Flir E4 Thermal imaging camera teardown

[jquinn11733]

FYI.... bought four of the E4 units for academic use (chemistry lab-class),
and just did the upgrade.  Thanks!
Model: E4 1.1
Serial: 63909758
Serial: 63909748
Serial: 63909700
Serial: 63909711
Part#63901-0101
Software 1.19.8
Calibration Date: 12/9/2013  (all four)

Big change from the my Tech-Jam TJ-200 which was $4K for 8x8 pixels,
though it does go to higher temperature (1000C).

thanks and regards,
- Jim

[tomas123]

only for information:

I found here this values of the noise generator.
I think the unit of targetNoiseMk is [milliKelvin]

Flir E30B https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg335974/#msg335974
.caps.config.image.targetNoise.targetNoiseMk int32 90
Datasheet Thermal sensitivity: 75 mK

Flir E40 https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg362889/#msg362889
.caps.config.image.targetNoise.targetNoiseMk int32 60
Datasheet Thermal sensitivity: 75 mK

Flir E60 https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg335988/#msg335988
.caps.config.image.targetNoise.targetNoiseMk int32 90 (it's true?? taucher)
Datasheet Thermal sensitivity: 50 mK

Flir E4 https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg313584/#msg313584
.caps.config.image.targetNoise.targetNoiseMk int32 135
Datasheet thermal sensitivity: 150 mK

you can self measure your real thermal sensitivity - see my post here

a compare of Noise Measuring in Service Menu
http://192.168.0.2/service/Diag/NoiseMeasure.asp

new Flir E4 with temp. range -20 to +250 °C
-> Pixel Noise  17.86 mK  =  1.66 digits
...

old Flir E40 with temp range -20°C bis +120 °C
-> Pixel Noise:   28.96mK =  5.73 Digits
...

and here

targetNoiseMk is a noise generator in mK (conforming with NETD/ thermal sensity in Flir Datasheets of the selected cam)
...

http://gs.flir.com/surveillance-products/surveillance-technology/imaging-technotes/IR_Technology_Parameters
Uncooled infrared cameras systems are typically a little noisier, in the range of 30 - 120mK. Noise in an image can be spatial or temporal.
Spatial noise is noise across the image at any given point in time. It is perceived as an unchanging fixed pattern on top of the image.
Temporal noise is noise at any point in the image over time. It is perceived as the static that moves in an image.
NETD is typically the measure of both these noise types.

...

[Fraser]

I do not pretend to understand the noise value used on these cameras. Even less so after seeing the Exx series values !. I remain unclear on whether the value is noise injection or noise compensation  If anyone understands this and can put it into plain English I would welcome such a post. I thought I saw '5 'as a value used in the E4 but that may have been something different.

[tomas123]

sorry for my bad english 

Flir adds additional synthetics noise to the sensor natural noise to reach the (bad) NETD, which specified in Flir data sheet

The Flir E4 uses a targetNoiseMk of 135 mK.
The hacked E4 config file reduce this value to 5.

Here I wrote about the differences of measured noise with targetNoiseMk on/off:
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg342315/#msg342315

[Fraser]

tomas123

There is nothing wrong with your English 

When I said please explain in plain English, in the UK that means in simple terms so that novices, like me, can understand    Sorry if it sounded like any form of criticism ... it was not intended 

[tomas123]

macro without macro lens

Testo has a feature called SuperResolution, to stack multiple images for better resolution

I tried the image stacking with some Flir Images.

As sample a took some images from a raspberry pi ( the dimension you see here http://www.pyrosoft.co.uk/blog/2012/01/15/raspberry-pi-measurements-dimensions/ )

unfortunately the lens of flir E40 at macro distances (0,4m) is only sharp in the center

connect with telnet to shell of Flir E40 camera and take a RAW sequence of 90 frames
I used a tripod and moved the cam a little while recording.

Code: [Select]

rset .image.services.rtrecord.filename \StorageCard\001.seq
rset .image.services.rtrecord.count 90
rset .image.services.rtrecord.frequency 30
rset .image.services.rtrecord.active true
delay 5
rset .image.services.rtrecord.store true
now we have the raw video file 001.seq

we can split the sequence in 90 images ( seq101.fff to seq190.fff) with this perl script
see my post here: http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,5279.msg25567.html#msg25567

Code: [Select]

$ perl -f split.pl 001.seq 

$ cat split.pl
#!/usr/bin/perl
undef $/;
$_ = <>;
$n = 100;
# FLIR camera E40         
$pat="\x46\x46\x46\x00\x00\x00";
# Flir Tools (comment out)
#$pat = "\x46\x46\x46\x00\x43\x41\x50";
for $content (split(/(?=$pat)/)) {
        open(OUT, ">seq" . ++$n . ".fff");
        binmode OUT;
        print OUT $content;
        close(OUT);
}
as next step  convert this 90 images with exiftool/imagemagick to 16 bit png images

