Fringing ripple

[Robin Leadbeater]

Hi Thierry,

Yes because there is an atmospheric glowing contribution.
What we saw in Robin spectra is essentially interference fringes due to optic/slit contribution, with regular pattern. But what we saw in the Robin eps aur potassium spectra is certainly due to atmospheric contribution in this very rich of tellurics wavelength domain. Most of the time it is a mix of the two phenomenon and cannot be removed even by professionals.

There are of course telluric effects around 7699A but after removal of Tellurics, the main effect in my 7699A spectra is still the ripples from interference in the CCD cover glass. (The wavelength spacing is exactly right for the glass thickness and agrees with the theory at 7699, Ha and Na D.)

I do not believe I am seeing a significant sky background contribution. I see exactly the same effect in short exposures on bright stars like Altair etc. The sky background contribution would be very low.

Conventional flats do not work well for me for removing the ripples for two reasons.

1. They are not taken at the same orientation of the telescope. Even small shifts can upset the flat division

2. More importantly, the optical path is different for the flat lamp. The result is the fringe position and intensity across the field is different. (Even moving the flat lamp a small amount can change the intensity of the fringes across the flat.)

interference fringes from the cover glass are generally less severe at shorter wavelengths because the spacing gets smaller and once it is smaller than the resolution they become invisible. That is why they were only first seen by amateurs when the LHIRES was introduced.

Cheers
Robin

[Robin Leadbeater]

Hi Bernard,

They say the problem is due to internal CCD reflections but also state that "fringing is not significant below ~7000A", which has been said by others (e.g. Howell, etc). Not sure what I should conclude as I certainly have a significant fringing problem at much shorter wavelengths.

Here Howell is talking about interference within the thickness of the sensor substrate, not in the cover glass that we are seeng. This can only happen where the substrate is transparent ie at IR wavelengths. Pro CCD's are generally thinned and back illumated too, different from our normally front illuminated "domestic" CCD. The CCD construction can cause ripples though, for example the KAF CCD have very long wavelength ripples in the spectral response. These are very different from the closely spaced ripples from the cover glass though.

Cheers
Robin

[Bernard Heathcote]

Robin,

Good points ... you are right. I was forgetting that the pro's and amateur's CCDs are usually very different beasts. I now need to find the time to go back through my spectra to see if I can more clearly define how, when and where fringing has impacted on my various setups.

Bernard

[Benjamin Mauclaire]

Hello,

Theis phenomenon comes from the plate that protect CCD chip: it acts as a parallel dioptric that makes interferences.
There are two web pages that make an overview about this problem and the way to solve it:

- http://www.astrosurf.com/aras/fringing/opus/index.html
- http://bmauclaire.free.fr/astronomie/sp ... s/fringes/

Benji

[Bernard Heathcote]

Hi Benji,

Happy to see you joining this discussion. Your, and others, work on this problem was mentioned in my original message ... We are now trying to progress our understanding a little more, specifically why the ripple appears to vary and how to reduce or eliminate it.

Cheers,
Bernard

[Bernard Heathcote]

Some more puzzling results ... if you look back to my original post (above top) you will see that, compared to my Lhires+Atik314L, my eShel+ST-8ME has a lot more ripple. The plot below, from current proam work I'm doing on eta Car, shows that the eShel+ST-8ME and Lhires+Atik314L now have very similar results. The only real difference between the two sets of results is that they are on different stars (zet Pup and eta Car), nothing else that I'm aware of has changed.

Any ideas whey the ST-8ME ripple ... or is it just random noise ... has gone? I'm also a bit puzzled why the eShel emission line amplitudes are 25% greater than those with the Lhires.

Cheers,
Bernard

[Martin Dubs]

Hello Bernard,

two comments on your Echelle spectrum:

Noise:
The higher noise of echelle spectra may have its origin in the fiber. In the fiber different modes are excited, which give an intensity pattern looking similar to interference fringes. I noticed it earlier in my flats where S/N is high. An more detailed explanation you find here (in german):
http://spektroskopieforum.vdsastro.de/v ... 5986#15986
Look also at other posts in the forum under Technics - LWL(fibers)
I found an english version:

fiber optic mode noise Muenchen.zip

Fiber optic noise

(202.93 KiB) Downloaded 405 times

One method to get rid of this interference is to move the fiber during exposure (gentle shaking).
Another method would be to use a mode scrambler (more complicated, additional losses).
The lower S/N may als have the cause in lower signal (smaller number of e- per spectral bin, with possibly different bin size)

Peak intensity
- the higher peak values of the emission lines with the echelle may have different causes:
- higher resolution
- better spectral purity (absence of scattered light from the grating) due to the cross disperser due to imperfections of the grating.
- eta Car is a speciaal star with an extended nebula around it. With your slit spectra you may subtract some background from the nebula, reducing the intensity of the emission lines. With the fiber this background is not subtracted, unless you make a control exposure with slightly offset position of the fiber on the sky.

These are just ideas, do not blame me if not everything is indeed the case.

Good to now that other people have similar problems.

Regards,

Martin

[Bernard Heathcote]

Hi Martin,

Thank you for your comments and the link to the very interesting paper on fiber noise ... the idea of shaking the fiber cables during an exposure is certainly counter intuitive, but fascinating. According to the first graph the predicted S/N at around 4700A, where I'm working, is only around 80, which is about what I'm getting. I still don't understand why the 'noise' is varying, unless it is related to where the fiber cables are positioned. There is definitely a need for further investigation on how it affects amateur equipment and how to reduce it.

Your ideas on the line intensities are also interesting and, even though I don't think I agree with some of them, I need to look into each of the points you make ... I won't hold you responsible for any errors

Cheers,
Bernard

[Martin Dubs]

Hello Bernard,

the expected S/N depends also on the fiber diameter, mode number is proportional to fiber diameter^2. In the FOCES spectrograph a fiber diameter of 100 microns is used, compared to the 50 microns we use. Therefore you should not expect the same S/N as in the paper. In addition it depends on how much of the light is vignetted. See here for a detailed description of the Foces:
http://adsabs.harvard.edu/abs/1998A%26AS..130..381P
By the way I mostly observed the effect with my lab setup, by dividing a flat spectrum by a flat reference. On the telescope other limitations to S/N prevented me from seeing this. In the flats the additional noise is smoothed out before using it for the blaze function, so it would show up only in the object spectra.

I did not mention incorrect subtraction of background as a cause for difference in peak intensity, a common beginners error, since I trust your correct treatment of spectra.

Regards, Martin