Flir E4 Thermal imaging camera teardown

[tomas123]

After dis- and reassembling my E4 to clean my dusty sensor

Can you please give as a dusty thermal image?

[nersut]

if anyone is still afraid to permanently hack his\her i3\i5 here is simple step by step guide

read post #524 https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg321956/#msg321956

1. power on camera

2. press menu button once

3. press play (gallery) button for 5 seconds

4. switch to RNDIS usb mode

5. install flir drivers on your pc

6. connect camera to pc, give it a minute to install and start

7. connect to camera's ftp server at 192.168.0.2 user:flir password:3vlig with ftp client capable of passive mode i.e. total commander

8. make full backup of filesystem to a secure location

9. download mike's crc01 tool from this https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg321956/#msg321956 post

10. install some proper text editor i.e. notepad++ (and enable showing all characters in view \ show symbols)

11. open backup\FlashFS\system\appcore.d\config.d\conf.cfg

12. remove last line that says

Code: [Select]

# CRC01 ????????

13. replace number at the end of   ||.caps.config.image.settings.resIR int32 X||   with value from   ||.caps.config.image.settings.fpgaResolution.resIR int32 X||*

14. save changes as new file

15. use crc01 to calculate checksum of this file

16. open backup\FlashFS\system\appcore.d\config.d\conf.cfg file again, do the same resolution modification and replace checksum with calculated one

17. save changes as new file that you will upload to camera (in extra step i have changed attributes and date & time of new file to be the same as old one, but this is most likely unnecesary)

18. connect to ftp again, and upload created file overwriting one that sits in \FlashFS\system\appcore.d\config.d\conf.cfg (or check mike's post just under this one and let us know)

19. unmount & disconnect camera, use reset button in battery compartment to reboot it

it worked for me, i don't take any responsibility for anything but merely sharing what i did described as good as i could, do it at your own risk


*

Code: [Select]

* Z-cam has a detector with 80x80, 120x120 or 144x144 pixels.

Thank you very much!

I managed to permanently hack my FLIR i3 to a FLIR i7, using the step-by-step guide.
The thermal images looks much better now, from 60x60 (3600) pixels to 144x144 (20736) pixels, that's 5.76 more resolution!  for free!

The installed firmware is v6.5.3 (newest version)

Used apps:

• FLIR device drivers
• Total Commander v8.01 x64
• Notepad++ v6.5.2
• FileDate Changer v1.1
• Mike's CRC01 tool

Thank you again! 

[Fraser]

The Patent document relates to the use of the thermal image in a projection system. I belivee that it is the projection of the thermal image onto the target surface that is the Patented idea and not the MSX functionality.

[tomas123]

from 60x60 (3600) pixels to 144x144 (20736) pixels, that's 5.76 more resolution! 

the flir hack is not new, but nobody published the hack yet (avoid closing the vulnerability)
I hacked my Flir E40 a year ago, after I read something about the the Flir i3/i5/i7 sensor (documents from year 2010)

The Flir i3/i5/i7 sensor size is 320x240 pixel
see http://www.i-micronews.com/reports/FLIR-ISC0601B-Micro-bolometer/1/297/

and page 3 in  www.i-micronews.com/upload/Micronews/Micronews_90.pdf

nice blog http://blog.naver.com/PostView.nhn?blogId=sgyee&logNo=10082558135


http://www.systemplus.fr/wp-content/uploads/2013/04/MEMSTrends_2012_10_N12_systemplus.pdf

The sensor used in the i3 has 20 times more pixels than necessary, and four times more pixels than the i7. Incidentally, the same sensor can be used in more expensive cameras.
...
Each pixel is composed of a thin resistance in vanadium oxide (VOx) covered with an absorber in silicon oxide which absorbs the infrared ray. The temperature of the SiO2 layer is proportional to the temperature of the IR radiation, and the value of the resistance VOx is proportional to the temperature of the absorber. A measurement of the resistance provides the temperature.

Have you tried the full resolution 320x240 pixel (or square 240x240)?
What resolution you get over the service menu?
see here
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg316173/#msg316173

flir published  the root password in may 2010 ;-)
see 1.3.3.2 in http://support.flir.com/DocDownload/Assets/75/English/T559504$A.pdf

PS: mike knew about flir i3/i5/7 sensor in August 2013 
https://www.eevblog.com/forum/crowd-funded-projects/m-thermal-imager-real-or-fake/msg281507/#msg281507

[Fraser]

OK, now here is an ugly thought for those who adjust their lens focus using a lens tool.

