Author Topic: Yet another cheap thermal imager incoming.. Seek Thermal  (Read 1012838 times)

0 Members and 3 Guests are viewing this topic.

Offline callipso

  • Regular Contributor
  • *
  • Posts: 87
  • Country: cz
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #350 on: October 24, 2014, 03:15:34 pm »
Could the dot-grid maybe be a sort of watermarking - protection against use without the supplied app?
I wonder... **BOOM
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #351 on: October 24, 2014, 07:25:44 pm »
Mike, neat stuff. I take that just powering up the device doesn't spew the raw data until it's initialized, right?

Can't wait for part two, it seems like tapping into that data stream would be a piece of cake to capture it on an FPGA at full frame rate and convert it to a 14 bit bus for DSP processing and then to an LCD.

But if it needs some kind of handshake (like the fixed invariant data you where probing as well) maybe it can be reproduced by the FPGA to start up the camera, maybe not allowing changes in the configuration but at least a a start.
 

Online mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 13695
  • Country: gb
    • Mike's Electric Stuff
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #352 on: October 24, 2014, 10:24:57 pm »
Mike, neat stuff. I take that just powering up the device doesn't spew the raw data until it's initialized, right?
Correct. But it's possible there might be a test mode...
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1528
  • Country: 00
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #353 on: October 24, 2014, 10:43:15 pm »
Maybe Mike will be able to capture another raw Seek Thermal thermal image for bad pixels analysis, while completed my software and made some automatic measurements how many those black dots are and which percentage looks like bad ones  :o

Attached also 208x156 pixels version-oryginal output from my software.
The only 5 black pixels which I marked by hand from red are those pink one on left side, while they are very close to those looking good-only a tiny 1 pixel difference detected by my automated tools  ;)


There are some statistics from this image.
Quote
Seek Thermal sensor 206 x 156 black pixels: count: 2169  good: 2138  bad: 31
However, when we count those 5 pink pixels and let them be good ones, than we have in 206x156 image:
* 2143 good black pixels (2138 green+ 5 pink ) and this is what I would expect on perfect Seek Thermal sensor image raw data if all their sensors shares the same black dots pattern
* 26 bad looking (red ones)
So, when we calculate percentages we'll  get:
2143/206/156 =  6.668% good black dots on perfect sensor
and
26/2143= 1.21% bad black dots in this case  >:(

It is interesting if this is really manufacturing defects or some kind of data capture errors?  :-\
Probably more samples needed from the same camera and also other Seek Thermal sensors to investigate it futher.

Maybe someone else will include there raw data from this sensor for similar analysis.
I haven't got this thermal camera yet.

BTW: On Mike video teardown it looks like there is very downgraded efective frame rate and some kind of werid averaging with horrible pixels around thermal image, so probably they forgot to cut off rest of blured image at the edges and it creates very bad looking rectangle around thermal image  :-DD
They could simply make those pixels... black if do not wanted resize output image  :palm:

Note: Just thinking maybe those red black pixels (they had 0 value) are simply noise from sensor itself?  :-//
Difficult to say without another sensor images analysis...


« Last Edit: October 24, 2014, 10:49:55 pm by eneuro »
12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #354 on: October 24, 2014, 11:00:49 pm »
The almost-black dots are not all 00., they are definitely intentional.

Refer to this Xilinx patent. This shows the coverage of BGA grounding pins arranged in a "sparse chevron" pattern which reduces SSN.



Each ground pin (or black dot) has 8 surrounding pins.
While the pattern is not quite the same here, similar concept.
It's my thought that the thermal sensor embeds these in the data to provide compensation for that small region of the image. What the compensation is for, I don't know.

Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1528
  • Country: 00
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #355 on: October 24, 2014, 11:13:08 pm »
The almost-black dots are not all 00., they are definitely intentional.
On PNG image provided by you (RGBA) 8 bits per channel those black pixels marked by my tools green, red and pink had 0x00 zero the same values in RGB channels and 255 in Alpha channel  and most of them fits into this hexagon pattern.

Now it is time to try dissasemble Seek Thermal Android JNI library (*.so) and maybe will be possible also look inside dissasembled code of this MPU on Seek Thermal PCB to see what is going on there  with sensor data  >:D

12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #356 on: October 24, 2014, 11:18:10 pm »
A paper has been written on the production of smaller microbolometer pixel sizes, namely 12um. It is dated October 2012 so 12um pixels have been a possibility for some time yet have not been developed by the likes of FLIR and I am not aware of any ULIS 12um microbolometers either. Could such small pixels have performance 'issues'  I wonder. I am not a subscriber to SPIE but if anyone her is, it may make interesting reading.

http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1387338

Aurora
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline OrBy

  • Regular Contributor
  • *
  • Posts: 220
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #357 on: October 24, 2014, 11:22:37 pm »
Perhaps the "black" pixels are for some sort of future shutter-less NUC that has not been implemented yet like the lepton?
 

