Author Topic: what are the most well respected companies in thermal? the ones that dont lie  (Read 1923 times)

0 Members and 1 Guest are viewing this topic.

Offline edubzTopic starter

  • Contributor
  • Posts: 37
  • Country: us
so I was reading how some companies cheat on the NETD test and was wondering which are the well respected companies?  has anyone ever heard of LMIR?


I know FLIR, SEEK, and infiray but arent there others? what about companies like this

http://laomarcx.com/

there are so many im surprised, but I dont know there reputation

what about companies AGM? they claim less then 35 netd


 like https://www.agmglobalvision.com/agm-asp-micro-tm384
« Last Edit: July 27, 2021, 04:43:04 am by edubz »
 

Online Bill W

  • Super Contributor
  • ***
  • Posts: 1102
  • Country: gb
    • Fire TICS
Much depends on what you mean by 'don't lie'.

There are relatively few sensor manufacturers in the commercial arena (ULIS/Lyred, FLIR, SEEK, guide IR, iRay, GST, DaliT).

They are the only people who can (sensibly) perform a true NETD measurement, and will do it at the best conditions of sensor biasing and integration times etc.
They will always use f/1.0 all-Ge lens with high transmission coatings.  They can also have some very impressive 19" rack driver electronics.

While some of these companies do build modules and cameras, there are many more people making cameras from sensors or modules with varying degrees of skill starting from a simple screwdriver operation upwards.  For them, they will usually just quote the NETD of the sensor as supplied.  People buying are also likely to want to know 'the NETD' and with proof - ie the sensor datasheet.

One question you may be trying to ask is 'is the NETD as running in the camera the same as in the sensor datasheet'.
It is probably fair to say, not often !
Any of the usual requirements of an end product camera (dynamic range, temperature variations, power and space, response time, frame rates, video timings) may prevent the use of the 'ideal' NETD settings of the sensor at test.
Using a different lens is not relevant at this point.

The other question is 'what temperature differences will I be able to see'?  Here you have a load more variables thrown in and this is where 'NETD' is less relevant and the 'MDTD' is more what you want to know about.
For that measurement the lens aperture, image processing, display, colour schemes come into play.

As an example, I can examine the use of a ULIS 03191 sensor in e2v Argus4 cameras.
Quoted NETD is '<120mK' for the usual f/1, 30Hz, 300°K

For a start the typical sensor was testing more around 65mK and later dropped to 55mK.  The spec only ever changed to <100mK, just the yields went up !

The 'fire' version (P7130, f/1) had 3 gain modes with different MDTD (60mK, 100mK and 300mK). 
The 'security' version (P7250 f/1) had a lower dynamic range, allowing a better MDTD (45mK)
The same settings with a narrower field of view f/1.3 lens came out at MDTD 60mK of course.

The next irony is that the same die in the ULIS 0336 sensor had a better NETD  (<80mK) but, being without a peltier, could not be kept at 300°K and so gave worse camera MDTD's even with the same lens / software / display!

So were any of those specs lying ?  Possibly when sales insisted we came up with a way to claim an NETD of 50mK, as the customers could not understand why MDTD mattered and NETD did not.  Tenders were being rejected by pen pushers for 'NETD not <50mK' although some thought the higher number meant better  :scared: :palm:

Bill
 
The following users thanked this post: IR_Geek

Online Fraser

  • Super Contributor
  • ***
  • Posts: 13140
  • Country: gb
The challenge you face is badge engineering where an OEM makes a thermal camera and sells it to 3rd parties for distribution under their branding. Then there is the OEM core market where a core is made by an OEM and incorporated into a 3rd parties camera design. The core can have excellent performance but may be coupled to a poor “host” that does not make the most of its capabilities. Where thermal imaging analysis is concerned, there can be a huge variation in the quality of the analysis software provided. This ranges from very poorly coded, pretty useless, software all the way up to science grade professional software costing many thousands of Dollars.

Regarding your question on specifications etc…….

1. A company is unlikely to reveal any negative attributes to its product so look for what is not detailed in specifications. Deception by omission ?  ;)

2. NETD is a pretty well defined specification that sets certain constants for the test, such as the use of an f1.0 lens. That said, there are ways to manipulate the test or it’s results to provide a better NETD than will ever be seen by be end user of the production product.

3. Judging a product like a thermal imaging camera based purely upon paper based comparisons is a very risky approach as the companies marketing team will have worked hard to make their product appear better than the Opposition. Specifications can be massaged to look good but the testing that lead to such specifications is not always equivalent to that of other, more honest, companies.

4. Always be wary of non precise specifications such as “Up to 15fps”. This “up to” specification is not often qualified in terms of what is needed to reach that highest specification. Real world performance rarely meets these “up to” specifications but the marketing team like to have a nice impressive number for the buyers eyes to lock on to and they cover themselves by saying “up to”. The stated spec is often only achievable under a certain set of conditions in a test lab.

