Yet another cheap thermal imager incoming.. Seek Thermal

[JcDenton]

Do you have a credible source to verify this?  This sounds like a 'my hairdresser has a friend who knows this beta tester that was told by an insider at [insert tech company] that the [insert tech product] will have usb 4.0 '

No verifiable source, I had only read a post on another forum talking about the flir1.  Could just as well be bs, I have no idea, which is why I was asking.

The second generation should be released this week and it is Droid compatible. I chatted with David a Flir representative last night about this and some other projects they are working on. They will adding a mil dot reticle on their scopes and it should be ready shortly. Then you can upload it from your PC. I prefer a mil dot and looking forward to this upgrade.

[Fraser]

@Mike,

I was surprised to see ZnSe lenses in 'high end' thermal camera lenses as well.

I first saw them used as internal elements in large military thermal  lenses and then in the FLIR PM 2xx and 3xx series.

I attach a couple of pictures.

The ZnSe material is relatively soft and easy to scratch. The close up picture is of a scrap camera that is currently for sale on ebay. Note the rings in the lens.... fresnel lens ?

I do not know why ZnSe was used but could be cost or in the case of military optics, transmission specs.

Aurora

[Fraser]

@-jeffB,

I regret I am not a chemist, just an end user.

The background to my knowledge about these lenses is that I have used ZnSe lenses for some years as a close-up auxilliary lens for my industrial thermal cameras. I was familiar with them and their fragility in terms of scratching. Unlike Germanium lenses ZnSe offered an affordable solution to close up work on PCB's

Move forward a few years......the FLIR E4 was released. Many of us on this forum discussed the various parts of the new Ex series, including the lens. There was some discussion about the size of the lens and whether it was Germanium. Without saying too much, I have a friend inside the industry and I discussed the E4 with him. He knows the design well  Being a great chap he directed me towards the Lightpath web site and said that a Chalcogenide lens was used in the E4 to save cost.

http://www.lightpath.com/images/Products/LightPath%20Infrared_Brochure_Jan2015.pdf

I then started reading about the Chalcogenide lenses used in modern thermal cameras. I read an article detailing a company that started production of the first Chalcogenide lenses that could replace Germanium lenses. These lenses were dark grey/black in colour and opaque to visible light. The clever bit was that they could be molded to shape which saves cost. No mention was made in any relates articles about ZnSe. My experience with ZnSe  is that it is yellow in colour and very wide in bandwidth so transmits visible light as well as IR and thermal wavelengths. It normally needs a filter between it and a microbolometer.

 http://ceramics.org/ceramic-tech-today/fraunhofer-imw-molded-chalcogenide-glass-approach-delivers-low-cost-infrared-lenses

http://www.gizmag.com/cheap-infrared-lenses-fraunhofer/23659/

I was not aware that ZnSe was a chalcogenide glass. What I do know is that GASIR is not ZnSe. Exactly what it contains may be covered by commercial secrecy ?

What I can say for sure is that the GASIR product is used extensively in modern thermal cameras and ZnSe is not, so GASIR must have some advantages over ZnSe.

A Google search on GASIR does produce results discussing it. Many are too technical on the Chemistry front for me.

i.e.

https://global.arizona.edu/sites/default/files/docs/experts/Lucas-Recentadvancesinchalcogenideglasses.pdf

Just found this article stating that Germanium is a significant element in GASIR, hence the dark colour of the lenses.

http://www.photonics.com/Article.aspx?AID=24318

Hope this helps.

Aurora

[WS-PI]

@Joe-c,
many thanks for sharing your software and sourcecode. I got it running today and I am impressed how good the pictures are already. Now we just need someone who is able to implement something like the TESTO superresolution technic http://www.thermalimager.co.in/super-resolution.php

[mikeselectricstuff]

If anyone's interested in poking about with the hardware, here are some pics of the bare PCB layers. all viewed from sensor side

Some interesting features on the die, but waiting for new bulb for my microscope to get some decent pics

[Monkeh]

How exactly did you get that apart so cleanly?

[SeanB]

Looks like it was milled down, likely with a surface grinder and a grinding wheel, or with a standard small end mill at high speed.

