Telescope: CTIO Blanco 4m
Date: 2016-01-24
Submitted by David James

Present: DJJ, Katelyn Allers, Everett Schlawin, Javier Rojas, Hernan Tirado, Sean Points (remotely)
Comments: Second night of last ARCoIRIS engineering run.

During the afternoon, mountain electronics teams changed out the Qz lamp in the internal 
calibration unit of the Cassegrain cage. When we tried to take flat field exposures, the 
fluxes were still very low. Even with 20% illumination (we were not advised to go higher), 
a 30-seconds exposure yielded a peak counts level of 750-counts ... much, much too low.

So, as with yesterday's engineering, we took domeflats with the COSMOS top-ring lamp system 
and the DECam flat-field screen with good results. The colour-balance of the COSMOS lamps 
is still not ideal for ARCoIRIS, with too little blue flux for a very strong red flux 
distribution; The COSMOS-lamp flat-field data are decent enough (not optimal, but good-enough 
for now) for data analysis and engineering. 

Later, when we tried to take internal arc lamp exposures with the hollow cathode (CuHeAr) 
unit, there was no light being emitted at all, and so were were not able to take fixed-format, 
high-flux wavelength calibrations. With such low flux in the hollow cathode lamp system, 
Javier R., Hernan T., and Claudio A., spent nearly three hours investigating its power supply 
unit and cabling in the Cassegrain cage. They found problems with the high tension power 
supply and so made some remedial repairs, and changed out an high-voltage cable (from the 
Xenon lamp) -- a replacement will need to be fabricated. Thanks to all involved.

As an afterthought, we have no more spare CuHeAr discharge lamps, and so I've requested 
of the ETS group to purchase new ones. I think Javier Rojas has today requested new lamps.

Several on-sky tests of the instrument control software and observing environment have 
been made this evening, and can be summarized as: 

(a) In an attempt to make a standard ABBA nodding macro, we saved the GRID positions to 
 a text file, cleared the grid, and then reloaded them. Unfortunately, the ASCII grid 
 file reloads but with some odd behaviour.

  (a-I) the grid files also include data for the two cursors stars on the GRID for 
        STAR 1 (red) and STAR 2 (green). These values end-up being displayed in the 
        nod list box, and should not be. 

 (a-II) the grid files does not remember the "number of visits" flag --> for instance, 
        for an ABBA nodding sequence, one typically flags the number of visits 
        to be equal to 2 (for the B nods). These are not recalled when the grid file 
        is reloaded. 

(a-III) the new "go to BB" flag is not saved to the grid file either. This would be 
        a good option to be recorded in grid file.

(b) Marco has added a "go to BB" flag in the GUI_tool_GRID GUI so that we can we nod 
    off-slit from a nod-position to a more remotely located Background Box, which the 
    user can specify. We made extensive on-sky tests of this feature; results are:

  (b-I) For one nod-position on the slit, slew to the BB-box works well, except that 
        the star does not slew to the centre of the background box, but only close (few 
        tens of pixels) to its upper right -- see GoToBB-offset-25Jan2016.png).
        
 (b-II) For two nod-positions on the slit, again, the feature works well, with a sequence 
        of nod-A --> BB, nod-B --> BB, however again the star did not go to the centre of 
        the Background Box for nod-A or nod-B (even more offset -- see attached image 
        GoToBB-offset2-25Jan2016.png). 

        When we tried moving the STAR 1 (red) cursor, it seem to change the amount of 
        BB-offset, and then froze completely the GUI_Tool_GRID GUI -- forcing a full re-start.

(b-III) For a three nod-position, this feature did work, but the offsets were again bad, 
        and one more time, the GUI froze when the STAR 1 cursor was not in the bottom left 
        corner of the Grid_Tool_GUI. 

(c) In order to test the sensitivity of the extracted spectra to nod positions on the slit, 
    we observed a bright M9V star at many positions along the slit, repeating the observations 
    with an A0V calibrator star. Data reductions are on-going.

(d) Once we added the options of Fowler16, and co-add=5, from spectrum SPEC_Eng0185 onwards, 
    the fits headers of the images looked very strange, with missing values from the nod-A and 
    nod-B images. A comparison of the nod-A and nod-B FITS headers shows considerable differences 
    and omissions.  Even though the "write expo" button was not selected, there were "accum" 
    files written to disk. Oftentimes, the header of the accum-fits files were not the same 
    as the difference science images. This caused many problems in the IDL data reduction 
    algorithms that Katelyn Allers is writing. 

    Upon investigation it looks like the Fowler16 selection is causing the problem. 

    Marco, can you please check that the FITS headers are be correctly written ?  Also, can 
    the "accum" files be written to a separate directory please ?

(e) The selection of the OBSTYPE=calibration option has caused a few strange actions. 

  (e-I) Several times, after taking an image, the software continued to take several images, 
        even though only one image was selected. The software just ran-away, taking image 
        after image, and had to be re-started.

 (e-II) The normal action of the "calibration" option is to take one arc lamp exposure, and 
        then, turn off the lamps, and then take an equal exposure dark frame. A couple of times, 
        this course of action was not enacted. Can Marco please investigate ?

(f) Katelyn A. and Everett S. performed many tests and upgrades to the IDL software that K.A. 
    is developing. 

A busy night !
saludos
DJJ, KNA, ES

p.s. GoToBB-offset-25Jan2016.png  -- https://www.dropbox.com/s/n221xssgapahw8u/GoToBB-offset-25Jan2016.png?dl=0

p.p.s GoToBB-offset2-25Jan2016.png -- https://www.dropbox.com/s/os123oousz8yo8a/GoToBB-offset2-25Jan2016.png?dl=0