Offline Fry-kun

  • Newbie
  • Posts: 8
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #358 on: October 25, 2014, 12:56:19 am »
Is there a convenient connector to plug this camera into a standard laptop USB port (type A)?
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #359 on: October 25, 2014, 01:52:07 am »
Is there a convenient connector to plug this camera into a standard laptop USB port (type A)?

A quick search found this:

http://usb.brando.com/usb-2-0-male-to-micro-b-female-adapter_p02840c0042d015.html

There has to be others as well.

My search was "usb a male to micro b female adapter" and click on images to make sure the micro b is female.

 

Offline cynfab

  • Regular Contributor
  • *
  • Posts: 175
  • Country: us
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #360 on: October 25, 2014, 02:49:41 am »
Is there a convenient connector to plug this camera into a standard laptop USB port (type A)?

A quick search found this:

http://usb.brando.com/usb-2-0-male-to-micro-b-female-adapter_p02840c0042d015.html

There has to be others as well.

My search was "usb a male to micro b female adapter" and click on images to make sure the micro b is female.

Or: http://www.amazon.com/gp/product/B009AWA3VK/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1

Probably made in the same factory. I got one of these and it took about 2 weeks to arrive. I figured for ~$1.50 why not...

It works fine with my Seek TC.

  ...ken....
« Last Edit: October 25, 2014, 02:51:14 am by cynfab »
 

Offline Fry-kun

  • Newbie
  • Posts: 8
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #361 on: October 25, 2014, 02:50:45 am »
A quick search found this:

http://usb.brando.com/usb-2-0-male-to-micro-b-female-adapter_p02840c0042d015.html

There has to be others as well.

My search was "usb a male to micro b female adapter" and click on images to make sure the micro b is female.

Awesome, thanks!
I tried searching but couldn't figure out the magic combination that didn't make me manually filter by photos (gave up after ~20min)
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #362 on: October 25, 2014, 03:01:40 am »
Or: http://www.amazon.com/gp/product/B009AWA3VK/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1

Probably made in the same factory. I got one of these and it took about 2 weeks to arrive. I figured for ~$1.50 why not...

It works fine with my Seek TC.

  ...ken....
Thanks for the link, actually that pushed me to get one for my seek to start probing the protocol, but it seems like it will come from the slow boat (delivery estimate Nov 20th to Dec 9th) but i'm in no hurry since I got this to winterize my home and I won't be messing with it until I'm done.
 

Offline Sensor Geek

  • Contributor
  • Posts: 15
  • Country: us
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #363 on: October 25, 2014, 12:40:53 pm »
This gradient issue is pretty bad. Does anyone else experience this? Seems almost unusable for most practical situations, although I can now use this $200 camera to verify that my teapot is hot vs. using my finger before.

Here's a door with a leak around the bottom and it has a dual-flap pet door that's a little cool too. Comparing E4+ with seek (no comparison of course - to be expected), but look at what happens when I turn the Seek upside-down. Now where is the door leaking?

It seems to me this has potential to be corrected by improved calibration routines in the app/firmware, but I haven't contacted them yet to see if that's in the works. Has anyone already had that conversation with them?
 

Offline ricksastro

  • Contributor
  • Posts: 30
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #364 on: October 25, 2014, 02:30:49 pm »
This gradient issue is pretty bad. Does anyone else experience this? Seems almost unusable for most practical situations, although I can now use this $200 camera to verify that my teapot is hot vs. using my finger before.

Here's a door with a leak around the bottom and it has a dual-flap pet door that's a little cool too. Comparing E4+ with seek (no comparison of course - to be expected), but look at what happens when I turn the Seek upside-down. Now where is the door leaking?

It seems to me this has potential to be corrected by improved calibration routines in the app/firmware, but I haven't contacted them yet to see if that's in the works. Has anyone already had that conversation with them?

Agree 100% with this.   This is obviously a thermal gradient since it gets worse the longer you are using it.  Makes the min/max thing useless under non-extreme circumstances.

Makes you wonder what exactly they are using the dark frame with the shutter for?   Wouldn't this record the thermal and non random noise to allow subtraction?    Hope like heck there will be a SW fix for this!
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #365 on: October 25, 2014, 03:02:14 pm »
Nothing a hacksaw and a multi conductor ribbon cable wouldn't solve  >:D

But seriously, it would be VERY useful to see if there is an issue with thermal creep on that PCB towards the microbolometer.