Mrflibble is doing some great work analysing anomalies in the E4 image maps. At the moment it is appearing that the lens used in the camera has some bearing on the contents of the maps and distortions are being observed. What is worrying is that some E4's have significant anomalies and others very little, if any evident. Consider this hypothetical situation for a moment :

From production to use:

1. E4 camera fitted with a Chalcogenide glass lens that is not perfect and contains random optical errors (not confirmed as fact but something is showing up in the maps suggesting this may be the case)
2. E4 lens is set to correct focus point and 'locked-off'
3. E4 camera is calibrated and correction tables produced to compensate for physical 'defects' in the sensor AND OPTICAL BLOCK.
4. The lens optical errors are captured and countered in software ? (unconfirmed)
5. As noted the errors in the lens may be random in nature. The user rotates the lens to adjust focus. All calibration data for lens defect correction becomes invalid as defects have now moved !
6. The result of lens defects moving and the cameras attempt to counter the defects in software tables that are static could cause interesting effects ?

This is all circumstantial and I do not profess to have any inside knowledge on this, but it just occurred to me whilst considering the lens structure and materials used. Take a look at the lens document that I posted earlier and see what you think. If I am right, it would not be a good idea to rotate the lens as significant errors in the lens defect correction would appear.

[tomas123]

google found the first reseller of a flir cam with hacked resolution 

"new Flir i3 Thermal Imaging Camera with resolution enhanced to i7 144x144 microbolometer"

[Fraser]

It was only a matter of time I suppose. E4's will soon be hitting the market in their 'enhanced' configuration. Human nature...the quick buck has great appeal.

[OrBy]

Which says that yes there is quite a bit of alternate line-ness and more so for colums. And when just eyeballing the normalized correction image you can also see definite stripes. Well, either that or I have been staring at this stuff for too long. 

I let the E4 warm up for a while with the shutter closed and then took a snap. With the RainbowHC color scale you can see this affect if you look closely. Does it bug me - nope.

[schdiewen]

Can you please give as a dusty thermal image?

I'm afraid I only have one picture with that dusty spot on it and it's hard to see here (top middle, the bluish dot in the black sky).

But here you find a small description of the symptoms and a link containing another sample from JockeT:
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg337392/#msg337392

My spot looked like this, maybe not that prominent but the same size. I managed to get a second identical one on my images before cleaning the sensor. So beware where you blow on your cam 

[Fraser]

I must say I was surprised at how open the internal optics and Micro-bolometer are on my PM695. I had expected the optical block to be 'sealed' to the Micro-bolometer to avoid dust ingress. Granted, the camera is fully gasketed and water resistant, but not when the battery/ memory card cover is open. You can actually see right inside the camera then ! FLIR charge a lot for lens cleaning...go figure !

I mention this only because members of this forum have suffered dust on their E4 Micro-bolometers..... it isn't a 'cheap camera thing'. A pity that FLIR did not make the optical block a little more resistant to dust ingress as that need not increase costs significantly. Dust on the optics isn't great, but dust on the Micro-bolometer window is a nightmare for anyone not able to dismantle the camera and clean it themselves.

[mrflibble]

I let the E4 warm up for a while with the shutter closed and then took a snap. With the RainbowHC color scale you can see this affect if you look closely. Does it bug me - nope.

For bonus points you can do the following. Warm up with the external shutter closed like you are doing. Then do:

Code: [Select]

# turn off auto-nuc + trigger nuc
rset .tcomp.services.autoNuc.active false
rset .image.services.nuc.commit true

# check settings
rls .tcomp.services.autoNuc.active
rls .image.services.nuc.shutter
rls .image.flow.shutter.position

# close internal shutter
rset .image.flow.shutter.position "in"

# wait 1 second, then take snap
bt -s

The internal shutter will open up again automatically after a short pause. You can also take a snapshot using the store command or with rset etc, but the bt -s method is pretty lazy and doesn't mess with regular operations. 

[Fraser]

I see large squares in your first image....wonder what that is ?

Mystery picture time  Love it !

[mrflibble]

OK, now here is an ugly thought for those who adjust their lens focus using a lens tool.

...

4. The lens optical errors are captured and countered in software ? (unconfirmed)
5. As noted the errors in the lens may be random in nature. The user rotates the lens to adjust focus. All calibration data for lens defect correction becomes invalid as defects have now moved !
6. The result of lens defects moving and the cameras attempt to counter the defects in software tables that are static could cause interesting effects ?

4: Yeah I'm pretty sure that lens is being corrected for. Some time ago I posted the relevant resource settings in the image flow with which you can enable/disable it.
5: Indeed. Both the GAN + CRS files show corrections that are almost certainly lens related and that are NOT invariant under rotation.
6: Only several ways to find out. 

