How to create small (mK) temp differences to test cameras?

[Fraser]

cnxunuo,

Thank you for your interesting comment on the Fluke Black Body. I have found exactly the same with AGEMA and other ‘proper’ black bodies. In British parlance, we call such construction techniques “Agricultural” in reference to farm machinery that tended to be basic, pretty ugly, engineering that just got the job done and could be easily maintained with baling twine and wire 

Black Body references are not rocket science but the designs do need to consider the quality of the emission plate thermal flatness and accuracy. There are many approaches to this, as you likely know. Some BB’s use cavities whilst others, like your Fluke unit, seem to get away with just a suitably coated flat plate. Air convection currents are an issue with simple flat plate types but maybe not serious enough to effect the intended use ? I have BB references that have their thermal emission plates driven by either a simple PI heater sheet or a Peltier module. I prefer the Peltier module units as they can generate temperatures below ambient, but they are limited in terms of maximum temperature. Specialist very high temperature BB’s appear to be built like a Kiln with lots of specialist ceramic insulation required and so are harder to DIY fabricate so as not to burn your house down 

I would love to hear more about what you found inside the Fluke  unit. It certainly “looks the part” but I always wondered what hid inside such units. The PID controller is often the key to a good design as its programming has to be capable of keeping the emission plate at an accurate temperature without too much drift. I have looked at the small BB references that are available from China and have always thought them expensive for what likely resides inside their casing. I expect to find only a PID controller, an emission plate cooling fan, an emission plate and a PI heater attached to the rear of the emission plate. Hopefully the PID controller uses a PT100 sensor but I would not be that surprised to find a thermocouple or thermistor being used ! I had hoped that China would offer cheaper units but being somewhat specialist, I suppose I can understand the pricing.

Fraser

[Fraser]

For anyone interested, this is the type of heater used on the rear of the emission plate of my very expensive AGEMA Black Body reference. It is called a PI heater and is now commonly found heating build plates of 3D printers.
This sort of heater has the advantage of excellent thermal spread over the contact area to produce a relatively flat thermal profile across the emission plate. These heaters come in many shapes and sizes and are inexpensive. They are also known as “Kapton heaters” in reference to the material from which they are constructed. Some come with high temperature 3M adhesive on one side to enable easy fixing to whatever needs to be heated. A great and very useful heater design for heating plates.





Fraser

[Fraser]

Regarding the original OP’s question....... I think this thread may have got a bit too ‘scientific’ in its approach to his needs 

I can understand what Ultrapurple was asking as I have seen great differences in thermal cameras real world imaging performance despite them having very similar specifications on paper. This is why I took the approach that I did in answering his question. Let me explain my experience of this issue from 28 years using this technology and a good many repairing it too.

When I was buying industrial thermal cameras for my employer I would invite the various companies to demonstrate their products and this also gave me the opportunity to speak with the sales reps about any performance issues I noted. It was quickly clear to me that the microbolometer based cameras were unable to produce the clean low noise images of the better cooled staring array cameras when low Delta T scenes were involved. Such was not a surprise though.

In 1997 I tested the AGEMA PM570 for its imaging performance. Compared to the AGEMA THV 550 cooled camera the microbolometer staring array appeared noisy and had an almost chicken mesh appearance to its images. The salesman acknowledged by observations and stated that this was a known issue that AGEMA hoped to address in later revisions of firmware and microbolometer scenes or arrays. The PM570 contained a 1st generation 320 x 240 pixel microbolometer and AGEMA were still learning how best to deal with its relatively noisy signal output. Their software engineers were tasked with taming the beast through image processing techniques.

Move forwards a few years and we saw improvements in the image processing software that was used to clean up the images that the microbolometer was producing. The PM series was, by then, a FLIR product. Over the years, the PM series saw better image processing software and lower noise 2nd and 3rd generation microbolometers. By the time the PM695 was released, FLIR had mastered the taming of the noisy microbolometer and provided the user with different levels of image processing based noise reduction to suit differing scenarios. The 3rd generation microbolometer was also a less noisy sensor array and ROIC combination. I witnessed this development of the PM series cameras first hand and I have multiple examples of each generation in my collection.

