Bin x 2 - problem with ISIS

Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Bin x 2 - problem with ISIS

Post by Peter Velez »

Hello

I'm back to spectroscopy after a break of 7 years and working my way up the learning curve with ISIS. I have my rig at Coonabarabran NSW, Australia adjacent to the Siding Springs Observatory and operate it remotely. Its been a challenging exercise but the first results are proving very positive.

My main issue at the moment is to process spectra taken with the imaging camera binned x2. I have a LISA attached to a Planewave CDK 12.5 which is stepped down to F5 with a reducer. My imaging camera is an Atik 460EX. Its a great combination and unbinned images are easily reduced and processed. However, to access fainter targets, I'd like to bin my images. And this is where I have problems.

I have taken images and Ar Ne calibration frames x2 and have prepared master darks and bias frames from images that are also binned x2. I have adjusted the pixel size - from 4.54 to 9.08. However, when I try and process the spectra, the spectral calibration fails or the RMS is surprisingly high and the results dubious.

The routine I am following is the same. I prepared a .lst file based on the file I use for unbinned images with the dispersion adjusted to reflect the change in pixel size to 3.652. Processing this, I generate a reasonable spectrum but with a RMS of about 19. I have managed an RMS as low as 0.2 with unbinned images. My reference star is a bright A0V star quite near the target. I then use the spectral calibration assistant to improve the calibration using the Ha line in the reference star. It generates an RMS of 5.18 and then reports that the calibration error is too high.

The .lst file method generates a reasonable spectrum but its clearly not correct - comparing it to the Miles A0V spectrum, my spectrum is shifted to the red at the blue end of the spectrum and vice versa.

As an alternative, I have used the Predefined mode for the LISA using argon IR, argon visible and neon and selected the appropriate lines in the calibration frame. When processing, the program seems to flicker between 2 spectral processes - checking the log I can see multiple attempts to determine the dispersion - primary and inverse. The primary RMS is a healthy 0.98 but the inverse RMS is an unhealthy 26.1! The resulting spectrum is worse than that generated by the .lst based method.

I am sure this is pilot error and not a bug in the software. But I can't see where I am falling down.

I'd be delighted if anyone is able to tell me where my method is wrong.

Thanks

Pete

PS I am sure I have some lovely data on 2020hvf - if only I could process it correctly!
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

Over the last few days, I have had the opportunity to play with the beta version of Demetra. It works well with my binned images - though I have yet to process any unbinned images.

So my work around is - use Demetra for binned images and ISIS for unbinned images.

Any assistance would be appreciated.

Pete
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

Here are some screenshots that may help showing what I am encountering:

(I have the screenshots in reverse order - but you should be able to see the issue)

The RMS flips from 0.98 to 15.5 for the inverse multiple times till the profile is generated. This is with the Predefined mode for the LISA - the same result with Ar IR, Ar Visible and Neon. I can't refine this with the Calibration Assistant as it generates a RMS of about 5 (or 8) and reports it as too high. The response curve generated with the Response Assistant is also unusable.

There may be a simple setting I am missing but I can't find it - the pixel size is correctly set at 9.08 for binning x2 (up from 4.54 for this camera).

Has anyone ever seen this?

Pete
Attachments
Screenshot 2020-05-21 13.55.27.png
Screenshot 2020-05-21 13.55.27.png (269.83 KiB) Viewed 6511 times
Screenshot 2020-05-21 13.55.11.png
Screenshot 2020-05-21 13.55.11.png (284.15 KiB) Viewed 6511 times
David Boyd
Posts: 161
Joined: Wed Dec 11, 2013 5:50 pm

Re: Bin x 2 - problem with ISIS

Post by David Boyd »

Pete,

I have a LISA+SX-H694 (same chip as the Atik 460) on a C11 operating at f/5 in the UK. I process with ISIS using the file mode calibration method and it works well. I generally use the 23 micron slit in the LISA, run the camera unbinned for resolution reasons and expose for longer on fainter targets. So I'm afraid I cannot shed any light on your ISIS problem.

What intrigues me is how you operate a LISA remotely and maintain good internal focus of the LISA. I have to manually refocus the LISA whenever the temperature here changes by more that a few degrees C if I want to keep the resolution above 1000.

David
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

David

good to hear from another LISA user.

