Mon, 12/08/2014 - 05:27
Following the recent interest in the SA100/ SA200 gratings and the obvious commitment from the members to buy one, I think there needs to be some serious thought given to the necessary initial support to users - "holding hands in the traffic" - experience in photometry, using filter wheels and your favourite camera doesn't automatically guarantee you will have easy and successful initial results.
What do you want to achieve with your grating?
Let's throw open the discussion................
I'll beak the ice -
I've never done any grating spectroscopic work in a converging beam before (like that coming from the primary mirror) and I have one main question:
Since the optical path will be longer the further from the zero-th image you are on the chip, where is the best place to have the focal point - the H-alpha image, or towards the blue? If towards the blue, how much?
Or am I all wet?
If you do the math for your system, I think you will find that the difference in negligible. That assumes you have a typical system. But your system would have to be well outside the norm for it to matter.
Regardless, it works best to use H-alpha or H-Beta to focus for other practical reasons (they are easier to see and stronger lines).
Hi Lew,
Yes the effect of curvature of the focal plane on focus along the spectrum is one of the limitations of such a simple system and can be significant, particularly with larger apertures which need longer spectra to overcome the limitation of star image size. This is one of the reasons these systems give better resolution on smaller aperture scopes. For this reason it is important (as explained in the manual) to focus in the spectrum rather than on the zero order.
For most setups focussing on the H beta line of an A star gives a good compromise in resolution over the visible range and is an easy line to focus on initially before moving to other targets but if you are particularly interested in H alpha and definitely if you want to study lines in the IR then it is best to focus on that end of the spectrum. (Note that if you have a system with a digital focusser, once you have determined the optimum focus you can focus on the zero order and then apply the apropriate offset)
The effect of this and other aberrationscan be reduced in part by adding a wedge prism which brings the centre of the spectrum onto the optical axis. Paton Hawksley can supply one matched to the SA100 but to be honest I find the disadvantages outway the small improvement for the low dispersion of the Star Analsyer. In any case it is not generally practical to do this with a filter wheel mounted setup.
Ken also introduced the separate issue of using achromatic refractors. This is also covered in the manual. I would recommend avoiding achromatic refractors for any serious spectroscopy. Even chromatism in focal reducers can be problematic.
Cheers
Robin
Though where the Star Analyser is concerned the use of a focal reducer to minimise star size in average seeing with typical f10 SCT above 8 inches aperture is often advantageous as the improvement in resolution generally outweighs any disadvantages of increased chromatic aberrations.
Robin
Some basic achromats give a "fishtail" spectral image due to the chromatic aberrations.
To get maximum resolution, you should focus on the spectrum in the area of interest. After initial focus, ignore the Zero )Order image and focus on the spectrum
As mentioned previously, using an A type star as a target will give you an ideal "learning" opportunity.
You will see, as you refine the focusing, that the absorption feartures (Balmer series of hydrogen) get more pronounced (and appear "deeper" in the 1D profile) - this is where you want to be.
Download one or more of the recommended processing software packages and start your "Spectroscopy 101" training.
Increasing the dispersion (i.e. larger/ longer spectral images) leads to fainter images with marginal improvement in resolution.
Hope this helps.
Hello: Happy Holidays
Got my SA200 installed, spaced, focussed, and almost aligned. It cleared last night for the first time in about 3 weeks, and I got 9 images of EE Cep thru the grating. Also got a short time series on it. I have attached the stack. Comments welcome. The only flat I had was one made with an ELU panel, and it looks like it did not work well. I will get some sky flats tonight thru the grating and retry.
Gary
Well done, Gary.
I loaded your image into RSpec and attached a screen shot.
What camera, spacing and exposure were you using.
I could not match the obvious emision lines on you image with the Biii star in the reference library.
What do you get as Angstra/pixel?
Richard
I suspect those are not emission lines, but background stars. Also, I think your zero order is wrong. The SA is not good for crowded star fields like this. This is where a slit spectrograph would work much better. For the SA to work in this situation a mask of some sort might be needed.
For a paper that I co-authored (I did photometry, not spectroscopy of the star) and information on the star and its spectrum see
International observational campaigns of the last two eclipses in
EE Cep: 2003 and 2008/9
http://arxiv.org/pdf/1205.0028.pdf
Jeff (187283)
HPO
Hopkins Phoenix Observatory
Hello Richard
Good to hear from you. Thanks for loading it into Rspec. I don't have SW yet. Waiting for AAVSO to recommend. Here are particulars.
