Forum der Vereinigung der Sternfreunde

To buy, or not to buy..

I have an expensive question in front of me, which is to decide if I really need to get a 1200 l/mm grating to contribute with meaningful data for the 134/5/7 campaign.
Robin has made very seductive and motivating progress with C11, Lhires + 1200 grating and a nice sensor with 9mm of width.
I do not have the 1200 grating yet, but I do have a nice sensor with 15mm width and a 2400 grating! (I also have a C11)

I know that in astro-photography we can keep the aperture, double the focal length, and use pixels twice as wide to get roughly the same information.. Does this apply to spectroscopy as well when increasing the number of lines in the grating?

I know the answer to the question can simply be "try it!", but the weather here just got worse, and I do not expect improvement before the next weekend. I have been playing with Simspec, instead!
I am almost getting to the conclusion that if I bin 3x3 in a KAF-3200 with 2400 l/mm grating, I get slightly worse SNR than binning 2x2 in a ICX-285 with 1200 k/mm. However, the dispersion is a bit greater (0.5 vs 0.3 A/pix), and spectral range a bit shorter (254 vs 356 A around 5400 A) (though I think I can still fit the interesting spectrum in the sensor).
Then there is the concern for calibration, and this I did not check if 254A is enough...
Another detail is I assumed the grating reflectivity was the same between 2400 (holographic) and the 1200 (ruled).. (?)

I know the Lhires was designed for sensors of up to 10mm, but since I would be binning 3x3 I do not expect issues in terms of resolution. Unless these may reveal themselves in calibration! It is not good the frame the interesting spectrum very tightly, I think.

I should probably try it when the weather allows, before buying a new grating, no? If I buy a new grating, I am tempted to get a 50mm square 1800 l/mm reflective ruled grating blazed for 500nm... Could this sound interesting? Ruled vs holographic should be a good advantage on reflectivity (?)
But since I know no one with a 1800 grating on a LhiresIII, I am afraid I might face unexpected problems, like observing overlapping orders that I do not want...

Cheers,

_________________
Fil.

Hi Fil,

I have used an 1800 l/mm grating with my Lhires, only once so far for a specific project, and while my experience with it isn't extensive I didn't have any problems.

Why are you proposing to buy a 50x50mm grating, as the 25x50 should be fine?

Working at Ha, or below, will result in 2nd order overlap but as it will be spectra up to around 3500A it will be weak due to the low CCD and grating efficiencies. I think the use of a blue cut-off filter would still be advisable.

If you have one of the original Lhires grating drawer designs then you only need to buy the grating from Optometrics which, at US$109 for the standard ruled 25x50mm, 3-7185, is a good saving. I also suggest the cheaper standard ruled grating which (at 6000A) is 55% efficient versus 50% for the holographic version. See details at http://www.optometrics.com/products_gratings_r.html

The problem with the larger sensors is that sharp focus, and hence resolution, is lost at the shorter and longer wavelengths ... but still worth having

Cheers, Bernard

_________________
Quicquid Nitet Notandum

Hello Fil,

the 2400 l/mm grating is a halographic one and has a poor efficiency compared with the ruled and blazed gratings 1800 oder 1200 l/mm.

I used therefore - erverytime if the strong resolution obtained with the 2400 is not required - the 1800 or the 1200 g/mm grating. They have also the advantage of more calibration lines on the CCD chip and therefore more exactness of the calibration function (3thrd degree). Therefore is also a bigger CCD chip better then a small one. Even if on the spectrum edge the resolution decreases.

_________________
Herzliche Grüße / best regards

Lothar

https://lotharschanne.wordpress.com/

Bernard,

The 50mmx50mm was the one I found "quickly", in Thorlabs.com.. They do not have other sizes for gratings with 50mm on one side. I want that to tilt it to my liking... The optometrics 25x50 is better, thank you!

Lothar, thank you as well !

_________________
Fil.

Hi Fil,

Optometrics are cheaper than Thorlabs, the supplier of Thorlab gratings and also where Shelyak get theirs from.