You pose an interesting question. I have been operating the spectroscope on the basis that internal focus is steady and that I need only concentrate on focus via the telescope. I haven't seen much drift in resolution as a result of a fall in temperature. That said, I generally only image for a few hours each evening and there is probably not a huge drop in temp in that time. We also set the focus on a particularly cold day so perhaps the rig cools down to a suitable temp in time.

I'll keep an eye on this and will certainly let you know if I encounter any issues.

I suspect that adjusting focus is more an issue for a high resolution spectroscope - there are a few posts on the forum about remote control of the Lhires internal focus. If I graduate to that unit, I will need to revisit remote focus.

Pete
etienne bertrand
Posts: 1040
Joined: Mon Aug 31, 2015 11:26 am

Re: Bin x 2 - problem with ISIS

Post by etienne bertrand »

For ISIS the pixels size of 9.08µm is not correct because you need calcul it with :
Pixel size = 3123/(Ha - Hb)
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Bin x 2 - problem with ISIS

Post by Robin Leadbeater »

etienne bertrand wrote:For ISIS the pixels size of 9.08µm is not correct because you need calcul it with :
Pixel size = 3123/(Ha - Hb)
This works for the ALPY predefined mode but I think it will be different for the LISA.

The pixel size is only used for the predefined modes. With the ALPY it sometimes needs adjusting slightly to get ISIS to lock onto the right lines. Perhaps the LISA is the same ?
With file mode, have you tried altering the dispersion in the .lst file slightly to see if that allows ISIS to lock onto the right lines ?
It is also worth checking the geometric corrected image to check if the reference line position has moved significantly and if so try entering the corrected pixel position (This is a problem some people have seen with the the ALPY if the spectrum is significantly slanted or tilted)

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

etienne bertrand wrote:For ISIS the pixels size of 9.08µm is not correct because you need calcul it with :
Pixel size = 3123/(Ha - Hb)
Thanks for the suggestion Etienne

Based on an earlier thread regarding the Alpy, I tried this method. Trialling this with unbinned images, I received a pixel size of 3.65 - well short of the reported 4.54 of this camera. I was unable to calibrate unbinned images this way and so assumed that this is for the Alpy only not the LISA.

Pete
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

Robin Leadbeater wrote:
The pixel size is only used for the predefined modes. With the ALPY it sometimes needs adjusting slightly to get ISIS to lock onto the right lines. Perhaps the LISA is the same ?
With file mode, have you tried altering the dispersion in the .lst file slightly to see if that allows ISIS to lock onto the right lines ?
It is also worth checking the geometric corrected image to check if the reference line position has moved significantly and if so try entering the corrected pixel position (This is a problem some people have seen with the the ALPY if the spectrum is significantly slanted or tilted)

Cheers
Robin
Robin - you were right. ISIS was not able to match onto the correct lines.

I tinkered with the dispersion in the .lst file to no result.

So I reviewed all of the lines in my .lst file. I stripped it right back to the very obvious lines (ie brightest and easily isolated) and managed to calibrate. By amending the .lst file and rerunning ISIS, I was able to improve the calibration and identify the lines in my lamp that ISIS doesn't like. I managed to get the RMS for my reference star down to 0.08 with which I was very happy.

Interestingly, when making corresponding changes to the .lst file I use for unbinned images, I was not able to improve the RMS by eliminating the "challenging" lines. I have no idea why.

Anyway, I am on my way now. Many thanks to all for the assistance

Pete
Peter Velez
Posts: 146
Joined: Tue May 12, 2020 7:36 am

Re: Bin x 2 - problem with ISIS

Post by Peter Velez »

So, I've been doing some more testing on this.

The adjustments I made to my .lst file to bring the RMS for spectra produced using binned images have involved removing problematic lines - mainly in the blue. This did reduce the RMS - but the actual spectra still looked dodgy.

I had been using the LISA with the 19 micron slit. This is fine for unbinned images but seriously undersampled for binned images. Changing to the 23 micron slit made little difference.

This week I managed to change to the 35 micron slit. Processing with the .lst file still yielded a RMS of about 8.5.

However, I was successful in refining the solution using the Calibration Assistant - managing a RMS of about 0.35 - which I am very happy with.

My suspicion is that undersampling degrades the Ar Ne calibration images and object images which confuses ISIS. By addressing this, I can now process in ISIS.

Does that make sense?

Also a question for the novice - I found that I needed longer exposures for my calibration frames to obtain a good calibration image with the larger slit. I would have thought that a larger would mean that more light would enter the spectroscope and the calibration images are more likely to be saturated, not less. Am I missing something?

Pete
Post Reply