I fully agree with Lothar on the value of the additional spectra. In particular in the first phase of the campaign the spectrograph in combination with the telescope has to be tested thoroughly. Unfortunately some tests of the echelle could not be done earlier, partly becouse of the terrible weather in middle europe before the campaign start.
After the first week the equipment is running, most problems have been sorted out and fully reduced spectra are produced. Once the remaining problems are sorted out, nothing except bad weather will prevent us from producing spectra in good quality at a higher rate than now.
The spectrograph is running as expected, the only problem is the signal of the calibration channel. I will try to explain this in a few words.
The flats are produced by sending the light of a tungsten halogen lamp or a ThAr HCL through a fiber into the guide unit at the focus of the telescope, where a mirror can be inserted into the light path and the calibration light sent through the fiber to the spectrograph. Ideally the calibration light and the telescope have similar f/ratios, therefore the light path of the fiber and the star are identical.
As a result the response curve which is calculated from the ratio of starlight flux / tungsten flux gives a fairly smooth curve as a function of wavelength. Both the blaze function and the detector response should cancel out and the result for the Shelyak echelle hav a shape similar to the one below, taken from the ISIS guide (typical eShel response).
http://www.astrosurf.com/buil/isis/eshel/tuto1.htm
measured with the same spectrograph and CCD
Going through the same calculation with the data from IAC, with the replaced lamp for the flat and zet Aql for the calibration star we get the following response curve (see below), quite different. Each wave represents a single order, the orders have been joined carefully. To everybody in echelle spectroscopy this is a well known problem, what is disturbing is the amplitude of these waves, not normally seen in an correct working echelle.
Having built an echelle and several other spectrographs I came across this problem also, so I checked several causes for these waves.
- misalignment of the spectrograph optics;
- poor background subtraction;
- poor alignment of the feeding of the calibration light into the spectrograph fiber.
The first cause I cannot completely rule out but think it is unlikely, anyway there could not be done anything about it now.
The second cause I checked thoroughly with the spectra taken so far and think I can exclude it.
Remains the third cause. By chance I had this problem about two years ago, with the same guiding unit as installed at the IAC. The culprit was the moving mirror, which was pushed out of its correct angle by an unknown cause.
Correcting only the slow variations of the response curve I got a spectrum of gam Cas which had almost the same waves as the response curve of the present data (see below).
Now you may say: so what. After all good spectra can be produced anyway by using this somewhat complicated function or the spectra can be reduced without using the flat files only for order extraction.
Correct, but there is another point I would like to make.
The calibration fiber is used not only for the flat but also for the ThAr calibration reference. And here comes in another point I found out some time later. When I made a careful wavelength comparison between FeAr lines of the calibration lamp and terrestrial H2O lines in the H alpha region I noticed reproducible, systematic calibration errors of up to 1 km/sec with a well aligned spectrograph, which I could not explain initially. I found the solution at the OHP at the T193, where the following reference was pointed out to me by the astronomer in charge.
http://arxiv.org/abs/1001.0794
There radial velocity errors of the order of meter/sec are discussed, however the instrument is quite a bit more advanced than our echelle, using state of the art mode scrambling. A rough estimate of the errors for my echelle, using the same lens as the Shelyak echelle showed that this was the solution for my calibration errors as well.
With a poorly aligned calibration feed I would estimate that the error could be considerably larger than what I measured. This would lead to additional difficulties if you cannot rely on the calibration produced by the ThAr lines. This error again can be corrected, by feeding in the calibration as is with a calibration produced by feeding in the light exactly on axis and measuring the calibration errors, later rescaling the individual spectra of each order.
The other solution would be to replace or fix the guiding unit, with the problem that fixing it has some risk of changing it for the worse. I could offer my guide unit on loan, but the problem is that the setup team is leaving in a few days and the following crews should do some spectroscopy on WR stars.
The third possibility is to ignore this possibility of error and work with the data as they come out, already very good.
I thought I bring this point into the discussion, with the intent to improve the quality of the spectra. We may not have to resolve it right away, but maybe check if it affects data quality. After all, also amateurs want to work to highest standards, as far as their means allow.
Martin Dubs
Registrieren
Anmelden
FAQ
Suche
