High speed thermal imaging cameras - When 60fps is just not enough !

[Fraser]

This is a post for those readers who are interested in high speed thermal imaging. As many will know, the common frame rate for consumer grade thermal cameras outside of the USA is <9fps. There are 30fps or 60fps pro-sumer cameras available and 120fps is to be found for more demanding tasks such as Motor Racing thermal analysis.

For some forum members, the <9fps frame rate is somewhat restrictive and they go in search of 30fps or even faster frame rate cameras. What some readers may not be aware of is that there are thermal cameras capable of significantly higher frame rates. Take a look at the details bellow:

320 x 256 pixels @ 432fps
128 x 128 pixels @ 1502fps
64 x 64 pixels @ 3948fps
64 x 4 pixels @ 43103fps

The above details belong to one camera that dates back to 2005 ! The camera offers the user full resolution, or windowed resolution. As the resolution is reduced, the frame rate increases. The frame rate may be set in 0.1fps increments up to the maximum for any selected resolution.

It is possible to capture high speed events with such a camera in order to analyse the thermal profile. The high speed recording is fully radiometric, meaning that every frame contains a full set of radiometric data

It will come as no surprise that such a camera was very expensive and a tightly controlled technology. The current equivalent models remain very expensive. These cameras were often used in scientific applications where budgets were healthy and the need for advanced high speed thermal imaging could be justified.

So why am I telling you all this ? Well because I will be using one of these high speed cameras next week

Watch this space 

Fraser

[Fraser]

I should add that high speed thermal cameras are often considered scientific instruments due to their advanced capabilities. They provide many configuration options and input/output ports. The unit I will be using has a trigger input for event synchronisation and variable integration time to suit the task. The radiometric image capture is via Camera Link or Gigabit Ethernet to a PC running scientific research software.

Before considering the purchase of such a specialist camera, factor in the cost of the software needed to drive it as such can be very expensive and/or hard to source.

Fraser

[Fraser]

A FLIR video showing production of their high speed scientific cameras :



Very nice scientific camera units and facilities 

Fraser

[Fraser]

The I/O panel of the camera. Plenty of options 

It can be used with a Gigabit Ethernet equipped laptop without the need for custom interfaces.

Fraser

[Fraser]

I am surprised no one has asked how it is possible to get such high frame rates from a thermal camera. You most certainly cannot achieve the stated frame rates with a microbolometer.

The answer is that these high frame rate cameras use cooled narrow band-gap semiconductor sensor arrays. The camera I will be using has an InSb sensor array with 30um pixel size in a 320 x 256 pixel format. The NETD is stated as <18mK but I have seen it also proven to be 13mk  It is not uncommon for high performance thermal cameras to exceed their stated specifications in some areas as the OEM only states the specification that they are certain can be achieved. Some cameras will perform better, but all will at least meet the specification.

Fraser

[Vipitis]

While I am jealous here, I also liked the video you linked, because that seems like exactly what I would like to do in the future.

I read about highspeed LWIR camera using MCT and achieving up to 1000fps uncooled. While the bandwidth of the readout limits resolution. There are a few example videos on YouTube, but many use rainbow palettes, I would love to see detailed macro heat transfer and heat loss in whitehot.

[Fraser]

Indeed there are different sensor array materials that can achieve high frame rates. Some need cooling to 77K with Liquid Nitrogen or a Stirling Coller, but others may only need a Peltier Stack to lower the sensor temperature to a point where thermal imaging is possible. The FLIR Thermovision 450 used such sensor technology.

The key requirement is a fast pixel response. This can be achieved with semiconductor pixels but microbolometers are just too slow. It comes down to physics. A microbolometer is just a variable resistor that responds to heat and varies the current that passes through it in response to its temperature. A thermistor by any other name  The pixel has significant thermal mass which will limit the frame rate. A semiconductor sensor array accumulates electrons in wells and is not constrained by the physical changes in an absorber material. Semiconductors offer very fast frame rates with ease as a result. As you have stated, the bandwidth of the system connected to the sensor array then becomes the limitation on frame rate. In the case of the FLIR camera I am discussing, the frame rate is decent at full resolution, but may be increased further at the cost of optical resolution where a window of pixels are processed, rather than the whole array. The fastest frame rate is only achievable when a 64 x 4 pixel window  is being processed. Now this might sound like a useless low resolution, but remember this is a well calibrated and accurate scientific camera. Sometimes the user needs a fast frame rate with accurate recording of changes, rather than pretty pictures.

Fraser

[TooQik]

The only question I have, is how fast of a frame rate do you really need to capture temperature change in an object? Or to put it another way, what's the fastest rate of temperature change that can physically occur in any object?

Like Vipitis, I'm also jealous. I'm yet to buy my first thermal camera and look forward to the day I get my hands on a full frame rate version. 

[Ultrapurple]

I rather like Flir's demo thermal video of popcorn exploding, recorded at 1000fps...