Scope is mewlon 300, fl = 3000mm, aperture = 300 mm, pixel = 24 micron, 512 pixels x 512, 1.65 arc secs / px, 14 Angstroms/px, 0-7228 dispersion, spacing = 85 mm.
Looks like I should drop my spacing. I can reduce it in 20 mm increments. From the calculator, looks like spacing of 65mm, 18.5 A/px, and 0-9452 A dispersion would work a little better.
BTW: I posted the image in PPT. I tried to attach the Maxim fit files to this forum and they would not take. Anyone know the issue?
Gary
Hi Gary,
Thanks for posting that EE Cep grating image. I'm learning RSpec so that I can decide whether it is the right path for the grating group; I should have something to report in a week or two.
Regarding flatfielding: you basically CANNOT flatfield a diffraction grating image. The closest you can do is to take an unfiltered flat and apply that. It would then correct the zero-order star images, but would not properly correct the first-order spectra, since they appear on the chip at a different location. This is very obvious for a field that has vignetting; the spectra may appear in the vignetted region, yet are not vignetted, and applying a flat will improperly scale them. Apply bias and dark and call it good. If your software requires a flatfield image, I suggest creating an image with a constant value at every pixel location and use that as a pseudo-flat.
As Jeff mentions, working in crowded regions is a challenge. Zero order star images will cause emission-like bumps in your spectral extraction; overlapping spectra will occur. There are two main ways to improve that situation: (1) have two gratings in your filter wheel, rotated different amounts, so that you can use the grating that has the least contamination in the spectrum of interest (or use some advanced software and images taken with both gratings to remove contamination); (2) use a "slit" placed above the grating to limit the number of stars showing up in your image. John Martin showed some examples of this at the grating workshop during the Fall meeting. The final solution is to avoid crowded fields and do spectroscopy on a more limited set of your target stars. As Robin mentioned, some of your first targets should be calibration stars, until you understand the limitations of the setup. He gave some references to lists of good starting stars.
The Bright Star Monitors have diffraction gratings. We only do spectra of bright stars, so there is less confusion in the field.
Arne
[quote=HQA]
Hi Gary,
Thanks for posting that EE Cep grating image. I'm learning RSpec so that I can decide whether it is the right path for the grating group; I should have something to report in a week or two.
snip
Arne
[/quote]
Arne
Just a comment about RSpec
Does it save 1D spectra as .fit files or just as .dat files?
I tried it some years ago and it wouldn't open 1D .fit files
fit files are the standard as they have useful info in the header that is missing from the .dat files.
Terry
Flat fielding with slitless spectrographs is indeed an extremely difficult and complex subject and there is as yet no easy way that iam aware of for the amateur to do this entirely. There are things you can do however to minimise the effect flat field defects may have on your data. The key thing to remember however is if you do not at least take a flat or take into consideration the effect of flat defects, you cannot rely 100% on the veracity of what you see in your data.
Vignetting can be a problem as the effect depends where in the optical chain it occurs.
Vignetting ahead of or at the plane of the grating is not a big problem as it has no effect on the spectrum (other than changing its overall intensity if the zero order is placed in the vignetted region)
Vignetting after the grating will affect the spectrum shape and this cannot easily be corrected for using a flat field. The best way to correct for the effects of vignetting after the grating is to place your reference star at the same position in the field as your target, then when you correct for instrument response, you will also correct for any gross vignetting.
Localised defects at/near the imaging plane eg dust on the CCD window etc can be identified and avoided by placing the spectrum in clean part of the field or corrected for in the usual way using a flat (normally recorded without the grating)
Robin
Terry,
I don't think Tom has managed to get RSpec to handle fits.....
I did ask a while ago about BeSS standard fits headers - RSpec is not there yet.
Try something like BASS, VSpec or ISIS.
[quote=Ken4optics]
Terry,
I don't think Tom has managed to get RSpec to handle fits.....
I did ask a while ago about BeSS standard fits headers - RSpec is not there yet.
Try something like BASS, VSpec or ISIS.
