The following list contains problem issues relevant for ISAAC data reduction up to mid 2008, which may have impact on the quality of science data. The page has not been updated since mid 2008. For further references see Paranal supported ISAAC web pages.

Odd-even column effect

P67: The odd-even column effect consists of variations of the signal between the columns of the array. This effect may be present in some frames. The effect is different in each quadrant and slightly more evident in the lower left one. Before the second week of May 2001 it was usually seen only on frames with signal > 10000 ADU. Starting the second week of May, the effect has been observed in frames with ~3000 ADUs. The effect can be eliminated software wise by means of Fourier Transform. It is recommended to treat each quadrant individually. The relative difference between even and odd columns is a VERY sensitive function of the flux, but it does not scale linearly. Therefore after sky subtraction, some residuals may still be present. An eclipse recipe is being coded and tested right now. It will be made available to the community as soon as possible. As of 24/07/2001 the read-out speed of the SW array has been slow down by a factor of two. Since that night the odd-even column effect has vanished.

P68: The odd-even column effect reappears for p68 (1 Oct 2001 - 14 Oct 2001) until the first intervention. While it is not present until Dec. 24, It reappears again on Dec 25 and exceeds the 1% level at the end of 2001. The odd-even column effect, as a non-linear detector response varies in time for each detector quadrant in a different manner . In Jan 2002 the odd even column effect vanished again. The residual gain difference, which is not flux dependent is less than 6%%.

Response structure of the SW imaging

Since 010408 the QE of the SW array decreased by about 4% for all pixels within x -gt 190 and 88 -lt y -lt 920. This effect can be eliminate by simply flat fielding IS_SIJC frames. This persistent structure faded during p67 and is no longer visible during p68.

Startrace Problem

In p67, medium resolution spectroscopic products like IS_SSCM and IS_SSSM suffer from an insufficient correction of the spectrum shape distortion. This effect becomes relevant for good seeing conditions ( -lt 0.6'' ) and slit_0.3_tilted. As a result the spectrum is not properly co-added. Startrace frames are taken now several times a period and co-adding should work properly.

Dispersion solution for Z (0.9 micron) and SZ (1.06) micron

For SW spectroscopy the sw_arc recipe fails for grating LR and filters Z and SZ, simply because the Xe, Ar arc line catalog used for the ISAAC pipeline does not contain sufficient lines for that wavelength range. The sw_arc products are nevertheless distributed, since they contain beside the invalid dispersion coefficients also the coefficients to correct the optical distortion in dispersion direction. The science and standard star spectra are reduced, using the sw_arc products to correct for the optical distortion, but they are not used for the wavelength calibration. The dispersion solution is either given by the built-in optical model or derived by using the sky lines if available. For any other wavelength range using filters J, SH and SK the dispersion solution is derived using the arc frames. More details on the SW_SSAL product page.

Dispersion solution for SK (2.2 micron)

The sw_arc recipe fails or converges to the wrong dispersion solution for grating MR filter SK and wavelength around 2.2 micron for the same reason. Since the arc recipe derives beside the dispersion solution the x-component of the optical distortion map, the product is used for later steps of the data reduction and is delivered with the ServiveMode packages, although the dispersion solution is not correct.

Optical distortion solution for SK (1.98 micron)

The sw-arc recipe fails also for grating MR filter SK and wavelength around 1.98 micron. For this setting all detected arcs lines are on one half of the detector and the algorithm exceeds the extrapolation limit for the optical correction function (WH 020517).

Spectrum extraction

For SW spectroscopy, spectrum extraction fails and no tfits file is provided in cases where the recipe cannot find the two negative spectra in the reduced science frames. Further details are given on the SW_SSCL product page.

