[ ESO ]
HAWKI Quality Control:

CAL | HC | refs | QC
Trending & QC1
   Detector: monitoring
Data Packages
Data Management
QC links:
hawki_detlin :: HAWK-I_QC_DETLIN
Daily Health Checks:    
QC1 hawki_detlin database (advanced users): browse | plot
   Click on CURRENT to see the last trending (Health Check).
   Click on HISTORY to see the historical evolution of the trending.

top Introduction

This tutorial provides information for the the Detector: monitoring Health Check trending.
The Detector: monitoring trending is based on the measurement of QC1 parameters of Detector Linearity calibration data obtained with a dedicated Health Check OB executed approximately monthly.
The Detector Linearity OB consits of 10 sets of data - each set is one DARK and two Imaging FLAT FIELD frames taken with the same DIT. The DIT is gradulally increased from set to set to increase the flat flux level

The data are obtained with the Ks filter using the Nasmyth shutter as a source: its flux is then dependent on the ambient temperature as expected from the Planck law.

As of 2016-09-28 only the effective linearity is monitored. The conversion factor (conad/gain) measurements were found unreliable. The corresponding HC plot was taken off-line untill template and/or pipeline recipe are improved.


QC1 parameters

  • effective non-linearity (QC1 database table hawki_detlin, column qc_lin_eff, product keyword QC.LIN.EFF)

    The effective non-linearity is the difference between the polynomial flux and the one predicted by the linear fit at user-defined flux level, normalized with the polynomial flux.

  • conversion factor: conad/gain [ ADU/e- ] (QC1 database table hawki_detlin, column qc_conad, product keyword QC.CONAD)

    For every pair of ON frames obtained with identical DET.DIT (in Ks filter using the Nasmyth shutter as a source), the difference of the average of the OFF-corrected frames is compared to the difference in rms between the OFF and the ON frames. The factor relating the two quantities is conad in ADU/e- or gain in e-/ADU. For method=PTC (Photon Tranfer Curve) a linear relation is fit to these quantities for the various DET.DITs.

    2016-08-30: It was noticed by Alain Smette that the template delivering the data uses NDSAMPLES (the number of times the detector is read within a DIT) and NDITs that are different depending on the DIT. The higher DIT, i.e. the higher flux, the higher NDSAMPLES. As the relation of the variance of a frame with the intensity depends on NDSAMPLES and NDIT, the gain determination is incorrect. It can still be monitored for trending purpose though. Further work on template and/or pipeline recipe design is to follow. In the meantime (starting from data on 2016-04-27), the recipe (detmon_ir_lg) parameters were set to exclude high flux frames from gain determination:

    • --filter=2500 (following the detector DET.SETLEVEL parameter)
    • --gain_threshold=1500


top History


mid-September 2014: indication of problem with non-linearity in all chips (followed in PPRS-05910); it was fixed and data taken on 2014-10-10 scored "green" again.

August 2015: all parameters nominal after GRAAL installation

2016-09-28: analysis of the detector monitoring data sets showed that the gain measurements cannot give reliable estimates. The QC_GAIN monitoring HC plot is taken off-line untill template and/or pipeline recipe is improved.