Flir E4 Thermal imaging camera review

[IanB]

A question about the camera I didn't catch an answer to in the video: does it have a tripod mount? It seems like it would be highly useful in many cases to mount the camera on a tripod.

[mrflibble]

A question about the camera I didn't catch an answer to in the video: does it have a tripod mount? It seems like it would be highly useful in many cases to mount the camera on a tripod.

I recall mention of there NOT being a mount point, and that being a source of mild disappointment.

[IanB]

This is not true.  I am a level III thermographier and I can tell you that most of the information I learned is not available anywhere except via training.  And the training is vastly different between sources.

This is a deep misunderstanding. There are two ways information can be learned: either someone tells it to you, or you discover it for yourself. Anything you or I may have been trained in was at some time deduced by someone who was first on the scene with nobody going before to train them.

It's similar to the question I often get asked, "Which book did you learn that from?" The questioner tends to forget that someone had to write the book...

In summary, information is always available to those who can teach themselves, especially and particularly in the field of science. It is not limited to what you can learn from others.

[mikeselectricstuff]

A question about the camera I didn't catch an answer to in the video: does it have a tripod mount? It seems like it would be highly useful in many cases to mount the camera on a tripod.

No. The Ex0 range have a tripod mount accessory option but not clear if this fits the Ex range. However the handle is nice and straight, so something 4" long with a tripod thread could be fixed to it with velcro cable ties

[mikeselectricstuff]

Not quite cheap-ass - it does come out at 75mbits/sec

Interleaved over 2 channels at half that rate, right? Incidentally, what IO standard is it?

Sort of - 2.5v levels, 2 simultaneous, pixel-interleaved streams that last for half the line period, the other half is idle.
Using a thin hot wire target to test, I am now 100% certain it is really 320x240

[Fraser]

With that information, it is evident that FLIR are charging a somewhat inflated price for the E8 as it is likely a software configuration difference and a slightly different lens between the models.

OR it could be that all units are identical hardware (including lens) and the software interpretation of the imager chip output provides the various resolutions. i.e, 4 pixelss combined (averaged) in the E4 creating 1 low resolution pixel, and the single pixels read in the E8.

Interesting times indeed. The 320x240 detectors must be getting cheaper to manufacture as FLIR would not waste money.

[mikeselectricstuff]

With that information, it is evident that FLIR are charging a somewhat inflated price for the E8 as it is likely a software configuration difference and a slightly different lens between the models.

OR it could be that all units are identical hardware (including lens) and the software interpretation of the imager chip output provides the various resolutions. i.e, 4 pixelss combined (averaged) in the E4 creating 1 low resolution pixel, and the single pixels read in the E8.

Interesting times indeed. The 320x240 detectors must be getting cheaper to manufacture as FLIR would not waste money.

The sensor design is very interesting - must be hugely cheaper to make. And has the ADC on-chip. I suspect this has been driven by potential future automotive applications.
If you look at the specs there is a sudden jump in sensitivity, particularly between E5 and E6 which I suspect it is a lens change.

[aargee]

A question about the camera I didn't catch an answer to in the video: does it have a tripod mount? It seems like it would be highly useful in many cases to mount the camera on a tripod.

I recall mention of there NOT being a mount point, and that being a source of mild disappointment.

Now there's a job for a 3D printer!

[mikeselectricstuff]

There is actually space in the bottom of the battery moulding to add a tripod thread, and this is part of the battery so any warranty issue would be avoided by not returning the battery (due to shipping regs of course!)

[amyk]

This is getting better and better... Flir must be making a huge profit on those sensors! In contrast from what I see about the Chinese cameras, they appear to really be using sensors of the resolution they claim - e.g. Dali make a 160x120, 384x244, and 640x480 and they're all physically different sizes.

And everything I could find about the i.MX25 series suggests that only the highest end '258 has the security module so the messages you're seeing are probably from the driver they left in the OS (maybe for some higher-end/military model?) --- given they left the wlan one in too, and who knows what else.