Consequences of the Intervention Oct 2001, SW Detector temperature variations

Before the first intervention of P68 in October 2001 (52188 < MJD-OBS < 52196) , the SW detector temperature was stable at 60 +- 0.002 K (according to the HIERARCH ESO INS TEMP-DETSW keyword). When ISAAC came again in operation at MJD=52203, the SW detector temperature was still cooling down, meaning TEMP-DETSW was variable for 52203 < MJD-OBS < 52205. For 52206 <= MJD-OBS <= 52207 there was a stable plateau at TEMP-DETSW 65.3+-0.002 K . After MJD-OBS = 52210, TEMP-DETSW is again stable at 63+-0.002 K.

As a consequence the dark level and the RON visible is highly variable in the period after the intervention. The zeropoints also show the same trend, meaning the detector efficiency as part of the whole instrument throughput changed. After MJD-OBS = 52210, when the SW detector was operating again at a stable temperature of TEMP-DETSW = 63+-0.002 K, qc parameters dark, RON, and zeropoint relaxed also, but now, as due to the new operation temperature at new levels. (WH, 011221).

Pattern in spectroscopic flats

In a sporadic manner spectroscopic master flats show a large scale pattern, related to a overfitting of the y-dependence of the flat flux level (perpendicular to the dispersion direction). From March 2002 on the -f 1 option is used for the sw_spflat recipe, meaning a polynomial of order zero (just a constant) is used to normalized the flat and the y-dependence of the product flat is the one of the raw flat frames. (WH 020304)

Fits header problems

Between Nov 2001 and Jan 2001 not all acquisition images taken with the SW MoveToSlit and MoveToPixel template are stored, since the fits headers are missing. These files are not distributed with the package, since they cannot be classified. The fits headers will be reconstructed as far as possible and will be ingested into the ESO archive, from where they can be downloaded in case these files are relevant for the scientific evaluation.

Due to some operational problems, some of the twilight flat stacks consist of frames where the TPL.START is not filled. Beside the twilight flat stacks also imaging darks (IS_SIDK) as well as jittered images (IS_SIJC) are affected. The missing TPL.START fortunately does not impact raw data classification, calibration association, data reduction and packing. These files are still part of the package; they can easily identified, from the attached GEN_INFO/<RUN.ID>_packsum.txt file where missing keyword entries are identified as 'oOo'. (WH 020201)

We found missing HIERARCH ESO INS keyword trees in some of the LW imaging (LIJC) frames. Following nights are affected 020111, 020112 and 020113 (WH 020204).

At least in Feb 2002, the temperature sensors are not properly logged and stored in the fits headers, meaning the HIERACH ESO INS TEMP-DETSW and others are given as 0.000 (WH 020301).

Due to a faulty software installation on Paranal all frames of 020328 have corrupt fits headers. The following branches are missing: TPL, OBS, DPR and OCS. Furthermore some core keywords like the world coordinate system are missing as well. Information on the publication policy of these data is given elsewhere. (WH 020422).

P69: From the beginning of P69 we include LW calibration frames in our QC process

Dispersion solution for second order arcs in LW

Variations of the row y=512 and y=1 of Hawaii using DoubleCorr

Consequences of the Instrument Intervention 2002-06-23 2002-07-05

Contamination of the LW-arm: Since the intervention, a (light/heat) contamination is observed in all Darks of the LW-arm. The contamination is correlated with DIT. It occurs at the left hand side and in the top area of the array. As a consequence the statistical noise composed by read-noise and shot-noise is no longer a constant over the array. This only affects the Aladdin SW J spectroscopy, as this is the only science observation that is read-noise limited. The contamination is stable up to the end of July, but becomes variable thereafter.

The SW-arm detector temperature control is not able to keep the detector temperature within 62+-0.5 K. Fluctuation of more than 1.5 K are observed after the intervention. On 2002-08-14 the operational temperature of the SW-array is modified from T= 62 K to T = 64 K. On 2002-08-26 the operational temperature of the SW-array is modified from T= 64 K to T = 65 K. As a consequence SW-arm Dark current levels for most of the DITs have changed as well.

