Camera Instrument Response

[Heian]

Hi all,
Does the instrument response of the ccd vary according to the class of object (star) being measured with a SA100?

I've been watching the video tutorials on RSpec and there is a comment that suggests you need an instrument response for each class of star.

I would have thought that the response is dependent on the CCD chip, and as such, is independent of the object being measured.

Comments?

Mark

[Merlin66]

It's more the "instrument" and atmospheric response which can/ does vary, rather than just the CCD response curve.
It's best, if you want to achieve best resolution, to 're-calibrate' with a A type star to confirm the response curve each night.....
It doesn't vary by star classification, more by altitude and general conditions of your set up.
What are you using for your spectral data collection?

[Heian]

Hi Ken,
I've just been getting back into spectroscopy after a break, and the images I got were through my 66mm APO with the mono DSI3 attached. The Hydrogen lines showed up really well, with a pixel spacing of 8.7A/px. I took some images of Fomalhaut (A3v) and the Hydrogen lines down to delta + possibly epsilon shwoed up fairly clearly. :) I've attached one of the images below.

My issue came about when I tried to process a spectra of delta scorpii, and when corrected with the IR I worked out from Fomalhaut, the processed curve was not a good match for the reference curve.

My other issue has been remembering how to normalise the corrected spectrum, I could never quite work that out!!

Mark

[Merlin66]

Mark,
What do you mean:
""when corrected with the IR I worked out from Fomalhaut"
Normally the response curve would cover 400nm through to around 750nm and provide 'corrections. over the same bandwidth.
Do you mean normalised to the continuum?

[Heian]

Ken,
I worked out the IR based on the A3v spectra, and then applied it to delta scorpii. My issue might be that the IR curve covers from 3100 out to nearly 10000 A. I'll try it over a smaller range and see what sort of curve comes out.:thumbsup:
Your comment about normalising to the continuum is an acion in Vspec that I've had trouble with over the years.

thanks for your comments, Ken.

Mark

[Heian]

Hi Ken,
After a bit of work and reading over the last few days, I've come across this info in Analysis and Interpretation of Astronomical Spectra By Richard Walker, pg 25 of 90;

For correction curves, obtained this way, the familiar term instrumental response" is misleading.
This curve doesn’t just reflect the erroneous recording characteristics of the spectrograph, camera and telescope, but also the
wavelength dependent filtering effect of the Interstellar Matter and of the Earth's atmosphere. Strongly different spectral classes with distinctly different temperatures must therefore inevitably generate strongly different correction curves. Therefore the often postulated “all purpose use” of just one single correction curve causes strong errors, if used for other spectral classes.
After a number of runs through VSpec, it seems the only way to get my curves to match up with the relevant library curves, is through the use of slightly different instrument response curves.
Does this make sense to you? Mark

[Merlin66]

Anything Richard Walker say's is worth listening to!
The difference for me is that the other variables he mentions depend on the object under observation and the altitiude etc etc.
There is a basic underlying "correction" you can apply to ALL your spectra which at least will correct/ compensate for the the scope/camera/ spectroscope issues/ problems. The other factors you can address later.
What's the concern with doing a continuum normalisation? I can help.

[sheeny]

Very Interesting. I struggled to get a single "instrument response" curve that could apply to different stars. I think my results are in a thread here somewhere. I ended up averaging my response curves collected from a number of different stars. This quote explains a lot I think.

Al.

[Heian]

Ken,
thanks for the offer of help, but I think I've managed to work it out while playing around with the instrument response issue.

I generate a continuum on the measured curve and then divided the curve by that new continuum, and the result is a horizontal line with the peaks and troughs in the right places.

Mark

[Merlin66]

Sounds OK to me!
The idea is to use the background continuum as the "zero" reference...
Similar to preparing the camera/ instrument response curve...