Code: [Select]

for i in *.fff; do exiftool -b -RawThermalImage $i | convert - -auto-level -resize 300% _$i.png ; done
now stack all png images with free tool Registax  http://www.astronomie.be/registax/ and save as 16 bit tif image
 
compare the images

a single frame

Code: [Select]

$ exiftool -b -RawThermalImage seq101.fff | convert - -auto-level resize 640x -clut single.jpg

the LQFP64 chip size in the middle is 8x8mm

a single frame sharpen

Code: [Select]

$ exiftool -b -RawThermalImage seq101.fff | convert - -auto-level -resize 640x ../iron-pal.png -sharpen 0,3 -clut single-sharp.jpg

and the stacked image of 90 frames

Code: [Select]

$ convert registax.tif -resize 640x ../iron-pal.png -clut stack.jpg

and now a (over) sharpen stacked image at 300%

look at the vertical chip labeling SMSC®
LQFP64 pin-distance: 0.5mm (16 pins at 8 mm)

[wilheldp]

Um...bad things are happening.

I got the resolution upgraded.  Went to install the BETA3 menu.  Temp enabled RNDIS, FTP'd in, backed up my FlashBFS folder, copied in the new FlashBFS files, and did a hard restart.  The Special Edition logo screen came up on restart, but when I pressed the menu button, I got an Application Error warning.  I did another hard restart.  The camera is now functional and takes pictures, but I cannot access the menu at all. 

I tried to temp enable RNDIS again, but without the menu, I can't get the camera into FTP-accessible mode (since I can't access the info screen).  Please help!

[Taucher]

Um...bad things are happening.

I got the resolution upgraded.  Went to install the BETA3 menu.  Temp enabled RNDIS, FTP'd in, backed up my FlashBFS folder, copied in the new FlashBFS files, and did a hard restart.  The Special Edition logo screen came up on restart, but when I pressed the menu button, I got an Application Error warning.  I did another hard restart.  The camera is now functional and takes pictures, but I cannot access the menu at all. 

I tried to temp enable RNDIS again, but without the menu, I can't get the camera into FTP-accessible mode (since I can't access the info screen).  Please help!

I wonder if people startet appending instead of overwriting the files or if they did not enable all the features...
There's a reason I said "advanced users" (hanging "advanced" pretty low btw) *sigh*

Try FTP while the camera is running - if that fails too, then try applying the normal flir "upgrade" or create a net_install package with the patched files and use the flir tools to upload the hack - that one should always work unless you really screwed things up (like hurting appcore.exe)

[mikeselectricstuff]

Um...bad things are happening.

I got the resolution upgraded.  Went to install the BETA3 menu.  Temp enabled RNDIS, FTP'd in, backed up my FlashBFS folder, copied in the new FlashBFS files, and did a hard restart.  The Special Edition logo screen came up on restart, but when I pressed the menu button, I got an Application Error warning.  I did another hard restart.  The camera is now functional and takes pictures, but I cannot access the menu at all. 

I tried to temp enable RNDIS again, but without the menu, I can't get the camera into FTP-accessible mode (since I can't access the info screen).  Please help!

It is possible that you have made an incorrect CRC01 of e8.cfg

That would just make it revert to E4
I've seen the very occasional menu/app crash but sorted itself after the next hard reset.
Even if the menu crashes, I would think the .fif option to enable RNDIS would work.
If the .fif installer works, then you should be able to make an install file to rewrite the factory default files.
If all else fails there's the serial console.

[Mavro]

Flir announced at CES an iPhone case/Thermal camera that will cost $350.

http://gizmodo.com/flir-just-turned-your-iphone-5-into-a-predator-like-the-1496554213

[Fraser]

Indeed. I would like to see the thermal imaging engine in that unit....it is so slim-line !

Sadly this gives the guys at Mu Optics a plausible excuse for their product failing.

[wilheldp]

Color me stupid.  I will accept all insults hurled.  I made a number of errors, and luckily didn't break anything permanently.

First off, the first time I tried to copy the new files over, I *did* choose overwrite BUT I had the damned INFO screen open at the time.  That made two copies of the menu files.  The reason I got the application error is that it had two copies of all 3 menu files and didn't know which one to use.

The second error occurred when I was trying to FTP back in to fix my first error.  I left FLIRInstall open while I was trying to access the camera by FTP.  I'm pretty sure that Install program hijacks the FTP connection so that you cannot have another active connection to copy the files over.

I got back in there and straightened everything out.  Thank you guys for the quick troubleshooting efforts.  I appreciate the effort that you have put forth in developing these hacks.  Even if I did brick my camera, I wouldn't have blamed you guys in any way.  I knew it was for use at my own risk.

