spectro-aras.com

You may have seen my lengthy exchanges on the ISIS subforum about calibration when binning x2. Over the last month I have managed a workaround that provided reasonable calibration results. However I have found that my calibration toward the UV is average at best and generally poor. So I did some experimenting this morning.

A few months back, I realised that ArNe calibration images yielded better results at lower temperatures. So I took some images at an ambient temperature of about 6.5 degrees C and have been using that to calibrate all of my images irrespective of the temp at the time of imaging. I also trimmed my .lst file to the essentials to achieve a reasonable fit. My file reads:

3
3.63
4158.59
4200.67
5849.07
5944.83
6143.06
6266.49
6334.43
6506.53
6965.43
7067.22
7272.93

I achieve an RMS of between 0.12 and 0.25 with this set up. I did this partly because the internet switch I was using to operate the calibration lamp was not working.

This week I managed to get the internet switch working - I have been taking Ar Ne images immediately before taking my reference star and target images. As the LISA is sensitive to temperature, I reasoned that if the focus was out due to the temperature, the effect should be the same for the calibration and reference/target images.

However, comparing spectrum calibrated with Ar Ne images taken at ambient to those using the stock 6.5C frames, I see a big difference in results. At around 3900A, the difference is around 25A. I've attached a screen shot comparing the 2 in ISIS. All reduction and processing steps were identical except for the calibration frames used. Interestingly, the RMS reported for both were broadly consistent - about 0.2. Comparing the images to a synthetic Pickles spectrum, its clear that the spectrum using the 6.5C frame is much better.

I suspect my focus is slightly out. I can adjust that (with the assistance of the on-site observatory technician). And I think I should do that when the temp is cold - to be closer to the ambient temperature when imaging. That said, it seems odd to me to refine my focus to match temp that in fact may be lower than the ambient when I am imaging.

My question is whether it is as simple as this or am I missing something?
Is part of the issue the fact that I have so few lines to calibrate with above 5800A?
If it is, would I be better served by changing back to the old Ne bulb and then using the two step calibration in ISIS using an A0 star?

I accept that the LISA is challenged at the UV end of its range - and the UVEX should be a much better unit at shorter wavelengths. I'm interested in hearing about the experience of other LISA users.

Also, has anyone successfully rigged up a motor to adjust the LISA focus remotely? I see that Shelyak are looking to incorporate this in the commercial UVEX unit. I suspect that there are space constraints.

Pete

Hi Pete,

I suspect your calibration problem is because you are using very few calibration lines at the blue end. This does not constraint the choice of lines which ISIS uses well enough so it is probably picking the wrong lines in the lamp spectrum to match these wavelengths in your file. This throws the calibration at the blue end off. The red end is calibrating well as it has lots of lines to match.

I don't recommend using the old calibration method. You will be better staying with the ArNe lamp and trying to use more lines at the blue end.

The LISA's capability at the blue end is limited both by its design and by the lack of calibration lines below 4158A. Extrapolating the calibration polynomial fit much beyond that becomes unreliable. I normally crop my spectra at 3900A for that reason.

Motorising the LISA focus would be difficult with the current design. I don't know of anyone who has done it. You would be better going to a spectrograph designed to be operated remotely such as the new UVEX.

David

Thanks David

I think you are spot on.

Interestingly, I get better results with Demetra. But it’s still not perfect.

I fear I will be lining up to purchase a UVEX when they are released by Shelyak

Pete

Hello,

If it works better with an other soft and that it's not perfect this tells you that the problem is in the setup of the hardware or the setup used for data acquisition, isn't it?

Bejni

Benjamin Mauclaire wrote:Hello,

If it works better with an other soft and that it's not perfect this tells you that the problem is in the setup of the hardware or the setup used for data acquisition, isn't it?

Bejni

It is

More work to be done - sadly.

Merci

Pete

Hi Pete,

As doing spectroscopy is like doing a physics experience, calibration frames have to be done in the same situations of target frames.
Focus of the Lisa spectro won't be possible on all the wavelength bandwith, so choose 4500 or 5000 A zone for best focus as blue part's signal is weaker than red part. Calibration frame has to be done as close as possible in time to the target frame and has to be done for all targets. As in low resolution lines are narrow, *** it's mandatory to work in binning 1x1 in order to have sufficient sampling ***. This fixes imaging setup.

Software setup for wavelength calibration depends mostly from lines used in low resolution.
Relco calibration lamp offers usable lines even in blue part of the spectrum. In your post it's not clear if you used Ne lamp or Relco Ar-Ne lamp. You will get better resutls with Relco lamp.
As calibration line choice can be tricky, Spcaudace software use a set of lines given by many tests and the whole pipeline is full automated after filled a simple form as shown in the tutorial video:
http://spcaudace.free.fr/docs/index.php ... cop2a.html

Benji

So this may be a very dumb question.

When focussing the LISA, on what lines do I focus?

I had thought to focus around 6145A - however anything shorter than 5852 is indistinct.

This afternoon I refocused with particular attention on the shorter end of the spectrum - but the red is now poorly defined. Trying to calibrate with this yields garbage.

Where is the sweet spot? I'm surprised I am finding this so challenging.

BTW I am using the Relco ArNe lamp that Shelyak supplies

Pete

Peter Velez wrote:When focussing the LISA, on what lines do I focus?

I take a Ar/Ne line close to the middle of the CCD field like 5852,49 Å
This allows to have a good compromise between the red and the blue of the spectrum.

Thanks Olivier

that is what I had started with - clearly I have made it worse by trying to improve it.

I will have another go tomorrow trying to sharpen it up

Pete

Hello.

I am reopening an old topic.

I wonder if this issue has been fixed. The topic entries stopped two years ago without a clear answer to Peter's troubles.
I have the same issue as Peter velez. My set up is pretty similar:LISA spectroscope, Atik 460, Meade telescope working near f/5.

I am working generally at 2x2 and with the 35μ slit, as my seeing normally is between 3,5 and 4 arc sec. So far i have been using Demetra for spectra proccess. Its learnng curve is not terrible and the software very user friendly. My problem, as Peter's, is calibration. With Demetra i get RMS around 0.24 and the lines drawn by the software are spot on.
With ISIS i got RMS ≈ 20, sometimes more, somtimes less, never under 15. No matter what set of lines i try calibration with. I even tried te same line set Demetra is ussing, to no avail.
I followed Peter's post and people advices and tried many times...

I tried also with some 1x1 lamp images to check. With those, i can obtain RMS ≈ 5 with ISIS. Much better but still far from Demetra.

I want to switch to ISIS as it is the standard software by most people, and is loaded with powerful features. So after trying a lot i managed to work with it, except calibration.
My work around is to first process with Demetra and then save Demetra's obtained polynomial coefficients into a lst file. I then load the polynomial in ISIS profile tab and use the predefined dispersion ecuation for calibration. It works!
But i would much preffer this to be done by ISIS directly.