Flir E4 Thermal imaging camera teardown

[Taucher]

Here (.image.flow.maps.combGainDeadMap.pixReplace) images for my cam - directly captured

Captured them in different modes as everyone seems to prefer another palette

Don't see anything that worries me there

[mrflibble]

Mike, with all the experimentation on your unit (for our benefit) maybe you have corrupted something ? The -40 degree temp could have been caused by such ? She's only spec'd to -20 !

I doubt it. It looks like he was just mucking about with the range since -40 happens to be the lowest you can get is by setting the level directly. And just for the fun of it I put mine on a span of -379.9 to 620 , and the dead pixels are in the exact same boring spot, as expected.

Dead pixels on/off does have an effect on the auto ranging of temp scale. The defective pixels on mine all have an loooooow value (corresponding to low temp). And if you have it show the dead pixels then the lower end of the range drops (when auto ranging the temp scale). But I doubt it is anywhere near enough to go from room temperature to -40 that way.  Just tested it and the lower end of the range went from 23.8 (no dead pixels) to 22.7 (show dead pixels).

[bean_dip]

ok, maybe the installation part needs a readup across this thread

- apply mike hack
- put camera into IP mode (RNDIS) (hold "right" button on version screen for >10s = hidden menu)
- access camera over IP (default gateway IP of "infrared" interface = camera - usually 192.168.0.2 ... but had seen own_ip+1 scheme too)
- use filezilla to backup anything readable - esp. the FlashBFS folder
- overwrite the FlashBFS folder with the attached one, do not remove the old one as the new zip just adds or overwrites certain stuff
- turn off camera, remove battery, repower

have fun, report back

Installed...works pretty well for a Beta.  Thanks for your hard work and keep it up! 

I haven't looked into it yet, but I wonder if there's a way to change the default temp for the Insulation Alarm mode...we keep our house between 66 and 68, so most of the outside walls test as "bad" since they'd only have to be 2-4 degrees colder to be below the 64.4 threshold.  I know I could also just do "Blue Below" and set my own but it's the principle. 

[olsenn]

All this talk about dead pixels; this is undoubtedly a stupid question, but is this referring to dead pixels on the screen, or dead pixels on the sensor? I'm assuming the latter.

[mikeselectricstuff]

All this talk about dead pixels; this is undoubtedly a stupid question, but is this referring to dead pixels on the screen, or dead pixels on the sensor? I'm assuming the latter.

The sensor, obviously - how would it mask dead pixels on the screen...?

[mikeselectricstuff]

Dead pixels on/off does have an effect on the auto ranging of temp scale. The defective pixels on mine all have an loooooow value (corresponding to low temp). And if you have it show the dead pixels then the lower end of the range drops (when auto ranging the temp scale). But I doubt it is anywhere near enough to go from room temperature to -40 that way.  Just tested it and the lower end of the range went from 23.8 (no dead pixels) to 22.7 (show dead pixels).

It will be using a histogram+threshold function to do the autoranging, so as the number of very cold pixels increases, it's plausible that it could suddenly jump to the cold pixel value if it finds more than a certain number of them

[mrflibble]

On the sensor. The bad pixels in the calibration pic all had a value that was more than 4 standard deviations below the average. I picked that factor of 4 stdevs fairly arbitrarily after eyeballing the distribution. At any rate, the sensor readout for dead pixels all have a way lower value than the rest of the image.

[mrflibble]

It will be using a histogram+threshold function to do the autoranging, so as the number of very cold pixels increases, it's plausible that it could suddenly jump to the cold pixel value if it finds more than a certain number of them

Good point. I found the -40 a bit too much of a coincidence. But then again the same limit might be applied for autoranging as it is for manual input of level value. And -40 (or lower) might very correspond with the loooow value on dead pixels. So if it does things (histogram+threshold) as you describe, that might cause it.

[Taucher]

-snip (MENU hack)

Installed...works pretty well for a Beta.  Thanks for your hard work and keep it up! 

I haven't looked into it yet, but I wonder if there's a way to change the default temp for the Insulation Alarm mode...we keep our house between 66 and 68, so most of the outside walls test as "bad" since they'd only have to be 2-4 degrees colder to be below the 64.4 threshold.  I know I could also just do "Blue Below" and set my own but it's the principle. 

In fact I'm investigating into that already - humidity/condensation/mold mode seems to have additional features where one can set humidity and tempterature and the camera seems to calculate the condensation point then.