[/quote]
Yes
I use all of those and Iraf as well. If there is plans for the AAVSO to recomend particular software then it needs to produce 1D fit files.
Terry
I have used most of the avaialble spectroscopy processing software, VSpec, ISIS, spcAduce, BASS, RSpec. My choice, while not free, is RSpec for ease of learning, use and power. I process all my spectroscopy image with it now, from SA100, ALPY600 and Lhires III. All the images produced are in FITs format and RSpec handles them fine. RSpec also had an option to allow reading the FITs header.
Since most of the AAVSO people who have purchsed SA200s are just begining spectroscopy, I highly recommend RSpec. There is a great deal to learn about hte processing of spectra. Starting with VSpec or ISIS will most likely produce great frustration and discourage many.
Jeff (187283)
HPO
Hopkins Phoenix Observatory
I have been blackening
Hello
I have been blackening some modified T adapters for my SA200 image train. Tried magic marker, but very poor. Krylon ultra flat black is my go to technique, but concerned about getting spray paint into the threads. I discovered that Krylon makes a paint pen, and it worked about as good as the original anodize. Does anyone have a source for ultra flat Krylon or equivalent in a can/bottle? I can paint with a brush if needed.
I have discovered Testors offeres a paint pen with "Flat Black" as a color. I have one on order, and I will give it a try.
I may try to mask the threads and use the spray Krylon Ultra Flat, but with capilary action, I expect to have overspray migrate into the threads.
Any suggestions?
WGR
Hi Gary,
We have two of the Testor's flat black paint marking pens, and they work well. We used them with APASS to blacken filter edges until Astrodon started doing that as part of their practice. Certainly better than magic marker; probably worse than Krylon spray paint. The markers and Krylon are the only things I've found to be effective so far.
Arne
Jeff, et al,
Yes, RSpec will handle fits image files....
The issue of having the profiles saved in 1D fits format with header info similar to the BeSS format, was discussed with Tom a while ago.
I've just contacted him, and he confirms that this feature has been worked on and will probably be included in the next RSpec update.
Hello
Based on the comments from the first step spectra that I posted, I have changed the spacing, and the rotation and taken another cut. I also used a bright star this time Alpha Cyg. Looks like I can see the Ha feature at about 400 pixels
WGR
Gary Walker
Hi Gary,
Some comments. I had to the rotate the spectrum 2.8 degrees, no problem. The jpg is only 2^8 ADUs so I cannot tell too much without seeing the fits file, but it looks like you are way way over saturted. The pixel map shows most of the spectrum pixels at 255 ADUs which is 2^8 or saturated. I'm not sure what happens when you save a 2^16 range file as a jpg 2^8 range. I suspect the program dose some scaling which means the spectrum is still way over exposed. Look at the pixel map for the fits spectrum. The maximm pixel ADU of the first order spectrum (don't worry about the zero order) should be below 60,000 ADUs. As it is no reliable features can be seen in the spectrum.
Keep trying.
BTW you really need to invest in RSpec so you can anaylse what you are doing better.
Jeff Hopkins (187283)
Hopkins Phoenix Observatory
Hello Jeff
The image I posted was the .jpeg and it looks like I need to learn a little more about the 16 bit to 8 bit conversion. The .jpeg was maxed out, but the fits was not. I will try to send the fits directly when I get back. I am on an airplane at 30,000 feet. Yes, Ipaid $10 for access--said I would never do that. I am on the way to AAS meeting.
The line plot attached shows the maximum adu's of the fits.
Gary
Hi Gary,
I looked at your plot. The peak for the summed columns (if that is what iit is) is ony around 2300 ADU. That looks like a summed column of pixel ADUs for an 8 bit image. The 2300 ADUs would be very low for even a single pixel for a 16 bit image. As I said, the maxiumum ADUs for single pixels should be under 60,000, ideally in the 40,000 to 60,000 ADU range. When several of the column pixels are summed the maximum summed ADU count for that pixxel column position can be in the several hundred thousand ADU . Something is not right. Your image looks saturated.
Keep trying and if you can, please send me directly the fits file and I can check it more. Send tp phxjeff@hposoft.com.
Have fun at the AAS meeting. I was there a could years ago.