[Fraser]

I stumbled upon a forum discussing another consumer FLIR offering, the FLIR SCOUT. The chap discovered that his TIC responded to the TAU core controller software and in fact contained the TAU engine. He didn't risk uploading the fully fledged Tau firmware though.


http://www.t2iforum.com/47/thermal-imaging-(flir)/

http://www.t2iforum.com/47/thermal-imaging-(flir)/15/

He refers to a 'document' that details FLIRs usage of common cores across their range, but sadly no link to that document.

[mikeselectricstuff]

This is getting better and better... Flir must be making a huge profit on those sensors! In contrast from what I see about the Chinese cameras, they appear to really be using sensors of the resolution they claim - e.g. Dali make a 160x120, 384x244, and 640x480 and they're all physically different sizes.

..and expensive hermetic metal packages.

[SeanB]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

[neggles]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

or, you know, the even simpler reason of "making lots of one kind of chip and changing some settings is a lot cheaper and easier than making four kinds of chip"

[mzzj]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

I have seen at least one thermal camera with 4x4 dead pixel group, this was "fixed" in firmware by interpolating the nearest pixels on both sides.
The dead pixel group showed up really nasty when sharp temperature gradient was over that area, one side white another side blue and red square block in the middle.

[mrflibble]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

or, you know, the even simpler reason of "making lots of one kind of chip and changing some settings is a lot cheaper and easier than making four kinds of chip"

I suspect that the simpler simpler reason is very close to what SeanB describes. Indeed make a one size fits all sensor. And then put the best ones in the high end gear, and the ones with a defect here and there in the low end. Then during image processing average/decimate a bunch of pixels and you have your low res result. And they probably truncate the LSBs as well if they really average 16 subpixels for one pixel on the E4. Because the temperature resolution on the E4 is quite a bit lower than on the high end gear. You'd think that with a bit of statistics they could get better temp resolution on the E4. Or is this also a matter of the lens material used?

[Fraser]

The FLIR TAU core has evolved into the new TAU2 

Take a look here:

http://www.flir.com/cvs/cores/view/?id=60983

http://www.flir.com/cvs/cores/view/?id=54717


http://www.flir.com/cvs/cores/view/?id=54717&collectionid=612&col=54726

You will see that the TAU has electronic zoom, how convenient for producing low pixel count products via software alone !

Tau is designed to OEM incorporation into both high and low cost products. It is not beyond possibility that Tau2 micro-bolometer and support chip set lives inside the new E Series. There is also the FLIR Quark to consider 

Controller software for the TAU and QUARK is here:

http://www.flir.com/cvs/cores/view/?id=51880

According to the text, the hardware is fault tolerant and a new firmware does not overwrite the factory firmware 

Some interesting reading on the FLIR OEM CORE pages especially under the documents tab.

http://www.flir.com/cvs/cores/view/?id=54717&collectionid=612&col=54721

Manuals available here:

http://www.flir.com/cvs/cores/view/?id=51878

[mikeselectricstuff]

Teardown thread
https://www.eevblog.com/forum/testgear/flir-e4-thermal-imaging-camera-teardown/msg309199/?topicseen#msg309199

[mikeselectricstuff]

The FLIR TAU core has evolved into the new TAU2 

Take a look here:

http://www.flir.com/cvs/cores/view/?id=60983

http://www.flir.com/cvs/cores/view/?id=54717


http://www.flir.com/cvs/cores/view/?id=54717&collectionid=612&col=54726

You will see that the TAU has electronic zoom, how convenient for producing low pixel count products via software alone !

Tau is designed to OEM incorporation into both high and low cost products. It is not beyond possibility that Tau2 micro-bolometer and support chip set lives inside the new E Series. There is also the FLIR Quark to consider 

Controller software for the TAU and QUARK is here:

http://www.flir.com/cvs/cores/view/?id=51880

According to the text, the hardware is fault tolerant and a new firmware does not overwrite the factory firmware 

Some interesting reading on the FLIR OEM CORE pages especially under the documents tab.

http://www.flir.com/cvs/cores/view/?id=54717&collectionid=612&col=54721

Manuals available here:

http://www.flir.com/cvs/cores/view/?id=51878

Data format I saw is very close to that shown in the Tau electrical spec. I noticed various sync markers, which are probably the same

[mrflibble]

What's also interesting is in the boot log it says:

FVD_Init: Downsampling setting: 0x00
...
FVD_Init: Downsampling setting: 80 60

Right after loading the fpga bitstream. And then further down you have:

FVD_Init: Downsampling setting: 0x00
FVD_Init: re-read the hardware information x: 80, y: 60

And the camera fps settings with "jitterlimit 134mS" is floating around in there as well. Or jitteblimit depending on random errors while grabbing the serial port.

