[ ESO ]
VIRCAM Quality Control:
Darks

CAL | HC | refs | QC
QUALITY CONTROL
    HOME
VIRCAM QC
Trending & QC1
   Dark
Pipeline
QC links:
noise parameters | dark level
 
QC PLOTS
  CURRENT HISTORY
noise parameters
dark level
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.


All VIRCAM Health Check Plots follow the same design:

 

Upper Left:

Scored aggregate value.

Average over 16 detector values to monitor changes of the parameter common to all detectors

Upper Right:

Values of all 16 individual detectors

Lower Left:

Scored aggregate value.

RMS over 16 detector values to monitor the dispersion within the sample and changes to only one or a subset of detectors.

Lower Right:

Internal scores of the 16 individual detectors.

 

Dark frames are detector calibration frames to measure the counts generate by the detector during readout, to monitor erratic light linearly increasing with exposure time and to monitor the read out noise. Dark calibrations are acquired with the same DIT as used by the science observations at the end of the night or as a Daily Health Check with a fixed set of DIT (discrete integration time) and NDIT (number of DIT) for monitoring reasons. The following setups are covered by the Health Check:

  • NCORRS=Double, NDIT=1, DIT=1.001100
  • NCORRS=Double, NDIT=1, DIT=120.000000

The following detector setups are monitored with at least one measurement a week:

  • NCORRS=Double, NDIT=2, DIT=5.000000
  • NCORRS=Double, NDIT=6, DIT=10.000000

Only four of the many quality control parameters extracted from dark calibrations are monitored within the VIRCAM health check plot pages and are explained in the following; all other QC parameters are available from the QC1 DB. In short the tags:

  • MED stand for the median dark level derived from the master dark pipeline product,
  • RON12 stands for read out noise derived from the first and the second dark raw frames,
  • RMS stands for the root mean square of the master dark pipeline product
  • STRIPE stands for the rms induced by the horizontal stripes
top dark level

QC1 parameters

parameter QC1 database: table, name procedure
qc_darkmed vircam_dark, qc_darkmed - QC.DARKMED is the median of the MASTER_DARK pipeline product frame. The MASTER_DARK is a clean mean of the five raw DARK input frames.
REFERENCE_DARK parameters vircam_dark, qc_darkdiff_med, qc_darkdiff_rms - For the QC.DARKDIFF_MED, a reference master dark ( PRO.CATG = REFERENCE_DARK ) is subtracted from the current master dark. The median and the robust noise, using MAD, of the difference frame is given


There are some general properties of VIRCAM detector quality parameters. The relation between the dark level and the DIT is given in the following plots.

Each Fig. shows the relation: dark level versus DIT for four detectors aligned in a row. Note the particular reset anomaly of detector #13.

There occurs a further subtle effect: Each NDIT adds about 12 ADU to the dark level. And each NDIT adds 1 ADU to the statistical noise. This means: A raw frame acquired with DIT=10sec and NDIT=5 has 60 ADU more signal and 5 ADU more statistical noise than a raw dark frame with DIT=50 sec and NDIT=1.

Updated 2018-01:
As a consequence: Only for the setup of smallest exposure time of 1.0011 sec and smallest NDIT=1, the measured dark level is unbiased by the thermal radiation slipping through the radiation shield. From the monitoring of the dark current (see related health check plot), a thermal component of ~0.8 ADU/sec is known. The dark level of ~35 ADU for the setup NDIT_DIT=1_120.0 is consistent with the mentioned dark current. Only the 1_1.0011 setup dark level serves as key performance indicator (KPI) for the instrument, as all other setups are contaminated by thermal component.

top noise parameters

QC1 parameters

parameter QC1 database: table, name procedure

darkrms

vircam_dark, qc_darkrms
- The QC.DARKRMS is a measure of RMS obtained from MASTER_DARK frames in ADU. It is the median of the absolute deviations from the median ( = MAD ) times 1.48.
striperms vircam_dark, qc_stripe_rms - Raw dark frames show statistical detector noise plus an additive pattern of horizontal stripes, called coherent controller noise which is varying in scale and amplitude from raw frame to raw frame (~ +- 5-20 ADU). The pattern RMS of the stripe pattern is monitored per detector. The master dark is collapsed in x-direction (parallel to the stripes) and the rms over the median column is retrieved. For details, please see the related section on the VIRCAM problems page.
short term variations vircam_dark, qc_RON12, qc_RON23, qc_RON34, qc_RON45 - subtract two consecutive DARK raw frames and retrieve a robust estimate of the statistical noise (read out noise) via a histogram fit. The QC parameter qc_RON12 is based on the difference of the first two raw frames. The robust statistical noise measured in qc_RON12 is dominated by the incoherent variable horizontal stripe pattern. The QC parameter is hence more sensitive to amplitude of stripe variations than the qc_striperms QC parameter. As the read out noise is composed by the supposed statistical detector noise and the correlated controller noise, the qc_RON12 covers both components.

 

The darkrms QC parameter measures the structural noise or fixed pattern noise in a master dark pipeline product. Only in detector #13 with its strong reset-anomaly, the nearly linear relation between the darkrms and the darkmed parameters ( = between the structural noise and the counts) is established. For all other 15 detectors with negligible reset-anomaly, the darkrms is rather constant within the small range of registered darkmed values.

Figure caption:
Upper left: darkrms QC parameter against darkmed for VIRCAM detector #13. Data are based on VIRCAM master darks taken between 2010-05-03 and 2010-08-03 with NDIT=1.
Upper left: darkrms QC parameter against darkmed for VIRCAM detector #10.
Lower left: darkrms as function of DIT (NDIT=1) for detector #13. As darkrms is linearly related with darkmed, this Fig. shows the dominating reset-anomaly of detector #13.
Lower right. darkrms as function of DIT (NDIT=1) for detector #10 as a representative for other detectors (except #13). The variation of darkrms over DIT is negligible, a consequence of the reset anomaly.

Updated 2018-01: While the statistical detector noise is supposed to be stable the correlated controller noise is not. The VIRCAM pipeline science recipe applies a destriping algorithm, which removes the horizontal stripes in the science product frames, meaning those stripes introduced by the calibrations frames and those coming from the raw science frames. The VIRCAM technical specification allows a read out noise (detector and controller noise) of up to 32 ADU.

History

When the instrument is warmed up and cooled down as part of a scheduled maintenance or an unexpected longer power cut on the site, the detectors show the following pattern: The dark level values and the dark rms values are higher than before the warm-up and decrease slowly in an asymptotic manner to the pre-intervention value. This might take weeks or longer. The strongest effect is seen in detectors #3, #13, and #14. The list of these events is maintained on the plot tutorial for the RESET frames.

top secondary parameters

The QC1 analysis scripts triggered after the pipeline processing generate QC reports (graphical information) and measure some additional parameters of lower priority. The measurement of most of these additional parameters might be discarded after some initial time of operations.

QC1 parameters

parameter QC1 database: table, name procedure
qc_post vircam_dark,
qc_post_man, qc_post_mdm
- evaluation of the histogram (distribution) of the master dark. qc_post_man is the histogram center and qc_post_mdm is the histogram width. The parameters are obtained by fitting a Gaussian to the distribution.
pixel property maps vircam_dark, qc_num_hp, qc_num_cp, qc_num_dp, qc_dev_mean, qc_dev_std

- use the qc_post_man and the qc_post_mdm parameters. All pixel with counts more than qc_post_man + 6 * qc_post_mdm are called hot pixel. All pixel with counts less than qc_post_man - 4 * qc_post_mdm are called cold pixel. Calculate the stdev of the counts from each raw input frame and calculate the histogram of all stdev.The two parameters that describe the histogram of all stdev is given in qc_dev_mean and qc_dev_std. Each pixel that deviates more than 6 times qc_dev_std from qc_dev_mean is called a deviant pixel. The number of hot pixel, cold pixel and deviant pixel is given.

cosmic particle rate vircam_dark, qc_particle_rate - The QC.PARTICLE_RATE is the average number of pixels rejected during frame combination. The number of rejected pixel is decreasing with increasing DIT, meaning the parameter is not an estimate of the cosmic ray hits.
hot pixel vircam_dark,
qc_nhotpix, qc_hotfrac
- QC.NHOTPIX is the number of detected hot pixels. A pixel is considered hot, when more than 2 of 5 raw input frames deviate from the mean. So far what is called qc_nhotpix by the pipeline are actually the number of deviant pixel. The threshold --thresh is a pipeline recipe command line parameter. QC.HOTFRAC is the fraction of hot pixels = QC.NHOTPIX / (2048 * 2048)

 
 
[ESO][Index][Search][Help][News]