If somebody could obtain a FTP dump (FlashBFS) from a T440 / T600 camera model, that could significantly help (or any other with that mode - the T4xx/T6xx seem to have time lapse, so I'd really like to take a look at that code

Code: [Select]

rls .image.sysimg.alarms.humidity.1
active                      true
atmTemp                293.14999
audioBeep                  false
dewpoint               282.41962
duration                       0
hysteresis                     1
isoCoverageActive          false
isoCoverageMBoxId              1
isoCoverageThreshold         0.5
isoIndicationActive         true
isoIndicatorId                 1
label                        "1"
relAirHumidity               0.5
relAirHumidityLevel            1
thresholdTemp          282.41962
trig                       false
type                     "BELOW"
visualBeep                  true

.image.sysimg.alarms.humidity.1.relAirHumidityLevel  <- this one can be changed, but marking color changes to reddish - while image gets still colored in green (see attachment)

dewpoint or humidity won't change as they seem to be auto-set by the camera

rset .image.sysimg.alarms.humidity.1.isoCoverageActive true <- will enable a Tmax pointer-cross (useless)

[G0HZU]

Here (.image.flow.maps.combGainDeadMap.pixReplace) images for my cam - directly captured

Captured them in different modes as everyone seems to prefer another palette

Don't see anything that worries me there

The dead pixels look quite big to me for a 320x220 image. The dead pixels suggest the screen is 160 pixels wide. But there are some pixels that look smaller. Can someone explain the image format to me?

[G0HZU]

The image looks like it is 320 pixels wide but I'm confused as to why the dead pixels are often in a neat block of 4 making the dead ones look like they are on a 160pixel wide screen.

Why are they in such neat blocks?

It probably magnifies them to 4x the actual size for convenience.

Maybe, but not every case is a neat block of 4. Some are smaller.

[mrflibble]

If somebody could obtain a FTP dump (FlashBFS) from a T440 / T600 camera model, that could significantly help (or any other with that mode - the T4xx/T6xx seem to have time lapse, so I'd really like to take a look at that code

You could take a look at the firmware updates for these models.

[Taucher]

I'd guess lens correction algo - we're not looking at the unprocessed image
Just turned off ONE step

[Taucher]

If somebody could obtain a FTP dump (FlashBFS) from a T440 / T600 camera model, that could significantly help (or any other with that mode - the T4xx/T6xx seem to have time lapse, so I'd really like to take a look at that code

You could take a look at the firmware updates for these models.

been there already.. unfortunately incomplete filesets - nothing (EDIT: useable) regarding any menu buildup whatsoever

[olsenn]

I don't think dead pixels are too much to worry about; Mike's camera seems to work quite well and he has quite a few dead pixels. Especially when dealing with heat, adjacent pixels will be similar. Take the pixel to the right and the one to the left, and the one between is probably going to be 50% in between the two

[mikeselectricstuff]

The pixels look much bigger in the captured images than onscreen due to JPEG compression artifacts.

[mikeselectricstuff]

I have seen Tauchers commands but do not have any idea how you deliver them to the camera. I am a self confessed IT programming/hacking newbie.

Enable RNDIS and enter them at the console prompt using Telnet

[zapta]

No matter about the dead pixels though.....anything would be better than 80x60   

In 80x60 mode, the E4 has 320*240-80*60=72,000 dead pixels  ;-)

[ixfd64]

Any recommendations for a Telnet client ? ....PuTTY ?

I haven't used Telnet much, but PuTTY was my client of choice whenever I worked with Telnet at my previous job.

[IanB]

Any recommendations for a Telnet client ?

Well, bugger. What is the world coming to?

[eliocor]

 Install PuTTY....

[Richard Wad]

Any recommendations for a Telnet client ?

Well, bugger. What is the world coming to?

Putty works or if you like the old school cmd line you need TCC/LE 64 bit if you're running Windows 7 64 bit.
http://jpsoft.com/tccle-cmd-replacement.html
Once installed you can either run TCC LE x64 13.0 or cmd.exe, then telnet.exe works fine.

[kaz911]

Any recommendations for a Telnet client ?

Well, bugger. What is the world coming to?

telnet client IS there - it is just not installed by default... http://elmajdal.net/Win7/Enabling_Telnet_Client_in_Windows_7.aspx

[madshaman]

A bit off-topic, but running VMWare workstation hosting VM running netbsd has greatly reduced my irritation with the Windows' shell and its default assortment of cmd-line programs.

You can do the same thing with your favourite linux if that's what you prefer.

Putty is pretty good as a comms terminal though.

[Loafdude]

Use vmware player, its free.

Hate to reduce the S/N ratio in the thread but...
Thanks Taucher for all the WinCE hacking!
Appreciate everything you can unlock in the FW!
Hopefully the T600 series FW holds some secrets yet

And thanks mike for testing some ZnSe lenses for us!
That looks like by FAR the easier way to handle focusing at short focal lengths!
What was the diameter of the lenses you purchased?
It looks like the field of view is catching the edge of the lens?
Can you confirm?