The odd-even-column effect of the LW-arm Aladdin array is below 0.005, but exceeded 0.01 on 2002-07-29 (to be continued). (WH 020819)

On 2002-08-01 the collimator got stuck and was moved to a compromise position. Therefore ISAAC image quality is limited by the current fixed collimator position optimized for the broad band filters. The point spread function of SW-arm spectra of point sources can show up two peaks, meaning spectra are not longer fully focused. (WH 020917)

Ongoing problems concerning the alignment of the LW-arm and SW-arm objective wheels the reduction scheme of LW chopping spectra is changed. Instead of the day-time calibration arc frames, the sky lines are used for wavelength calibration. In case the sky lines cannot be used, the physical model is used for for wavelength calibration. Note that LW chopping spectra are not subject of optical distortion correction using startrace frames and arc line curvature. LW chopping spectra products are also not flat fielded. (WH 020920)

Tilted telluric absorption lines can occur in spectra of point sources. They can appear for both gratings LR and MR and both arms LW and SW. The tilt is about 0.187 pixels per row, while sky emission lines are not tilted, apart from the optical distortion which is several orders smaller. Since P70 we find more and more examples, where the tilt can become curved, meaning the dispersion is no longer a constant for all of the rows. The effect is not understood and the pipeline does not correct this feature. (WH 020925)

Persistence can occur in dark frames of DIT greater than 100 sec. The amount is usually a few counts. They occur when due to operational reasons the last science frame of the night contains bright sources which saturate the array. The proceeding day-time calibrations are slightly contaminated by a persistence effect. (WH 021023)

LW sky flats products (IS_LISF) are generated using the twflat recipe. The calibration template takes several exposures at three different air mass, but the flux range is not large enough to allow for proper zero level offset determination. (WH 021105)

After a detector intervention the arc persistence disappeared. On 28 Nov 2002, The read speed of the Hawaii changed from 6 to 12 to minimize the odd-even column effect. Min. DIT has increased from 1.77 to 3.55 s (WH 021202)

There is since Nov 2002 a big dust grain on the SW-detector. By chance it is on a row where the SW-arm telluric star is jittered. As a consequence all SW-arm stars taken with the SW-spec_cal_StandardStar template suffer from the obscuration by the patch at pixel 585,630
There is a small patch also on the LW-arm detector and standard stars by chance are positioned by the template also on that patch as well. (WH 030103)

The jitter recipe showed recently a high rate of frames which could not be correlated when applied to SW-arm imaging stacks. We found that the recent occurrence of the big bad pixel area appears in the list of suitable correlation objects and is used to correlate the image stack. To overcome these problems, we modified the jitter ini file. We use from now on (Dec 2002) 'AutoThreshold = 10.0;' as a peak detection sigma threshold instead of the default value of 2.0. So far we allow only for objets brighter than 20 sigma above the background. Another solution, would be to set 'ObjectSource = file ;' and write the x and y pixel of the desired correlation object into a file called 'objects.in'. Jitter will take the user-define position to search for the correlation object. (WH 030110)

The big bad pixel area is gone. All science images taken after 2003-02-20 are pipeline processed with the default 'AutoThreshold = 2.0' parameter again.

There occurred a misalignment in the SW-arm from 2003-02-14 to 2003-02-18 showing strong gradients in the SW-arm twilight flats in the x-direction. There occurred temperature fluctuations of the SW-arm array in the same time interval. A detailed analysis has shown, that the science observations are useful but suffer from the variable slope of the twflats and the unknown spatial distribution of the zeropoints over the array.