Bernard

_________________
Quicquid Nitet Notandum

Ok, next question:

I can observe with higher resolution spectral range capturing He II + C III with a 1800 L/mm grating
However, with a 1200 L/mm, I might be able to capture a He II + C III and also He I in the same frame (5350~5950). I figure it could be interesting to observe all three points of interest simultaneously, no? The S/N should degrade roughly in a proportion of 180 to 150.. and resolution would change from 1.7A to 1.1A.. Dispersion should be near 0,5 A/pixel in either case (binning 2x2 vs 3x3).
(I am not sure that I have a correct value for the resolution)

_________________
Fil.

Hi Fil:

Experience shows that the He I 5876 emission behaves a lot like CIII 5696, so if the 3-pixel resolution degrades from 1.1 to 1.7A going from 2 to 3 spectral lines, then I would say stick with the shorter range and higher resolution. The HeI emission is also often crowded out by the neighbouring CIV complex at 5800A, while CIII 5696 is quite isolated, making it a cleaner test. On the other hand, you also get HeI absoprtion from the O companion (if there is one, as in WR137, but not WR134 or 135, although the orbital period for WR137 is 13 years, so you would only get a small snapshot of that even during 4 months). Of course, one has to fold the exposure time into the argument here: if this becomes prohibitively long to get sufficient S/N, then I might have to rethink the above argument.

I?m not sure why you say that the dispersion would be the same with both gratings, however.

Best, Tony

From: Filipe Dias (fg-spek-convento@vds-astro.de)
Sent: Friday, September 28, 2012 6:39 PM
To: fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)
Subject: [fg-spek] Re: Grating choice for campaign




Ok, next question:

I can observe with higher resolution spectral range capturing He II + C III with a 1800 L/mm grating
However, with a 1200 L/mm, I might be able to capture a He II + C III and also He I in the same frame (5350~5950). I figure it could be interesting to observe all three points of interest simultaneously, no? The S/N should degrade roughly in a proportion of 180 to 150.. and resolution would change from 1.7A to 1.1A.. Dispersion should be near 0,5 A/pixel in either case (binning 2x2 vs 3x3).
(I am not sure that I have a correct value for the resolution)



Fil.

_________________
Tony Moffat

Hi Tony,
Sorry if I did not make it clear. The two cases are:

1200 l/mm grating
binning 2x2
dispersion; 0,54 A/pix (1092 pixels)
range: 5350~5700
S/N estimated from simSpec: 215
simSpec presents resolution to be 1.7A

1800 l/mm grating
binning 3x3
dispersion: 0,50 A/pix (728 pixels)
range: 5350~5900
S/N estimated from simSpec: 190
simSpec presents resolution to be 1.1A

_________________
Fil.

Hi Fil,

It sounds like simspec is a bit pessimistic. I get 1.3A resolution using a 1200 grating and a 35um slit. I could perhaps even close the slit a bit and get a bit more resolution but I would struggle then to get good SNR in 1 hr. Maximum resolution with your bin size would be ~1.1A with a 29 um slit. This would give you both good resolution and coverage

Cheers
Robin

Hi Robin, Fil,

I?ve always achieved a significantly better resolution (based on calibration line FWHM) and a lower SNR than the spreadsheets predict. I can understand the lower SNR as the model assume perfect guiding, but the reason for the better resolution is not clear.

Cheers,
Bernard


From: Robin Leadbeater (fg-spek-convento@vds-astro.de)
Sent: Saturday, September 29, 2012 12:00 PM
To: fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)
Subject: [fg-spek] Re: Grating choice for campaign




Hi Fil,

It sounds like simspec is a bit pessimistic. I get 1.3A resolution using a 1200 grating and a 35um slit. I could perhaps even close the slit a bit and get a bit more resolution but I would struggle then to get good SNR in 1 hr. Maximum resolution with your bin size would be ~1.1A with a 29 um slit. This would give you both good resolution and coverage

Cheers
Robin





No virus found in this message.
Checked by AVG - www.avg.com
Version: 2012.0.2221 / Virus Database: 2441/5297 - Release Date: 09/28/12

_________________
Quicquid Nitet Notandum

OK, Fil, thanks. In this case, taken at face value the 1800 line grating does look best all round. However, I still have some questions:

1. Is this dispersion of close to 0.5 A/pix after binning? If so, then given the similar values of dispersion for both gratings, I don?t understand why simSpec gives such a large difference in resolution (normally about 3 pixels) between the two gratings (1.7 vs. 1.1A). 1.1 is acceptable, 1.7 not.