5. All companies employ a marketing person or team and their job is to help sell product in a competitive market. It is no surprise that these marketing teams push the limits on a products claimed performance. The marketing team will interrogate the design and production teams to establish just how far they can massage the specifications and performance claims whilst remaining ‘legal’. An example might be to claim a high performance from a particular microbolometer in the full knowledge that the host in which the microbolometer is installed cannot actually reproduce the situation in which that specification is produced. The claimed specification OF THE SENSOR ARRAY remains true but the marketing team are clever as they inform the potential buyer of the SENSOR ARRAY specification and not the complete cameras specification, that is inferior due to power supply quality, cooling and lens quality. No lie is told, but deception by omission can occur  ;)

6. There are some large and well respected long term ‘players’ in the thermal imaging system manufacturing market place. Some have been bought by other companies but names that immediately come to mind are….. AGEMA, Inframetrics, Indigo, Hughes, GEC, E2V, FLIR, Raytheon, Amber, Lockheed Martin, Racal, Selex, SCD and some publicity shy military equipment providers. Would I trust the specification claims of these manufacturers ? Yes and No. They are in the business of selling product after all. Whilst they would be unlikely to deliberately deceive as that effects reputation, their sales persons will always raise the profile of the good bits whilst lowering the profile of anything not so impressive about a product. Standard marketing and sales stuff. It is for the intelligent buyer to burrow into the cameras specifications, ask the awkward questions and get the required answers. Then there is the testing phase where a potential buyer borrows a demo camera to test. Some demo cameras are ‘tweaked’ to perform better than the generic stock ! A manufacturer will just claim that the demo camera was just a random unit that performed slightly better than others due to parts tolerance  ;) So, can a salesman or sales brochure be 100% trusted ?  I would say no, you need to do your own research and testing. What I can say is that the above detailed reputable thermal camera and core manufacturers are unlikely to lie in their specifications as that effects reputation and they are used to selling to Governments so the implications of lying are serious on multi Million Dollar contracts. Smaller companies with less of a reputation have less to lose by misleading the customer and in some societies it is considered OK to mislead the customer based upon the belief that a fool and their money are easily parted and the customer should be smart enough to not be conned or else suffer the consequences ! Yes that is a true attitude in some parts of the World. This attitude filters through into some thermal camera marketing and specification documentation from certain companies. The classic ‘trick’ with cheap thermal imaging cameras was to state the “displayed resolution” rather than the actual microbolomter resolution. A low resolution FPA was interpolated up to the more respectable resolution of the LCD display panel. Was this a lie ? Nope, just misleading the customer by the exclusion of a meaningful specification and insertion of a misleading specification. Caveat Emptor ! You would not find the likes of Raytheon or FLIR doing this to a customer as it is blatantly misleading with the intention to deceive.

So with the above comments considered, whose specifications can you trust ?

If you are buying from a well known thermal imaging system manufacturing brand you will likely receive a product that meets or exceeds its specification but you do need to read the ‘fine print’ if any specification appears really good, or too good to be true ! That said, I would expect some specifications to actually be understated, such as measurement accuracy. In my experience, well known manufacturers of thermal cameras often produce a product that not only meets the measurement specifications, but is far more accurate than indicated in said tolerance specification. The manufacturer is being cautious due to production tolerances and aging of the system. The contrast to this is a budget thermal imaging camera from a relatively unknown manufacturer who is using a 3rd party imaging core of dubious quality yet claiming the same measurement tolerance as the ‘big boys’ like FLIR and Raytheon. Testing often shows these ‘bargain’ cameras to have very unpredictable measurement errors that often far exceed the stated +/-2C or +/-2% tolerance claimed. The issue is often a combination of imaging core stability and poor firmware implementation in the host as such cameras are often little more than a reconfigured development board !


I know this has not answered your question but I hope you can see that this is a complex topic with no simple answer. You just have to look at specifications as a guide rather than absolute fact when it comes to lower priced thermal imaging equipment. You tend to get what you pay for with thermal imaging systems. When corners are cut to save cost, specifications tend to suffer.

Fraser
« Last Edit: July 27, 2021, 03:23:17 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 
The following users thanked this post: Bill W, IR_Geek

Offline IR_Geek

  • Regular Contributor
  • *
  • Posts: 91
  • Country: us
Very well said by Fraser and Bill W !!!

Very rarely have I seen a flat out lie from any vendor / sales rep.  I've seen more innocent mistakes by the sales person not understanding a product as deeply as they should.  Usually bypassing (nicely) the sales person to get to the designer/engineer yields much better results.  That is impractical for 'cheap' one off sales with with 3rd party vendors or faceless website orders.  If you are buying a bunch for a business, you would normally have the ability (and equipment) to perform independent measurements.  That is the only way to truly understand what sensor/system can actually do.  If not, then relying on journal papers of by people other than the vendor is the only way.