Jeff Hopkins (17283)
Hopkins Phoenix Observatory
Hello Gary,
Attached is a screen shot of your image of Alpha Cyg loaded into RSpec. The H Beta line looked like it is at pixel 431 resulting in a dispersion of 25.3 Angstroms/Pixel. The screen shot has the "Star Type A" lines from the element library superimposed. The blue reference line on the right is H alpha.
Richard Sabo
With the holidays and bad weather here in Charleston, I have been unable to play with my new SA200. However, there was about a 1 hour window last night so I decided to give it a shot. I was only able to get one image of Vega before the clouds rolled back in, but at least that gave me something to work with other than the practice files. In my haste, I made several bone-headed mistakes. I overexposed Vega and did not place the star far enough to the left side of the image, so a good chunk of the right side of the spectrum was chopped off. Nevertheless, I was able to use RSpec to calibrate the raw spectrum and create an instrument response curve (see attached). This enabled me to generate a crude final spectrum that at least appears to match the H beta line at 4861 angstroms when compared to the spectral type a0v reference series (see attached). I found the RSpec software very intutitive and easy to use, especially after watching the helpful tutorial videos. I only hope that I don't have to wait another month to try again.
Blake
Here is the .jpg instead.
The fits file
Blake,
The fits file downloads as an HTML which needs to be renamed....
This shows the "artifacts" are actually caused by the saturation of the image!
Using the lower intensity 2nd order image and doing a rough profile from AstroArt shows some detail...
Setting the target star over to the left and having the spectral image central and horizontal across the chip, will dramatically improve your results.
welcome to the world
Blake,
welcome to the world of spectroscopy!
I'm sure with a bit more clear skies and practise you will obtain better results....
As an initial trial, it looks like the spectral image may have been sitting diagonally across the CCD chip and when rotated horizontally for processing you've introduced a series of "sawtooth" artifacts....
What dispersion were you achieving? It's difficult to confirm your calibration from the profile uploaded.
Can you upload a copy of the original spectral image?
Onwards and Upwards.
Yes, I definitely had to rotate the image. ***My dispersion was roughly 900-1400 and angstroms/pixel was 19.3. I have attached the original image. I realized my mistake as soon as I saw it, but unfortunately that was my only shot last night. Also, I cropped a good deal of the data points in RSpec because there was so much unused space. I felt like I did it in a way that it did not affect the results, but I could be wrong.
Thanks for your help,
Blake
I think the "feature"
Gary,
I think the "feature" around the pixel =400 may be the atmospheric line at 7600A
The spectrum seems to cover 180 pixel (220-400) giving a rough dispersion of 7600/180 = 42A/pixel.
This is very low and the resolution obtained makes line identification difficult.
You should consider increasing the spacing between the grating and the CCD, probably to double the current distance.
Using Excel (or a spreadsheet graph) will restrict your processing results... moving to one of the recognised spectral processing packages (BASS, ISIS, VSPEC or RSpec) will certainly help.
Onwards and Upwards
Hello Ken
I agree with your id of the pixel 400 id. My setup gives 18.5 A/px according to the on line calculator for 300 mm aperture, f10 optics, 65mm spacing and 24 micron pixels. This was the best compromise after trying larger and smaller spacings. This gives full spectrum and the first order image across the chip.
So 400 px * 18.5 = 7400A, pretty close, uncalibrated.
My setup is optomized for PT, and so have few options other than a second scope and spectra graph/ccd/etc. My FOV is 14 arc mins which is ideal for VS. I am at 1.6 arc sec/px and my seeing is about 3 most nights.
I think this is about the best I can do, unless I am missing something. Suggestions welcome. Rspec or equivalent is a good one. I will try it soon.
WGR
Gary Walker
Hi Gary,
There are much better results to be had from this setup yet. Discounting optical aberrations, your setup will have a resolution of ~40A which should give excellent results with the Star Analyser.
The feature you see is the O2 telluric band not H alpha, which is confirmed by your dispersion calculation.
Alpha Cygni (Deneb) is not the best beginners target as the Balmer lines are very narrow in this supergiant so do not show up as strongly at the Star Analyser resolution as those in a main sequence star. The lines will show much better in a main sequence star like Vega etc
Even so, the fact that you can see little of the Balmer lines at H beta, gamma suggests that either your focus may be optimised too far to the red end of the spectrum or your spectrum is saturated in the bright central region around H beta/gamma.
