LRIS Observing Procedures
Draft
There are some observing procedures that we (the DEEP group) use to reduce
the effects of fringing and to obtain calibrations needed for the prototype
reduction pipeline.
Calibration Mask
In order to solve for the alignment of various optical elements in LRIS,
so that the pipeline software can run easily, it is necessary to use
the calibration mask
"cal_grid.eng" with as many different gratings, tilts,
and orders as practical. Below is a typical data set:
For each grating, use appropriate internal arc lamp(s). One exposure is
sufficient; slight saturation is not a problem:
- go to approx. observing angle/wavelength (1st order)
- set GRANGLE about 1-2 degrees larger
- set GRANGLE about 1-2 degrees smaller
- set GRANGLE to zeroth order position (approx. 13.5 degree, see
table )
Do this for at least 2 gratings, preferably 3.
Also get image with the mirror in imaging mode (zeroth order).
Dome Flats
When dome flats are available (needs OA) get the following images:
- spectroscopic dome lamps, "calgrid" + grating (at observing tilt).
Spectra should be traceable across full image, in order to provide a map of the
"y-distortion".
- imaging or spectroscopy lamps, "direct" + grating (at observing tilt) NB:
This should be a set of 3 or more, with good S/N throughout and NOT SATURATED.
These images provides the pixel-to-pixel variations across the whole CCD.
Wavelength Matching for Best Fringe Division
The LRIS CCD finges at a level of +/- 5% or so in the red. The fringe pattern
is imprinted on the pixels, and moves through a full phase in about
50 Angstroms, meaning that the wavelength scale of dome flats must agree very
closely (ie, << 50 A) with the science images in order to remove fringing to
a high precision. Unfortunately, LRIS flexes in both the spatial and wavelength
directions, resulting in wavelength shifts (bad phase match) and spatial shifts
(different slit illumination) as the telescope is moved around the sky. The
following procedure can be used to reduce these uncertainties.
The general procedure is this: from each science image (or set of images,
using the same mask), use the night sky lines to figure out where a given
arc line will fall as seen through a particular slit when the grating is
tilted at the same angle.
(Example for 600/7500 grating, using night sky lines
at 7572 A and 7316 A to predict location of 7635 Ar I.)
Then, at the end of the night, return to the nominal setup used for
the science observations -- that means similar
elevation (EL),
physical rotator angle (ROTPPOSN) and
grating tilt (GRANGLE)
(these are all found in the image headers). Using the arc lamp,
note where the chosen arc line in the chosen slit falls. If it doesn't
land in the right place, tweak the grating angle (GRANGLE) to move it to
the right spot. When it is close enough (within about 1 px or so), you have
recaptured the same configuration as well as you can. (HINT: if you
know how to window the CCD, reading out onlyi, say, 100 rows encompassing the
chosen slit really speeds up this process!)
PROBLEMS: The encoders on the gratings seem to be fairly coarse, so small
GRANGLE moves will be ignored. You may find it best to set GRANGLE to a very
different value and then reset to the updated value. This can be a very
time-consuming process.
How much to change GRANGLE? I have a couple of numbers in my head, namely,
0.00192 degrees/A for the 600/7500 grating, and around 0.004 deg/A for the
1200/7500 grating. I should calculate these in terms of pixels sometime ...
(oops, got to go for now ...)
Standard Disclaimer
These procedures were developed by Drew Phillips and others at Lick Observatory
for the private use of the DEEP team at UCSC.
They are not official Keck procedures,
so please don't bother Keck folk with questions/complaints relating to the
above.
Andrew C. Phillips / Lick Observatory
Last modified: 18 sep 2001
phillips@ucolick.org