Forum Replies Created
-
Posts
-
Hi Tony,
A couple of ideas:-
It looks like there is a huge hole in the spectrum from ~4500-6600 where the response should be highest. Could there be saturation in these regions ?
Alternatively could you have set the binning zone limits very narrow or offset from the middle of the spectrum so it is missing part of the spectrum in the middle wavelengths ?
Cheers
Robin
Hi Kate,
There is still something wrong I am afraid. The RMS fit should be better than 1 Angstrom. For example I have just run a Balmer line calibration on one of my recent reference stars and get the following results (ALPY 600 2xbin 4.54um pixels, pixel size set to 8.97A (3123/348) )
————————————————————————–
Wavelength fit deviation
point #1 x = 143.059 lambda = 3835.148 dlambda = 0.242
point #2 x = 154.045 lambda = 3888.984 dlambda = 0.066
point #3 x = 170.621 lambda = 3970.476 dlambda = -0.396
point #4 x = 197.253 lambda = 4101.992 dlambda = -0.242
point #5 x = 245.161 lambda = 4340.028 dlambda = 0.452
point #6 x = 349.413 lambda = 4861.473 dlambda = -0.133
point #7 x = 696.362 lambda = 6562.787 dlambda = 0.023
point #8 x = 762.564 lambda = 6872.013 dlambda = -0.013
————————————————————————–
Coefficient a4 : 3.601046E-10
Coefficient a3 : -1.409022E-06
Coefficient a2 : 1.183699E-03
Coefficient a1 : 4.63613
Coefficient a0 : 3146.763
————————————————————————–
RMS : 0.408720
————————————————————————–
If you email me your spectrum image fits file, I can see if i can get a better result if you like
EDIT: typo in pixel size corrected (4.54 not 5.45)
Cheers
Robin
Hi Kate, The pixels size is usually not too critical whe using the H alpha lines as there are no other lines nearby but it can be super critical with the calibration module where it needs to distinguish between some very closely spaced lines
I find Christian Buil’s tip here
http://www.astrosurf.com/buil/isis/guide_alpy/resume_calibration.htm
works well for calculating the pixel value to use:
“A tip for calculate the scaling factor p (or virtual pixel size). Use the formula:
p = 3123 / dx
where dx is the distance between the H alpha and H beta lines measured in pixels along the horizontal axis in a raw image. Try to find the distance to the nearest pixel (the reading of the mouse pointer is sufficient). In the example, the H alpha line is at x = 882, while the H beta line is at x = 535. So dx = 882-534 = 347, and thus the pixel size to adopt is p = 3123/347 = 9.00 pixels.”
Another trick is to keep an eye on the RMS value that ISIS generates in the running commentary on the “go” page when it is running the calibration fit. If you keep on repeatedly running varying the pixel size slightly, it should be obvious when the program locks onto the right lines as the RMS will suddenly tumble to a very low value.
Finally double check you are picking the H alpha line and not the nearby Telluric band as the reference point. (It has been known !)
Cheers
Robin
Is this for spectroscopy? If so you can’t use twilight or LEDs. You need a light source with a smooth spectrum with no features.
EDIT: you might be able to get away with LEDs in some parts of the wavelength range in higher resolution spectroscopy but I dont know of any LEDs which have a smooth enough spectrum or wide enough wavelength coverage for the ALPY for example
Cheers
Robin
Hi Nick
The defect (or cosmetic as it is called in ISIS) map is to catch the hot and warm pixels which will have unreliable values after the dark subtraction (ie the pixels will either already be saturated or will become prematurely saturated during the exposure, so the dark subtraction will not work correctly.) The same procedure is sometimes used in imaging where it prevents dark holes appearing in the image after dark subtraction where fully saturated hot pixels are for example. The cosmetic file is generated from the master dark and a threshold is set above which the pixels are considered defective. I have mine set to 500 currently which is well above the noise and generates around 100 pixels in a 10 min exposure with my ATK314
Cheers
Robin
Hi Steve,
At least the ALPY is stable enough not to have to go to the extreme of taking flats with the scope still aimed at the target like you have to do with the LHIRES if you have a camera with small scale ripples.
http://www.astrosurf.com/buil/fringes/index.html
Before I had the calibration module, I injected the light from the RELCO calibration lamp via a reversed flip mirror. Perhaps a diffused incandescent light source could also be used this way to produce a flat.
Cheers
Robin
Hi Kate,
I found my Gert Neumann panel and put it in front of the ALPY 600 spectrograph with an ATIK 428 camera. Here is the comparison between the EL panel and the internal ALPY lamp.
(This is effectively the output from the light source x the response of the grating and CCD camera)
As claimed by the supplier, the spectrum of the EL panel is smooth with no obvious lines (The small scale ripples in both spectra are from the camera QE and demonstrates one of the reasons why flat correction is important).
The EL panel could potentially be used as a flat light source for spectroscopy, even though the spectrum shape is very different from the Incandescent Halogen lamp black body curve but unfortunately there is no light from it below 4000A. (The internal Halogen lamp in the ALPY also struggles in this region but in practise it still gives enough output down to ~3600A provided a large number of flat exposures are combined to reduce the noise)
A key difference between imaging and spectroscopic flats is that it is much less important that the field is evenly illuminated since you are only sampling a small slice of the field. Even some variation in illumination along the slit is not important if you are measuring at one position eg as for a star. This means even a crude flat illumination setup can be used. I bought my ALPY before the calibration module came out and used the same flat lamp setup as I used for my LHIRES – A halogen lamp waved a couple of metres in front of the telescope aperture which was covered by a diffuser (actually a pillow case!), similar to Christian Buil’s setup here
http://www.astrosurf.com/buil/isis/guide_alpy/tuto_en.htm
This worked well and I saw no difference when I eventually got the calibration module
Cheers
Robin
EDITED: to mention the grating response
Hi Kate,
Yes I know Lothar Schanne who made those tests. I think the Neumann panels work well for high resolution spectroscopy within a limited wavelength range, particularly H alpha with the LHIRES III spectrograph which was what Lothar was doing there. For that particular application they worked better than the usual halogen lamp, matching the defects caused by dust for example. Note that some panels from different manufacturers have some sharp emission lines which make them no good for spectroscopic flats.
