Calibration - LISA and ISIS v 5.1.0

[pvelez]

I've returned to spectroscopy after a 6 month break and am having some difficulty with calibration. I had been using ISIS v4 but have become lost amid v 5.1.

To get back into the swing of processing, I took the usual calibration frames - dark, bias, flat and neon images. I then took a spectrum of the sun sitting the LISA in a bright room. I then loaded the images and tried to calibrate using the LISA internal neon option in the General tab.

I then looked carefully at the Neon frame and identified the 5944 line and entered that as the X location in the Calibration tab.

Pressing Go, I have a reasonable looking profile but with a RMS of 26.07. I'm surprised as I thought I'd achieve a better result than this.

Then seeking to use the Predefined dispersion equation and the Calibration Assistant in the Calibration tab, I receive the following error code:

The calibration error is too high, verify parameters and data

Having said that, the RMS error has been reduced to 23.65. But looking at the log, it states that The horizontal computed position of the Ne 5944.83 A line, X= 648.
However, looking at the Neon calibration fits file, there is no line at that position. Rather, the 5944.83 A line is at 593. I have input that value in the requisite place on the Calibration tab. When I run the calibration assistant, it adjusts the X position to 648.

I am sure I am selecting the right position for the neon line. There must be an explanation for this but I am struggling to work it out.

Should I be expecting a lower RMS than this? Reading the tutorial, I believe that the result should be much lower. I downloaded the sample images with the tutorial and was able to process them easily as directed by the tutorial.

Last night, I imaged Fomalhaut - using the same technique, I achieved a similar result as I had with the sun.

I am using an ST-10XME on the LISA binned x2. Pixel size is 13.8 microns. I am sure I have missed a critical step - which is hard to address by remote. I can post some images if that helps.

Pete

[Christian Buil]

Pete,

First, the base pixel size of ST-10 ME is 6.8 microns, and for binning 2 x 6.8 = 13.6, not 13.8 microns. Correct if necessry.

Second, if you select the good calibation option (LISA internal neon) the spectral calibration assistant button disapear in "Calibration" tab
(this assistant is now not necessary). Look screen copy for a typical run (note, I use an Astik460EX camera with 9.08 pixels size) - in the order
calib1, calib2 and calib3.

A final point: my neon calibration image is well exposed for the purpose. Here a 30 seconds exposure for extract the faint blue/UV line of
gaz trace in the neon lamp.

Check the differences and give the resultat.

Christian B

[pvelez]

Christian

thanks for your prompt reply.

I'll check all this later this evening when I am home from work.

I believe I mistyped in my post and have the pixel size set at 13.6 rather than 13.8 microns. I'll check this.

I'm pleased to say that I have taken the steps you have outlined - at least I have understood the process. I will post some screen shots if that is helpful.

Is the R field in General completed by the program following calibration? Should I clear this field when first extracting a profile?

I have not been including the CCD details in the Output field in the General tab. I had assumed that it was not necessary. Is that correct?

My neon images were taken as 20 second exposures binned x2 which I believe is sufficient to achieve a good SNR. I experimented with a longer exposure with no different result. I can try that again this evening.

[Christian Buil]

Yes the R field is completed automatically after the processing operation.
For the first guess, enter for example R = 0 (not important).

ISIS is very tolerant. For example CCD, observatory, ... infos are not
mandatory for first test. Also for response file, flat-field, dark and so on. But for
a well documented result, and for an ultimate processing, fine
calibration images set is welcome of course.

Christian

[pvelez]

Thanks Christian

I'll let you know how I go

Pete

[pvelez]

Thanks again for your help Christian

As you might expect - there was a combination of errors. For those wanting to learn from my mistakes, here they are:

1. My arithmetic is lousy. 6.8 x 2 = 13.6 not 13.8. I had entered the incorrect pixel size - as you rightly spotted

2. I had been using a neon image taken from the day prior to the target images. I think there was a different tilt in the neon image to that of the target image.

3. I was trying to refine the neon only calibration with the Sun rather than an A star so the calibration was never suitably improved.

Putting all this together, I calibrated an image of Fomalhaut with neon only. This yielded a RMS of about 0.2 pixels which translated to a RMS of about 23 A. Not a great result. This was without applying bias, dark or flats. I then used the Calibration Assistant with the Ha line from the Fomalhaut image - which was easy to spot in such a clean profile. The RMS dropped dramatically - to around 0.2 A from memory. I used the Response Assistant to produce a response curve, added bias, dark and flat frames and combined 5 images rather than the single image I used initially to calibrate and produced a better profile.

To test this, I then produced a profile for a nearby A2V star - around mag 5 and produced a very clean profile for that too.

All going well, I'll attempt a few fainter targets with longer integration times and see how it all runs.

Once I had the hang of the new ISIS, I found it much easier to use than version 4.

So thanks Christian for all your efforts

Pete

[Christian Buil]

Excellent news Pete,

Christian

[Francois Teyssier]

Hello Pete,

It's nice to see there is a new LISA well controlled.

From where dio you observe ?

Best regards

François Teyssier

[pvelez]

Hi Francois

I bought my LISA in March this year. I was working on a university project requiring measuring the redshift of quasars. After completing the project, the spectroscope has sat in the back of my study until I finished the academic year. With the holiday season approaching, I thought it woudl be worthwhile dusting off the LISA and turning it to other pursuits.

I'm based in Sydney - about 7km from the centre of the city so my skies are quite light polluted. I have a PMX mounted permanently on a pier in my yard so setting up time is minimal.

Now all I need to do is remember how to drive the LISA effectively

Pete

[Francois Teyssier]

Hi Pete,

The best way to tune the autoguiding cdd is to proceed indoor, on a table, with daylight, like explained in the Lisa Documentation provided by Shelyak
For memory :
http://thizy.free.fr/shelyak/Lisa/DC001 ... -%20En.pdf , pages 12/13

The acquisition CCD can alos be tuned on the table :
1. Focalisation on neon lines. You can use a very efficient function of ISIS : the FWHM of a neon

neon_fwhm.PNG (194.29 KiB) Viewed 10882 times

2. Rotation of the CCD in order to get an horizontal spectrum
I use for that the pinhole 19 mcm of the additional slit for LHRES/LISA. You can use tape behind the slit
On this recent image (CI Cyg), I have a 0.09 angle from horizontal. Almost good ...

Tilt.PNG (121.2 KiB) Viewed 10881 times

Once this tuning achieved, the spectroscop is mounted on the scope.
You don't have to change the tunings, (unless focalisation of neon is temperature has changed)

You focalize the scope with a type A star
1. First an approximate tuning with the image of the star on the CCD autoguiding camera
2. Fine tuning of the spectrum with the image of the spectruml on the acquisition CCD
For that i launch the autoguiding on the slit and proceed by very slight adjustments
Not easy to get a perfect focalization for all the continuum, either in blue and red range. This tuning is very sensitive to scope temperature and must sometimes be operated several times in a night (don't forget to take a new neon image !)
See above image of CI Cyg

This is this tuning which is important. While tuning the focalisation of the spectrum you can have a slight defocalisation of the autoguiding image.

Look forward to the first spectrum ...

François