Is it possible yet to buy sub-$300, 30Hz, (US only) thermal camera modules?

[David1234]

TL;DR; Is it possible yet for a US citizen to buy a lag-free, sub-$300 thermal camera sensor without having to buy them in bulk?

I've done a lot of FPGA work with the FLIR Lepton 2, 2.5 and 3 but the problem with them is that the only thing I can find and buy online is the 'international' version, with the built in 0.5sec lag, which currently makes my product pretty useless. The lag is really obvious when you overlay a camera feed over it. It's been about a year so I was wondering if FLIR, or someone else, has finally made a lag-free thermal camera sensor available in quantities of one.

I have bought other thermal cameras that fit the requirements, like a $300 320x240 tank scope on eBay, which is great, but it is a one-time-buy kind of thing. I can't build a product around it because I don't know how many I can get.

[Vipitis]

The Lepton from TLP are 30hz.... But I don't think you can get only the Lepton

[Fraser]

I have to admit that FLIR's marketing policy on the Lepton cores mystifies me.

They have developed a very capable (for its size and cost) thermal imaging core. It has the potential to find its way into many useful products and systems which you would think FLIR would want to facilitate.

What I have seen though, is significant reservation on the part of FLIR to release the cores onto the market in a manner that makes them a popular sensor. Let me expand on that comment though.... some bullet points...

1. The Lepton was/is only available through a few specific third party component suppliers.
2. The cores appear to be limited to <9fps even when sold within the USA. >9fps cores are to be found in finished products as we know.
3. Many products from FLIR and 3rd parties limited themselves to the lower resolution Lepton 2 that is unlikely to impress the retail market. Why did they not use the better 160 x 120 pixel Lepton 3 when it became available ?
4. The Lepton core may be capable of decent imaging yet FLIR appear unwilling to release its full imaging potential in the software they use with it. The F1G2 and F1G3 phone applications are prime examples.
5. FLIR do not, as far as I can tell, produce a Lepton Demo or Development board. They only provide a breakout board. I find this particularly surprising as FLIR did produce a marketing gimmick in the form of a demo camera for trade shows etc. I am used to companies producing a development or evaluation board for their front line products that they wish OEM's to adopt. FLIR obviously think differently 
6. The Lepton is supplied as a complete core with lens and FLIR do not offer a lensless version for OEM's to marry up to their preferrred optical block. If you think about Visible light PCB cameras, they come as the bare bones PCB with imaging chip mounted but no lens holder, PCB with lens holder mounted (M12 or C mount) or the complete PCB camera with lens fitted of your choice. Why FLIR do not wish to follow this well trodden path I do not know. The Lepton is hampered by its relatively basic silicon doublet optics.
7. There is no official PC support for the Lepton core ..... no reason given. Crazy, plain crazy. Windows and Linux users would love such an option in order to develop amazing products. FLIR prefers to leave any and all such PC connectivity development to 3rd parties.
8. FLIR do offer an SDK for the Lepton, but from what I have heard, it is somewhat 'lacking'

So from the above bullet points I am left feeling underwhelmed by the Lepton core. It is almost as though FLIR created a great product and is either too scared that it will damage other camera sales or truthfully do not know what to do with it ! An unusual situation. I would have hoped that FLIR would sort out their production issues during the F1G2 release and then move to mass production of the Lepton 160 x 120 core with a view to placing it in any market that could potentially have a use for such technology. It should be made as common as the low resolution pyrielectric sensors used in PIR sensors or at least the Panasonic and Melexis sensors.

There is, of course, the possibility that the profit margins on Lepton production are so poor that FLIR does not really want to sell the cores on their own and would rather incorporate them into more expensive products to increase the profit margin ? We know a retail Lepton core costs almost as much as a FLIR One Gen 2 so OEM's would want significant discounting in order to incorporate such a core into their designs. All conjecture of course.

Fraser

[Fraser]

For info, FLIR produces a Fire fighting camera called the K2. It is equipped with the Lepton 160 x 120 pixel core and two temperature ranges (sensitivities). IIRC that camera is also limited to <9fps. Fire fighting cameras are normally 30fps or 60fps with an option for the <9fps where regulations force such. I have to wonder why a 30fps K2 is not marketed. Then again, maybe it is available and I missed it.

FLIR wanted to produce an affordable Fire fighting camera that could lead to all fire fighters having one as personal equipment. Cost is around $1200 each.

https://www.flir.co.uk/products/k2/

https://www.flirmedia.com/MMC/CVS/Firefighting/FF_0016_EN.pdf

Fraser

[Fraser]

OK, I am going to float a really way out thought.....

What if the Lepton Core is not actually available from FLIR with a >9fps frame rate, for whatever reasons ?

We know at least one gun sight manufacturer provides a Lepton based sight running at >9ps for sales within the USA. Do we know for a fact that those Leptons are "as supplied" by FLIR or has the company employed someone to reverse engineer and hack the cores ? Such would not be illegal if sales were controlled as required for a high frame rate core.

Just thought I would float the possibility as the K2 example above suggests to me that FLIR do not ship >9fps cores even in the USA and even when a product needs such, like the K2 Fire fighting camera.