I don’t think they work as low resolution wide wavelength range flats (for the ALPY or LISA etc) though as they have a limited wavelength range with no output at the UV and IR ends compared with a halogen lamp. (Francois Cochard commented about this at the recent workshop)
I have a small one knocking about somewhere. I will see if I can dig it out and test it with the ALPY
Cheers
Robin
OK I found a description of using the response assistant here in the LISA tutorial (about half way down)
http://www.astrosurf.com/buil/isis/guide_lisa/tuto_en.htm
it appears it reruns the reduction of whatever spectrum data is currently set up in ISIS, but without a flat correction.(This is all hidden in the background) It then combines this with the flat and the reference spectrum (which has to be of the same object which ISIS has reduced in the background) to produce the instrument response. This is effectively the same as the usual “long hand” way of doing an instrument response but with some automation. It does seem a slightly confusing way of going about though. All very odd !
Cheers
Robin
Hi Paul,
I am at a loss to know what it is doing. It could perhaps be using the flat lamp as a reference assuming a nominal black body spectrum for the flat lamp including allowance for typical atmospheric extinction, but then why would it need the reference star library spectrum at all, and what is it comparing the library spectrum against ? It does seem to work though. I shall have a play, but I am not sure I would ever fully trust it !
Hi Paul,
That looks interesting
I have not used this before but cannot find a description of exactly what it does on Christian’s website (It is not mentioned in his ISIS change log for example where I normally look). I had assumed it just automated the usual “differential spectroscopy” procedure, using the “response” and “continuum” functions under the profile tab to calculate the response using a reference star at a particular air mass which is then used with the target star spectrum at similar air mass. Does it do something different to this?
EDIT:- rereading your paper, am I correct in saying that the response assistant calculates the response using only the flat and the reference spectrum? ie you do not need to measure the reference star?
Cheers
Robin
Looking good, Tony
I can see a couple of interesting places where the two are slightly different at ~5750 and ~6300. I wonder if the differences are real? There is a telluric O2 band at 6277A but nothing as far as I know at 5750. Christian Buil has an annotated Vega spectrum here which shows the Tellurics.
http://www.astrosurf.com/buil/us/vatlas/vatlas.htm.
If you divide your measured by the filtered library it will show up very clearly any discrepancies. It is particularly worth doing with the reference star itself which should of course match perfectly. You can use any differences to see where you might need retune your response calculation.
Cheers
Robin
Impressive stuff but my money for the “big one” (detecting life tracers in earth-like exoplanets in the goldilocks zone) is still on transit measurements rather than direct imaging eg
https://www.spacetelescope.org/news/heic1603/
Robin
There was a piece on this on Radio 4 Front Row program on Monday.
http://www.bbc.co.uk/programmes/b081l7zt#playt=0h10m55s
Some fuzzy reassurances made concerning environmental impact. eg “we don’t want to dazzle the fishes”
Robin
I measured these stars (and added a few more to better cover the spectral classes) on Saturday.
See the attached report.
I have also attached the reduced spectra as fits files if anyone wants to play with them. (Note the extension has been changed from fit to fits to be acceptable for upload to the forum. I also tried zipping them together but the forum does not allow zip files unfortunately)
The results are generally very encouraging with typically less than 2% error in relative flux from 4000-7300A. There are some larger systematic errors below 4000A which seem to be due to a small flux offset and will need further investigation. Extra precautions also need to be taken against scintillation when measuring first magnitude targets.
Cheers
Robin
Hi Steve,
You are effectively taking a spectrum of the flat lamp when you take a spectroscopic flat so you need a light source with a smooth spectrum with good output and no sharp features in the region you are interested in, which for low resolution spectra effectively means incandescent lamps only, running as hot as possible to give maximum uv, hence halogen are favoured. LEDs definitely do not meet these criteria. (At higher resolution some EL panels are ok over at some wavelengths but not from ~3600-7500A that we need) As far as I know Xenon torch bulbs (as opposed to Xenon car headlights) are incandescent filament lamps so should work ok so I dont understand why the flat using the torch looks essentially identical to your LED flat. I know it sounds silly but could you possibly have mixed up the files somehow? eg could ISIS be recalculating using the same raw files. I suggest checking back to the original raw flat images. Other than that, I have no idea what is going on !
Robin
Hi Steve,
Looking again they both look rather like typical blue white LED spectra to me eg
https://upload.wikimedia.org/wikipedia/commons/thumb/4/48/White_LED.png/350px-White_LED.png
Are you sure your torch is a halogen incandescent ?
Cheers
Robin
Hi Steve,
Those do indeed look strange and very similar considering they were taken using very different light sources which should have produced very different looking flats. Are these images raw or are they the master flat after processing in ISIS? If so do the raws show the same effect? Here is a flat from my setup last night (ATK428) using the halogen lamp in the calibration module. (reduced in size)
As far as I can recall flats taken with a halogen lamp plus white sheet over front of the scope as a diffuser before I had the calibration module produced a similar flat.
Robin
