[ ESO ]
SINFONI Quality Control:
DARK frames

CAL | HC | refs | QC
Trending & QC1
Data Packages
Data Management
QC links:
general information | parameters trended
DARK parameters
QC1 database (advanced users): browse | plot
   Click on CURRENT to see the current trending (Health Check).
   Click on HISTORY to see the historical evolution of the trending.

top General Information

Dark frames are used to monitor the health and technical performance of the detector. For data processing they are used to correct the pixel-to-pixel dark current variations and to flag those pixels with very high dark currents (hot pixels). Dark frames are routinely obtained for each night that SINFONI is in operation. They come in stacks of 3 raw frames, and generally mirror the DIT's of the science observations. The SINFONI pipeline processes each group of 3 dark's into a master dark and a hot pixel map.

master DARK

A SINFONI master dark as computed from 3 input raw frames (here DIT = 300 seconds). The circular feature of bad pixels near the center of the frame (made of up ~1000 pixels with elevated counts) is clearly visible and intrinsic to all SINFONI images.

hot pixel map

A SINFONI hot pixel map. A value of 0 indicates the hot pixels (as shown in black in the adjacent image); while normal pixels are set to a value of 1 (as shown in white in the adjacent image).


top DARK parameters trended

QC1 plots

The following properties of SINFONI DARK's are trended. The QC1 plots are divided into 6 html pages for each of the following DITS: 300, 10, 60, 600, and 900 seconds. Each page (DIT) monitors the following plots:

  • plot 1: median dark level of master bias (ADU)
  • plot 2: standard deviation of median dark level (ADU)
  • plot 3: read-out noise (ron) (ADU)
  • plot 4: fixed-patttern noise (fpn) (ADU)
  • plot 5: total number of pixels with a high dark current (hot pixels)
  • plot 6: ratio of odd/even column strips (for slitlet #25)


QC1 parameters (trended for DITS = 10, 60, 300, 600, and 900 seconds)

Parameter Table: Name (QC1 database) Description of Procedure
Plot 1: median dark level

sinfoni_dark: qc_darkmed_ave

The median level is computed from all pixels in the master dark frame .
Plot 2: standard deviation of median level sinfoni_dark: qc_darkmed_stdev The standard deviation is computed over all pixels in the master dark frame .
Plot 3: read out noise sinfoni_dark: qc_ron1 and qc_ron2

The RON is computed over the entire detector and is monitored as a function of time and DIT. Two consecutive frames are subtracted from each other and the median standard deviation is determined from a number of sample areas and then normalized to DET.NDIT = 1. The RON is computed from the difference of the first two raw darks (qc_ron1) and the second pair of raw darks (qc_ron2). Since the very first dark of a series can sometimes have an elevated count level, qc_ron2 is used in the trending plots.

Plot 4: fixed pattern noise sinfoni_dark: qc_dark_fpn The fixed pattern noise is computed by fitting a Gaussian to the histogram to the master dark. The FPN is simply the standard deviation (s) of this Gaussian fit. The FPN should scale linearly with the number of counts.
Plot 5: number of hot pixels sinfoni_dark: qc_bpm_nbadpix The number of pixels having an intensity greater than a pipeline-defined threshold (default threshold is 10sigma).
Plot 6: odd/even column ratio sinfoni_dark: qc_oes1, qc_oes2, qc_oes3

SINFONI frames sporadically show vertical stripes near slitlet #25 (see the images below). This potential feature is monitored by determining the median of all even columns (med(even)) and all odd columns (med(odd)) near slitlet #25 and then computing:

oes25 = med(odd) - med(even) / 0.5 * (med(odd) + med(even))

The odd/even ratios are computed for each of the 3 raw dark frames (hence the values qc_oes1, 2, and 3).


QC Reports for DARK Frames

QC Report 1

QC Report:
UL: the mean collapsed columns of the current master dark (blue), the reference master dark (red), and the difference between the reference and current master darks (black).
UC: the mean collapsed rows of the current master dark (blue), the reference master dark (red), and the difference between the reference and current master darks (black).
UR: a logarithmic histogram of the current master dark (blue), the reference master dark (red), and the difference between the two (black). The difference histogram should symmetrically straddle the zero axis.
LL: an image of the current master dark.
LC: an image of the current hot pixel map. Hotpixels are assigned a value of 0 (black), while normal pixels have a value of 1 (white).

LR: logarithmic histograms of the difference between adjacent raw dark frames. The second raw dark minus the first raw dark (purple), and the third raw dark minus the second raw dark (turquois).


Trending and Issues

Sporadically, the SINFONI detector will show an electronic effect whereby vertical stripes are readily visible. This occurs near slitlet number 25 (near columns 1922 to 1985) with even numbered columns showing elevated counts (of a few ADU) above their neighbour odd numbered columns. This feature is relatively rare and is best seen in frames with longer (> 200 second) DIT's, but can be found in the dark, standard star, psf star, and science frames. It is only trended in the DARK frames using the odd/even column ratio. Clearly, frames not showing the odd/even vertical stripes have oes25 values near 0.0.

master DARK with odd/even stripes

A SINFONI master dark (DIT = 300 sec.) showing the odd/even column electronic feature in the zoom near slitlet #25.


The SINFONI detector is afflicted by persistence if an overly bright source (generally having counts > 15 000 ADU) is observed. This effect decays with time, but if a DARK is obtained after observing such a bright source remnant flux will be visible.

master DARK with persistence

A SINFONI master dark (DIT = 600 sec.) showing persistence from a bright star observed in an earlier exposure.

Sometimes the source of persistence can be an over-exposed flat field image. In that case, the SINFONI master dark, instead of displaying residual source spectra, will show larger scale image residuals. For example, between March 20 and 25, 2011 an over-exposed flat produced the following effect on subsequent master darks:

master DARK with persistence

A SINFONI master dark (DIT = 600 sec.) showing persistence effects from a previously over-exposed flat field frame.

For seriously over-exposed data, the decay time until SINFONI DARKs return to their normal levels can take as long as several days. This persistence decay time can be seen in a plot of the flux ratio of slitlet17/slitlet32 as a function of time:

persistence decay



SINFONI DARKs have been very stable over the current lifetime of the instrument. This can best be seen the the full history DARK trending plots shown here: FULL DARK history