Interesting calibration problem

[Stu]

Hi all,

I'm wondering if someone can help with a calibration issue I seem to be having on an Ha master.  In the screenshot below you will see a calibrated Ha light (bias, dark, calibrated flat master) that is clipped with an average of just over 1 count.  I tried it using the default WBPP, BPP, and manual and I get the same results.  I'm unsure if it's just that it's low signal data, or if I'm doing something wrong.

I've given a link to a folder that has several files in it.  A single bias, dark, and flat, and light (the same light that was calibrated in the screenshot).  I also included the calibrated light in the screenshot, the bias, dark, and calibrated flat masters. 

The Ha data isn't critical as there is LRGB data, but I've been thinking about this problem all day trying different calibrations (omit bias, omit dark, etc.) 

My general calibration routine if I do manual is:
bias/dark integration:  Average, no normalization don't care/ linear fit clipping
flat calibration:  Check master bias (do not check calibrate), check master dark (check calibrate and check optimize)
flat integration:  average, Multiplicative, don't care, Windsorized sigma
light calibration:  bias (calibrate unchecked), dark (calibrate and optimize checked), flat (calibrate unchecked)

then I do cosmo, SFS, star alignment, and light integration
Here is a link to a folder containing the .xisf and .fits files noted above

https://drive.google.com/open?id=1FFBsaAFkUJB76woOui4rBcNe4lDTpXR_

Thank you so much! 
Stu

[bulrichl]

Hi Stu,

why do you set the data range in Statistics to '14-bit [0-16383]'? The camera has a bit-depth of 16 bit.

Despite an exposure time of 30 min the signal in the Ha light frames is very low. As you can see in the histogram of your calibrated light (see appended Light_cal.JPG), the noise distibution effects that about 25 % of the pixels are clipped. In order to avoid that you should use an output pedestal in the calibration of the light frames (see the effect of an output pedestal of 100 ADU in the appended Light_cal_output_pedestal.JPG).

In the calibrated light frame I see a stripe pattern (dark vertical stripes in a zone 1250 - 1950 pixels away from the upper border). Don't know what that is.

Bernd

[Stu]

Thank you Bernd,
I have never used the pedestal before, but it looks like it will solve the issue.
The vertical bands are some artifact in the bias frames--I may try calibrating without bias.

The 14 bit thing was an oversight-- a few weeks ago I was showing someone awhile back the difference in output between 16 bit and 14 bit output and the final number and I accidentally left it at 14 bit.  The setting stuck. 

I really appreciate your response, thanks again!
Stuart

[bulrichl]

I fear that this will not succeed. I see the artifact (in the zone 1250 - 1950 pixels away from the upper border) only in your light frame. In the MasterBias and in the MasterDark, there is a different artifact (similar as in the light frame, but only in a zone 0 - 140 pixels away from the upper border).

I also tried calibrating only with MasterDark (no 'Optimize', no 'Calibrate') and MasterFlat -- expectedly the artifacts remain the same.

Perhaps this is a Residual Bulk Image (RBI) which could be avoided by a NIR preflash? (I don't have experience with this effect, so this is speculative.)

Bernd

[Stu]

Thanks Bernd,

It's old data, and not my equipment, so I really don't know the details on the hardware.  I don't seem to see the artifact in the broadband frames so I'll just keep going.  My real question was on the pedestal issue, and you answered that.  Thanks!

Now I'm wondering if there is a way to add a pedestal in on the WBPP or BPP script.  It would, of course, be added into any light frame I dumped into the script but I'm just curious.  I don't see it at first look. 

Stuart