So to the matter at hand..... manufacturers often provide impressive specifications for a thermal cameras performance in the hope of attracting a purchased of their wares. Nothing wrong with that and it is the way of the trade, like many others. For the buyer, it is wise to consider such specifications just a guide to a cameras maximum achievable performance and not always its real world behaviour whilst in general use. A proper assessment of a cameras real world performance is wise and that is why I never ordered an Industrial thermal camera based on a paperwork exercise. I always required a demonstration and testing period in our hands before making a purchase decision. The USA Fire Brigades found that they had to do exactly the same as the cameras specifications on paper were only a small part of the story and many other aspects of a cameras design needed investigation. Any manufacturer who would not loan me a demo thermal camera did not get further consideration.

When testing the various industrial thermal cameras it became very clear to me that the paper based specifications provided little indication of how well the camera performed in my chosen application and also it was clear that some camera designs were superior to others despite having the same, or very similar specifications. Did I carry out NETD tests ? No I did not. Why Not ? Well I was not using the cameras in a scientific lab and needed to test them in the real world use scenarios for which they were being purchased. Very much like the Fire Fighters scenario. I used the cameras in various test situations and assessed the image quality and accuracy that each produced for me as a user viewing the built in display. That is how the cameras would be used, so that is how they were tested. Numbers are meaningless compared to practical use in my scenario.

During the many camera tests that I undertook, some side by side with others, some on their own, it became clear to me which manufacturers had mastered the art of taming the microbolometer based technology and achieved excellent thermal imaging performance. Some cameras were quite frankly appalling in their imaging performance due to the manufacturers inexperience within microbolometer or their use of building block cores that were not suitably tuned for the best imaging performance in my usage scenarios.
It quickly became clear that FLIR (nee AGEMA) were masters of the image processing ! There was little to compete against FLIR at that time and they remain a manufacture of high performance thermal imaging systems that deliver on their promises. Over the years, other manufacturers have caught up on the image processing front and we are fortunate to have a good selection of thermal camera manufacturers to choose from.

In my testing of thermal cameras, I tested the complete system from lens through to the display. NETD was meaningless to me if the camera proved incapable of producing the required high quality imaging in my usage scenario. Many cameras did indeed fail my testing, some badly so ! The choice of lens and the display on which the image was viewed could have a dramatic effect on the performance of the cameras during my testing. Some actually produced better images when connected to an external broadcast quality monitor, but that was still a fail as the user would not be using such a monitor in the field ! At the end of the day I needed to buy thermal cameras that the users would find ergonomic, accurate and a joy to use. It was very important that the low Delta T imaging was as good as possible with the lowest possible noise for the camera technology. Cooled imagers were the best, but Microbolometers did manage to produce acceptable images. 2 Celsius temperature spans were common in my usage scenario.

So in my humble opinion,  it is best to test a thermal cameras imaging performance in the real world and in the scenario in which it will be used, rather than getting too excited about some numbers on a specification sheet. Such numbers might impress some, but not me. I want real World performance as experienced by the average user.

I attach a little example of a real world situation that applies to the Fluke Ti450.

Fluke realised that their Ti450 user experience was lacking on the display front. The LCD panel was just not performing as well as they wished. FLUKE used a better LCD display panel in later models of the camera. The result is, from a users perspective, two cameras of the same model and specification, yet one produces better images than the other, depending upon the age of the model ! Remember, on paper they are the same camera with the same sensitivity and NETD noise performance 



Fraser

[Fraser]

An interesting article on comparing thermal cameras imaging performance........

https://www.ndt.net/article/ecndt2006/doc/Tu.4.6.4.pdf

Fraser

[cnxunuo]

AGEMA makes nice cameras, I once picked up an older than me scan camera, something 900, with a huge external processor and a box full of floppy, it was less noisy than my best camera at that time, the 25um photon 640. But due to its size, I ripped the cryocooler out and throwed away everything else.

