At last ... the ultimate FWHM-tool ...... not.

[georg.viehoever]

Hi Hans,

In theory (and ignoring some practical issues such as coma, diffraction spikes, tracking errors, ...), FWHM would depend only on the optical characteristics of your telescope, and would be independent of the measured star, see for example http://www.astro-imaging.com/Tutorial/MatchingCCD.html.

I think I understand why the tested tools give different values for different stars: most of the selected stars have a center that has a brightness close to saturation (K>0.95). This of course distorts the usual bell shaped distribution of intensity values. This probably also invalidates the bell shape assumption that some FWHM measurement tools make. Therefore you get high FWHM values for bright stars, and low values for dim stars.

If you want to measure the seeing or the optical quality of your telescope, you would of course need to measure stars that are not saturated.

(I hope I got that right...)

Georg

[Carlos Milovic]

Here are the new measurements, in arcsecs:
5,71
5,70
5,17
7,24
4,72
7,55
7,36
5,19
6,69

And the graph comparing the values is attached.

I used a constant sky value of: 0.028290, or 1854 ADU (the median value of the image). The threshold used was 0.05, which is close to 3*standard_deviation, so almost no background pixels were included.

[Hans Pleijsier]

---
Georg,
I see your point but I think the color of the star will play a role here or will this only be at sub-pixel levels?
Anyway, ImagesPlus offers the user a possibility to calculate FWHM per color and I ve seen different FWHMs in excellent pictures (not mine).
My two cents on saturation: in the case when one and only one pixel is approaching but not exceeding saturation there is no problem in drawing a gaussian bellshape. When to or more pixels are saturated the maximum pixel is unknown in position and brightness -> equals no fit.
---
Carlos,
Stareval is using the exact same background estimate around 0,028
Watching your diagram I decided to check upon the first Sextractor value because the difference is quite remarkable at star 1.
And I must admit that I made a mistake: the correct first reading for sextractor at star 1 should be: 2,48 pixels and thus 5,00 arcsecs.
The final diagram shows a strong coherence of the tools.

This concludes our little test. I leave any conclusions to you or other forum-readers.

[georg.viehoever]

Hey, with this correction for Sextractor the 4 tools match quite well 

...
I see your point but I think the color of the star will play a role here or will this only be at sub-pixel levels?
Anyway, ImagesPlus offers the user a possibility to calculate FWHM per color and I ve seen different FWHMs in excellent pictures (not mine).
...

You will have different FWHM for different colors for refractors - because color dispersion (caused by the different refractive indices of for different colors of light) is one of the factors that may have an influence on the FWHM (just like coma, tracking error, diffraction spikes, ...). Reflective telescopes should not be affected by dispersion.

Could you try to measure the FWHM for stars that are less bright than star 5? I am pretty sure that you will measure values between 4-5, which would be the real FWHM for this shot. I think that for brighter  stars, the top of the intensity profile is flat (more like a mesa top than bell shaped), and most tools will produce wrong FWHM estimates in the case. For certain, such FWHM values are meaningless if you interpret them as the optical quality of seeing or instrument.

Georg