The less savoury possibility is that the microbolometer ROIC causes the localised heating and the camera simply cannot cope with the temperature gradient in its FFC routine. Such an issue should be correctable in software / firmware but it may be why the shutter is already operating so frequently. There is also the weird possibility that the shutter itself is the cause of the issue ! Most shutters on professional TIC's are metal. It may be that the shutter, when sitting in the rest position, is exposed to localised heating across its surface that in turn generates a temperature gradient in the FFC table and so screws up the images. Worth investigating for those who have opened their SEEK unit already and are willing to experiment.

Some simple experimentation with some freezer spray would be interesting. Spray it along the PCB between the micro and the microbolometer to create a thermal barrier to heat then see what effect it has on the image. It may well send the microbolometer in the opposite direction due to the transmitted cold ! A metal heat shunt fin attached via an electrical insulator (but thermally transmissive) to the PCB between the micro and the microbolometer is another possibility to play with but I am uncertain as to its effectiveness.

The trouble with any high sensitivity sensor is local 'pollution' of the signal it is processing. We could be seeing thermal pollution of the microbolometer and that can really only be best treated by thermal separation from the pollution source.

It will be interesting to hear what SEEK have to say on the matter as this is schoolboy physics stuff and I cannot believe that the issue went unnoticed in development. Most odd I must say.

As has been stated, an FPA with a thermal gradient across it that cannot be compensated adequately is pretty useless for anything more than toy applications.

Another test I have just though of would be to blind the camera lens with a piece of thermally opaque material. Does the thermal gradient show itself ? Remove the shutter mechanism with the blinding plate still in place to ensure that a FFC image is available to the sensor array. Does the gradient still remain ? Then cool the PCB with freezer spray in different places and watch the effect on the temperature gradient over time. Lots of basic physics fun to be had  :)

AND another thought....is heat pollution entering the shutter access port and polluting the image collected by the microbolometer ? I recall that there are open ports in the lens holder. Does the unit behave itself better if operated outside its small metal case with a fan blowing any radiated heat away from the sensor array ? Damn I wish I had my SEEK to play with !

Aurora
« Last Edit: October 25, 2014, 04:17:05 pm by Aurora »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #366 on: October 25, 2014, 03:46:16 pm »
For those still following.... I have had another thought on the 'plume' on the LHS of the image.... magnetic field interference.

The issue appears to occur on the side that has the shutter solenoid adjacent to it. We know that the FFC shutter is a NC type so needs power to the solenoid to pull it out of the signal path. A weird way to drive the FFC you may think, but this is sometimes done in an effort to protect the microbolometer from long term exposure to energy through the lens when not in use. Say if you left it pointing at a radiator or fire. I know that the solenoid is energised but is it a DC or high frequency AC signal ? Using an uneven duty cycle ac signal would save power but would create an alternating field in very close proximity to the microbolometer and ROIC on the FPA. Such an alternating field has the small potential to cause interference on the image data. This is normally only experienced in very high field strength environments and often presents as noise bars on the image. It is still a possibility with this new sensor technology however. The above test involving blinding of the lens, then removal of the shutter solenoid should reveal any interference issues from that component.

There is a potential flaw in my thinking however.....is the SEEK thermal lens inverting or non inverting ? If it is inverting, the 'plume' is actually occurring on the side of the microbolometer furthest away from the solenoid. The same applies to heat pollution through the shutter port etc.

Aurora
« Last Edit: October 25, 2014, 04:10:52 pm by Aurora »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8240
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #367 on: October 25, 2014, 04:10:51 pm »
After seeing the pictures posted earlier I thought the sensor would be on some elevated platform above the PCB, but Mike's teardown confirms that the die is actually bonded on the PCB itself.
Quote
As has been stated, an FPA with a thermal gradient across it that cannot be compensated adequately is pretty useless for anything more than toy applications.
I think this really was intended as a "toy" product, built down to a price. (And as the comparison with Flir One shows, "resolution isn't everything" either...)
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #368 on: October 25, 2014, 04:27:40 pm »
@Amyk,

I am starting to think the same BUT for those of us willing to experiment, the SEEK may offer a good platform on which to build. It has a potentially decent microbolometer with ROIC and USB connectivity (albeit some mobile phone OTG dedicated version). That does stack up well against the FLIR One or even the bare LEPTON in terms of bang for your buck.

As a hobby project I would happily hardware modify the SEEK to see what I could get out of it in terms of performance improvement, Sadly coding is not my skill set so I would leave that to others more able than me.

Noise has always been the challenge with microbolometers.... they are a naturally noisy sensor technology that is often tamed with signal processing. This new 12um sensor array may be suffering from a flaw in its basic design however.... if so, it will always be a compromised solution. The SEEK is the MKI product and we may be at the beginning of some excellent development so no need to be down hearted  :)  I was/am more than happy to spend $200 on such a 'building block' module  :-+
« Last Edit: October 25, 2014, 04:29:46 pm by Aurora »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Online mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 13695
  • Country: gb
    • Mike's Electric Stuff
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #369 on: October 25, 2014, 04:45:11 pm »
Surely any gradient should be corrected by the NUC?
 