2. The best way to obtain the resolution is to focus the spectrograph and actually measure the FWHM of well-exposed (but not over-exposed) unblended comparison lines.

Cheers, Tony


From: Filipe Dias (fg-spek-convento@vds-astro.de)
Sent: Friday, September 28, 2012 9:08 PM
To: fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)
Subject: [fg-spek] Re: Grating choice for campaign




Hi Tony,
Sorry if I did not make it clear. The two cases are:

1200 l/mm grating
binning 2x2
dispersion; 0,54 A/pix (1092 pixels)
range: 5350~5700
S/N estimated from simSpec: 215
simSpec presents resolution to be 1.7A

1800 l/mm grating
binning 3x3
dispersion: 0,50 A/pix (728 pixels)
range: 5350~5900
S/N estimated from simSpec: 190
simSpec presents resolution to be 1.1A



Fil.

_________________
Tony Moffat

The formula that simspec uses apparently devides the resolving power R by the wavelength.. And R is calculated using angle of incidence from the objective and diffraction angle, and based on the image of the slit on the CCD focal plane.. I also prefer calculating an effective R from observing FWHM.. The formula behaves slightly differently if the sampling factor is below 2.. Nyquist stuff. We bump into the problem of how to define "resolution"..

The dispersion mentioned is after binning yes. The 1800 increases the dispersion compared to 1200.. then the binning 3x3 decreasis it compared to binning 2x2, and that is why the final dispersion is similar.I made another mistake telling the range: 5350~5700 A is for the 1800 grating. 5350~5900 A is for the 1200, as it should be..


Measuring FWHM is the way to do it, yes. I do not have either grating so I cannot compare. However, Robin measured it at 1.3 (below any calculation by simSpec), with pixels similar smaller in size compared to mine (6.45 vs 6.8 micron). Berthold seems to get smaller values for resolution compared to simSpec, as well..


My concearn was that maybe observing the "same" thing on two different lines would help in compensating for low S/N or low resolution..  It would be like binning at the time of writting the paper
If there is no need for this, I will get the 1800 grating.



On Sat, Sep 29, 2012 at 2:38 PM, Tony Moffat <fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)> wrote:

_________________
Fil.

Hi Fil,

Aren't you undersampled at 3x binning and a 35um slit? If you are, you could increase the slit width until you are optimally binned without reducing the resolution and perhaps get a bit more light to impove the SNR.

Like Bernard, my resolution of 1.3A was measured on the internal neon lamp lines. The star spectra are sometimes a bit better as the internal neon beam is faster than the telescope so the aberrations are a bit higher but can also be a bit worse due to smearing caused by spectrograph flexure during the exposure. I rechecked my recent results using the interstellar Na D lines in the star spectrum which give 1.2A

With my LHIRES, I just estimate the resolution based on what the slit width corresponds in Angstrom. eg in this case I have a 35um slit and the dispersion is 0.5A/13um binned pixel so the expected resolution is 0.5*35/13 = 1.34A. This always seems to give a pretty good estimate of what I achieve in practice provided I am not undersampled.

Note that Simspec will not give a reliable estimate of SNR on WR stars because it estimates the flux from Vmag values assuming a given spectral type (which needs to be entered) and a black body continuum. This not valid for WR stars with a significant proportion of the flux in the emisison lines.

Cheers
Robin

Hi all.

As I won't be able to observe the WR134/135/137 targets until early 2013 I've used WR111 (WC5), of similar magnitude (8.25), as an observable test star for my C11+Lhires+Atik314L using an 1800 l/mm grating, 35 micron slit, with 2x2 binning. Unfortunately the weather isn't very cooperative so I've only been able to grab one 600s spectra, not instrument corrected, which is attached. While SNR<100 the resolution is around 0.83A and I'm confident of achieving SNR>=200 with multiple exposures. Given that the star was down to 30 deg altitude I'm happy with this initial test with the 1800 grating.

If Tony and Thomas are reading this, is there any interest in me doing some followup work on WR111, maybe at other wavelengths, a star that I know Tony has published on previously?