That PDF on chalcogenide glass was pretty interesting. If the lens assembly in the E4 is indeed as shown in figure 2, then you can easily see how small manufacturing imperfections can cause some of the things we see on those calibration pics. For example when the two elements are tilted slightly relative to each other. That would seem rather silly to me, but what do I know? More likely would be to actually have them properly aligned (that's not all that hard these days I would hope), but then have a slight imperfection in the negative element. I don't know squat about it so can very well be wrong, but I would guess an imperfection in that mold pressed chalcogenide glass element is more likely than not aligning the two elements properly. Any ideas?

[Fraser]

The good news is that the E4 appears to used a singlet and not an air gapped doublet lens. The singlet is concave on the target side and convex on the micro-bolometer. This is a reverse telescope design for wide angle combined with decent distance between the lens and detector face.

Lens alignment is unlikely the issue. Lens material or moulding imperfection is more likely.

[mrflibble]

The good news is that the E4 appears to used a singlet and not an air gapped doublet lens. The singlet is concave on the target side and convex on the micro-bolometer. This is a reverse telescope design for wide angle combined with decent distance between the lens and detector face.

Lens alignment is unlikely the issue. Lens material or moulding imperfection is more likely.

That's good to know. I already thought it a bit unlikely that in 2013 we would still get that large a misalignment. I mean given the relative costs of the camera and the distances involged it would be really dumb (IMO) if it was out of whack that much to show the imperfections that we see. But me no workey in manufacturing, so what do I know. 

So this singlet is just one single piece (singlet, duh) of BD-1 with a coating applied to one side?

[G0HZU]

OK, now here is an ugly thought for those who adjust their lens focus using a lens tool.

Mrflibble is doing some great work analysing anomalies in the E4 image maps. At the moment it is appearing that the lens used in the camera has some bearing on the contents of the maps and distortions are being observed. What is worrying is that some E4's have significant anomalies and others very little, if any evident. Consider this hypothetical situation for a moment :

From production to use:

1. E4 camera fitted with a Chalcogenide glass lens that is not perfect and contains random optical errors (not confirmed as fact but something is showing up in the maps suggesting this may be the case)
2. E4 lens is set to correct focus point and 'locked-off'
3. E4 camera is calibrated and correction tables produced to compensate for physical 'defects' in the sensor AND OPTICAL BLOCK.
4. The lens optical errors are captured and countered in software ? (unconfirmed)
5. As noted the errors in the lens may be random in nature. The user rotates the lens to adjust focus. All calibration data for lens defect correction becomes invalid as defects have now moved !
6. The result of lens defects moving and the cameras attempt to counter the defects in software tables that are static could cause interesting effects ?

This is all circumstantial and I do not profess to have any inside knowledge on this, but it just occurred to me whilst considering the lens structure and materials used. Take a look at the lens document that I posted earlier and see what you think. If I am right, it would not be a good idea to rotate the lens as significant errors in the lens defect correction would appear.

I think you are getting a bit paranoid here....

If there were 'significant errors' caused by rotating the lens then you would see this on basic tests on targets with uniform temperature. eg point it at a desk and rotate the focus control.

If errors were 'significant' then you would see obvious thermal error patterns appear as you rotate the lens for each focus point. Mine seems completely uniform but then I'm not trying to look at tiny uniformity errors of a fraction of a degree.

Also, I took my E4 into work today to compare to the old Irisys 4010 we have there. Basically the modded E4 completely outclasses the 4010 when used for closeup PCB work. The 4010 gives a very noisy and dithered image and the 160x120 sensor isn't as sensitive and a lot of detail is lost.

But if the cameras are used for looking at stuff a couple of metres (or more) away then the lens of the E4 lets it down because it has the wide angle lens. So everything looks so tiny. The 4010 easily wins here despite the noisy image. eg you can see facial detail far better at several metres with the 4010 and the E4 is a poor second. Maybe it's possible to fit an external lens to make it better for distance work?

[mrflibble]

I think you are getting a bit paranoid here....

If there were significant errors caused by rotating the lens then you would see this on basic tests on targets with uniform temperature. eg point it at a desk and rotate the focus control.

Not if the error is something that can be corrected for (and is being corrected for) electronically. I am all for logic based on "lets see what it does when I rotate this", but unfortunately this line of reasoning doesn't work.

In your example you rotate the entire imaging system. In Aurora's example you rotate 1 single part of the imaging system.

