[an error occurred while processing this directive]

from 2009-12-09 (P84) sinfoni pipeline version 2.1.0: This pipeline version incorporates a major improvement in the sinfo_rec_jitter recipe. For all grism (J, H, K, and H+K) and for the pixel scales 25 and 100 mas, the data cubes are now corrected for atmospheric dispersion. Depending on the airmass and ADA rotation angle the source centroid can shift from one wavelength end of the cube to the other by several pixels. For the largest wavelength range cube (i.e. for the H+K band) and at the smallest pixel scale, this shift could be as large as 6-8 pixels. This shift is now corrected within sinfo_rec_jitter and is applied to the telluric STD star cubes, the PSF standard star cubes, and to all science cubes. For the 250 mas pixel scale, no correction is applied. This is due to the fact that the shift from atmospheric dispersion is less than 1 pixel at this pixel scale. The atmospheric dispersion correction relies on additional fits tables that provide the size of the correction based on airmass, waveband, and pixel scale. These files are available in the pipeline distribution (2.1.0).

2009-01-01 to PRESENT DATE: Important note for science observations with a large number of input frames:
The sinfoni pipeline coadds its input frames into data cubes in computer memory. Therefore, if your OB contains a large number of input images, or has large offsets, the default parameter that defines the maximum allowed cube size may quickly be exceeded. In general, pipeline versions older than 2.0.8 has the maximum allowable cube size <= 130 x 130 x 21 (X * Y dimension * number of input frames), depending on the amount of RAM available, and on the pipeline parameter configuration. If the science OB is larger than this, will produce a pipeline processing warning and create no product files. To increase the default maximum cube size, edit the parameter:

sinfo.objnod.mosaic-max-size in the pipeline configuration file sinfo_rec_jitter.rc.

The maximum allowed size for the cube mosaic is computed as 100 * sinfo.objnod.mosaic-max-size. Therefore if you projected cube size is 150 x 150 x 50, you will need a sinfo.objnod.mosaic-max-size of at least 11250. For the sinfoni pipeline versions 2.0.8 and newer, the default value of sinfo.objnod.mosaic-max-size has now been set to 14196.

2007-11-13 to PRESENT DATE Occasionally, the SINFONI distortion frames will fail. This is not apparent at the time the OB is executed, but will only become evident when the distortion OB is pipeline processed. This failure manifests itself in fibre traces that are missing in some of the pseudo slits, do not cover the full length of the detector, or are too faint. As a result the pipeline cannot fit the full fibre traces and fails. These faulty distortions are found in roughly 10% of all distortions obtained. These bad distortions are detected by QC and a request is then made to Paranal to repeat them. To prevent users from downloading faulty distortions, the associated raw frames are hidden in the archive.

Failed distortions come in a number of flavours. Here is a representative sample showing the co-addition of all (~75) fibre frames (click to view the full size images):

For comparison, a usable, co-added, distortion should look like the following (click to view the full size image):

2006-07-25 (P77) The SINFONI pipeline corrects for optical distortion, but differential refraction can cause spectrum shears in the final cube. Although differential refraction is not important in the IR, with AO it can become important again. For this reason differential refraction is calculated according to Fillipenko1982, PASP 94, 715.

Table 1 is the repetition of the results of Fillipenko (1982).

Table 2 contains the results for the 1-2.5 micron range, where the shifts are given in arcsec with respect to 1.9 micron (the central wavelength of the H+K grating that provides the largest wavelength range).

At airmass sec z = 1.35 a spectrum taken with the 1.9 H+K grating will show a shift of 0.025 arcsec (= 2 pixel for the 0.025 camera in the final cube-reconstructed image)) in the H band (1.6 micron) and a shift of -0.015 arcsec in the K-band (2.2 micron).
At airmass sec z = 3, we get 0.076 and -0.047 = 0.123 arcsec = 1 pixel in the 0.250 camera and 10 pixel in the 0.025 camera.

In addition we give two trails of the peak position (x and y in pixel along the wavelength) of telluric standard stars with setting 0.025 H+K.

2005-09-15 (P75) Since August 2005, there appear additional hot/cold pixels in the area corresponding to slitlet #18. The additional hot/cold pixels are aligned in columns. The number of hot/cold pixels is correlated with DIT. Long-DIT darks show more hot/cold pixels than short-DIT darks. The hot pixel are no longer visible in lamp flat frames. The detector linearity products show, that the hot/cold pixels are independent on flux (it is just an additional signal, the linearity of the pixel is retained). The pattern varies in time, hence the hot/cold pixels cannot be eliminated via dark frame subtraction.

