Commissioning of the new CCD15 for FIES
On 30/9/2016 we installed the new CCD15 onto FIES. New vacuum ducting
was installed to allow for pumping from the front-room while the CCD
is on FIES. The new CCD comes with a new controller, and the current
detector+controller setup for FIES is now very similar to that of
ALFOSC+CCD14, with similar readout times and characteristics.
CCD15:
CCD15 is a deep-depleted CCD from E2V, with fringe suppression,
with, in comparison to CCD13, improved QE in the red, and somewhat less QE bluewards of 400nm.
The pixels are 15 microns large, and consequently the CCD is about 10%
larger than the previous one. This allows to sample further into the
infrared part of the echellogram, and we now reach from 3700 to 9100
Angstrom for the high-res fiber (F4), while for the other fibers we
sample few hundreds of Angstroms less into the IR. For the previous
CCD the sampled range was 3700-7300 Angstrom.
With CCD15 there are small gaps in wavelength coverage between the reddest few orders, i.e.
for wavelengths larger than 8300 Angstrom.
The high-res fiber's range extends to 9100 Angstrom only if you are
happy to extract the reddest order that is partially clipped by the
edge of the CCD. It is clipped at the center of the blaze, away from
the blaze the curvature brings it back onto the CCD. See the table
below for wave ranges sampled by the individual fibers.
fiber | wave range | wave range including
additional clipped order |
F1 low res | 3620 - 8580 | 3620 - 8700 |
F3 med res | 3640 - 8840 | 3640 - 8980 |
F4 high res | 3640 - 8980 | 3640 - 9110 |
Controller and main read-out characteristics:
The new controller allows for faster readout (50 sec in
default read speed, versus 90 s for CCD13), while the best achievable
read-out-noise (3.3 e-) is a bit worse than for the old CCD13 (2.8
e-).
For CCD15 with the default amplifier B and default read speed (108 kpix/s)
we now have: RON=3.3 e-, gain=0.142 ADU/e- , linear to within 0.5%
between count levels of 1000 to 160000 e-.
Bias level and structure:
Although there is little high-order structure in the bias level we
have found there is a variable slope in the Y direction. This slope
can be seen in and modelled for the wide X underscan region.
The typical bias level is around 10000 ADU.
For B amplifier, the pixels in the high-X overscan region are affected
by bias structure ramping up to the right-most pixels. These are
closest to the read-out register B.
Overexposure:
The new controller reads into 4-byte pixel values, which means that
the 65000 ADU counting limit does not exist any longer. The pixels can be exposed
upto the full well linearity limit, which is about 160000 e- or 1.1M ADU.
When overexposing more than the linearity limit, we see three features appearing,
listed below in order of increasing overexposure
- horizontal banding of the background (bias) level, resulting in
bands of CCD rows with count levels depressed upto a few hundred ADU.
This horizontal extent of the banding is minimal when using the slowest read speed.
The vertical extent of the banding depends linearly on the overexposure level (or exposure time).
- vertical blooming
- remnants in subsequent images.
Overexposure should not happen on stellar frames, unless stars brighter than 4th magnitude
are exposed for a long time. The halogen flat-field exposures are kept well within
the linearity limit when using easyhalo and fies-calibs.
The bright red Argon lines in the ThAr exposures will always lead to
overexposure, even when exposing to the standard level
using easythar. For this reason we have chosen to build into
the expose and dark commands a cleaning procedure
(remove_remnants) that clears the CCD in only 3.5 seconds
before the requested exposure. This procedure only removes the remnants in
the image following the overexposed one. This procedure does not get rid
of the horizontal banding and vertical blooming in the overexposed image itself.
Cosmic rays:
As the deep-depleted silicon is more prone to pick up cosmics, we
advise our users to consider to split up long exposures (e.g. longer
than 20 minutes) in to two or three shorter-exposed ones, in order to
better account for the cosmic ray hits. Not only are there more hits,
they also appear with longer streaks on the CCD.
Echellogram orientation:
With the default amplifier B, the read-out orientation of the
echellogram is the same as for the previous CCD:
blue is on the left and at the top.
We have 100 pix X underscan, 50 pix X overscan, 38 pix Y overscan.
Furthermore the first and last few rows are affected by 'edge effects'
and may be stripped from the useful CCD region.
CCD15 PSF:
For the high-res fiber the ThAr lines have FWHM=2.0 pixel in the
spectral direction over a large portion of the CCD, resulting in a
spectral resolution of R>65000.
In the cross-dispersion direction the FWHM of the echellogram-order
profiles is larger than 4 pixels on the blue side of the CCD going
down to FWHM=3.6 pixels for the reddest orders.
Fringes:
The excellent fringe suppression of the new CCD results in that we do not find
any fringes short of 8000 Angstrom, and have a peak-to-peak fringe
amplitude around 5% at 9000 Angstrom.
FIEStool:
Our reduction tool is being upgraded to handle the new CCD features.
Until then the online FIEStool is switched off. We will run all
FIES frames obtained so far with the new CCD through FIEStool once
the updated version is available.
John Telting