Cheers,
Bernard

_________________
Quicquid Nitet Notandum

Hi Bernard:

Nice spectrum of WR111 ? being WC5, a quite broad-line star! CIII 5696 is nicely covered, but CIV 5802/12 is cut off on the red side. Any way to increase the coverage a bit, also to include HeI 5876 + some continuum beyond that and below CIII 5696? Do you have a 1200 l/mm grating? It would probably give about 1.2A (3-pixel?) resolution, which would still be OK, just. But this would give about 300 A coverage which would be a bit better, instead of 200 A.

As for doing WR111, I?d hate to have you waste your time. Given a choice, I would prefer that you continue intensively on zeta Pup, even if Chandra won?t be observing it in the next round. And better to join the campaign on WR134, 135, 137 next June-Sept, even if they are quite far north for you. On the other hand, if you?re looking for targets, WR111 would be worthwhile, given its favourable position for you, but only if you can monitor it frequently enough (several times a night, well spread out in time, and on many nights). It shows lots of moving clumps, but requires good S/N. Just keep the same wavelength region, enlarged to 300A if possible.

Best, Tony

From: Bernard Heathcote (fg-spek-convento@vds-astro.de)
Sent: Thursday, October 04, 2012 10:40 AM
To: fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)
Subject: [fg-spek] Re: Grating choice for campaign




Hi all.

As I won't be able to observe the WR134/135/137 targets until early 2013 I've used WR111 (WC5), of similar magnitude (8.25), as an observable test star for my C11+Lhires+Atik314L using an 1800 l/mm grating, 35 micron slit, with 2x2 binning. Unfortunately the weather isn't very cooperative so I've only been able to grab one 600s spectra, not instrument corrected, which is attached. While SNR<100 the resolution is around 0.83A and I'm confident of achieving SNR>=200 with multiple exposures. Given that the star was down to 30 deg altitude I'm happy with this initial test with the 1800 grating.

If Tony and Thomas are reading this, is there any interest in me doing some followup work on WR111, maybe at other wavelengths, a star that I know Tony has published on previously?

Cheers,
Bernard



Quicquid Nitet Notandum

_________________
Tony Moffat

Hi Tony,

If the lower resolution of a 1200 grating is acceptable then I can certainly use that. My main reason for the 1800 grating test was to show Filipe, who was pondering its use earlier, what could be expected.

I certainly plan on resuming my zeta Pup work as soon as possible, which will only be in a couple of months time from my urban location. Meanwhile I will see if I can monitor WR111 along the lines you suggest ... good experience for WR134/135/137 next year.

Cheers, Bernard

_________________
Quicquid Nitet Notandum

Hi Tony,

I have switched to a 1200 grating, giving a 1.24A resolution and 345A window and have used it to monitor the CIV 5805 line on WR111, a test target. The attached series of three 800s exposures shows some, presumably wind, variability of CIV, a line so strong that I have to limit exposures to 400s to avoid saturation! Is there any value in trying to detect such short variations? If not, what is the shortest cadence we should aim for, or is it as long as it takes to have SNR=200?

Cheers, Bernard

_________________
Quicquid Nitet Notandum

Hi Bernard:

Yes, CIV is VERY strong in this star. But it is a doublet (5802/5812) and does not vary (% wise) nearly as much as its neighbour CIII 5696, which is THE line to observe, even if it?s a lot weaker! Try it and you?ll see relatively much larger variations over timescales of several hours (full timescale is about 10h). Normally, observations shorter than about an hour do not show much variability, although that does depend on the details (e.g. which star).

Best, Tony

From: Bernard Heathcote (fg-spek-convento@vds-astro.de)
Sent: Friday, October 12, 2012 11:06 PM
To: fg-spek-convento@vds-astro.de (fg-spek-convento@vds-astro.de)
Subject: [fg-spek] Re: Grating choice for campaign




Hi Tony,

I have switched to a 1200 grating, giving a 1.24A resolution and 345A window and have used it to monitor the CIV 5805 line on WR111, a test target. The attached series of three 800s exposures shows some, presumably wind, variability of CIV, a line so strong that I have to limit exposures to 400s to avoid saturation! Is there any value in trying to detect such short variations? If not, what is the shortest cadence we should aim for, or is it as long as it takes to have SNR=200?

Cheers, Bernard



Quicquid Nitet Notandum

_________________
Tony Moffat