Sidebar Rant:
I've never liked NETD as a benchmark for FPAs ... boiling a sensor down to a single value make me  |O.    Anybody remember D-star as a figure of merit?   :palm:   NETD had meaning for scanning systems but not very relevant for modern FPAs.   Measuring 3D noise is more insightful to a sensors potential ... also next to impossible to 'game' the system with 3D noise. 
 

Offline AkiTaiyo

  • Contributor
  • Posts: 30
  • Country: ie
In the industrial imaging world, there is a few companies that make their own thermal imaging cores and tend not to lie too much, eg Dias, Ametek Land, Optris. 

I know Land has a well-respected history in the industrial market (more so in single point temperature measurement) with providing measurement accuracy.  They also used to publish some really detailed and interesting white papers on non-contact temperature measurement, less so recently though..  I’ve worked with some Land gear, it’s been pretty hardy.

Some industries such as hydrocarbon processing or glass melting like to have accurate temperatures withing a few degrees, so these types of cameras tend to focus on getting thermal accuracy across the range rather than pushing NETD numbers..  A lot of processes don't change temperature that quickly so multiple frame averaging isn't really a problem.
These smaller companies often cover the MIR bands too which are a lot less market saturated than LWIR!  Also helps in certain processes for viewing through certain types of flame.  Industrial imagers also have much higher temperature ranges than 'typical' LWIR cameras to suit to target applications, sometimes with quite exotic optics too, eg very wide angle 'edoscope' type lenses for poking thought furnace walls, or very narrow lenses for monitoring flare stacks from afar.

The higher temperature focus also helps with noise - more signal!


Downside is, these types of imagers are typically not useable is the second hand market due to the use of licenced software and proprietary protocols... And the *massive* cost of these cameras from smaller companies and specialist application focus makes them not a good place to look for a new camera!
 

Online Bill W

  • Super Contributor
  • ***
  • Posts: 1102
  • Country: gb
    • Fire TICS
Sidebar Rant:
I've never liked NETD as a benchmark for FPAs ... boiling a sensor down to a single value make me  |O.    Anybody remember D-star as a figure of merit?   :palm:   NETD had meaning for scanning systems but not very relevant for modern FPAs.   Measuring 3D noise is more insightful to a sensors potential ... also next to impossible to 'game' the system with 3D noise.

A sensor NETD is of some use as it splits the pure 'sensor' aspects from the camera aspects, even if it is simply taking an average of the best 99.5% of pixels by treating each as a single point detector.

3DN at video output still has some flaws.  It ignores the display and scene (MDTD does not) and some video outputs differ from the screen due to compression (MP4, MJPEG streams) - is that for better or worse though ?

Equally used at a detector level 3DN would still be susceptible to 'supercharged' electronics and image processing and choosing an eye-brain integration time.

For me MDTD (ie MRTD at low frequencies) is one that matters along with MTF.  The standard deviation of a still frame of 'closed shutter' is useful as a comparative measurement when optimising things to reduce the noise.

Bill

Bill

Offline IR_Geek

  • Regular Contributor
  • *
  • Posts: 91
  • Country: us
Bill W
  More good points!    Once you fold in image processing and a display then I agree that a MRTD or similar is the only option to gain better insight into the full system.   Image processing and the right display can make a less capable sensor look.  The reverse is also true ... apply some crappy algorithm and the best sensor in the world can look like garbage.    I'm normally only looking at a sensor from a radiometric calibration standpoint while dealing with raw counts, hence 3D noise means more to me.   Really boils down to exactly you want to use an IR camera for.
 

Offline edubzTopic starter

  • Contributor
  • Posts: 37
  • Country: us
The challenge you face is badge engineering where an OEM makes a thermal camera and sells it to 3rd parties for distribution under their branding. Then there is the OEM core market where a core is made by an OEM and incorporated into a 3rd parties camera design. The core can have excellent performance but may be coupled to a poor “host” that does not make the most of its capabilities. Where thermal imaging analysis is concerned, there can be a huge variation in the quality of the analysis software provided. This ranges from very poorly coded, pretty useless, software all the way up to science grade professional software costing many thousands of Dollars.

Regarding your question on specifications etc…….

1. A company is unlikely to reveal any negative attributes to its product so look for what is not detailed in specifications. Deception by omission ?  ;)

2. NETD is a pretty well defined specification that sets certain constants for the test, such as the use of an f1.0 lens. That said, there are ways to manipulate the test or it’s results to provide a better NETD than will ever be seen by be end user of the production product.

