The Great Optris Debate

I have started this thread to provide a home for the Optris posts, which are derailing RossiVDarden2. Apologies in advance if I missed one, or move one too many (for some). Alan

Not quite the right thread, but it will do.

How to check emissivity calculations

Wyttenbach has for a long time promoted the idea that Optris engineers have told him that the camera software incorporates emissivity e into temp measurements as:

T = T0/e^(1/3)

T is the camera calculated temperature for emissivity = e where T0 is the camera calculated temperature for emissivity = 1.

So that T scales as exponent 3 of 1/e independent of T.

Rossele, Levi and others have promoted the idea that total emissivity must be used in these calculations and hence exponent = 4 uniformly of temperature (see below).

As opposed to what the camera must do to be accurate, which is to vary the exponent smoothly with T according to the Planck function evaluated over the Optris sensor passband. We expect an increasingly higher exponent at lower temperatures as the passband gets higher in frequency relative to the Planck peak. At very high temperatures the IR passband occupies the Rayleigh-Jeans part of the Planck curve - much lower in frequency than the peak - where the exponent is 1. For the Lugano test higher temperatures in the active test we had approximately an exponent of 2. An exponent of 4 would be what Levi would I guess expect since he claims that total emissivity is the only relevant parameter. We know that this scales at T^4 because total radiation scales like this.

So let’s do some limited checking, courtesy MFMP:

(1) Download the Optris PI Connect software and a sample camera ravi file from MFMP dogbone 2 thermal assessment experiment.

(2) view the ravi file - note that segment AL with defined emissivity 0.95 temperature 59.7C in the layout (go to tools->configuration to see this).

(3) change the layout emissivity for AL area from the tools->configuration screen from 0.95 to 0.475 (1/2 of old value) and save

(4) Note the changed temperature for AL

(5) repeat to get Optris calculated temperatures for a range of emissivities.

I've tabulated some values below, and the calculations I use to work out the corresponding exponent (the site here imports cut and pasted excel spreadsheets nicely, which is great)

I don't easily have a way to generate additional Optris camera ravi files so cannot in this way test higher temps: 0.1 is an emissivity limit in the software. If someone can unearth and post some of these (Paradigmnoia?), we could see precisely how the Optris camera responds over the entire temperature range, and check further.

Of course, TC's paper has done this theoretically, assuming the Optris camera software works correctly. MFMP have confirmed it experimentally. But it seems some here: specifically ele and Wyttenbach will be more convinced by their own checks than what other people say? Perhaps this would help Levi, too?

After all, were total emissivity to be the relevant parameter as Levi says we would need to integrate radiance over all frequencies and therefore have an exponent uniformly of 4 regardless of temperature. And were Wyttenbach to have correctly understood the Optris support guy he spoke to we would have exponent uniformly 3.

I needed to check this since I could not otherwise be certain that Optris were not selling camera software that just did not work for grey-bodies, though this is a bit unlikely!

Wyttenbach - please contact your Optris support person who told you the exponent was always 3 and set them right - giving this simple check that anyone with a PC and web access can make.

It is perhaps interesting from these figures that at a temperature around that used in the Lugano dummy test we have exponent=4 consistent with what Levi thinks!

Quote from LDM

Maybe some additional information to work with:

The Optris IR basics manual on page 9 gives the following formula for the sensor signal :

U = C · [εTobⁿ +(1 – ε) · Tambⁿ – Tpyrⁿ] Temperatures in K

They state that for long wavelengths the factor n is between 2 and 3.
For some analysis I did I found that n setting at 2.72 gave me good results, but I have no proof that this is the correct value to use with the Optris.
Comparing two situations of the same measurement means that U stays the same.
For two different emissivities you then can equate the two formula's and solve the other temperature.
Note that at lower temperatures the factor (1-e)*Tambⁿ becomes significant.
That is a possible explanation that for low temperatures the approximation formula given by Optris does not work. At high temperatures the formula becomes more accurate.

I am curious what your results will be with the formula above.

Display More

Well, i think this specific formula works always, it is just that Optris are giving vague data about what is n and the camera does not go up to high enough temperatures to get to n = 1 (I'd guess the highest allowed temperature will around 2). If we can find a higher temp ravi file from an Optris we can see exactly what is n for all these higher temperatures.

Anyone want to post one?

EDIT - I take your point about the ambient correction. IF the camera does this then both the Lugano authors and I are getting it wrong by doing it again. That will make very little difference at high temperatures but more difference at the control temperatures. But I'm not clear from this that the camera software makes this correction (how does it know Tamb?).

Quote from ele

Dear THH

I apprecieate your efforts if favor of IH.

But now you are making confusion beetween Stephan Boltzmann law and the internal calibration of the Optris Camera that is quite complicate.

I note also that from your table the change of temperature vs emissivity is quite smooth a change of about 50% in emissivity lead to a change in temperature of only 8% !

This means that also the calculated energy must be changing smoothly and means that the emissivity factor is not really relevant.

Ele - that is a perfect deliberate Rossiesque deceit - or you are much more incompetent than you appear.

Try again.

Quote from Wyttenbach

@THH: I thought that physics is done by measuring first...

I have no lab no camera and optris just refers to their internal formula, telling us that the IR bolometer follows a T³ law.

It's up to you to make measurements like mfp did. MFP confirmed the T³ dependency...

Wyttenbach. You don't need a lab. You need a PC, you can download the Optris software, the mfmp ravi data, and use it without a camera. As I did, and gave the download link. you will then discover that your hypothesis about fixed e^3 is wrong.

And I'm calling you on this. If you don't have the balls to check it for yourself - or the courtesy to look at my results which do this - you can make no valid contribution to any discussion of this matter.

MFMP did not confirm e^3 at 1400C (it would be so perhaps at around 800C). As you can see the exponent n varies with T, something you have not yet admitted.

Quote from LDM

The ambient temperature is measured by the camera but can also be manula set.

See text of PI connect manual below

2.4.3. Emissivity, transmissivity, ambient temperature

Using the menu Tools, Configuration and Device the Emissivity and the Transmissivity (IR-window compensation) can be set. The transmissivity is referring to the loss of radiation if an object’s surface is measured by looking through a suitable window. Knowing the Ambient temperature is inevitable for a correct temperature measurement. The ambient temperature value is delivered by the camera’s internal sensor by default. Alternatively, the value can be set as a fixed value

Display More

Well done. So it looks like we have a correction:

This will not change the powers deduced from temperatures.

It will slightly change the way that emissivity affects temperature.

It applies to TC's correction - it is a v small additional correction: but relatively much larger for the dummy than the active case.

It does raise the possibility of an error in the Lugano data. We do not know how this ambient adjustment was set. There was in any case an error due to the local ambient temperature being hotter during long tests than normal.

If I had the strength of will I'd redo TC's calcs including this - might be interesting in the dummy case.

Quote from Alan Smith

Could you get his permission? He's not around much anymore.

LOL. His stuff is all open published on lenr-canr. I can do python 2.7. (That is a bit old now - v3 seems significantly better).

I wonder how we can predict the correct n value for a measurement (The Optris has to do this also)

You can do it integrating the Planck function over the Optris sensor BW and take the derivative of this integral wrt T.

TC's code does this numerically with a horrible hack (the integration part). You could easily enough get a numerical solution using matlab, or an approximate polynomial approximation using maple etc.

without using proper tools it is a bit difficult because though the integration can be done numerically in a hacky way following this by differentiation is a bit dangerous. Better to use proper numerical tools.

when I feel really bored I'll see whether there is a nice way to do it.

regards, THH