Author Topic: Rhetorical question - Will there be a QVGA or VGA Lepton?  (Read 562 times)

0 Members and 1 Guest are viewing this topic.

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 8538
  • Country: gb
Rhetorical question - Will there be a QVGA or VGA Lepton?
« on: September 28, 2019, 11:34:30 am »
As in the topic subject, this is a rhetorical question but worth considering.

To date FLIR have designed and produced the Lepton 2 with 80 x 60 pixels and the Lepton 3 with 160 x 120 pixels. There is no doubting FLIR’s skills in miniaturising a thermal imaging core down to a size smaller than a sugar cube, but why have they stopped at 160 x 120 pixels ?

Let us consider the possible issues involved in the Lepton development process.

1. To create a miniaturised thermal core the microbolometer needs to be as small as possible whilst still maintaining enough sensitivity to operate in an acceptable manner. The introduction of 12um pixel Microbolometers made the Leptons diminutive size feasible. Could even smaller pixels be used to increase the resolution on a similar sized die ? Well thermal sensitivity would decrease significantly so great improvements in the processing of the tiny output signal would be required to produce an acceptable image of a scene. This would be especially important when viewing low Delta T scenes containing little thermal energy.

2. A Microbolometer requires optics to present the thermal scene to all of its pixels. Thermal imaging cameras traditionally used conventional optical elements made from Germanium, Zinc Sulphide, Zinc Selenide or, more recently, Chalcogenide IR glass. The Lepton uses Silicon diffraction type optics. This approach was chosen to reduce production costs and to simplify the manufacturing process. The Silicon Diffraction lens is far from ideal for LWIR Imaging but it is adequate for the relatively low resolution of the Lepton 2 and Lepton 3. Would it be adequate for a QVGA or VGA Microbolometers ? That is not known but the current lens design would need to to be revisited as it has inadequate resolution in its current form. Could a conventional lens design be used instead ? Yes, most definitely, but it might increase the size if the lens assembly and would increase the production cost.

3. One of the enemies of the thermal imaging camera is instability and self heating in the Microbolometer assembly. The Microbolometer die is usually bonded to the top of the Read-Out Integrated Circuit (ROIC). The ROIC contains all that is needed to read the rows and columns of the FPA followed by the required stages to create the required serial or parallel data output signal. The ROIC contains pixel bias, pixel timing and multiplexing stages. All these semiconductors in the ROIC create thermal energy that could adversely effect the Microbolometers performance. Self heating of the Microbolometer pixels (from read-out current) and the thermal output of the processing stages within the ROIC are carefully managed. Any increase in the active component count in the ROIC changes the ROIC thermal budget and must be well managed. Adding the ADC stage to the ROIC was achieved some years ago. This moved the normally external Analogue to Digital conversion stage onto the ROIC die. The move was not without technical challenges as the ADC creates an increase in thermal output from the ROIC into the Microbolometer die. Increasing the number of pixels on the die, and associated read-out electronics to handle such, would require careful thermal management in the miniaturised confines of the Lepton core.

4. The Lepton 2 outputs it’s complete thermal scene frame in one ‘packet’ via a serial stream. The higher resolution Lepton 3 has to output its thermal scene frame in four ‘packets’ that are then reassembled into a single frame by the host system. If this ‘complication’ were to be scaled up to a QVGA resolution then 16 ‘packets’ would be required for a single thermal scene frame. At VGA resolution you would be dealing with 64 ‘packets’ to recreate just one scene frame. If the speed of the serial data output were to be increased, this could increase the thermal output of the serial data stage on the ROIC. Such a serial data speed increase may, or may not, be needed with a <9fps camera core however.