[Radget]

Open question: Is it hackable? (Isn't that what we all are thinking?)

Flir announced at CES an iPhone case/Thermal camera that will cost $350.

http://gizmodo.com/flir-just-turned-your-iphone-5-into-a-predator-like-the-1496554213

[daves]

Open question: Is it hackable? (Isn't that what we all are thinking?)

Flir announced at CES an iPhone case/Thermal camera that will cost $350.

http://gizmodo.com/flir-just-turned-your-iphone-5-into-a-predator-like-the-1496554213

There must be some "SNAG", otherwise no one would buy any FLIR product anymore. Why would you pay 6000$ or more, when this crap can do 640x480 for $350. So, maybe they presented 640x480 on CES, but for that $350 you can buy "commercial" version of it, with 80x80 @ 1fps version... hope...

[mikeselectricstuff]

Open question: Is it hackable? (Isn't that what we all are thinking?)

Flir announced at CES an iPhone case/Thermal camera that will cost $350.

http://gizmodo.com/flir-just-turned-your-iphone-5-into-a-predator-like-the-1496554213

There must be some "SNAG", otherwise no one would buy any FLIR product anymore. Why would you pay 6000$ or more, when this crap can do 640x480 for $350. So, maybe they presented 640x480 on CES, but for that $350 you can buy 80x80 @ 1fps version... hope...

They do not quote the IR resolution anywhere - only the visible  camera.

[gabriel.klein]

They quote it here:
http://gizmodo.com/your-iphone-could-soon-get-predator-vision-1177511480

But it's the resolution of the "normal" camera, not the infrared one!

From what I see the resolution seems to be 60x40, something like that.

What could be interesting is that it's easy to write an application to create panorama...

[wilheldp]

Sc_Logo_FlirHardEdges.png <- that file got edited (into totally transparend) in the hack's RCC archive

Did you make other changes to the Facet RCC archive?  If not, couldn't you just replaced the hacked version with the original version of that particular RCC archive to get the logo back?

[daves]

Who knows. Maybe the photo is fake, I mean the resolution is pretty good, I have measured 220x160 just from that photo:


I saw some photos from CES of real item and the resolution seems much worse.

[gabriel.klein]

http://flir.com/flirone/#case-4
The picture is not that good on this page...

[Fraser]

@daves

Your first picture looks like a different camera model or a prototype.

As has been said before, images on the adverts can be a little misleading and there is usually a caveat warning that it is a simulated image. The image content suggests that FLIR are aiming this unit at gimmick addicts and those who wish to look 'cool' in front of the ladies. A toy by any other name.

[Taucher]

Sc_Logo_FlirHardEdges.png <- that file got edited (into totally transparend) in the hack's RCC archive

Did you make other changes to the Facet RCC archive?  If not, couldn't you just replaced the hacked version with the original version of that particular RCC archive to get the logo back?

You could, but your menu would lack lots of icons for the new stuff then as I merged Exx and Ex together in order to get the missing icons

[wilheldp]

Sc_Logo_FlirHardEdges.png <- that file got edited (into totally transparend) in the hack's RCC archive

Did you make other changes to the Facet RCC archive?  If not, couldn't you just replaced the hacked version with the original version of that particular RCC archive to get the logo back?

You could, but your menu would lack lots of icons for the new stuff then as I merged Exx and Ex together in order to get the missing icons

Gotcha.  We've already established that I lack the expertise to do the more complicated hacks (since I failed at the easy ones).  I didn't know what else was in that file.  I like the FLIR logo on the images, but I will certainly deal with it not being there to have icons for all the new menu items.

[Taucher]

sorry for my bad english 

Flir adds additional synthetics noise to the sensor natural noise to reach the (bad) NETD, which specified in Flir data sheet

The Flir E4 uses a targetNoiseMk of 135 mK.
The hacked E4 config file reduce this value to 5.

Here I wrote about the differences of measured noise with targetNoiseMk on/off:
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg342315/#msg342315

I completely turned off noise in my config from day one on ...
Your quote "after setting targetNoiseMk to zero, stacking of images don't improve results" confirms this step

[mrflibble]

Your quote "after setting targetNoiseMk to zero, stacking of images don't improve results" confirms this step

Yup, tomas123 did a pretty good job analyzing the noise. Still have a few tests for the temporal noise on my TODO list, but so much stuff, so little time.

[vapid2323]

So I am kinda a security nut, I wanted to see some folks thoughts on incorporating this into a security system. I would just want a warning triggered on movement and then if the movement continues map it against some thermal profile to hopefully rule out small critters. I saw someone post a histogram based on the thermal profile and that might do the trick

My camera is not here yet but, am I crazy or do you guys think this is within reach?