A simple convection finite element analysis can be made to find out how serious an impact air can be to overall uniformity. but even if it does has big enough an impact, nothing beats a 1inch thick copper plate, unless someone spray water on it before measurements...

Black body paint can be use on hobbyists level temperature, eg from no condensation to 100degree c. which provides a nice uniform emissivity with low reflection. it's just a bit expensive, in China it cost around 80 GBP per spray can.

[Fraser]

AGEMA ThermoVision 900 RIP 

[Max Planck]

Black body paint can be use on hobbyists level temperature, eg from no condensation to 100degree c. which provides a nice uniform emissivity with low reflection. it's just a bit expensive, in China it cost around 80 GBP per spray can.

Instead of paint it is possible to use a thin layer of microporous rubber with open pore surface, something similar to some mousepads. It does a great job as long as the temperature drop on the rubber layer is taken into account. IIIRC the French BB manufacturer, HGH, uses some kind of rubber/foam with a micropyramid surface.

An interesting article on comparing thermal cameras imaging performance........

https://www.ndt.net/article/ecndt2006/doc/Tu.4.6.4.pdf

Fraser

I wonder what was the purpose of comparing MRTD curves for two IR imagers designed for complitely different tasks and thus, having a complitely different field of view/optics.

Max

[yertle]

You can use a spray can of Krylon from non-condensing to very hot. Should be at most hardware stores - high heat, flat black. The emissivity should be quite good.

[Fraser]

Max,

From what I read in the article it would appear that this was a comparison of a standard issue military observer thermal scope with the sort of civilian thermal camera that is normally used for Non Destructive materials testing. They are talking about active modes of NDT testing such as flash thermography. The military use thermal cameras for NDT work in their maintenance activities. If the standard issue military thermal scope was comparable to the FLIR SC3000 they were hoping to avoid buying the additional SC3000 cameras.

As you say, they are very different cameras in many respects but I suspect someone made the suggestion and they decided to investigate the possibility.

Fraser

[Max Planck]

Fraser,

you may be right, but for me this article is an example of what could go wrong.

First of all, there is a lot of information missing.
What was the intended use of the camera (just NDT is more than vague)? Sample size, distance, etc.?

No technical data for both cameras, so I had to check it.

FLIR SC3000
LWIR QWIP 320x240 FPA
NETD 20 mK
FOV 20˚x 15˚
In fact you have a camera from the same familly.

Thales Sophie
Image size 754 x 576
LWIR
Wide FOV 8° x 6°
NETD <50 mK
Because of the weird resolution, I made some digging and the imager is a scanned 288 x 4 MCT LWIR FPA from Sofradir. It means there is probably a digital x2 zoom with some image processing and smoothing to get the final resolution. 

My educated guess is that FLIR's camera wins easily in all departments as long as it gets the right lens for the intended job.

As a sidenote, I also have some doubts about the final conclusions, for two reasons. The depth of the defect that can be detected is closely related to its size and NETD is not a limit if lock-in technique is used.* Again, the question about a more precise description of the intended use.


* Would it be possible to use lock-in thermography? Probably yes with SC3000, probably no with Sophie.
Max 

[Fraser]

Max,

I totally agree with you. In hindsight, whilst an interesting exercise, it was severely flawed from the start. As I say, probably someone’s idea of a ‘quick win’ to use equipment already in the inventory rather than buying the correct tool for the job. I was more interested in their approach to comparing two cameras performance though. I still found it an interesting, if flawed, test

I have a FLIR SC3000 cooled Quantum Well camera in my collection
A very nice piece of thermal imaging technology The SC4000 is a very different beast though. The SC3000 is a development of the AGEMA Thermovision THV 550 basic design with a different sensor array. The SC4000 is based on the Indigo Phoenix cooled camera.

Fraser