5. Could the LEPTON design increase in physical size to accommodate a QVGA or VGA 12um pixel based microbolometer? The simple answer is yes it could. FLIR has proved that they can produce a miniaturised thermal imaging core but customers who require higher resolution will likely understand and accept a small increase in size to accommodate that greater resolution. The Microbolometer and its optics will dominate the size increase equation. The resultant larger Microbolometer area provides greater real estate on its rear for the ROIC and other electronics behind it. As such the self heating aspect of resolution increase may be more easily managed. The Military use of the Lepton cores should not be ignored. Military contracts and finance are wonderful drivers of product development. The Military need miniaturised thermal imaging cores for personal weapons sights, personal thermal vision and drone use. They must operate in the realm of what is possible with current technology however. They will likely accept a small increase in the Lepton core dimensions if the resultant imaging enhances the capability significantly. Where recon’ drones are concerned, more pixels are always a good idea !

6. Now the fly in the ointment ...... competitive products within the FLIR range. FLIR have a history of segmenting their product ranges to avoid their lower cost products competing too much with their higher cost offerings. They deliberately ‘detune’ their lower cost offerings to degrade imaging performance and capabilities. The Ex series is a perfect example of such marketing techniques. It is common to limit a products capabilities in line with its price via software configuration etc. The Lepton core is a little different however. FLIR have tried to achieve the best image possible out of the Lepton because it is inherently a poor thermal imaging solution. As already detailed, it uses a sub optimal pixel size and lens system. FLIR did not need to degrade its performance as it was already marginal, at best. There is, however, an impediment to improving the capabilities and imaging performance of the Lepton core series........ FLIR’s new ‘baby’, the BOSON core. FLIR created the BOSON as the new 12um pixel thermal imaging core of choice for professional use. I suspect the military were the intended largest customer of this new core. The Boson is a little like a Lepton on a lot of steroids and it uses a powerful video processing chip that could have produced superb imaging from the less than ideal 12um pixel Microbolometer. How do I know this ? Well DRS are already supplying 12um pixel based cores with excellent imaging, far beyond that produced by a Lepton. The BOSON has the advantage of a larger form factor for a larger optical system and easier thermal management. Has the BOSON lived up to expectations ? ..... for me.... No. It requires further development.
FLIR will likely invest heavily in making the BOSON a success. They will not want to ‘muddy the waters’ with a new higher resolution LEPTON core until the BOSON is well and truly sorted and rooted in the marketplace as an alternative to the excellent DRS 12um core offerings. Internal politics and the marketing team will decide the future of LEPTON development and not the design team.

I hope this little commentary on the LEPTON core and it’s future was of interest

Fraser



« Last Edit: September 28, 2019, 01:18:36 pm by Fraser »
 

Online Vipitis

  • Frequent Contributor
  • **
  • Posts: 564
  • Country: de
  • aspiring thermal photography enthusiast
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #1 on: September 28, 2019, 05:49:53 pm »
strangely enough, I had a similar though in my fever dream last night. Due to another recent forum post I asked myself if phone integrated thermal cameras can ever get better than 160x120. Looking into the market- only Leptons have been integrated into phones. Dongles exist in resolutions up to VGA already (ThermExpert V1, ThermApp Pro, Xtherm T66 - 14um - if this even exists?). I already compiled information for a different thread I am currently writing but some fits here.

Current offerings for QVGA "miniature" thermal cores include(most numbers for lens and shutterless options):

all miniature cores I know off:



This list does not include all products available. It is just to give a small overview. Legacy FLIR cores like the Quark2 and Muon were 22x22x12 and 22x22x7 respectively(lensless). The Boson is one of the best options in terms of form factor for the core alone - but it is not miniature enough to fit into a phone. Opgal has a few small cores as well, but their website isn't running correctly for me right now. On the Chinese end there seems to be more news on the horizon - but no English coverage on the industry. This press release has a Tiny 1 camera and a S0 core that would be interesting for the topic.

I believe optics have to make the most impressive jump to allow for further miniaturization. The wide aputure required for LWIR makes it very difficult to built compact optics, the Lepton still has f/1.1 for example. But I have never seen microlenses or aspherical elements for thermal cameras.
Sensor size is not an issue - in phones cameras you find almost 2.5 times the area visible sensors compared to what a QVGA sensor would take in 12µm pixel pitch.