I suggest Vega, or another main sequence (luminosity class v) A star next time. The H beta, gamma lines should already be obvious in the spectrum image and can be tuned for best focus (under expose to start with rather than risk saturation) but if you are having problems with Vega (which needs a very short exposure to avoid saturation) you could try P Cygni which has bright H and He emission lines and is the target I used to test the SA200 performance here. (you should be able to match these results with your setup)
http://www.threehillsobservatory.co.uk/astro/SA200_resolution_PCyg_anno…
Robin
I wanted to see what I could get by just dropping an SA200 into the filter wheel for my ST-9. I didn't expect to get a dispersion I would like, but I wanted to see what was possible with minimum of alterations to my setup.
The first two targets were Aldebaran and Rigel. (Betelgeuse was behind a tree). On an 8-inch Celestron with a tele-compressor I learned quickly what little it took to saturate the spectrum for those bright stars, so on a whim I shot Zeta Gem.
I'm using a trial version of RSpec, found a reference spectrum that seems to work. Zeta Gem is hardly a standard star, so I'll try again with one that is more constant when it clears in about 3 weeks - given my luck this winter. I've added the profile to this post. I'd appreciate any comments and corrections.
Also, I've been directly coupling the CCD camera to the Celestron as opposed to using a nose-piece for a more secure connection. It would seem to me that a nose piece may be just the spacing needed if I place the grating into the threaded barrel, but I've always had problems with the set screws holding the CCD in place when using a nose piece. Any suggestions on how to better mount a CCD using a nose-piece into the visual back provided by Celestron?
Thanks and Clear Skies!
Neil Simmons
Hi Neil,
The fact that you are picking up features in your spectrum looks encouraging but your wavelength calibration is definitely out (your CCD will not have any response below ~3600A)
Start with a main sequence A star making sure you can see the zero order and some Balmer lines and use this to calculate your dispersion. You can then use this information to calibrate any other star where the lines are not obvious or easy to identify.
Note that the broad shape of your spectrum will not match that of the same type of star from the (Pickles) library until you also do an instrument response correction but the features should line up well
Robin
Hi Neal,
I have secured my ST9 camera a using a nose piece by first placing the nose piece (no camera) all the way into the visual back and marking the spot where the set screw on the visual back comes in contact with the nose piece. After removing the nose piece, I drilled an indentation into that spot on the nose piece so that the set screw would come up tight in the indentation. For added security, I drilled and tapped a second hole in the visual back, marked that position on the nose piece, and again drilled an indentation. This way I had the benefit of 2 set screws holding the camera in place. Never a problem with the camera slipping out or around.
Cheers,
Keith
Traditional test targets like Vega and P Cygni are setting too early for me now but bright main sequence A/B stars like Delta Cas, Castor, Regulus all show clear Balmer lines and can be used for wavelength calibration.
Gamma Cas is also a good first target as it shows H alpha in emission which will appear as a bright spot in the spectrum. You can see examples of what Star Analyser spectra of Delta and Gamma Cas typically look like here
http://www.threehillsobservatory.co.uk/astro/spectra_12.htm
Robin
The attached plot contrasts the H alpha lines in Deneb and Vega measured at high resolution (0.4A) using a LHIRES III spectrograph. The typical Star Analyser resolution (60A) is also shown on the plot.
Robin
I would like to ask people to start new threads, rather than continuing to add to this "first steps" one. It is far easier to find new posts and to respond to the current question rather than one several days old if we segregate things a bit. Thanks!
Arne
Sounds like a good
Keith,
Sounds like a good idea.
Sometime in spectroscopy it's necessary to re-orientate the camera/ grating combo to minimise overlap between the target spectrum and background stars....
By adding a 1" dia. flat prism atop the grating, the defracted image can be brought back more in line with the plane of the sensor. You can learn more at "Grism Conversion".
As the designer of the Star Analyser I would like to comment. The grating surface is sensitive to moisture and is assembled as a sealed unit with its cover glass. This modification will break this seal leading potentially to moisture ingress. (There are other reasons why this modification to the Star Analyser is problematic and anyone considering this route is welcome to contact me direct)
Paton Hawksley offer a separately mounted wedge prism for use by advanced users with either the SA100 or 200 and can also supply bespoke solutions.
Robin