spectro-aras.com

I've reproduced similar (with RMS=17 as just given 5944.83A = 687 (pixels).
There are 2 cases:
1) with cosmic ray enabled, I get this: Result is very similar to what we seen above.
Reason is obvious, at undersampled lines filtered (wrongly).
You can repeat checking your _V250.FIT downloading ds9 (I used the windows version for now) selecting "Scale -> Min Max -> Sample" menu.
I think it is mandatory for everyone to look at that file in such a doubt.
This is the critical phase where ISIS processed 2D and right before 1D.

2) cosmic rays disabled (as you mentioned and tried): Well, here is also an error just a little left from the green line, but the reason is different: you have an unreproduced hotpixel left (ADU=65535).
Note that cosmic ray won't remove it anymore.
Now it is a dark calibration error.
Solution:
1) redo 1 dark, at the same exposure time if needed (lowering the threshold didn't help)
2) add that point to your cosme.lst (a bit guesswork of 2 coords... question how compatible ds9 and ISIS indexes, I tried 1-2 but guessed wrongly...)
3) move away the star on the slit next time from critical / wrong behaviour pixels (by 1st binning zone)
4) this the reason I don't use binning in general.... it bins them with good pixels, too... And cosmetic rays would become usable then (at 20 min exposures I do have a lot of them...).

But with this 2) error, I can see only a less big ripple.

Peter

One more thing, I may forgot to mention.
When you get over the hotpixel / cosmic ray problem, your green line will still leave a nasty trail (I have such only above 13-14 mag resulting as a ripple):
As you can see, there's a little rotation error, as well as some pixel arithmetic error along the sharp line.
If I'd do it in ISIS, I'd transform it first to a much higher resolution and do the rotation in there, using double float precision at least.
But other practices (re-modeling such line by Gauss fit, ...) may exist.

All in all I suspect room for improvement on this in ISIS (not done probably because of an edge case scenario), until that I suggest avoid binning.

Thank you Peter to have passed time on my question. Next time I will try to shift the Spectrum track on an other part of the CCD Chip and use 1x1.

Hi Hubert, also consider that, from V5.5, ISIS allows to dose the cosmic rays filtering, useful for those who work under "near-undersampling" conditions (not only for sky lines but also for spectrum image - it might be suffer artifatcs). The parameter is located in the Settings tab: "Coef. cosmic rays filter". 100 is the standard value (used internally in previous versions). You may increase the effectiveness using a lower value (maximum effectiveness at zero value) or decrease it using a larger value.

Trying with your files, I found that a value of 1000 could be a good compromise. The sky line at 5461A totally disappear but the other at 4047A is still present. This narrow residual feature, although weaker, is superimposed to a very low intensity (ADU) continnum which is then enlarged by the instrumental response.

Concerning the saturated hotpixel (projected near 540 nm in the spectrum) I agree with Peter, you should recreate the cosmetic map from a new master dark or add the specific entry in actual cosme.lst: P 555 396.

A possible alternative to the use of no binning camera mode: since you use a relatively long focal lenght (C11 scope), you might want to change to 31 micron slit aperture. I think you'll gain in general efficiency (throughput) reducing the actual strong undersampling of spectral lines.

Let us know!
Ciao
Paolo

Grazie Paolo per questi consigli, li mettero in atto per i prossimi spettri.
Hubert