Quality Control: LINE frames

SINFONI Quality Control:
gain & linearity

QUALITY CONTROL HOME

SINFONI QC
SINFONI QC HOME
TOC
Trending & QC1
general info
trending & QC1

dark
HCAO
lamp flat
wavelength
gain & linearity
pupil
distortion

SINFONI Health Checks
Pipeline
general info
science data
calib data
calib recipes
assoc map
product names
Service Mode
general info
directories
list files
problems, issues
Data Management
general info
FAQs
QC links:
QC home
Cal Checker
Health Checks
Reference Frames

QC1 database
Paranal autrep database
(ESO internal)
[best viewing with style sheets and javascript enabled]

general information | parameters trended


QC PLOTS
LINEARITY:	CURRENT	HISTORY
detector gain
number of non-linear pixels
linearity coefficient C0
linearity coefficient C1
linearity coefficient C2

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

Linearity frames are obtained in order to monitor the pixel-by-pixel response properties of the detector as a function of flux intensity and to determine when the detector becomes non-linear. The linearity calibration is made about once per week in each of the four gratings and consist of a series of flat lamp pairs (lamp-on and lamp-off) of increasing exposure lengths with DIT's from 1 to 45 seconds. Pipeline distortion products included the BP_COEFF cube which contains the non-linear fit coefficients for each pixel, the LIN_DET_INFO table (containing the coefficients of the non-linear fit), the GAIN_INFO table (containing the detector gain values), and BP_MAP_NL (the non-linear bad pixel map image). All products are quality controlled and trended.

The three planes of the SINFONI non-linear polynomial coefficients cube (BP_COEFF). From left to right are the offset CO (dark), the linear part C1 (flat), and the quadratic term C2.

top WAVE parameters trended

QC1 plots

The QC1 plots for the SINFONI linearity show the:

detector gain for the J, H, K, and H+K gratings (in e-/ADU)
total number of non-linear pixels for the J, H, K, and H+K gratings
linearity fit coefficients for the J, H, K, and H+K gratings

The first two panels (most important) of the LINE QC1 plots.

These four panels show the gain (e-/ADU) for each of the four gratings (J, H, K, and H+K). Linearities are always made at the 25 mas pixel scale.

These four panels show the total number of non-linear pixels found for each of the four gratings (J, H, K, and H+K). Non-linear pixels generally are less than 0.2%.

QC1 parameters trended

Page	Parameter	Table: Name (QC1 database)	Description of Procedure
1	detector gain	sinfoni_line: qc.gain	SINFONI pipeline: Pairs of consecutive lamp-on flats (F¹_on and F²_on) and lamp-off flats (F¹_off and F²_off) are selected from the input data frames and their differences are computed: diff_on = F¹_on - F²_on and diff_off = F¹_off - F²_off. The mean of each frame (F¹_on, F¹_off, . . . ) and the standard deviation of the differences (s(diff_on) and s(diff_off)) are computed. The gain is then given by:
2	number of non-linear pixels	sinfoni_line: qc.bpm.nbadpix	SINFONI pipeline: For each pixel a median flux level as a function of DIT is known and is compared to a linear fit (as given below to the coefficients C0 and C1). Those pixels that diverge from the linear fit by a given threshold are flagged as non-linear. The non-linear bad pixel map (BAD_MAP_NL) is combined with the hot and cold pixel maps to create a master bad pixel map (MASTER_BP_MAP).
3 to 5	linearity fit coefficients: CO, C1, and C2	sinfoni_line: qc.bpm.lin0.med, qc.bpm.lin1.med, & qc.bpm.lin2.med	SINFONI pipeline: Each DIT in a linearity series is made up of four frames (two lamp-on and two lamp-off frames). The median of each image is computed and the lamp-on and lamp-off frames are subtracted: The mean of all N med_dit_i is computed as: Finally, a parabolic fit of the product of (DIT_i * mean), as a function of med_dit_i, is made with the coefficients C0, C1, and C2.

Trending and Issues

The overall lamp intensity of the flat used to make the linearities has sometimes been set too high (> 40, 000 counts), with the result that an excessive number of pixels enter the non-linear range. This can become apparent in large jump in the number of non-linear pixels detected (see above figure), the number of bad pixels visible in the master bad pixel map, and even in stacked wavelength lamp (seen as areas masked pixels), hence affecting the wavelength calibration solution.

LINE failure due to lamp intensity set too high

Frames indicative of a linearity flat lamp that has been set too high.

Left panel: A master bad pixel map showing an excessive number of non-linear pixels.

Right panel: A stacked arc lamp image with a large number of masked pixels evident.

This effect was observed during the period between 2006-11-24 and 2007-02-28 in the J-band and was discovered by the trending of the lamp intensities and number of non-linear pixels. Users dealing with this over-exposed lamp flats can circumvent this problem by either removing the raw frames with the highest value DIT's from the linearity pipeline processing or, even more effectively, by increasing the pipeline parameter bp_lin-thresh_sigma_fct from its default (10) to 15.

History

SINFONI is currently in the process of moving from its current linearity pipeline recipe to that of the detector monitoring pipeline recipe. This new recipe will have a number of different parameters that will change the trending that is monitored.