Flir E4 Thermal imaging camera teardown

[janekivi]

If I change contents then I delete this number.
Tried many calculators in the beginning with no success to get
same number but there wasn't big point to have it there.

[sky@eevblog]

Found that if you delete #CRC line but leave <cr><lf> in the end of file than you may 7-Zip it and then look inside new archive. CRC there will be exactly what you need. All this is not relevant here as common_dll.dll was patched to skip this test.

[Bud]

Wow, interesting finding on using a Zip-er to get CRC32. I was using a Python script found elsewhere in this thread.

[Myrulken]

Hey everyone that's got troubles with not getting the full menu.

Access your camera with FileZilla or similar, go to FlashBFS/system/appcore.d/factory.d folder and look at the files there.
The menu hack adds a file there with the name "Ui_control.rsc". But i noticed that i had another file there, named "xxxui_control.rsc". I deleted the xxx one and suddenly everything worked like a charm.

Big thanks to janekivi that helped me getting this solved! 

[janekivi]

OK. Easy it is http://crc32-checksum.waraxe.us/
Paste text and done. But without dll hack many things not working anyway.

[sky@eevblog]

Wow, interesting finding on using a Zip-er to get CRC32. I was using a Python script found elsewhere in this thread.

it is useful in quick comparison of several versions of the same file - just drop it to 7-zip window and one will see CRCs in real time

[janekivi]

As I said, 2.3.0 is easy to hack version now.
Couple of  things you need:
-use Set_RNDIS_permament
-connect with telnet and ftp
-you need restartapp.bat from old software
(in telnet I must type first command to introduction and after second
command all working and I can't reuse that command, I must retype it)
Now you can start experiments. Change something in file, no need for CRC,
upload it, type restartapp, no full restart, see results.
And so on.

(Old software does not need CRC too if you hack the same procedure in dll)

[Bud]

I bricked my 2.3.0 camera the other day when I played with USB/RNDIS modes in terms of I could not connect to it anymore by either way, though the camera still worked fine. It took me 4 hours to find a way to get it back, but could took half a minute had it had the hidden USB mode selection menu working. In native 2.3.0 you can get to the menu but cannot change selection. If we could find a way to unlock that menu that would be great.

[Bud]

I have a question for Aurora or other experts here. What would be the right way to test E4 sensor gradient ? I have attached a image from Flir Tools of my E4 thermal pic after a couple of minutes after power up with the lens cover closed . The gradient seemed to be 0.6 degree C from coldest to hottest and a round blob can be seen in the center of the thermal image. The palette span was tweaked for 1 degree C in Flir Tools. I am not sure if the 0.6 degree is OK or an indication of a poor thermal compensation, and if testing like this with the lens cover closed would be a right thing to do or not, so seeking your advice.

[Fraser]

Thermal gradient is best tested by aiming the camera at a surface of uniform temperature and emissivity. A Black Body source is specified by manufacturers for such tests.

House walls can have a thermal gradient across their surface. A painted sheet of metal at room temperature would likely do for the tests that you are interested in however. Paint the metal sheet with matt black cellulose, acrylic or enamel paint from a spray can. Try to keep the paint thickness thin and even to avoid differences in the surface emissivity.

I have tested my E4 previously and it DOES have a thermal gradient at narrow spans. I am not concerned with such at the E4 price point though. From memory, early testing of the E4 showed that the transmission range from the outer to centre of the E4 lens had to be compensated for in the calibration tables, and an area at the centre of the camera image did display a gradient compared to the periphery of the image. Such gradients only become apparent when working at narrow spans however and are not of concern to many.

I should state that I do not consider the E4 to be a truly radiometric thermal camera as its accuracy as you move away from a target at 30C decreases. Accuracy at low temperatures close to 0C is appalling ! For the money, the E4 is great value and we have to forgive some of the compromises that had to be made to keep the BOM cost down, such as not using a Germanium lens. If I was the owner of a genuine, ex factory, E8, I might not be so forgiving however !

Aurora

[Co6aka]

-use Set_RNDIS_permament

FWIW, permanent has never worked for me, only temporary.

[Bud]

A painted sheet of metal at room temperature would likely do for the tests that you are interested in however. Paint the metal sheet with matt black cellulose, acrylic or enamel paint from a spray can. Try to keep the paint thickness thin and even to avoid differences in the surface emissivity.