Thanks for the tear down! I loved the bit where you stuck a metal plate in front and then slowly heated it. Watch that binary reading go up on the scope.  That video may very well end up costing me EUR 995. 

[Psi]

Wonder if the I2C eeprom has any checksum 

[Corporate666]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

or, you know, the even simpler reason of "making lots of one kind of chip and changing some settings is a lot cheaper and easier than making four kinds of chip"

I think you are right.  It would be very hard, I would think, to manage supply if you were using QC-failed chips in lower end units.  You will almost inevitably have too many failed chips, or too few.  For a company like Intel who sells their chips to other vendors, this isn't a big deal, as if you have too many, you offer them at a sale price, and if you have too few, the price goes up.  Vendors using Intel chips also have a lot of alternatives to choose from.  But whereas FLIR uses the chips themselves, they don't really have those options.  So I tend to think they just use the exact same chip and I bet most if not all pass the "full" QC check just fine.

[mikeselectricstuff]

A simple reason for the sensors being the same is the lower cost ones failed final test in some way that made them no longer usable at the full resolution. Like stuck pixels in the outer section, very hot or very cold pixels in the image that will be out of the calibration range at full resolution but will be fine when averaged with neighbours, or dead rows or columns in the outer areas. Averaging 4 pixels to make the lower resolution unit or 16 to make the very low res unit makes sense to use an otherwise wasted chip, bonding, 2 die stack and package that has been up till final test a functional unit electrically but a poor thermal performer.

or, you know, the even simpler reason of "making lots of one kind of chip and changing some settings is a lot cheaper and easier than making four kinds of chip"

I think you are right.  It would be very hard, I would think, to manage supply if you were using QC-failed chips in lower end units.  You will almost inevitably have too many failed chips, or too few.  For a company like Intel who sells their chips to other vendors, this isn't a big deal, as if you have too many, you offer them at a sale price, and if you have too few, the price goes up.  Vendors using Intel chips also have a lot of alternatives to choose from.  But whereas FLIR uses the chips themselves, they don't really have those options.  So I tend to think they just use the exact same chip and I bet most if not all pass the "full" QC check just fine.

I suppose it depends on the number and nature of the typical faults that come out of the process. A lot of random bad pixels would be hard to deal with, but Mapping out the odd few bad pixels  from a 320x240 when downscaling to 80x60 could probably be done automatically by looking for differences from neighbours.
If this is done, then some simple production criteria for numbers of bad pixels and  minimum distance between them could be used for selection. Considering the big price difference between models, If they have significant numbers of only-just-bad die it would probably be worth using them.

From what I've seen so far there is not enough NV memory to store a full bad pixel map. I doubt it would be in main flash as all other production data (including sensor serial no) is in the eeprom.

As & when I get time to look at the raw data it should be fairly obvious. It will also be interesting to see what the output looks like before flat-field correction

[mrflibble]

From what I've seen so far there is not enough NV memory to store a full bad pixel map. I doubt it would be in main flash as all other production data (including sensor serial no) is in the eeprom.

How big is that eeprom? A sparsely populated 76800-bit map with some RLE doesn't take up all that much space I'd think.

But as you say, in the end it would depend on the nature of the defects, of which I am blissfully ignorant.

[mikeselectricstuff]

From what I've seen so far there is not enough NV memory to store a full bad pixel map. I doubt it would be in main flash as all other production data (including sensor serial no) is in the eeprom.

How big is that eeprom? A sparsely populated 76800-bit map with some RLE doesn't take up all that much space I'd think.

But as you say, in the end it would depend on the nature of the defects, of which I am blissfully ignorant.

2kbit. Not looked at all of it but there's quite a lot of ASCII strings in there so no room for more than addresses of a very few pixels. Highly doubt it.