The odd-even-column-effect reappeared on 2003-08-06. Usually, the odd columns are usually 1.0002 times higher and the even columns are 0.9998 times lower than the total flux of a twflat image, at least for three quadrants. The upper right (UR) quadrant shows the inverse relation. This behavior is independent on the flux and can be flatfielded. Since 2003-08-06 the odd/even columns are 1.02 times higher/lower and the relation is strongly dependent on the flux, hence flat fielding becomes difficult. On 2003-08-08, the detector read speed was reduced by a factor of 2 and as a consequence the OECE reduced from 2% down to 1%. (As a further consequence of the lower read speed, MINDIT, the smallest possible discrete integration time is now double as large at 3.5451 sec).
From several data reduction tests we conclude: For SW imaging the 'isaacp oddeven' recipe reduces the OECE nearly completely, when applied to the science raw frames before submitted to the jitter recipe. ServiceMode packages with SW imaging data taken after 2003-08-05 will be reduced in that manner. In addition the oddeven column values derived from twflat raw frames are distributed as well.
For SW spectroscopy, the 'isaacp oddeven' recipe does not reduce the OECE perfectly for for bright telluric standard stars as well as for the usually fainter science target, when applied to the raw science input frames. Using spectroscopic flats, corrected for the OECE does not improve the S/N in the spectra, but enhances the OECE in the reduced spectra. The quality of SW spectroscopy data is reduced by the odd-even column effect. ServiceMode packages with SW spectroscopy data taken after 2003-08-05 will be reduced in that manner, meaning the isaacp oddeven recipe is applied to raw science frames before they are submitted to the spjitter recipe.Uncorrected spectroscopic flats are used for the data reduction. (WH 030926)

Tilted telluric features Since 2003-08-14 the shearing problem reoccurred, meaning absorption lines and telluric features imprinted in a spectrum appear tilted with respect to the dispersion direction. This wavelength shift across the PSF of the spectrum is not corrected by the pipeline. The collimator is no longer at a position for minimum tilt of the telluric features in the center of the spectrum. (WH 031010).

The isaacp arc recipe was improved and put in operation for P72. One setting (SW MR 1.71) cannot be handled, since there are only two arc lines detected by the recipe to be used for the image distortion part. The recipe stops without reaching the wavelength calibration part. Science products with this setting are reduced without arc product (= without optical distortion correction in x-direction) and the wavelength calibration is done using the sky lines imprinted in the science raw frames. (WH 2004-03-03).

The oddevencolumn effect decreased 2003, but the read speed of the SW-arm Hawaii array was changed as part of the intervention in early 2004. As a consequence the OECE increased again and became variable. Science raw frames of the SW-arm are furthermore preprocessed by the eclipse oddeven recipe before they are submitted to the jitter/spjitter recipe. (WH 2004-03-03).

Startrace products were recently found to suffer from a too narrow extraction range, resulting in wrong polynomial fits for the curved shape of the raw spectra. All startrace products have been reprocessed using the '-w 40' option, and all spectra of P72 have been reprocesses using the correct startrace product. The corrected products apply to all Service Mode packages delivered after Feb 2004. (WH 2004-03-04)

50Hz noise appears in the SW darks in DoubleCorr read mode for DIT=5 and lasts for about 2 weeks in Feb 2004. The RON increased from 2.5 up to 3.5. 50Hz noise occurs in a sporadic manner for DIT=5 and DIT=2 as well. (WH 2004-03-18).

Persistence can occur in long-DIT SW darks in NonDest readmode. In most cases these are afterglow images of bright point sources taken during the night and amount to a few counts. They cancel out during pipeline data reduction via the double subtraction. (WH 2004-03-18).

The grating calibration has been updated in December 2004. As a consequence the grating offsets (measured central wavelength minus nominal wavelength) is no longer +20 pixel but ~0 pixel. (WH 050104)

The sp_flat recipe flags all pixel of the master flat with less than 0.5 and more than 2.0. The spjitter recipe flags the master calibration flat in the same manner resulting in a too restrictive handling of flat fielding, in case when the master flat is steep in wavelength direction. As a consequence science and std spectra have been cut at both end of the spectra. These thresholds have been updated to 0.01 and 3.0. (WH 2005-01-01)

On 2005-01-08 the collimator position changed, resulting in a broader PSF for the science product spectra. The tilts of the telluric features and the whole science spectra continue. (WH 2005-02-09)

