Yet another cheap thermal imager incoming.. Seek Thermal

[heavybarrel]

Has anyone been able to find the emissivity settings in the IOS app? They were there when i downloaded the original app months
ago. Now that i finally have the camera they seem to have disappeared with an update.

[Seeker]

Hi guys and gals...my first post, season's greetings. Just ordered an Android one of these (Amazon US, via an intermediary...not on sale in UK yet, harrumph). Bit disappointed to read of the gradient issues found with it--according to Mike'selecstuff YouTube video it's been put down to lens alignment, is this the definitive answer?
I've read the last few posts of this thread, the gradient issue seems to have slipped out of the consciousness somewhat. Without trawling through the entire thread, is this issue finally down to the lens?

Because a thought occurs--if it's down to sloppy assembly (misaligned lens tube) then some of these units must be perfect, surely...is the consensus that all shipped units have the same issue, or some do and some don't?

Looking forward to receipt (couple of weeks hopefully). Like member Aurora I have the Moto G and have just downloaded the Google Play (1.4.0.2) app for it.
Interesting thread! I've done a bit of electronics myself, but faddingly. Sometime enthusiast. If I need to mess with the shutter I may be able to stretch to that...quite nifty with tweezers and mini engineering and whatnot.

[-jeffB]

I certainly haven't investigated in as much depth as Mike and others, but my impression from the thread has been that the gradient issue is due to non-uniform heating of the internal shutter -- part of the shutter is close to a warmer component when it's open, so when it closes, the camera sees a non-uniform surface temperature, and thus does a uniformity "correction" that's incorrect.

I've wondered whether it would be possible to add a small blackened copper or silver shield that sits between the open shutter and the electronics, distributing the temperature more evenly. I've got no background in that sort of thermal design, though, so I have no idea whether it would be practical -- I don't even know whether there's room to do it.

If that's not practical, it may be possible to develop a per-camera post-processing correction. That would be messy, though; it would vary as the camera warmed to operating temperature, and might be strongly dependent on ambient temperature as well (since that would affect the internal temperature differentials).

I've been spending more time working with a modified FLIR E4 lately. It doesn't have the gradient issue, of course, but I also notice that it gives much more accurate temperature readings in general -- it looks like the Seek does very poorly with temperatures below ambient, although it does better with temperatures at or above ambient.

At this point, I'm thinking that the Seek is great for entertainment, and for some qualitative tasks -- it can certainly show hot-spots, thermal footprints, and the like. It's not much use for quantitative measurement, though. I still think $200 is an amazing price point, but if I'd paid (say) $400, I'd probably be pretty unhappy with its issues.

[Uho]

Gradient large. Move the tube - reducing the gradient of up to 4 degrees.  . Need an external shutter. With external shutter works better.  The noise is large. No manual adjustment range. Without modernization difficult to use.

[Seeker]

I certainly haven't investigated in as much depth as Mike and others, but my impression from the thread has been that the gradient issue is due to non-uniform heating of the internal shutter -- part of the shutter is close to a warmer component when it's open, so when it closes, the camera sees a non-uniform surface temperature, and thus does a uniformity "correction" that's incorrect.

Hmmmm, if you skip to 18.00 (of the 21min) video Mike seems to have ruled out the shutter as the culprit, and blamed the lens alignment after having moved the lens housing around, to virtually eliminate the gradient.

Video link--(can't find the editor option)

[Uho]

I have seen. That's what I did. I completely disassemble the unit. Leaving only the shutter. Gradient disappeared. So many reasons gradient simultaneously. Heating of the electronic board. Reflection from the walls of the tube.

[-jeffB]

Hmm. I'll confess that I hadn't watched this. (I have a hard time watching videos, as it's often difficult for me to understand the narration, and it's strenuous to follow along at someone else's pace for a long period.)

If I understood an earlier part of the video correctly, Mike was saying that the gradient seems independent of unit warm-up. My impression with my unit is that it's very warm-up dependent -- I see little gradient at the beginning, but a strong gradient once it's warmed up. That's what led me to suspect a self-heating issue, either unevenly warming the shutter or unevenly warming the optical assembly beyond the shutter. Mike's torn-down investigation might not accurately capture the effects of self-heating within the housing.