Thank you Aurora, I guess you mean paint it black ? Probably a copper plate or may be as simple as a door brass kick off plate would work better than aluminum in terms of temperature evenness across the plate. Or may be a aluminum black oxidized finished sheet would work too, being already black and having very thin coating.

[Fraser]

At room temperature a small (A4 size) metal passive plate should not have significant temperature gradient across its surface. A heated plate would be a very different matter.   For the test you requested any thin Matt paint should work as it is the paint chemistry that provides the emissivity to a greater extent than colour. white and metal colours should be avoided however. Gloss paint can produce reflectivity issues. Be aware of narcissus effect where the camera sees itself on the target surface and its warm micro bolometer creates a hot spot. To see this effect look at a window or mirror with the camera.
Corroded aluminium will have better emissivity  than shiny aluminium but I would not like to predict its evenness and a blotchy image may result.

Aurora

[Co6aka]

Had the resolution hack installed and working. Installed the menu hack and the noise generator seems to be back and I have the missing zoom/pip menu items problem.

PS- The following is NOT the problem; no such files in that folder:

Hey everyone that's got troubles with not getting the full menu...

[Bud]

Installed the menu hack and the noise generator seems to be back

Intersting, That confirms my observation, i too noticed noise, though it is much milder than on a factory 2.3.0. When have time i would like to go back and step through the process trying to find where noise comes back.

How much noise do you see comare to the original state?

[Co6aka]

How much noise do you see compared to the original state?

I think it's less. I may have another E4 to compare it to over the weekend. Could it be due to "more sensitivity" because of the palette? FWIW, I installed the "V4" resolution hack about a month ago, and the "2014 Menu" hack this evening. Maybe previous versions of the menu hack installed/configured something that's omitted from the latest hack?

Attached is a capture with the shutter closed, so you can see the noise:

[Co6aka]

And here's a capture with rainbow high contrast... Might not be less:

[Bud]

The one with the lens cover closed looks very similar to what I have, same cross pattern and identifiable vertical columns of a darker color. If I move the camera the darker colored columns remain in place as if they exist because of calibration deficiency or something.  They are barely visible and do not typically spoil images but I can still see them on the live screen when pointing on walls. When the camera runs I can also see (pseudo)random faint horizontal strips of noise similar to what you could see on a analog TV screen when was a interference signal nearby. I think I saw something like this discussed in SEEK camera topic, do not know if the two are of  a same nature, since some SEEK people seem to be former FLIR folks.

[Co6aka]

I don't think the (post-resolution-hack) pre-menu-hack camera exhibited that noise. Maybe something in the config file is munged, and I wonder if it involves the nonworking zoom/pip issue.

Janekivi is very close but no magnum of Champagne yet. Wish I had time to hack away at this too.

PS- I also don't have manual temp range; it's just auto and lock.

[stefbeer]

[...]
In native 2.3.0 you can get to the menu but cannot change selection. If we could find a way to unlock that menu that would be great.

Just a thought for a workaround:
You were able to set the image quality to 100 with those measuring spot preset files. Maybe it's possible to use this for RNDIS as well? Might not be the best solution but I think that would be still better than using FIF files all the time.
Something like: No spot = RNDIS+UVC, Center spot = UVC+MSD

In Set_RNDIS_Permanent.fif a new file with following commands is generated:

Code: [Select]

.system entry
.system.usbmode text "RNDIS"So you should only have to add those lines to the preset files as you desire.

And in the ressource tree the following possible values are listed:

Code: [Select]

.system: (17)
[...]
rw--r---r---1p*2 root   root   <a> usbmode              "RNDIS_UVC"
                 ["RNDIS" "MSD" "UVC" "RNDIS_UVC" "RNDIS_MSD" "UVC_MSD" "RNDIS_UVC_MSD"]
[...]
I think there's a good chance this will work

[janekivi]

If you do not remember what the noise is you can put factory conf back and see it with rainbowHC.
I didn't touch all in conf file and you can test them all by telnet
When I connect it I give couple of DIR and then test this:
rls .caps.config.image.targetNoise
rls .caps.config.image.zoom
rls .caps.config.ui.fusion.PIP
...
and all other
But conf hack is tested successfully 100 times now... I think : )
I use Total Commander from it's beginning an it has built in file compare and ftp and... all things.
If I'm not sure I can always test my file in right window against camera ftp file in left window with
compare and see if they not the same.
So replace files in camera with ftp and compare with your factory backup if there is made something
extra files. If fif can't replace file it renames old and then adds new (how I too replace files what in use)
After that there is small chance with conf.cfc in ui.d directory. Delete it.