The intervention in April 2005 and a further intervention in May 2005 resulted in slightly changed instrumental calibrations.For data reduction both interventions define 'breakpoint', meaning the complete cascade for a science OB should not cross the intervention periods. (WH 050512)

ISAAC SW-arm science raw fits headers taken between 2005-03-023 and 2005-05- 27T03:30 with ADA.POSANG != 0 contain wrong word coordinate system (WCS) values. The world coordinates are expressed by the CD matrix since 2005-03-23 and the correct values can be reconstructed from other fits header key values. CDELT1=0.148/3600; CDELT2=CDELT1; (old WCS : CROTA1=CROTA2=ADA.POSANG);
CD1_1=CDELT1*cos(ADA.POSANG * pi/180);
CD1_2=-CDELT2*sin(ADA.POSANG * pi/180);
CD2_1=CDELT1*sin(ADA.POSANG * pi/180);
CD2_2=CDELT2*cos(ADA.POSANG * pi/180)

2005-07-01 ... 2005-08-01 is a period of strong 50Hz noise in SW-arm DoubleCorr readmode darks. Darks with DIT=2, 3.5 and 10 sec were impact.

2005-09-01 ... 2005-10-01 the strong 50Hz noise continues.

2005-10-19 A temperature increase by about 100K in the vessel caused an increase of the Aladdin dark current and a decrease of the Hawaii dark current by a few 10 counts. After two days the regular dark currents were retained again. start 2005-10-19T22:00 ; peak at 2005-10-20T11:00. No further side effects have been detected.

++++++

2006-01-01 A migration of the ISAAC pipeline from eclipse to CPL occurred. As a consequence twflat products and startrace products are not compatible between both pipelines. Therefore run.ids with data taken during P76 in 2005 and 2006 will receive two packages. The 2005 data package will contain calibration and science products generated by the eclipse pipeline. The 2006 data packages contain calibration and science products generated by the CPL pipeline, although some calibrations like the semi-annual startrace were probably taken in 2005. Note that the ISAAC master calibrations in the ESO master calibration archive taken in 2005 and earlier are generated by eclipse pipeline and the 2006 data are generated by CPL pipeline. The mixture should be avoided.
As mentioned on the pipeline pages, SW-arm spectra are no longer wavelength calibrated using the day-time arc lamp calibrations but by using the embedded air glow sky emission lines.
As in 2005, the SW-arm photometric standard stars are reduced without being corrected for detector (much smaller) non-linearity of the Hawaii array. LW-arm photometric standard stars are reduced by correcting the detector non-linearity of the Aladdin array.
A new naming scheme is used for calibrations and science products from 2006 on.
The package contains for a given science spectrum the corresponding telluric standard star observed with the same instrumental setup. A further telluric standard star observation frequently acquired using the 2arcsec slit, is no longer distributed within the service mode data package. It can be downloaded from the public ESO archive if required. For SW-arm spectroscopy using the LR grating, only the Argon lamp arc frame products are distributed, the Xenon lamp products are no longer part of the package but are available on the ESO web.

2006-03-01 Several unusual grating shifts of more than 17 pixel have been observed in the second half of Febr 2006.

2006-06-06 The isaac_spc_arc recipe fails due to "not enough lines found" for the following settings

• SW-MR-1.2222-slit_0.8
• SW-MR-1.2726-slit_0.8
• SW-MR-1.3418-slit_0.8
• SW-MR-1.6625-slit_0.3_tilted
• SW-MR-1.7375-slit_0.3_tilted
• SW-MR-1.7089-slit_0.3_tilted
• SW-MR-1.71-slit_1
• SW-MR-1.7375-slit_0.3_tilted
• SW-MR-2.25-slit_0.6_tilted
• SW-MR-2.212-slit_0.3_tilted
• SW-MR-2.31-slit1
• LW-MR-3.09-slit_0.3_tilted-SH
• LW-MR-3.55-slit_0.6_tilted-SL
• LW-MR-3.9925-slit_0.3_tilted-SL (the J+Block filter works fine)
• LW-LR-3.55-slit_0.3_tilted-SL

2006-09-18 The default value of the new --arc_kappa command line parameter (control of the threshold for arc line detection) of the isaac_spc_arc recipe is 0.33. Using --arc_kappa=0.05 for the LW-MR-3.55 setting, recipe doesn't fail anymore.

