First spectra with Star Analyser

[malclocke]

Well, it's not really much to show, especially with the amazing results people have been sharing of Nova Del, but I'm fairly pleased with it.

It's actually my third attempt, first couple of efforts were with a DSLR which I need to use with a barlow to achieve focus on my scope and hence pretty miserable results, plus the setup gave me about 4 angstrom/pixel.

Tonight I tried it with my QHY5, a huge improvement. Much broader response curve, able to achieve focus without a barlow so the scope running at f/5, and the spacing gives me a more reasonable 12 angstrom/pixel.

Getting focus and framing with the dob is infuriating. One day maybe I'll own a tracking mount :(

10" GSO dob, QHY5 mono and a Star Analyser SA-100.

Malc

[sheeny]

Looking good, Mal.

Al.

[Merlin66]

Certainly you seem to be getting somewhere....
How did you apply the instrument correction you generated?
The calibration seems to get a bit "lost" on the blue side of Hb....
Onwards and upwards.

[malclocke]

Thanks guys.

Ken, the calibration is from a home brew Python script I wrote, hence the somewhat dodgy results. Basically dividing the raw spectra by the reference, and then applying a smoothing algorithm on the resulting curve to eliminate the peaks. Obviously the results are not quite right.

I'm on Linux, and have tried using VSpec under an emulator but it is somewhat 'crashy' and not every feature works. I've since found a nice tool called splat-vo (http://star-www.dur.ac.uk/~pdraper/splat/splat-vo/) that allows me to do the calibration process pretty much as described in your book (which has been a fantastic help by the way, thank you). I still need my script to do the wavelength calibration, but I think I have the maths correct for that. Here's the result of using doing the camera response through splat-vo, which gives a better result.

Still have a few things to iron out, but getting there. Now just need this cloud to stay away for 10 minutes so I can have a crack at Nova Del.

Malc

[Terry B]

Well done!

[malclocke]

Making slow progress with my processing pipeline. I think I can work on the response curve a little, all my spectra include an oscillation in the continuum which seems to be fairly uniform.

I now have a command line driven workflow to dark subtract, crop, bin, detect zero order and calibrate to predetermined A/px, wavelength crop, normalise and stack which works OK for me. I then do the response correction in splat-vo.

I must say I was sceptical of how effective the response correction was going to be against anything other than my reference star, but have been pleasantly surprised to see at least a vague match between my spectra and the corresponding pickles references.

Attached are Fomalhaut (A3V), Alpha Cen A (G2V - contaminated with Alpha Cen B) and Antares (M1.5Iab). My spectra are in red, pickles references in grey.

Still can't seem to do anything about the cloud though ...

[Terry B]

Are you using flats?
This might get rid of the oscilation in the continuum.

[Merlin66]

Malc,
that may well be worth trying.
Also in your work flow you don't mention background sky removal...

[malclocke]

Terry, I'm starting to experiment with flats, I'm not convinced it is the cause of this oscillation though, although it could be. Because I'm doing all this with a dob the target is moving all over the place in the frame, but the oscillations seem to come out at the same frequencies. So I think it is probably either the camera response or something inherent in the star analyser, or both. There is a thread on the astronomical_spectroscopy yahoo group currently which seems to suggest that a grating may produce this kind of effect.

My main ailment which I'm trying to cure with the flats is a bayer type pattern from my QHY5, see the top right of the attached screenshot. Even though it is a mono camera it still exhibits a bayer style grid. I haven't been able to find an answer as to why it does this. I think possibly the chip is the same as is used in the colour QHY5 and has a different gain based on whether it would have been R,G or B. Not sure. Of course, the real answer is to get a proper camera, but I have to make do with what I have :)

Ken, background sky removal I hadn't even thought of. I'll add it to the list!

Thanks for the help guys, it's really appreciated.

[Merlin66]

Malc,
If you are using the original QHY5 it has the mono MT9M001 1/2" CMOS chip. I have one which I now use for guiding...
With a pixel size of 5.2 micron it normally takes a hellava magnification to pull out the individual pixel.
I use AstroArt V5 for all my cameras DSLR, QHY5, Lodestar, ATik 16 and 314L. I don't see a Bayer type issue with my QHY5.

[robin_astro]

Hi Malc,

The grating is a possibility but not all users see it or see it to differing degrees. (I use Sony CCDs and do not see it but some Kodak CCD for example are renowned for these sorts of effects and there is currently a QHY5 user (CMOS sensor) in a thread on the RSpec_Real_Time_Spectroscopy yahoo group who is seeing the same effect so I am tending towards it mainly being a camera response issue.

With slit spectrographs a flat removes them but this does not work with slitless spectrographs as the flat does not contain any wavelength dependent information. The solution is to do less smoothing in your instrument response to try to preserve these features in the instrument response

Cheers
Robin

[robin_astro]

Here is an example showing instrument response correction of a spectrum of Altair using a Star Analyser and a QHY5 camera (Altair, taken by John Berman and processed by me using Visual Spec) I have taken care not to over smooth the instrument response so the ripples are preserved and correct the spectrum

Robin

[robin_astro]

And for comparison the same sort of analysis for an ATIK 16IC-S camera (Sony ICX429AL CCD) showing the absence of ripples in the instrument response.

This reference shows the ripples in the response of Kodak KAF CCDs
http://www.gco.org.au/st8/KAF-1603ME.jpg

Cheers
Robin

[malclocke]

Thanks Robin, that's really useful.

I smoothed my A3V (pickles) reference slightly, then used that to create a response curve complete with ripples. Result is a lot better match to the original pickles ref I think.

[Merlin66]

Malc,
That seems to be a much better outcome.
I must say I'm surprised to see the instrument response curve - I can't remember seeing anything like that with my mono QHY5, I'll need to check!
(The data sheet shows a smooth efficiency curve, similar to the ATiK16)

[malclocke]

Yeh, me too. It seems that these kind of things are exhibited variably by the QHY5 cameras, and it could be something driver specific.

I've got a couple of things to look at with regard to the bayer pattern, and I guess it's possible the bayer pattern is causing this issue although I don't think so. The bayer pattern causes a fluctuation between adjacent columns of pixels when binning my spectra, see attached, but this goes away when I co-add a few frames. I can't see how the bayer grid would cause the wider range response fluctuations though.

As stated flats would probably get rid of this, but I'd rather find out the root cause. Keep on keepin' on. :)