is anyone else still trying to improve the seek's image output with hardware mods?

[Ultrapurple]

I'm not sure whether work is still in progress anywhere but it does give interesting food for thought on other TICs. I'm thinking of trying something similar with one of my Therm-App cameras when I have time.

[Fraser]

If considering using a Peltier cooler on a small thermal camera a convenient TEC controller is the type commonly found cooling tuned Lasers. They are normally very compact and the temperature is set on a multi turn trim pot. Nice and portable 

Fraser

[IwuzBornanerd]

If one is stuck using the camera with a stupidphone the benefit of using a cooler will be limited to getting the camera down to the temperature at which Seek created the NUC table.  While cooling it down farther might get you less noise you will not likely see the improvement without having a NUC table created at that temperature, and the Seek app does not provide a way to do that.

I do not have any on-going effort to reduce the noise, but I am not done trying.  I have tried to figure out what Seek did to reduce the noise from the original atrocious level to it's current level but have had no success there.  There is at least one thing I want to try when I get around to it...but those round to-its are hard to come by. 

[Ultrapurple]

those round to-its are hard to come by. 

Nonsense, sir. You're simply spellling it incorrectly.





I simply asked my stupidphone to find me some. Perhaps my strupidphone is more stupid than your stupidphone...?

(quick help-ask: I tried to use the {img=200x200}file_url{/img} construct (but with square brackets of course) to size the above images, but to no avail. Is there a different function I should be using?)