Fringes pattern in LHIRES3's spectra

This page describes interference fringes that may appear with a LHIRES3 equipped with a 150 g/mm grating in combination with an ST8 XME camera.

Since the fringes don't keep the same position it is impossible to remove them by means of a flat field correction. CCD sensor and its protection glass seem to have a preponderant importance in fringes's formation. The fringe pattern thus seems to depend on the camera.

Netherless, with 2400 g/mm grating, we may observe a clenched fringes pattern.

The combinaison of LHIRES3+150 g/mm grating and an ST8 with an KAF1603ME sensor manage to cover theo whole visible spectrum.

In order to obtain the instrumental response of a {telescope+spectrograph+camera} system, it is classical to acquire the spectrum of a reference hot star that is also available in profesionnal's collection [A Stellar Spectral Flux Library, 1150 -- 25000 A (A.J. Pickles, PASP 110, 863, 1998]. Such a star is called a reference star. I usually choose Altair.

The profile of Castor (one of these reference stars) displayed below has beenacquired with an Audine KAF1602E camera and the same telescope (SCT 0.3m) and spectrograph. It shows a sleek continuum and some Balmer's lines:

castor

Now, here is the same type star's lines profile (Altair) obtained with the same spectrograph and telescope but with an ST8 XME camera. We can see fringes pattern in the continuum:

altair 1c

I first suspected the problem to arise from errors in the instrumental response (RI) correction. But the pattern is also visible on level 1b ( i.e. the profile prior to the RI correction):

altair 1b

So the problem would come from the flat field correction:

flat 150t ST8
We can see a fringe pattern with an irregular period and with positions that don't match those observed in the aquired Altair spectrum.
Flat's fits file.

profil franges
Normalized lines profile highlighting the intensity of high frequency ripples.

As the flat field has been taken the day after, fringe patterns couldn't be removed: there is an horizontal offset and the ripples show different periods.

Fringe patterns can also be seen in unprocessed Altair's spectrum:

altair bleu
Fits file of unprocessed Altair's spectrum.

This small study leads to the following conclusions:

Fringe patterns seem to be impossible to remove when flat-fields aren't done the same day.
A KAK1602E with the same spectrograph and telescope hadn't such fringe patterns. So, my ST8 XME is likely to be the problem. A test with another ST8 XME and my spectrograph would help to make conclusions.
With a 2400 g/mm gratting, fringes may appear with some CCD sensors but are closer together.
On the spectro-l mailing-list, some users of LHIRES3 spectrographs have such ripples in their spectra. One explanation would be that CCD sensor's cover glass produce interferences. However, fringe formation with LHIRES3 isn't yet totally undestood as there are many camera-spectrograph combinaison.
It would be appreciated that spectrograph providers help their customers to understand what happens. In particular would be very useful a methodic study involving:
- Tests of many cameras with the same spectrograph;
- Test of the same camera with many samples of the same spectrograph model;
- Influence of some tunings (mirror for instance);
- The study could be carried out at high and low resolution.