The Ohio State University
College of Mathematical & Physical Sciences
Department of Astronomy
R4K User's Manual
- CCD Properties
- Readout Modes and Times
- OSMOS at the 2.4m
- Direct Imaging at the 2.4m
- Direct Imaging at the 1.3m
The R4K is a 250um thick, fully depleted LBNL CCD with 4096x4096, 15um pixels.
This detector is particularly notable for its extremely high red sensitivity.
For example, it is approximately a factor of three more sensitive than the
MDM4K between 900nm and 1 micron.
See this QE comparison figure for the expected
performance of the MDM4K and the R4K CCDs (these two QE curves are based on
manufacturer data). The R4K also exhibits substantially reduced
fringing compared to thin CCDs. Its two main disadvantages are
a higher rate of radiation events (a.k.a. cosmic rays) and lower
response in the blue region.
The R4K is interchangeable with the MDM4K device and may be used
in direct mode on both the 1.3m and 2.4m telescopes, as well as with
OSMOS on the 2.4m telescope. Observers should specify their preference
for detector at the time proposals are submitted. Should a conflict arise
between observers for the R4K (or MDM4K), the 2.4m observer has traditionally
received precedence in similar situations with other instruments.
The R4K detector system was commissioned at the MDM 2.4m Hiltner telescope
in September 2011.
The R4K detector is LBNL CCD #127298-20-4. The detector electronics
were built by Bruce Atwood at Ohio State and read out array through 8
amplifiers (2 per quadrant). The readout of the columns in each quadrant
are alternated between the 2 amplifies such that there are effectively
separate amplifiers for the odd and even columns.
As the bias (and gain) levels for each of these amplifiers are somewhat
different, the raw data will usually exhibit a striped pattern.
There are three ways to remove this even/odd pattern in the quadrants (and
the differences between quadrants). The first is with a measurement of the
8 different bias levels from the overscan region. This is described
further below in the software section below.
The second method is to simply bin the data 2x2 after they are
obtained. Finally, the bias level is fairly constant and consequently just
subtraction of a bias frame appears to produce good results.
Note that the gain level also varies slightly between amplifiers. The
gain variations may be removed with flat field data. Once the gain
has been measured, those data will be posted here. The gain is
expected to be about 2 electrons per count.
Technical details are available from the
LBNL Manual for the device.
Various region of interest (ROI) readout modes are available, similar to
those available for the MDM4K. These modes may be used to readout a
subset of the full array, or to return to full frame readout. These may be
set in the same manner as for the MDM4K, namely with the following
- PR> call rroi1k.pro
- PR> call rroi4x1k.pro
- PR> call rroi2k.pro
- PR> call rroi4k.pro
These change the ROI to 1024x1024, 4096x1024, 2048x2048, and 4096x4096
pixels. In all cases the ROI is centered on the array. Note that these
scripts start with rroi and not roi to differentiate them
from the scripts for the MDM4K (the scripts are slightly different because the
arrays are slightly different sizes). Readout times for
a representative subset of these modes are listed in the table below.
The binning may also be changed. For example, to change the binning to
- PR> ccdbin xbin=2 ybin=2
The current binning is shown in the Prospero Status Window. Readout times
for 2x2 binning with various ROIs are also listed in the table. Note that
binning 2x2 will NOT remove the different even/odd bias levels, as the
array is still read through all 8 amplifiers.
Standard overscan subtraction codes, such as IRAF's colbias
task, do not have an option to calculate the bias level separately
for even and odd columns. Paul Martini has written a python code
called proc4k.py that calculates and subtracts
a constant bias level for each of the eight quadrants. This code
works reasonably well, although it has not been exhaustively tested.
Users are encouraged to report problems and requests for additional
features to Paul. One future addition is likely to be to remove the gain
variations between the amplifiers (once they have been precisely
Obviously, use of this code requires that there is an overscan region.
Good results have been obtained with 32 columns of overscan. To set the
overscan region in Prospero type:
Operation of the R4K and MDM4K are effectively identical. Should any
differences be identified, they will be described here.
The R4K may be used as a direct imager at the 2.4m telescope with the
12-position filter wheel. The plate scale is 0.17 arcseconds per pixel
and the field of view is about 11.5 x 11.5 arcminutes. The image quality
at the edge of the field should only be marginally degraded in the best
seeing. Specifically, the images will be elongated by approximately 0.5
arcseconds in the corners of the field. The main advantage to direct
imaging with the R4K compared to OSMOS is somewhat higher throughput.
This has not been measured, but is expected to be on order 30 percent.
The main disadvantages are smaller field of view and poorer sampling.
Note that while it is possible to bin 2x2 to approximately obtain the
same plate scale as achieved unbinned with OSMOS, the radiation event hit
rate per binned pixel will be larger than for unbinned data.
The R4K has only experienced limited use at the 1.3m telescope to date, and
not yet by the commissioning team. The operation should be identical
to the MDM4K. The scale is 0.315 arcseconds per pixel and the field of view is
about 21.3 x 21.3 arcminutes. Users are encouraged to refer to the detailed
for further information. Note in particular the sections on image quality and
the shutter correction.
There are saturated pixels shaped like 'icicles' at the top and bottom of the array
- PR> overscan overx=32
This is due to accumulated charge in the horizontal shift registers after either the saturation of the device or the device sitting idle for a long time. The Prospero script 'snapclean' will aggressively attempt to clean up the residual charge:
If you encounter this error, please report it in the Trouble Report and mention if snapclean resolved the problem. Note that snapclean will leave the readout mode as full frame and 1x1 binning. If you were using a different mode, you will need to reset it, as well as make sure you are in object (not snap) mode.
- PR> call snapclean
The Red 4K has been generously funded by the National Science Foundation, The Ohio State University, Dartmouth College, the University of Michigan and Ohio University.
- LBNL 4K Manual
- MDM4K Manual
- MDM Observatory
- MDM Page at Ohio State
Return to: [
OSMOS Home Page |
MDM Observatory |
Ohio State Astronomy
Updated: 2013 Nov 3 [pm]