If errors were 'significant' then you would see obvious thermal error patterns appear as you rotate the lens for each focus point. Mine seems completely uniform but then I'm not trying to look at tiny uniformity errors of a fraction of 1 degree.

And in fact I am expecting there to be slight differences under rotation, because not everything can be compensated away perfectly. Unfortunately I don't have a decent black body source, so that is just a gues for now...

Any chance I could take a look at your calibration files? The four files in ./FlashFS/system/maps/ I mean... Because not all camera's show this effect. It could for example very well be that your calibration and correction image doesn't show any significant banding etc.

[G0HZU]

In your example you rotate the entire imaging system. In Aurora's example you rotate 1 single part of the imaging system.

I'm using my homemade focus tool to adjust focus from a few cm to several metres and making a comparison on a uniform surface. 

[mrflibble]

In your example you rotate the entire imaging system. In Aurora's example you rotate 1 single part of the imaging system.

I'm using my homemade focus tool to adjust focus from a few cm to several metres and making a comparison on a uniform surface.

I was mainly responding to the logic, not to the case of your particular camera that may very well have a lens where this is not an issue. Several other cameras don't show this effect either, just some.

Don't get me wrong, I am all for the effect being negligable. My camera does show some banding, and I intend to use it for some closeup work too. So being able to get away with lens rotation with zero issues would be a plus. 

[Mikesus]

Realize this might not be the place for it, but playing around with the flir tools, I haven't been able to get it to record.  (No record button is showing on the app?)  Has anyone had success? And yes its the +

I did find this that might interest folks that don't have Tools+

http://flir.custhelp.com/app/answers/detail/a_id/332

It talks about using VLC to view/record...

[G0HZU]

I was mainly responding to the logic,

Can you explain why you think I am rotating the whole imaging system when I use my focus tool? I've not looked inside my camera but I kind of assumed I was just rotating a lens with the focus tool and not the 'whole imaging system'. Just wondering if there's something wrong with my reasoning (or yours )

[Fraser]

Oooops I didn't intend to instigate a major discussion on the issues of lens rotation.  I Just wanted to raise it as a thought for those who are studying the finer detail of the camera system. I keep in mind that I am using a very cheap thermal camera that was supposed to operate at 80x60 resolution, and I am using it at 320x240. So should be both very pleased with its performance and less concerned about any minor anomalies that it presents in its excellent images. I am a very happy owner of my E4(E8+) 

I do appreciate the work that mrflibble is doing as I am learning from it.
 

[G0HZU]

Well you did use words like ugly and "significant errors" with respect to using the focus tool.

I thought my reasoning was sound in that you can use the focus tool and if the errors were significant then you would easily see them. i.e. they would appear in a similar way to spotting debris on the top lenses in a microscope. You rotate the eyepiece lens in the microscope and if the dirt rotates you know there is a problem with the top lens and the artefacts are dirt on this lens and are not on the subject.

In this case it would be 'corrected errors' introducing a spontaneous rotating dirty error effect as you rotate the tool away from the factory position.

If they were that significantly bad you could easily set the camera back (exactly) to the factory position 

[mrflibble]

I was mainly responding to the logic,

Can you explain why you think I am rotating the whole imaging system when I use my focus tool? I've not looked inside my camera but I kind of assumed I was just rotating a lens with the focus tool and not the 'whole imaging system'. Just wondering if there's something wrong with my reasoning (or yours )

I cannot explain that. I can however explain why I think that pointing the camera at something, taking a picture, then rotating the camera and then taking another picture is rotating the whole imaging system. Which is what I thought you said. But given your reaction now that you mean just rotating the lens, and reading it back "... eg point it at a desk and rotate the focus control" I take it you mean that you keep the camera steady and only rotate the lens. In that case no arguments from me, because that is exactly the same thing that Aurora was talking about. So my bad, I cannot read. 

And all that after I just typed up a response about the signal path, but after I reread your response that is now a moot point. *pout*

Going back to the original subject, you have pointed the camera at a steady uniform surface, then rotated the lens and no differences to be seen? If yes, then that is good news. How good the news is for those of us with banding etc would depend a bit on if your calibration/correction pic shows any banding. If your .gan/.crs shows banding + you get no adverse effects under rotation ==> joy! If you don't have banding to begin with then this good news for you doesn't automatically translate into good news for those of us that do see some banding in the calibration picture.

[mrflibble]

Well you did use words like ugly and "significant errors" with respect to using the focus tool.

Heheh, I have also used words like "crappier sensor" when comparing the various E4 camera's. Doesn't mean I think they are actually crap. This E4 is damn good value for money IMO. Especially as an E8+.