Again, though, I've spent a lot less time at this than Mike has, and I'm not sure I fully followed his video presentation. I'd recommend weighting his opinion over mine for the foreseeable future.

[Uho]

I have a maximum gradient at the bottom left. When heated, it increases. Do you  the same?

[Seeker]

From what I read so far, there are conflicting observations as to the possible cause of the gradient...damn. Would be most annoying to go down the "realign the lens tube" route only for it to not solve it. Be interesting to hear MikeStuff's latest conclusions. 

[mikeselectricstuff]

From what I read so far, there are conflicting observations as to the possible cause of the gradient...damn. Would be most annoying to go down the "realign the lens tube" route only for it to not solve it. Be interesting to hear MikeStuff's latest conclusions. 

Not had time to look further & unlikely to anytime soon, but I'm pretty sure it's lens alignment.

[Uho]

I took off the lens. There is only the shutter. Remained gradient. Take picture later.

[mikeselectricstuff]

Of course it's also possible that there are multiple contributory factors.

[efahrenholz]

@Mike,

You are correct. There are several factors, and each unit is a unique case. The gradient different for a lot of people, so if the automatedmanufacturing process is the way they claim it is, every unit should have the same exact gradient. Alignment of the lens is what is causing the problem. Depending on how the lens is out of alignment, it cause the inner wall of the lens housing to reflect thermal radiation back to the sensor. Keep in mind the sensor heat is what we are seeing. To fix this issue, there is a software fix, by taking a calibration frame while shutter is stored, pointing at as uniform surface. The other fix is redesigning the lens housing.

The problem all stems from miniaturization. Things still get hot, even small. The sensor is small, so a matching lens is small, and must be placed close to the sensor to achieve focus. So there isn't enough space to allow for radiant heat from the sensor to diffuse before it hits a surface and reflects back.

[frenky]

It's great to see that this thread has not gone cold...

Anyone care to guess what this thing will do?



Hint it's not a part of DeLorean.

[Seeker]

Something to do with creating a perfectly perpendicular flat field?

[frenky]

Well actually it's a "thing" for making precession movement.
I'll mount the seek module in the center of it. The result should be low noise high resolution images...
(Those little fellas on the edges are micro stepper motors)

[efahrenholz]

You are going to move the camera in circles to increase the effective resolution. Basically this is a delta platform.

[Seeker]

I took off the lens. There is only the shutter. Remained gradient. Take picture later.

I look forward to seeing the pics Uho.
Anyone think the inside lens tube would be better done in matt black? Might be worth a try.

[Uho]

Photo gradient. Only the shutter. Indoors. Dog at a distance of 7 meters. Use the object 12 degrees. The air temperature was -7 degrees Celsius. Noise in the cold anymore. Using the lens at 12 degrees to Seek termal.   http://youtu.be/XoiNYlhJ7Xk  At the beginning of the video gradient evident.

[efahrenholz]

@Uho,

Is the photo of the dog through a different lens? That is incredibly blurry. Can you explain a little what is going on in the last photo?

One of the things I wanted to do was use a different lens over the sensor to see if the gradient would show up. My suspicions were that the lens housing was of low emissivity. Basicall the white paint is reflecting the radiation emitted from the sensor. The sensor see's it's own heat. The tight space of the lens housing puts it very close to the detector, so this makes it very easy to essentially feel it's own heat. To give a good example, you put your hand next to a soldering iron. You can't feel the heat till you are just an inch away. It gives off heat no matter how far away you are, but it dissipates (loses focus) and you can't feel it. It behaves like visible light, so the rays scatter the further from thr source they get. But like proximity focus, you get close enough to the source, and you suddenly become aware of the heat without having to touch it. Well the seek detector is within proximity focus of the lens housing, so it can *almost* see the bottom inside of the lens housing. But all it looks like is a circular gradient. It wouldn't normally see the lens housing, it doesn't normally get very hot. But the detector heats up, reasonably warm, and the inside bottom of the lens housing becomes a reflected source. It then becomes the proximity focused heat we see on our screens. It's close enough to resolve the circular cut out area of the lens housing, but too far away to be sharp.