To me the next generation Lepton can work in 160x120 easily. They just have to give us framerate. For all we speculate - the Lepton runs at 30hz (TPL/Black Hornet), but consumer products only ever get the 8.7hz. Even the K1 - which I doubt is useful in any situation Kx and Kxx is the firefighting line but the K1 is called "situational awareness camera". They very likely use the extra frames to stack an output image. This brings reduced noise but also lag - I never bothered to test it out. If they drop the blurred upscale to VGA resolution, enable 30hz and and in app superresolution model (as has been demonstrated by 3rd party apps and dongles), the Lepton 4 can look years better than the 3.5 without new pixels or a considerable change in size.



[/list][/list]
 

Offline Bill W

  • Frequent Contributor
  • **
  • Posts: 394
  • Country: gb
    • Fire TICS
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #2 on: September 29, 2019, 04:47:39 pm »

I believe optics have to make the most impressive jump to allow for further miniaturization. The wide aperture required for LWIR makes it very difficult to built compact optics, the Lepton still has f/1.1 for example. But I have never seen microlenses or aspherical elements for thermal cameras.


There are many aspheric elements in use, I'd go as far as saying the majority today, even if they might not look that crazy.  All moulded (Chalcogenide) lenses are likely to be aspheric simply because it is easy when moulding and provides a saving in element count. 
A typical fire camera lens would be 3 elements using spherical surfaces, but only two if using two aspherical surfaces.
The wierdest was in the Argus 2 'GMEO' variant, where the lens was two plano-aspherics, and the aspheric was a crazy shape to look at.  The logic behind this is that the lens maker  :-X did not have any simple spherical lens manufacture so only did flats or aspherics.

As you say the saving would be if f/4 was practical, especially when looking for narrow field of view.
Now where did I put that 150mm f/0.7 ......... :D

Bill

Online Vipitis

  • Frequent Contributor
  • **
  • Posts: 564
  • Country: de
  • aspiring thermal photography enthusiast
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #3 on: September 29, 2019, 07:08:18 pm »
there is a DARPA project for non planar uncooled Infrared FPAs to allow for non spherical lenses. They likely want it for advanced survaliance and targeting I stealth aircraft. The monetary insentive here will be helpful into new ideas that might be applicable for something like a new Lepton.
Given all the other DARPA projects into uncooled Infrared, WLP, and SWaP. It always ends up benefiting the consumer in one way or another.

Relevant link: https://www.darpa.mil/news-events/focal-arrays-for-curved-infrared-imagers-proposers-day
 

Offline bap2703

  • Regular Contributor
  • *
  • Posts: 64
  • Country: io
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #4 on: September 30, 2019, 05:13:06 pm »
That's for non planar sensors, not 'non spherical' lenses :p
 

Online Vipitis

  • Frequent Contributor
  • **
  • Posts: 564
  • Country: de
  • aspiring thermal photography enthusiast
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #5 on: September 30, 2019, 11:10:57 pm »
It never specified the non spherical lenses, it even explicitly has optics as an their party option to solve.  I miss have read it somewhere else.

The attached document shows very interested graphics, also has some hints to what they want.
 

Offline Ultrapurple

  • Frequent Contributor
  • **
  • Posts: 484
  • Country: gb
  • Just zis guy, you know?
    • Therm-App Users on Flickr
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #6 on: October 01, 2019, 09:44:27 am »
Thanks Vipitis - that's a very interesting document indeed, not least because it lists the current state of the art at cooled MWIR as 16Mpix (4096x4096, 10µm pitch).

One of those is now on my Amazon wish list, along with a nice big zoom lens.
Rubber bands bridge the gap between WD40 and duct tape.
 

Offline frogg

  • Regular Contributor
  • *
  • Posts: 53
  • Country: us
Re: Rhetorical question - Will there be a QVGA or VGA Lepton?
« Reply #7 on: October 01, 2019, 01:32:13 pm »
Perhaps a fair summary of Fraser's post is:

"Nothing technical prevents FLIR from selling a QVGA or VGA lepton"

The obvious corollary is....
« Last Edit: October 01, 2019, 01:36:04 pm by frogg »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf