SA Notes On PFIP Setup

The following instructions describe how to determine the RC and do the 7-star calibrate:

1. Determination of the rotator center:
   • Ask the TO to move to a bright star (~11 mag), and set the sky position angle (PA) to 180 degrees.
   • Translate the pointing of the telescope to CCD1 (to avoid the gap):

     SYS> user "ENTER APERTURE 0 250 0"
     SYS> user "AP 0"

   • For speed, set the CCD readout speed to fast.
   • Take an exposure and measure the position of the star (x1,y1) on the chip (should be close to the centre of CCD1).
   • Ask the TO to move the sky position angle to zero degrees and measure again the position of the star in the chip (x2,y2)
   • Compute the position of the RC: (xc,yc) = ((x1 + x2)/2, (y1 + y2)/2)

2. Pointing calibrate with 7 stars:
   • The sky PA should already be at zero degrees, but check that this is the case. Also keep the same aperture as that defined when determining the rotator centre.
   • Make a window of 301x301 pix around the rotator center using the following command:

     SYS> window pfip 1 "[xc-150:xc+150,yc-150:yc+150]"

     Remember that the coordinates of the RC in this new window will change (for the example above, the RC will be at pixel (151,151) of the windowed frame).

   • Now tell the TO that you are ready to run the pointing calibrate with seven stars and that you would like to use the faint point grid. Remind the TO that the sky PA should be maintained at zero. Sometimes these stars are too bright for PFIP, so use a narrow band filter (or the U filter with short integration times) to avoid saturation. If all else fails, then you may close the primary mirror petals by a small amount.
   • For each star, take an image and calculate the telescope movement needed to put the star at the RC. Normally one, or a maximum of two, small offsets (<15'') for each star are needed.
   • Once finished, disable the window and reset the aperture:

     SYS> user "ENTER APERTURE 0 0 0"
     SYS> user "AP 0"

     Also remember to reopen the mirror petals fully if you closed them.

Removal Of Mosaic Tilt

1. Point the telescope to an open cluster or any star field with stars well distributed across the 16'x16' area. You will need to have at least 10 (ideally 30 or even more) stars with some 3 to 30 kADU in the peak of their profile.

2. Ensure that the CCD mosaic is not windowed, and for speed, set the CCD readout speed to fast.

3. Put a PFIP broad band filter in the beam (e.g. R or V).

4. Set the previous PFIP run focus or the past night focus as a start point to focus the telescope.

5. Take a series of 7 (or 9) images with different telescope focus, centred at the start focus and stepping the focus by 0.05mm using:

   focusrun pfip <no. exposures> <exptime> <focus start> <focus step>

6. Use the IRAF task mpf_head from the pfip_ql package to create new images (from the focusrun frames) with some new necessary header info in order to analyze the mosaic tilt (this can be done in the /obsdata/whta/YYYYMMDD directory):

   pf> mpf_head input=r??????.fit output=a1.fit
   pf> mpf_head input=r??????.fit output=a2.fit
   

   and so on... Note that mpf_head does not work with image lists.

7. Run the IRAF task mpf_tilt:

   pf> mpf_tilt images=a1,a2,a3,a4,a5,a6,a7,a8,a9
   

   The task will prompt you to select an image to display and you should choose a5.fit, which is the image closest to the best focus. Now you should select at least 10 (and up to 30) uniformly distributed stars from the whole displayed image mosaic using the letter "m", and you may use the usual imexam options of "e" and "r" to select "good" stars, i.e. PSF objects with no saturated pixels.

   Note that mpf_tilt will crash if it finds even a single saturated pixel in one of your chosen star profiles on ANY of the images to be analyzed. For this reason it is a good idea to pick stars with peak counts of less than 10 kADU, which are unlikely to saturate in any other image. Especially in case of bad weather (sky changes within few minutes during the setup procedure), it is advisable to choose faint stars, so no star from any image could saturate and crash the script! Stars may be removed from the list of stars to be analyzed by removing the corresponding entries in the file /obsdata/whta/YYYYMMDD/tmp_coord1. When you have finished your selection, you may press "q" to finish.
   
   IMPORTANT IN CASE OF CRASH: in case that mpf_tilt crashed or if you need to force it to stop it because of mal-functioning, then you must search for any remaining temporary files (pfip_ql> ls *tmp*) in the current folder (observing date) and delete these files manually (pfip_ql> del tmp*) before launching mpf_tilt again; otherwise a bug will occur and will prevent the script to work properly.