[sky@eevblog]

Using 'preset_centerHotColdSpot.rsc' code instead of any other working *.rsc is fine but why not to find corresp menu icon?

Wonder where 'preset_*.rsc' file names are connected to ToolBar name="measure_*"?
One suspect is 'facet_ui_qml.dll'. It has some interesting strings inside:
facet_ui_qml.dll from 2.3.0

Code: [Select]

ID_AUTO_COLD_SPOT   Ic_MeasureToolPreset_SpotMinBox measure_spot_min_box   

ID_AUTO_DELTA_TEMP  Ic_MeasureToolPreset_DeltaTmeasure_delta_t 1addDeltaT(int, const QVariant&, const QVariant&)   1updateDeltaTRefTemp() 

ID_AUTO_DELTA_SPOT  Ic_MeasureToolPreset_DeltaS measure_delta_s

ID_THREE_SPOTS  measure_three_spots

ID_FIVE_SPOTS   Ic_MeasureTool_Spot3    .image.sysimg.measureFuncs.spot.5   measure_five_spots  userPresets noUserPreset   

ID_AUTO_HOT_SPOT    Ic_MeasureToolPreset_SpotMaxBox .image.sysimg.measureFuncs.mbox.1   measure_spot_max_box    caps.config.ui.measure.presetException  chino1  Ic_MeasureToolPreset_SpotCenter measure_spot_center

ID_NO_MEASUREMENTS  Ic_MeasureToolPreset_SpotHide   measure_spot_hide   

compare it to facet_ui_qml.dll from 1.18.7:

Code: [Select]

ID_AUTO_DELTA_TEMP  Ic_MeasureToolPreset_DeltaT measure_delta_t 1addDeltaT(const QVariant&) 2triggered(const QVariant&) 1updateDeltaTRefTemp()  2selected()

ID_AUTO_DELTA_SPOT  Ic_MeasureToolPreset_DeltaS measure_delta_s

ID_AUTO_CENTER_HOT_COLD_SPOT    Ic_MeasureToolPreset_MaxMinCenter   measure_spot_cold_spot 

ID_AUTO_COLD_SPOT   Ic_MeasureToolPreset_SpotMinBox measure_spot_min_box   

ID_AUTO_HOT_SPOT    Ic_MeasureToolPreset_SpotMaxBox .image.sysimg.measureFuncs.mbox.1   measure_spot_max_box   

ID_CENTER_SPOT  Ic_MeasureToolPreset_SpotCenter measure_spot_center

ID_NO_MEASUREMENTS  Ic_MeasureToolPreset_SpotHide   measure_spot_hide   

ID_TOOLBAR_MEASUREMENT  Ic_MeasureTool_Spot measurementMenu
This can explain why 'measure_spot_cold_spot' doesn't work anymore when used in 'toolbar-config.xml'. Any ideas where to look? May be one knows good resource viewer to look inside those *.dlls for icons? If they stored as bitmaps it could be possible copy-paste icons from 1.x.x file to 2.3.0 file.

[Isaac-1]

Sky, is that lack of a space between DeltaT and measure in the 2.3.0 file on the line ID_AUTO_DELTA_TEMP  Ic_MeasureToolPreset_DeltaTmeasure_delta_t .... in the original file, or did that accidentally get inserted in the copy/paste?

[janekivi]

I have a problem with the camera Flir E4. 2.3.0 I deleted a file by mistake common.dll.dll.
At this time, the camera shows only the logo and nothing else. Is there any possibility to
save the file in the system? Earlier I made a backup of the system but I can not upload.
Is there any chance to run the camera?

Good if you have backup. Some files you can restore with 2.3.0 update from flir.
If you have files use ftp with
user:   anonymous
psw:    NcFTP@

[mariuszko]

Many thanks for your reply it all works