I welcome anyone to challenge this. Use a longer focal length, something that puts some distance between the lens and sensor. You'll lose the gradient.

EDIT: In addition, as Mike pointed out, there appears to be an alignment issue. This changes the way the gradient appears. I personally removed the lens housing, cleared the glue, and got the lens closer. This gave the sensor better focus out further. It also reduced the gradient. In theory, according to what I stated, this would cause the gradient to get sharper (proximity focus, remember?) But what is happening is the bottom inside of the lens housing is moving onto a blind spot. Take both your hands and make a circle with your index fingers and thumbs. Close one eye and make the circle in front of the open one. Start a few inches away. See the fingers? Of course, its reflected light. Image how bright they would be if your eyes actually created light. Start moving the circle closer. Notice something? You suddenly cant see certain areas of your circle. Move the circle around, theres your alignment issues. Keep moving them in. Too close and suddenly you can't see them at all. This is what's happening.

[Seeker]

Problem with that: beginning of the video (linked, bottom of previous page) shows the gradient from stone cold on MikeStuff's example.
And if the inside heat of the lens housing is being reflected and screwing up the calibration, wouldn't it show the shape of the housing, be symmetrical? I'm presuming the inside lens housing is circular...that should (I would've thought) be shown on the calibration image: instead it's always one corner.
Did I read that the shutter is before lens because of patent concerns? It wouldn't help temp uniformity that the lens tube has a hole in the side of it...
Just thinking out loud, dispose of as desired.

[efahrenholz]

The inside of the lens housing is not circular. It's a box shape with rounded corners. When I moved my lens housing around, the gradient shifted. The lens housing is so close that any movement of the lens housing greatly moves the detected gradient by a large factor. You could replicate this yoursself, if you are willing to remove the lens, let alone crack open the case. It's a physical design flaw, but it can be solved with software. Seeks software development team could allow an option to capture an external calibration frame and stop internal shutter events. Just give simple on screen instructions and a disclamer that temperature readings may be inaccurate. They have instead chosen to go the 'dummy' route and lock down the software, so it's a one size fits all solution. This is, in my opinion, the wrong way to go. The unit is aimed at the general consumer market, nothing commercial, and mostly for the average joe. You don't need to even understand what thermal is. It's priced low enough that even curious people with a little extra cash who want to see in heat will plunk down the doe. They can't risk the unit being too complicated, or they will risk sales. Basically, the gradient will never be fixed correctly.

Also, I wouldn't think there would be a patent against using a shutter in front of the lens. That seems like it's on the border of using a black body for calibration, standard practice, which any patent for that is long expired, since thermal camera inception. That would be like the first automobile to use rubber wheels. Then patent it so nobody else can use rubber tires. It's a lame way to control your competition but its probably expired long ago and is considered standard practice now.

[Uho]

In the last photo dog outdoors. It was snowing. Frost . Distance 7m. At the same distance in the room less noise. The video link used this lens. Focus 15mm. The sensor has a strong side illumination. He reacts to heat the side. Place the soldering iron from the gate. Gradient change. Gradient prevent us from seeing little difference in temperature objects.
  In the software enough to make manual removal of the gradient. Front of the lens to put a uniform surface and make balance. Or an external shutter. I bought a bolt from fotoaparata. To it is necessary electronic control board. Or return spring. In SEEK applied magnet.

[Seeker]

Apologies in advance if this is tripe (this stuff is above my paygrade), but is it possible the shutter is too small, or not quite the right shape for the sensor? There is indeed a big lump off the corner (presumably for clearance when moving)...if this is allowing a corner of the sensor to see the housing when calibrating, then moving the lens about (Mike's video (about 20.00min) when viewing a flat field could be merely cancelling out that calibration error, thus giving the (virtually) gradient-free image of his white card?
I can't see (sorry!) how a smaller, offshape shutter can calibrate the thing...sure, it (hopefully) can block the lens image completely, but the sensor must be able to see past the shutter, when the lens is blocked off... My little brain hurts.   

EDIT--I mean see the housing past the shutter...

[Seeker]

A pic of the MikeStuff housing for reference...may jog someone's ingenuity.