Fraser

[Fraser]

Out of interest, has it been proven beyond doubt that the high frame rate TPL thermal sight contains a FLIR Lepton core ? Other, similar cores are available from China. Just not usually as good in terms of performance.

Fraser

[Fraser]

I just did some googling and BH photo definitely state it uses a 80 x 60 Lepton and the frame rate is 30fps.

https://www.bhphotovideo.com/c/product/1431156-REG/torrey_pines_logic_tp_t20_01_t20_thermal_imager.html

Fraser

[David1234]

I did some research on Torry Pines Logic as well, about 6 months ago. Apparently the only 30Hz core they have is the 80x60 shutterless lepton 2. Their products that use the lepton 3/3.5 are 9Hz. I contacted FLIR as well at the time, and IIRC, they don't sell the 30HZ version of the lepton 3/3.5 at all.

I did some oscilloscope tests on a disassembled Seek XR and the core is actually outputting frames at 30Hz. They just use software/firmware to limit the frame rate. At the time, I didn't feel like continuing to reverse engineer the seek sensor would be worth the trouble, as it would just result in a really finicky, hacky product.

The Korean company, Thermal Expert, sells a lot of interesting versions and configurations of 30Hz and 60Hz thermal cameras but I'm unsure of what their prices actually are, how to buy any of their products, and whether or not I would actually need some sort of license to buy one as a US citizen.

Another option could be a large stock of used, cheap 30Hz imagers on eBay but I only ever see one or two of any given model.

[Bill W]

Much depends on whether you need a current generation core.  The main source of older cores is redundant fire service imagers which (to you in the US) would offer various Lockheed/BAe VOx, Raytheon BST and Raytheon ASi (2000 and 3500).

All are 30Hz
Most are 320x240

The 'stock' in the UK/EU is slightly different, skewed as it is by the respective core sales to the local suppliers.  Less VOx more Raytheons and ULIS.

The problem is that as a US person there would be export issues from the UK/EU, not a problem getting approval in principle but more that as a potential supplier I am not sure I would/could do the paperwork as a private individual.

Bill

[cq-317]

This is an interesting topic. Products exported to China need to be limited to 9Hz according to ITAR. However, many manufacturers in China can produce their own thermal image sensors. The highest resolution is 1024*768, the lowest frame rate is 25Hz, and the Products  with resolution of 1920*1080 also coming soon.  thermal image sensors below 25 Hz can't be produced In China.

[mikeselectricstuff]

It's a while ago but I did get an email from someone claiming they were buying 30Hz Lepton cores from Flir.
Agree that their marketing was dumb - it took GroupGets ( Who I think had a chance contact with someone at Flir) to get them available, and then a long time later they were in Digikey - I have a feeling even that might be via Groupgets.

If you look at all the chip-level reverse engineering that's been done in the last few years, it wouldn't surprise me if a way to get a standard Lepton running faster could be found, but it needs someone with teh skills & resources and sufficient desire to do it.
Obviously it's slightly complicated by the whole export regs nonsense.

BTW has anyone researched the legal position on someone other than the manufacturer releasing technical info on a thermal imager?

[mahony]

OK, I am going to float a really way out thought.....

What if the Lepton Core is not actually available from FLIR with a >9fps frame rate, for whatever reasons ?

We know at least one gun sight manufacturer provides a Lepton based sight running at >9ps for sales within the USA. Do we know for a fact that those Leptons are "as supplied" by FLIR or has the company employed someone to reverse engineer and hack the cores ? Such would not be illegal if sales were controlled as required for a high frame rate core.

Just thought I would float the possibility as the K2 example above suggests to me that FLIR do not ship >9fps cores even in the USA and even when a product needs such, like the K2 Fire fighting camera.

Fraser

Probably but what I do not get here is the fact, that internally the core seems to be running at least at 3x the final output frame rate. I mean, you even get an actual Output Frame rate of 3x8.3=26.4 fps (according to the engineering datasheet) but only every 3rd image is unique/valid. Why would anyone do this if the internals were not meant to perform at this framerate?
I also did some Tests with the Lepton3's I have to see if the read-out is done as snapshot or line-by-line - from the Images you could see that it is snapshot, BUT: not just one but I think 3 or 4 of them that may be averaged together into 1 single output frame. I do not have an image at hand but may provide one tonight but anyone with a Lepton3 based camera may be able to test it with a pen or similar item moving it fast up an down in front of a cooler background - you will likely see multiple 'pens' in a single output frame if you capture one...

[Fraser]

Most Microbolometers operate at 30fps or higher frame rates. The frame rate reduction occurs in the image processing stages. Microbolometers are read out in Parallel or serial mode. Each has its advantages and disadvantages.

IIRC the Lepton 3 outputs 4 quadrants to form a single frame which is novel. The Lepton 2 outputs single complete frames. It is almost like the Lepton 3 uses the Lepton 2 image processing technology that can only cope with 80 x 60 pixels at a time. I have to wonder if higher frame rates overtax the Lepton 3 image processing engine, hence >9fps is limited to the 80 x 60 Lepton 2.

I have not spent much/any time investigating the Lepton so others likely have a much better understanding of its design and operation.

Fraser