8. A few seconds after pressing "q", a window will appear with a plot of FWHM versus focus. The options of the mpf_tilt task may be accessed by pressing "?". Use the option "b" to bring up a plot of the best focus for each star as a function of position in the CCD mosaic. Red and green circles show stars whose focus is below or above the mean focus, respectively.

   From the distribution of stars in this plot it is easy to see whether or not the CCD mosaic is tilted with respect to the focal plane (here is an example of a clearly tilted mosaic and a non-tilted mosaic). Another useful option, "a", displays the FWHM and ellipticity of the stars as a function of position across the mosaic for the mean focus. Note that a star can be deleted by pressing "d" with the cursor on top of it.

   On pressing "q", the task will report a recommended movement for capstans A, B and C:
   

   • If the recommended capstan adjustments are around one quarter of a turn or less, then it may not be necessary to apply them. In this case, check that there is no clear trend of FWHM across the field of view, and that the best focus value for each star is within ~0.05mm of any other value, indicating that the best focus is independent of position in the CCD mosaic.
   • If the recommended capstan adjustments are larger than one quarter of a turn, and you can see a clear gradient in the best focus estimate for each star across the field of view, then you will need to apply the adjustments and check the mosaic tilt again:
     1. Ask the TO to move the telescope to the access park position.
     2. Adjust the capstans in the order A, B and C. If adjusting only B, leave A and C alone, but if adjusting only A and C, release and lock B (but without turning the capstan) between adjusting A and C.
     3. Once the capstans are adjusted and locked again, make sure the cryostat is tight again on its mount.
     4. Return to the same star field and repeat the process from step 4 (NB: after adjusting the capstans the telescope focus will be different. The mpf_tilt script gives a prediction of this change).
9. When the mosaic tilt has been properly removed, then communicate to the operations team that the capstan settings should be remeasured during the next day and their positions noted in the CCD Capstan Settings folder.

1. TO BE DONE ONCE IN THE BEGINNING: after you have selected the stars (min 10 and up to cca 30) within mpf_tilt, in DS9 finish normally the process by pressing "q" which prompts you the irafterm plot. Right here, open another terminal and drive to the observing folder (ex: /obsdata/whta/20121108) then make a copy of the selected stars temporary file,
   
   > cp tmp_coord1 copy-tmp_coord1
   
   Come back to the irafterm plot and finish normally the mpf_tilt process, then correct the recommended tilt at the telescope.
   
   
2. TO BE DONE ANY NEXT STEPS: Next time(s) point at exactly the same position (could be accurate up to a few pixels) and before running mpf_tilt, in the same terminal copy back the saved temporary file to be used by the new mpf_tilt run
   
   > cp copy-tmp_coord1 tmp_coord1
   
   then run mpf_tilt normally, which will prompt you in DS9 with the same marked stars and avoid re-selection.

Filter Focus Offsets

Filter Focus Offset (mm) = F1 - F2

where F1 is the best telescope focus (mm) for the filter without a filter focus offset, and F2 is the best telescope focus (mm) for any filter with a filter focus offset.

The newly measured filter focus offset must now be entered into the ING filter database, and in order for the PFIP control GUI to pick up the change, then the GUI should be closed down (File => Quit), and restarted by typing:

SYS> 4MSControl -panels PFIP &

Check that the list of filters and the focus offsets are correctly displayed on both pages of the control GUI.

Below are a few notes on the symptoms and origin of the WHT PF-camera 
light leak, and a suggested cure (basically 3 baffles, indicated by the
sheets of card currently in the camera).  Where I give orientations, they
are as would be viewed by someone standing at the filter-wheel access
door, with the camera pointing down at the floor.


Symptoms
--------
The leak is visible as a V-shaped patch of light on exposures made 
through narrow-band filters with the 2-EEV mosaic.  The apex of the V
is on the join between the CCDs (at the bottom as usually displayed,
= opposite B capstan), with the arms reaching out to about halfway up
the left and right edges of the displayed frame.  The intensity in
dark of moon peaks at about 0.1 counts/sec, which is about 4% that
expected for sky as seen through a V filter.  Hence for narrow-band
imaging (dlambda ~ 30 A), the magnitudes of the leak and the sky
background are similar.  The data taken by this month's narrow-band
imager (Steve Smartt) were seriously compromised by the leak.
The leak is faintly visible on broad-band images.
The leak does not change position when the camera is rotated.
It is unaffected when wrapping the entire camera in black cloth,
so is due to light coming through the main optics.  
It is unaffected if all the filters are blacked out, so must be light 
scattering round the filter wheel.

It happens that when only one EEV CCD is used at PF (EEV12), it is oriented
with the long axis perpendicular to the join in the 2-EEV, so the light
leak is not then prominent.


Origin
------
I replaced the 2-EEV cryostat with a translucent screen at the approximate
position of the CCD, and covered all except the entrance aperture with
black cloth.  With a near-opaque filter in position,
an arc-shaped leak was immediately visible on the screen, with its
apex in about the right place.
With the front aperture covered, the leak is not present.
The leak is sky light passing through the main optics and bouncing
off the edge of the top of the corrector, skimming between filter wheel and
autoguider head, and landing on the mosaic opposite the capstan B end of the 
CCD join:


   ---- 2EEV
       .         *
         .      **
-F-wheel---.   *** autoguider head
             .****    
               .    
----------------- top of corrector (illuminated by sky)  


I was able to reduce the leak by a factor of 200 (as measured on the
CCD) by blocking the edge of the top of the corrector with card.

There is only about 1 cm horizontal clearance between the filter wheel
and the autoguider head at the limit of its travel.  Any baffling
has to fit in this space.

There is a large vertical gap between the top of the corrector and
the filter, and the diameter of the corrector is much greater than that
of the filter, so the complete absence of internal baffling means that there
are many other routes for light to get onto the CCD.
When the main leak is blocked, many other distinctive leaks are visible
on a short exposure e.g. light passing through an unprotected cable
entry port on one corner of the camera appears on the CCD as a diagonal
streak.  The obvious leaks need blocking up, and internal baffling
is needed.


Suggested cure
--------------
Servando and I looked at the camera yesterday, and concluded that
the following baffles are needed (I've taped pieces of card in the approx
positions required):

(1) A baffle running left-right between filter wheel and autoguider head,
from side to side of the box, supported on the box.  In cross section
as viewed from left or right of the camera:

___F.wheel__   -----
              / **
             / ***
            / ****

where * = the autoguider head, ___ = filter wheel and ///--- = baffle.
This baffle blocks the main leak.
The shutter bar projecting below the base plate which sits on top
of the camera may get in the way of part of this baffle.  In this
case, a similar shaped baffle should also be stuck to the autoguider head,
so that it moves left and right with the head.  The length of this
extended hood will be determined by the left-right travel of the
autoguider.
The front-back position of the main baffle needs checking with the 
autoguider driven to its limit in that direction to check no fouling.

(2) A circular baffle blotting out all of the top of the optics 
except the area used by the autoguider and a central circular area
below the filter X mm in diameter, where:
    X = 78 mm (corner-to-corner diameter of CCD) + H/2.8 (to allow 
        for spreading of the beam)
    H = distance between top of optics and front of CCD
I think H is about 100 mm, which gives X = 114 mm, but it would be a good
idea to measure H (to an accuracy of a few mm), and then add about 10 mm
to the calculated value of X, as a safety margin.
This should be just above the optics, and if mounted on them should 
be removable when the PF optics are used with AF2, 
This baffle blocks most of the unwanted light emerging from the optics.

(3) Two baffles just underneath the filter wheel, i.e. about 10 cm vertically
above (2), symmetrically left and right, between them blocking the whole 
area except under the filter in use.  These baffles are bounded by the
box, by baffle (1) and by the central circular structure of the
filter-wheel holder.  They need to be mounted on the box (not the
filter wheel, which is where I've taped the pieces of card).
These baffles prevent light from the aperture left in (2) passing 
through one of the filters not in use (e.g. a much more transparent one) 
and being scattered about inside the box until it lands on the CCD.

In addition, the following modifications are desirable:
- Light leak at unprotected cable entry point top left needs blocking.
- Small light leak at cable entry point about 30 cm below filter wheel
  needs blocking.
- There are 6 bolt holes to fix the top-plate on the camera, but only
  4 bolts.  Light leaks through the remaining holes.
- The flaps between the petals of the filter holder are loose, and I'm worried
  that one of them will bend, and foul against something.  Can't think
  of an easy solution though.
- The inside of the main box should be black everywhere (e.g. I noticed
  there's a white sticky label on the underside of the shutter plate).

A few other things which have nothing to do with light leaks:
- Is there any possibility of extending the slide that the filter wheel
  goes into?  On the telescope, this is normally withdrawn vertically,
  by someone working alone.  The handle isn't accessible until the 
  holder has fallen out of the slide, so there is no safe way to 
  take it out.  One day the entire filter wheel is going to leak
  through someone's fingers and make a large dent in the floor.
- The cables passing through the filter-wheel door have to be disconnected
  every time a filter is changed.  This seems to me a recipe for broken
  connectors, but I can't think of any easy solution.
- There are some flecks of black paint visible on one of the optical
  surfaces, looking down through the camera.  Can they be removed,
  or is it sealed?

CRB