... but then I've always been into thermal imaging for the sake of the pretty pictures rather than the scientific application of the technology. If a high speed thermal camera came my way I wouldn't refuse it, though.

[Ultrapurple]

How fast you need your thermal imaging to be depends on what you're doing. Melt spinning has a rate of temperature change of tens of thousands to tens of millions of degrees per second. Somewhere I've seen a slow motion video (probably by FLIR) of a gun firing, seeing for example where hot gases escape, plus various bits of projectile research. And I suspect the folks that test explosives have some pretty rapid temperature changes to deal with, too.

[Ultrapurple]

@Fraser - may we ask what you'll be doing with such a high frame rate camera? Or is it another of those fifty-quid eBay bargains that we're all jealous of?

[Fraser]

50 Quid eBay special ? .... I should be so lucky  .

I will tell more of the camera next week.

It may interest readers to know that the user manual is not available from FLIR without answering some in depth questions.

Fraser

[Ultrapurple]

One of these?

[yertle]

The only question I have, is how fast of a frame rate do you really need to capture temperature change in an object? Or to put it another way, what's the fastest rate of temperature change that can physically occur in any object?

Like Vipitis, I'm also jealous. I'm yet to buy my first thermal camera and look forward to the day I get my hands on a full frame rate version. 

I used to work on these science cameras (a competitive product), and people used high framerate for all sorts of interesting applications, such as looking at combustion chambers in engines.

[Vipitis]

working on scientific thermal cameras to develop rocket engines, that is a place I would love to be at... I shouldn't look at this thread any longer...

a note to add. Looking at local specialists with thermal cameras, I found one very interesting processing technique with is asking me more question.

https://youtu.be/fRN6ooSuRfY

they call it binning while increase framerate and resolution.

[eKretz]

This one is pretty good:

https://youtu.be/y-Hid8yriiA

[Fraser]

I personally like this video from FLIR. Sadly there are no humming birds anywhere near me.

I have seen thermal imaging of honey bees though and that shows the work their wing muscles are doing and the resultant heat.

https://youtu.be/pRaR72v3L98

Fraser

[Fraser]

Then there is the artistic stuff 

https://youtu.be/flAtYQvriAI

Fraser

[Fraser]

And not forgetting the impressive automatic pistol imaging showing heat distribution in the ejected cartridge casing and hot material coming out of the ejector port !

https://youtu.be/roddQPTZ-J4

Fraser

[Ultrapurple]

@Fraser  - I think the pistol video you found was probably one I was thinking of.

If anyone from FLIR or any other thermal imaging equipment manufacturer is reading this, I believe i could bring a lot of interesting new idea and would make an interesting addition to your demo team / stock footage crew.

[Fraser]

FLIR have a good car airbag deployment thermal video somewhere too. Lots of hot expanding and venting gasses involved !

Fraser

[Fraser]

The air bag deployment imaging is in this FLIR presentation......

Airbag Deployment video is at 1 minute 32 seconds

https://youtu.be/kpZ_rqODNXk

Fraser

[Fraser]

Compare the FLIR airbag video to that of a standard thermal camera......

http://youtu.be/_Zw744kN_aM

Fraser

[Fraser]

A better FLIR video that also includes the the Airbag deployment ......

http://youtu.be/gC6f_3debvg

Fraser

[Fraser]

Well I collected the FLIR SC4000 today 

The camera is even better than I had hoped. It is a true scientific thermal camera with masses of functionality that is not normally found in generic thermal imaging equipment.

I was surprised when I found that FLIR could not provide the cameras user manual on the public servers due to ITAR regulations    Now that I have the two user manuals in front of me, I understand. As this is a scientific instrument, it’s user manual contains an awful lot of design detail that is not normally public information. This is the first thermal camera user manual I have seen that contains multiple block diagrams and technical desciptions. You may wonder why such technical detail is needed, well this camera offers a lot of configuration options and the user needs to understand how the camera is designed in order to change key settings in its configuration. You can even change the voltages applied to the sensor array ! Thermal cameras do not normally offer such deep access to the system to the user.

The camera cooler is also better than expected. The SC4000 contains a linear Stirling Cooler that contains an oscillating linear stroke piston. The linear piston compressor design has a longer operational life than the conventional rotary crank, connecting rod and piston type.

The camera came with all its originally supplied accessories plus a specialist high temperature glass filter. The supplied lens is a Janos 25mm (21 degree HFOV) wideband unit of excellent quality. The supplied software includes FLIR Researcher 2.8, FLIR Quickreport and the SC4000 GUI for specialist configuration.

This is likely the highest performing thermal camera in my collection   I am a very happy chap tonight.

Sadly the story behind this camera appears quite sad. It belonged to a company that appears to have gone out of business due to family issues and has languished in a corner of a workshop for two years. The owner wanted it to go to a good home where it would be appreciated. Well it has certainly found that home with me

More about the camera soon. Maybe in a fresh thread ?

Fraser