[Fraser]

@Mike,

Nice pictures. thank you 


I will be interested to compare your pictures with what can be revealed by the Faxitron. Guaranteed X-Ray will not show the layers as well as physical disassembly but having you images will show me the Faxitron's limitations.

Aurora

[mikeselectricstuff]

How exactly did you get that apart so cleanly?

grindy wheel on dremel and 600 grit paper. 

[stak]

I don't know any FLIR reps, but I got my information from their press release...

    http://www.flir.com/flirone/Press/FLIR-ONE-CES-Press-Release.pdf

It's in the middle of page 2.

Are you sure about this?  I read a post where someone said he recently talked with a flir rep, and said the android version was coming out in a week or so.

[miguelvp]

The FLIR ONE will be available for pre-order online from FLIR.com mid-year and will be available
through retail partners across the globe later this year. For more information, please visit
www.flir.com/flirone2015

So Aug/Sept is about right.

SDK announced in that press release to be available for Android developers in spring and iOS already was released back in Dec last year.

I guess they are playing ball with Seek since Seek also announced an SDK that we are all waiting for, let us hope that Seek releases their SDK before FLIR does, because being first is important because if you are last then your developers might be too busy playing around with the competition's SDK.

But the ball is in Seeks possession they have 4 months to be first on the SDK front and 7 on a 2nd generation device.

[JcDenton]

Says the same thing on their facebook as well, so I guess that guy was smoking something.    Oh well, just bought the seekXR, so i'll see how that performs, and possibly get the flir android later.

[efahrenholz]

I don't think seek is really pushing the sdk issue. It's definitely something they are working towards, as it shows in delayed app updates, but there isnt much motivation to get it out. Since there is already hardware support, and the sdk will be designed for the android and iOS version, there probably won't be a lot of developers clamoring for this kit. I don't even see them providing an sdk for Windows, but that would be very cool of them.

[miguelvp]

I think there is potential in the Android front, there is OpenCV for Android and that can do image registration between the image and the visual camera. Also a better interface amongst other features.

[mikeselectricstuff]

@Mike,

Nice pictures. thank you 


I will be interested to compare your pictures with what can be revealed by the Faxitron. Guaranteed X-Ray will not show the layers as well as physical disassembly but having you images will show me the Faxitron's limitations.

Aurora

Although you can see the layers, you can't tell which track is on which layer, so not as useful as you might have thought it would be

[efahrenholz]

Mike it's only a matter of time and someone will have mapped the circuit with components in place. Having the layer images available was the key.

[frenky]

I'll try to make vector drawings of all layers so that we can stack them on top of eachother.
With semi transparent layers it should be easy to reverse engineer whole circuit.
Layer 2 is in the attachement...

[Rasz]

If anyone's interested in poking about with the hardware, here are some pics of the bare PCB layers. all viewed from sensor side

Some interesting features on the die, but waiting for new bulb for my microscope to get some decent pics

Vias to nowhere?
thermal sensor connector pin 8, 9 and 12 goes to dummy vias?
that leaves pin 7 10 and 11 as the only real data connections
one of supplies is pin2
ground is pin 4 6 and 18

[eneuro]

With semi transparent layers it should be easy to reverse engineer whole circuit.

It is more interesting if new Seek... has the same PCB and circuit... didn't checked latest software changes if any (I mean MD5 check sums and binary comparisions of *.bin MPU files),
to see if there were changes in binary MPU code with new Seek released 

BTW: It could be nice to put somewhere, maybe even in first post dates of new software relases and links do archive *.apk  (ZIP files) with reported hardware versions and software available.
I wish Seek hasn't got any release notes aka README file with official software releases and links to archive pointing to his own website to ensure this what you find in playstore (or somewhere in internet) is what we need and is not tweaked by someone....it should be digitaly signed and official PGP public key provided on Seek website to be able to verify if needed that we downloaded official relased Seek software....

In this thread on Seek website we have:
http://support.thermal.com/hc/en-us

Where do I find the Seek app?
The Seek Thermal apps are available in the Apple and Google Play app stores.
    Seek Thermal for iOS
    Seek Thermal for Android*

*The Android APK is not available for direct download outside of Google Play, and cannot be provided by Customer Support.

  in modern internet age is it?
*The Android APK is not available for direct download outside of Google Play, and cannot be provided by Customer Support. 
This f*ing app should be included on small USB or CD in the format of credit card inside box when you buy this device? 

[efahrenholz]

@eneuro,
I assume they didn't supply a disc or USB flash drive because of the cost and fragmentation it would cause. IOS devices in stock working order can't even access apps outside the app store. Why spend the money for something that will only work for half the user base when both platforms have a successful infrastructure in place to do all the hard work of getting the software to the device? Supplying a disc means that you have to update the software for each batch being sold when an update is released, which means fragmentation. That's a nightmare. I don't know what Internet age you are in, but the days of people having to hop on a computer to download software for devices to work has left long ago. Even TVs are hooked up and populate the app market for you.

[mikeselectricstuff]

I'll try to make vector drawings of all layers so that we can stack them on top of eachother.
With semi transparent layers it should be easy to reverse engineer whole circuit.
Layer 2 is in the attachement...

Note some copper is missing on the ground plane (l3) due to over-Dremelling

[miguelvp]

@eneuro
The Android APK is not what you think it is

http://en.wikipedia.org/wiki/Android_application_package

[frenky]

I've drawn middle two layers in Inkscape. File is in the zip if anyone wants to try drawing the seek scheme...

[eneuro]

The Android APK is not what you think it is

I do not know what you think what I think about *.apk, but... I simply upack it using unzip and get what I want- *.bin Seek MPU raw code 

$ md5sum *.apk
4f46662903727d09079da60ea27a92c0  com.tyriansystems.SeekThermalSeek-Thermal-1.6.0-20141127v102900.apk
$ unzip com.tyriansystems.SeekThermalSeek-Thermal-1.6.0-20141127v102900.apk
Archive:  com.tyriansystems.SeekThermalSeek-Thermal-1.6.0-20141127v102900.apk
...
inflating: com/tyriansystems/Seekware/SUBI_LPC43XX_LPCOpen_1.2.0.0.bin
inflating: com/tyriansystems/Seekware/SUBI_LPC43XX_LPCOpen_1.3.0.0.bin
...

This is this fake version with broken Seek app - If I have MD5 of *.apk I can easy verify if this downladed Seek app is what I want to install or not, because of I'm not idiot who install everything which is published in Google Play store or on other site...

[joe-c]

Hello again,

You did not try to eliminate the gradient? I have a large gradient.

I hope this gradient disappears while using a kind of reference or gain map.

Well I tried to implant a gain map...
I have taken a averaged cold image (Tec with around 15°C) and a averaged hot image (heating plate with around 120°C).
Than I just take the value of each pixel with HotVal-ColdVal as a gain map. I experimented a little bit, and make than a processed Gain map (float array).
With here the values are based on 1... 1.3 will make the pixel higher and 0.8 will reduce the pixel after processing.
But I cant get it working now... maybe some of you has a better idea or find a mistake.

Now the reference image (the similar way to miguelvp) seems the only good way now.
That means... take a averaged image of an even heated surface and pres on "take actual Frame as Reference".
Than the new image will be processed like NewFramePixel-ReferencePixel+MinValue+Offset (the MinValue was from NewFrame). The output has no gradient and very less pixel artifacts... but just for a short time period.
I take a picture from my raspberry Pi... and a picture with my Ex... well here a huge difference.

Like the first version, the frames could be stored in 2 ways (both in tab "TestArea"):
- a big file as readable value separated with a #
- a *.dat file with binary saving (2byte 1 value) it has lower size
for viewing a *.dat file (or *.SAT file from some thermal cameras) you can use the DevTools (C# open source):
http://joe-c.de/pages/posts/programm_dev_tools_133.php#bilder

and now just a small success... the usb reconnect works now.
The defect pixel map was stored in editor readable format (Pixel ok: - pixel defekt: #).
but this map is not used now.

Btw some idea... has anyone tried to use another byte command to get a frame?
Now we calculate some processed image to process our own pictures.
But a better way was a true raw sensor data stream (maybe just another byte command?). This will end in a nicer frame rate, because you can use each frame you get.
Ok, you have to implement your own shutter for NUC... not a huge thing, old camcorders has nice shutters, that easy could placed in front of the lens.

@Mike... do you sacrifice your seek just for getting this layer images?
Well... thanks.
It would be interesting if this sensor could be used with another µC...like a stm32f4xx.