Hello Peter,

I'm not sure who is writing the fits headers, is it the driver or kstars/ekos ?

After a data aquisition with kstars/atik (v0.3) combo, I tried to stack the images.
For this purpose I'm using the theli software www.astro.uni-bonn.de/theli/gui/

Since I got some errors during the stacking process, I contacted the developers.
The main developer of Theli pointing two issuses inside the data:

(1) Negative exposure time (EXPTIME) inside fit
(2) Changing point or comma as separator inside the value itself. Again EXPTIME

The original link to the issue description , it's in german :
www.theli-forum.info/viewtopic.php?f=11&...9abbe8235f3c11#p4458

thank you

I'm using atik driver 0.3 with latest stable KStars (as part of KDE apps 15.04) as well as bleeding edge KStars on Fedora 22 and never got negative exposure times. I use Atik 314L and 383L+. So can you describe your setup and a way how you can reproduce this issue?

Hello lupnix,

I captured around 30 flats and some of those have negative EXPTIME values.
The headers (DS9 viewer) are from the attached example:

EXPTIME = -0.250456988811493 / Total Exposure Time (s)
CCD-TEMP= -4.95970678329468 / CCD Temperature (Celcius)
PIXSIZE1= 4.53999996185303 / Pixel Size 1 (microns)
PIXSIZE2= 4.53999996185303 / Pixel Size 2 (microns)
XBINNING= 1 / Binning factor in width
YBINNING= 1 / Binning factor in height
FRAME = 'Flat Field' / Frame Type
DATAMIN = 65535. / Minimum value
DATAMAX = 65535. / Maximum value
XBAYROFF= 0 / X offset of Bayer array
YBAYROFF= 0 / Y offset of Bayer array
BAYERPAT= '' / Bayer color pattern
DATE-OBS= '2015-06-14T00:07:52' / UTC start date of observation
COMMENT Generated by INDI
END

Only flats had those wrong exposure times. Lights and bias were fine.
What ADU percentage are you setting for flats ?
I'm using Neumann panels for my flats. I think I was playing with values between 20-30% ADU.



thank you

Hi Carlos,

driver is not setting any FITS headers, it is all done by underlying INDI library. What was requested exposure time?

Peter

Hi Peter,

Ah ok, it's INDI layer. I'm using 15sec.

On the other hand I see no bug in INDI::CCD as well... Jasem, any idea?

BTW, I don't think it is a good idea to use localised number formats in FITS header, decimal points is correct, but EXPTIME value is wrong anyway, it is not rounding or formatting problem.

Peter

I'm not sure what's the problem exactly... at any rate, I set setlocale(LC_NUMERIC,"C"); when saving FITS header values. Maybe that will help. But I don't have the sources for the ATIK driver, when do you set the exposure duration?

I'm setting it just in StartExposure(). I don't think it is a locale issue, then number is completely wrong. BTW, I believe that cfitsio uses "C" locale anyway, slovak locale uses "," instead of "." as well and I never noticed this problem before.

What is the difference between lights, darks and flats in Ekos? It seems that only flats are corrupted...

I take flats all the time with QSI without any issues. I have no idea what could be causing this.

Hi guys,

I'll try to take flats with different exposures & ADUs to narrow this issue.

Hi,
I just look into the code and noticed that ImageExtention is defined as a MAXINDIBLOBFMT sized array but in the SetImegeextention method, the strncpy uses MAXINDINAME as max size. Furthermore ImageExtention is defined just before the ImageExposure number vector in the header file. This suggests an overflow but I do not have looked at how ImageExtention is used.
Maybe this could be investigated.

Jean-Luc.

Hi Jean-Luc, both MAXINDIBLOBFMT and MAXINDINAME are defined as 64. So this is potential issue, but not this time. Peter