Instument response question!

[Gasman]

Hi guys

I took a spectrum of CH Cyg the other night and wondered about the final profile after correcting for instrument response!.

This is a grab of the raw but stacked spectrum before calibrating/correcting you can see the H alpha point which is the brightest point on the spectrum about 29000 ADU.

so you would think this would be the highest intensity point on the final corrected curve thus - and clearly it isn`t

so as the Halpha point intensity on the final curve is not the highest does this indicate that my Atik 314L shows certain points brighter than others and the software is simply doing its job in correcting accurately or have I missed something along the way??.

I get the same final curve using Rspec, BASS and ISIS so I assume the software IS doing its job!!

cheers

Steve

[Merlin66]

Steve,

I think it may be due to the low camera response as you move into the NIR. The QE curve drops pretty fast to some low numbers and zero around 1100A

[jnp]

Hi Steve,

This could also be affected by how tilt and active binning was applied.

cheers John

[Gasman]

Thanks for the helpful replies chaps . I didn`t notice any difference with the other spectra taken that night just the CH Cyg one?? Same processing with each one with just a -0.39 degs correction !!

Steve

[robin_astro]

Hi Steve,

the instrument response will be significantly lower at 7500A compared with 6563A. Also it looks like the spectrum may be wider in the IR than at 6563A (chromatism in the telescope and spectrograph?) so the integrated counts over the full width of the spectrum may be higher for a given maximum value (A check of the raw binned counts would confirm this)

Cheers

Robin

[robin_astro]

Also is the image flat corrected as this affects the apparent instrument response? ( If you flat correct, the Instrument response effectively becomes  the flat lamp spectrum. If not you are seeing the effect of the true instrument response which is mainly a combination of the response of the grating and camera, both of which will drop into the IR end)

Robin

[Gasman]

Hi Robin

I didn`t take any flats this particular evening , would it make that much difference to the response?.

Steve

[robin_astro]

Hi Steve,

Yes the "instrument response" ie the difference between the measured and the true spectrum is completely different if you do a flat correction. 

For simplicity let's ignore a few things like the atmosphere and the  flat defects we normally see in images. 

If we do not make a  flat correction , then as expected,  

the non flat corrected measured spectrum = the true spectrum * the spectral response of the instrument

A spectroscopic flat though is effectively a spectrum taken of the flat lamp

ie the flat = the lamp spectrum * the spectral response of the instrument

This means if we apply a flat correction by dividing the measured spectrum by the flat, the spectral response of the instrument cancels out and we are left with

the flat corrected measured spectrum = the true spectrum / the flat lamp spectrum

ie the "instrument response" effectively becomes 1/flat lamp spectrum

which is quite different of course.  This effect of using a flat is quite useful as it replaces what might be a quite lumpy instrument response with what is generally a much smoother black body curve from a Halogen lamp which is easier to smooth to produce a final "instrument response"

Note this is only true for slit spectrographs.  Flats for slitless spectrographs are much more complicated and unfortunately cannot be used to solve the "lumpy instrument response" problem which is why you sometimes see ripples in Star Analyser spectra, particularly those taken using Kodak CCD or some CMOS sensors which have more severe ripples in their spectral response.

Robin

[robin_astro]

Other than being aware of this, you dont need to worry though as the data reduction is the same in both cases. You cannot mix instrument responses calculated with or without flats though. Just remember the Golden Rule   - Provided you  record and reduce the target and reference spectra in exactly the same way  it will all work out correctly in the end

Robin

[Gasman]

Hi Robin

I have sometimes found that flats seem to add more noise than they help to reduce, at least the flats I take . Maybe flats might have a mention at Sidmouth ?.

Steve

[Merlin66]

Steve,

The original question....

Put it down to the loss of QE?

[robin_astro]

Yep, since there is no flat correction, it is just the Instrument response correction doing its job. Both the camera sensitivity and the response of the grating roll off at the red end beyond H alpha so the H alpha looks stronger in the raw spectrum than it really is relative to  the far red end. (The instrument sensitivity will probably peak around 5500A)

More about flats at Sidmouth but the main thing with flats for low resolution is you need to sum  a lot of them to get enough counts at the blue end to keep the noise down while avoiding saturation in the middle wavelengths. I normally take 30 or more well exposed flats, particularly if I want to push as far as possible into the violet  (It only takes a few minutes with the ALPY calibration module)

Robin