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #370 on: October 25, 2014, 05:00:51 pm »
@Mike,

Should but Isn't by the looks of things  ;)

The NUC calibration table is the COURSE correction element and the FFC is the FINE correction element that make up a decently equalised output from the Microbolometer. The FFC is supposed to correct for SMALL drift levels in the microbolometer. Large external thermal stimulators polluting the microbolometer can push the required correction beyond that programmed for the FFC event. The result is a non flat field  :( I would expect an 'FFC out of range' error event to be produced in a professional TIC but this is not such a camera. This may be correctable in a new firmware or software release by increasing the FFC pixel offset range but that is sticking a band aid on the problem as the cause of the non uniformity across the FPA should really be addressed.

If the FFC shutter performance is the cause of the issue then that should be fixed as it destroys the FFC effectiveness and CANNOT be compensated for adequately. It is supposed to be a flat field not a gradient field  ;D

For those wondering, a TEC attached to the PCB at the rear of the FPA is unlikely to solve the problem as it would be a relatively uniform heating/cooling area on a non uniform temperature area so a gradient would still exist, just shifted up or down in centre temperature. This would only be an idea for addressing thermal creep in the PCB anyway.

This is a neat little challenge to get ones teeth in to. First we need to identify the source/cause of the non uniformity. Hopefully it is an influence EXTERNAL to the FPA and ROIC sandwich.

Aurora
« Last Edit: October 25, 2014, 05:08:48 pm by Aurora »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline eneuro

  • Super Contributor
  • ***
  • Posts: 1528
  • Country: 00
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #371 on: October 25, 2014, 05:16:13 pm »
... but look at what happens when I turn the Seek upside-down. Now where is the door leaking?
Maybe there is more trivial explanation to this- some kind of weird software bug, so while such use case scenario (turning Seek up-down) is not implemented well and missed something during testing. ;)

It looks like that we need simply more raw data from this sensor to see what sensor sends to its main PCB MPU and than also capture what is send via USB and third final output image presented on Android device screen.
When we had three such images, than we could find maybe even some bugs in Seek Thermal app software, which looks like they made to be able to sell this thing, while lack of temperature scale on screen with pallete colors mapped to min/max temperature makes this thing simply useless  ???

I would'nt be supprised if it were some kind of software bug.
Did you tried very slowly rotate in the rate for example 45*degrees/min  CW and see what happends?
Maybe you flipped it too fast and they did not notice it is rotated 180*  :-DD
Another thing when you face this cam perpendicular to the region of interest-so this door claps on the bottom (not parallel to doors), so those cold claps in the middle on the screen-than still you have such strange gradient effect?
12oV4dWZCAia7vXBzQzBF9wAt1U3JWZkpk
“Let the future tell the truth, and evaluate each one according to his work and accomplishments. The present is theirs; the future, for which I have really worked, is mine”  - Nikola Tesla
-||-|-
 

Online mikeselectricstuffTopic starter

  • Super Contributor
  • ***
  • Posts: 13695
  • Country: gb
    • Mike's Electric Stuff
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #372 on: October 25, 2014, 05:40:03 pm »
BTW as the coil only pulls about 2mA I doubt there is any thermal influence from it
Youtube channel:Taking wierd stuff apart. Very apart.
Mike's Electric Stuff: High voltage, vintage electronics etc.
Day Job: Mostly LEDs
 

Offline callipso

  • Regular Contributor
  • *
  • Posts: 87
  • Country: cz
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #373 on: October 25, 2014, 05:44:30 pm »
The thermal gradient on the shutter itself seems like the most plausible scenario to me. We know that the FFCing can make BIG changes in the imagery, as was seen when Mike operated his without the shutter.

It's probably heating nonuniformly from the outer shell and transferring this gradient to the imagery.

@Mike,

as you already have one with the shutter removed, could you try to FFC it against something you know is thermally uniform? Or maybe FireFlir the back of the pcb?
I wonder... **BOOM
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Yet another cheap thermal imager incoming.. Seek Thermal
« Reply #374 on: October 25, 2014, 05:48:32 pm »
@Eneuro,

The image from Sensor Geek is the one I am concerned about. A non uniform thermal gradient across the image. That simply should not exist after a FFC event. It is accepted that such would occur over a period of time, normally around 3 minutes but not as quickly and permanently as it appears to be in the SEEK unit.

I am borrowing Sensor Geeks image for information.

As you have stated though, we need more comment and examples from other SEEK owners.
« Last Edit: October 25, 2014, 05:55:50 pm by Aurora »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf