Quality Control and
Data Processing

HC PLOTS
UVB gain:
VIS
NIR
AGC
UVB non-linearity:
VIS
NIR
AGC
UVB counts in brightest flat
VIS
NIR
AGC
QC1 database (advanced users): browse | plot

To measure the detector parameters for UVB and VIS (and AGC) a sequence of image flat field pairs is observed for the standard readout modes (100k/1pt/hg and 400k/1pt/lg). An appropriate set of bias frames is included in the template. For NIR on-off Echelle flat field pairs are observed instead. Due to the inhomogeneous illumination in the NIR we specify a window within one of the brightest orders to determine gain and linearity. This means, however, that also the bad pixel map is limited to that small window. Detector monitoring data have been taken regularly only since April 2010 (no 2x2 data for UVB and VIS before November 2010 ).

QC1_parameters

FITS key	QC1 database: table, name	definition	class*	HC_plot**	more docu
QC.GAIN	xshooter_detmon..gain	measured gain (e-/ADU; inverse of CONAD, but the same quantity as DET.OUT1.CONAD)	HC		[docuSys coming]
*Class: KPI - instrument performance; HC - instrument health; CAL - calibration quality; ENG - engineering parameter **There might be more than one.

Trending

For every pair of flat fields with identical exposure times within the stack of frames the difference of the average of the bias-corrected (off-corrected) frames is compared to the difference in rms between the bias/off frame and the flat field difference. This comparison is limited to the illuminated parts of the detectors. The factor relating the two quantities is the gain in e-/ADU. For method=PTC (Photon Transfer Curve) a linear relation is fit to these quantities for the various exposure times.

For UVB, VIS and AGC, using imaging flats, the full chip is analyzed, for the NIR arm with spectroscopic Echelle flats, one rectangle region within a central illuminated order is analyzed. Flats with more than 40000 ADU are rejected for the photon transfer curve.

Scoring&thresholds Gain

Relaxed thresholds are used, since the accuracy of the gain is usually dominated by the stability of the lamp (within a pair of flats and within the full sequence) and the coarse sampling of the exposure time.

History

Date	event
2010-02-22	change from xshooter_lingain to detmon_opt_lg ( UVB / VIS ) and detmon_ir_lg ( NIR ) recipes
2010-11-20	set --autocorr=TRUE for NIR data (slight changes in gain and conad)
2013-08-01	the thresholds for the linearity have been moved closer to 65335 ADU (saturation level) in order the measure better the non-linearity. Drawback is that the gain is based also on flats with more counts, generating a larger scatter in the gain values. A smaller slope in the PTC results in higher gain values.
2015-02-06	the new detmon recipe allows for raw frame filtering for linearity and gain separately. All flats with > 40000 ADU are no longer used for the gain determination.

Algorithm Gain

The photon transfer method is used.

QC1_parameters

FITS key	QC1 database: table, name	definition	class*	HC_plot**	more docu
QC.LIN.EFF	xshooter_detmon..non_linear	relative deviation from linearity at the flux level given in flux_non_linear	HC		[docuSys coming]
*Class: KPI - instrument performance; HC - instrument health; CAL - calibration quality; ENG - engineering parameter **There might be more than one.

Trending

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

The generic detector monitoring pipeline is used. For the processing, pipeline parameters are tuned for each detector (UVB, VIS, AGC. NIR) and each read mode separately.

Scoring&thresholds Linearity

Relaxed thresholds are used, since the accuracy of the effective non-linearity (the deviation from linearity at 65000 ADU) is usually dominated by the the coarse sampling of the exposure time, the choice of the flat with the largest number of counts not yet saturated.

History

Date	event
2010-02-22	change from xshooter_lingain to detmon_opt_lg ( UVB / VIS ) and detmon_ir_lg ( NIR ) recipes
2010-11-20	set --autocorr=TRUE for NIR data (slight changes in gain and conad)
Before_2013-08-01	UVB,VIS: the whole detector was used for the analysis and the effective non-linearity was given at LIN_REF = 50000 ADU.
Since_2013-08-01	UVB,VIS: a smaller window size of X=1500:1600, Y=1700:1800 pixels is used. For 1x2 binning and 2x2 binning corresponding smaller pixel window sizes are used. The rejection maximum ADU in flat frames was increased to ~ 65000 ADU, meaning more flats with more counts are taken into account, (also flats that might saturate outside the analysis window), and the non-linearity is reported for LIN_REF = 65535 ADU. Drawback is that the gain is based also on flats with more counts, generating a larger scatter in the gain values.
Before_2013-08-01	NIR: the effective non-linearity was given at 10000 ADU. A small window within a single echelle order of X=871:1065, Y=709:734 was used.
Since_2013-08-01	NIR: the non-linearity is given at 42000 ADU.
2015-02-06	the new detmon recipe allows for raw frame filtering for linearity and gain separately. All flats with > 40000 ADU are no longer used for the gain determination.

Algorithm Linearity

The accumulated counts in the flat is fit to the exposure time with a polynomial. The relative deviation from the linearity at a configured counts level (~65000 ADU, close the the saturation level of 65355 ADU) is retrieved.

QC1_parameters

FITS key	QC1 database: table, name	definition	class*	HC_plot**	more docu
QC.COUNTS.MAX	xshooter_detmon..max_count	maximum counts used for linearity estimate	HC		[docuSys coming]
*Class: KPI - instrument performance; HC - instrument health; CAL - calibration quality; ENG - engineering parameter **There might be more than one.

Trending

For each detmon sequence, we monitor the counts of the brightest non-rejected flat used for the analysis of the the detector linearity

Scoring&thresholds Counts in brightest flat

Relaxed thresholds are used, since the accuracy of the gain is usually dominated by the stability of the lamp (within a pair of flats and within the full sequence) and the coarse sampling of the exposure time.

History

Date	event
2010-02-22	change from xshooter_lingain to detmon_opt_lg ( UVB / VIS ) and detmon_ir_lg ( NIR ) recipes
2010-11-20	set --autocorr=TRUE for NIR data (slight changes in gain and conad)
2013-08-01	the thresholds for the linearity have been moved closer to 65335 ADU (saturation level) in order the measure better the non-linearity. Drawback is that the gain is based also on flats with more counts, generating a larger scatter in the gain values.
2015-02-06	the new detmon recipe allows for raw frame filtering for linearity and gain separately. All flats with > 40000 ADU are no longer used for the gain determination.

Algorithm Counts in brightest flat

The median counts of the brightest non-rejected flat is retrieved.