Hot pixel map showing the normal features (circle and randomly distributed pixels) and the additional new cold/hot pixel columns in the area corresponding to slitlet #18.	Black : Column @1234 (this is one of the hot pixel columns in the area of slitlet #18) of a raw SKY frame from a telluric standard star template on 2005-08-13 (250H). Red : The same column of another SKY frame with the same setting taken in the same night. Note the pixel pattern variations. Blue: Column @1236 of the same frame (outside the hot/cold pixel columns)	Raw telluric standard star frame in the central region covering also the area of slitlet #18 with the hot/cold pixels.

2005-06-15 (P75) The transmission of SINFONI is not a constant over the FOV. It varies up to 20%. There is a minimum for slitlets #8, #9, and #10. Empty SKY frames have been pipeline processed as science OBJECT frames (flat fielded, corrected for optical distortion ..., but no sky subtraction) and cube reconstructed. The table shows the median collapsed cubes. The sky emission lines serve as a spatially homogeneous illumination source. The transmission as function of the FOV is not corrected by the pipeline.

250 J	250 H	250 K	250 H+K
100 J	100 H	100 K	100 H+K
25 J	25 H	25 K	25 H+K (with odd/even signal in slitlet #25)

2005-06-02 (P75) Not a problem, but more a limitation: Science products (PRO.CATG = COADD_OBJ) are only generated when the cross section of the final coadded cube does not exceed a certain number of pixel. The upper limit is at about 128 x 128. Mosaic is not supported by the pipeline.

2005-05-23 (P75) The flat fields taken after 2005-04-01 show a flux reduction in slitlet #32 of about 40% correlated with a partial flux increase in slitlet #1 by 5-10%. Both slitlets build the lowermost and uppermost slide in the FOV.

2005-05-23 (P75) On 2005-04-04, several telluric standard stars saturated the chip. Hence a persistent signal occurred for about a week, mostly detectable in long DIT (e.g 300sec, 600sec, 900sec) dark frames. The persistence appears as several vertical stripes in all the raw frames, most striking in high DIT frames. The stripes are partly flagged as hot pixels in the hot pixel map. The largest gradient of fading occurred in the first three days after this event. All science raw frames taken up to 3-6 days after this event are more or less impact. The typical excess in a DIT=600sec dark in ADU is

date	excess [ADU/pixel]
2005-03-22	0
2005-04-04	9
2005-04-05	5
2005-04-06	4
2005-04-07	2
2005-04-08	1

The following images demonstrate the temporal persistence of the array using DARKs with DIT=600 sec (cuts=20..30):

2005-03-22	2005-04-04	2005-04-05	2005-04-06	2005-04-07	2005-04-08

2005-04-28 (P75) Science raw frames taken before 2005-03-30 have been reduced using pipeline version 1.0. Reconstructed and coadded science product cubes suffer of two inconsistencies. a) slitlet #1 of individual cubes is not expanded from one spaxel to two pixel. b) the plane size of the cube is fixed to 80 x 80 pixels, hence large offset stacks will not fit completely in the fixed size cube (pipeline version 1.0.1).

2005-04-25 (P75) From science stacks without embedded SKY frames, that are pointing to the clear sky (no source), we derived typical sky subtraction residuals of about 60 ADU in the stacked product frames and about 200 ADU in the coadded product frames for DIT=600 sec. These residuals are due to sky emission line variations from exposure to exposure.

2005-04-25 (P75) The pipeline version 1.0.1 is used to reduce stacks of science frames. The dark products are not part of the cascade. Science data stacks with embedded SKY frames are well reduced. Science stacks without embedded SKY frames are reduced using one of the three offered reduction methods. We use a pair-wise subtraction of consecutive OBJECT frames to eliminate the SKY, before co-adding. Hence science products contain the negative spectra as well (similar to ISAAC LW chopping). If the user-specified jitter offsets are less than or of the order of the PSF, this method becomes critical.

2005-04-25 (P75) : The dark frames, with DIT>200 show in a sporadic manner a few ADU higher signal of the even columns in the area of slitlet #25, (~ column 1922 ... 1985). The same behavior has been found for telluric standard star observations, psf calibrations and science frames. As en electronic effect, it may occur in all kind of exposures of the SINFONI array. The effect is monitored in DARK frames. Slitlet #25 can therefore contaminated in a sporadic manner by this effect.

The following two images show the upper right part of two raw dark frames (DIT=300 sec), once without and once with high signal effect
t.
The effect is trended in raw dark frames. The following two images show the difference of two sky frames of which one is contaminated by the sporadic oddeven signal in slitlet #25. The other image shows the impact of the oddeven signal in a cube-reconstructed and co-added science product. The median collapsed product is shown. There was no source in the FOV on the pointing position.

2005-03-13 (P75) : Start of operations