3. Judging a product like a thermal imaging camera based purely upon paper based comparisons is a very risky approach as the companies marketing team will have worked hard to make their product appear better than the Opposition. Specifications can be massaged to look good but the testing that lead to such specifications is not always equivalent to that of other, more honest, companies.

4. Always be wary of non precise specifications such as “Up to 15fps”. This “up to” specification is not often qualified in terms of what is needed to reach that highest specification. Real world performance rarely meets these “up to” specifications but the marketing team like to have a nice impressive number for the buyers eyes to lock on to and they cover themselves by saying “up to”. The stated spec is often only achievable under a certain set of conditions in a test lab.

5. All companies employ a marketing person or team and their job is to help sell product in a competitive market. It is no surprise that these marketing teams push the limits on a products claimed performance. The marketing team will interrogate the design and production teams to establish just how far they can massage the specifications and performance claims whilst remaining ‘legal’. An example might be to claim a high performance from a particular microbolometer in the full knowledge that the host in which the microbolometer is installed cannot actually reproduce the situation in which that specification is produced. The claimed specification OF THE SENSOR ARRAY remains true but the marketing team are clever as they inform the potential buyer of the SENSOR ARRAY specification and not the complete cameras specification, that is inferior due to power supply quality, cooling and lens quality. No lie is told, but deception by omission can occur  ;)

6. There are some large and well respected long term ‘players’ in the thermal imaging system manufacturing market place. Some have been bought by other companies but names that immediately come to mind are….. AGEMA, Inframetrics, Indigo, Hughes, GEC, E2V, FLIR, Raytheon, Amber, Lockheed Martin, Racal, Selex, SCD and some publicity shy military equipment providers. Would I trust the specification claims of these manufacturers ? Yes and No. They are in the business of selling product after all. Whilst they would be unlikely to deliberately deceive as that effects reputation, their sales persons will always raise the profile of the good bits whilst lowering the profile of anything not so impressive about a product. Standard marketing and sales stuff. It is for the intelligent buyer to burrow into the cameras specifications, ask the awkward questions and get the required answers. Then there is the testing phase where a potential buyer borrows a demo camera to test. Some demo cameras are ‘tweaked’ to perform better than the generic stock ! A manufacturer will just claim that the demo camera was just a random unit that performed slightly better than others due to parts tolerance  ;) So, can a salesman or sales brochure be 100% trusted ?  I would say no, you need to do your own research and testing. What I can say is that the above detailed reputable thermal camera and core manufacturers are unlikely to lie in their specifications as that effects reputation and they are used to selling to Governments so the implications of lying are serious on multi Million Dollar contracts. Smaller companies with less of a reputation have less to lose by misleading the customer and in some societies it is considered OK to mislead the customer based upon the belief that a fool and their money are easily parted and the customer should be smart enough to not be conned or else suffer the consequences ! Yes that is a true attitude in some parts of the World. This attitude filters through into some thermal camera marketing and specification documentation from certain companies. The classic ‘trick’ with cheap thermal imaging cameras was to state the “displayed resolution” rather than the actual microbolomter resolution. A low resolution FPA was interpolated up to the more respectable resolution of the LCD display panel. Was this a lie ? Nope, just misleading the customer by the exclusion of a meaningful specification and insertion of a misleading specification. Caveat Emptor ! You would not find the likes of Raytheon or FLIR doing this to a customer as it is blatantly misleading with the intention to deceive.

So with the above comments considered, whose specifications can you trust ?

If you are buying from a well known thermal imaging system manufacturing brand you will likely receive a product that meets or exceeds its specification but you do need to read the ‘fine print’ if any specification appears really good, or too good to be true ! That said, I would expect some specifications to actually be understated, such as measurement accuracy. In my experience, well known manufacturers of thermal cameras often produce a product that not only meets the measurement specifications, but is far more accurate than indicated in said tolerance specification. The manufacturer is being cautious due to production tolerances and aging of the system. The contrast to this is a budget thermal imaging camera from a relatively unknown manufacturer who is using a 3rd party imaging core of dubious quality yet claiming the same measurement tolerance as the ‘big boys’ like FLIR and Raytheon. Testing often shows these ‘bargain’ cameras to have very unpredictable measurement errors that often far exceed the stated +/-2C or +/-2% tolerance claimed. The issue is often a combination of imaging core stability and poor firmware implementation in the host as such cameras are often little more than a reconfigured development board !


I know this has not answered your question but I hope you can see that this is a complex topic with no simple answer. You just have to look at specifications as a guide rather than absolute fact when it comes to lower priced thermal imaging equipment. You tend to get what you pay for with thermal imaging systems. When corners are cut to save cost, specifications tend to suffer.

Fraser



what about software, is there one companies software that is better then others?

for example someone told me to make sure I get one that can set the tempatures of highs and lows to find things better, are there any companies that make superior software?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf