Typically, after one has aligned the beam directions coming into the lab,
and is satisfied with the image overlap; the two combined beams output
by the beam splitter, and focused by the parabola,should be placed on the
two NICMOS pixels of choice.
This is accomplished with the help of a program which images the NICMOS
quadrant, and zooms inon a 21x21 pixels region around a specified center
pixel, and by steering the beams by adjusting the tip and tilt of the mirrors
labelled NICMOS1 and NICMOS2.
In the Pentium controller, and from DOS, go to the "c:\nic_code\ directory"
and type "q_read" toexecute this program. After selecting the various choices
and parameters you will get a display which shows the part of the quadrant
that is being read, the 21x21 zoomed region, and information on the location
and signal level for the two maximum pixels found in each half of the 21x21
region.
Typing "Shift-S" on the Pentium keyboard will stop the program, and ask
you whether you want to continue with different parameters or just quit.
The "save to file" option is something I use to diagnose the image quality,
you will typically not need this, but feel free to store images to your
own floppy disk if you like.
After moving to a new long delay position and correcting the resulting
yaw error, the beams go back to the correct nicmos pixels, as they should,
and it is in principle not necessary to re-do this adjustment; but feel
free to be cautious and check this from time to time.
IMPORTANT NOTES :
The max. pixels are searched in each half of the 21x21 region, so you must
put each beam on either side of the dividing central column. If you are
completely lost, you can image the whole quadrant first to find the two
spots of light, and then steer the beams where you want them and relocate
the 21x21 box around them.
Unless the whole quadrant needs to be searched, I recommend doing this
using the mode 1 readout (64x64 pixels. Also, for most sources of say 1<K<4,
an "integration time" of 0 is ok, because the real integration time includes
the readout time (249 ms for mode 0, 69 ms for mode 1 and 28 ms for mode
2). For bright sources (K<1 or so) this readout may take too long and
the pixels may saturate (which results in their value rolling over to zero
and they appear black on the image!), in that case you can use the faster
mode 2 readout, or a fainter source.
Our experience has been that we can put between 60 and 80% of the light
in one pixel. So that is what you could shoot for. You can judge that by
using the color code and max. pixel intensity information.
The color code autoscales to the max. pixel found in the whole 21x21 region,
it does not autoscale each beam separately.
The two pixels where you put the beams must BE IN THE SAME ROW and they
must both be on EVEN columns. Our current prefered pixels are (row,column)
= (36,22) and (36,36) ; so that the center of the 21x21 region would be
(36,29). The new focusing lenses added in October now define a fairly narrow
range where the beams get focused on the pixels with no occultation,
but you may explore an area of +/- a few pixels near the above suggestion,
as long as you respect the "same row" rule and the "even columns" rule.