2007-04-16 SW-arm broad-band filter twilight flats are acquired, with the telescope in operation. Point sources in raw frames are eleminated by the piepline recipe. SW-arm narrow band twilight flats are acquired very early in the evening. The telescope is not tracking/pointing, hence star trails are dominant features, in raw frames and the pipeline recipe can eleminate star trails down to about 1-2% in the master twilight flat product, but not below that.

2007-08-17 SZ-band wavelength dispersion using skylines sometimes failes (the QC.XCORR is below 30%) because the spectroscopic flat introduces artificial sky line beyond the band defined by the narrow filter curve. To ovecome this, the default recipe parameters of the isaac_spc_flat recipe --thresholds=0.1,3.0 can put to less tolerant values (closer to 1.0); e.g. --thresholds=0.4,1.8 meaning the resuling MASTER_SP_FLAT will be flagged for all normalized intensities below 0.1 and above 1.4. Using this master flat with the modified masked area will improve the dispersion solution of the isaac_spc_jitter recipe.

2007-08-17 Spectroscopic flats of the setting LW LR M 4.75 use window read out with NAXIS1=576 and NAXIS2=1024, however the fits header keys HERARCH ESO DET WIN NX|NY and DET WIN STARTX|STARTY are wrong. The correct values should be:
DET.WIN.NX = 576 ( == NAXIS1)
DET.WIN.NY = 1024 ( == NAXIS2)
DET.WIN.STARTX = 225.000
DET.WIN.STARTY = 1.000

2007-09-11 There were two instrument interventions, one from 2007-06-26 to 2007-07-01 and a second one from 2007-07-19 to 2007-07-30. On 2007-08-11 an electric power cut and UPS failure lead to an uncontrolled pressure and temperature increase of the Hawaii detector. After the recovery the dark current, the reset anomaly are reduced considerably; the readout-noise remained for both operational read modes DoubleCorr and NonDest. The characteristic jig-saw pattern of the DoubleCorr Darks is gone. Parameters like structural noise and hpt pixel number changed accordingly. The Aladdin detector is not affected. SW-arm science data taken before/after 2007-08-11 should not be reduced using calibrations taken after/before 2007-08-11.

2007-09-18 The dark current (DIT) relation in NonDest readmode has changed considerably. For DIT=900 the dark current measured from individual raw frames shows a decrease in a series of consecutive raw dark frames, starting with about 100 ADU and decreasing to about 30 ADU with an additional scatter term of about 10-30 ADU from raw dark frame to raw dark frame. The same behaviour can be measured in the background of science spectra, where the background is measured between the sky lines of MR grating science raw frames.

From 2008-02-03 on lasting up to 2008-05-01 the ISAAC Hawaii detector showed a feature called the odd-even-column effect. The odd-even-column-effect (hereafter OECE) is described on the qc page for twilight flats based on the experience collected during the last occurrence in 2003. The OECE is a non-linearity feature. It shows a long-term variability of the order of a week and sporadic amplitude changes of the order of 10 ADU, which can occur several times a day. From 2008-03 on linearity frames are taken several times a week instead of twice a year to improve the calibration plan. The quality of all ISAAC calibration and science data acquired with the SW-arm Hawaii detector suffer from the OECE, and the pipeline products of the service mode data package are not corrected for the OECE.

Around 2008-05-01, after the intervention, the Hawaii detector cosmetic degraded, due to a new piece of something in the central part of the detector. Since 2008-05-26 the debris became larger in size. On 2008-06-12 a further large bad pixel area appeared. The evolution is documented by four twilight flat master calibration frames: flat_080413K.png, flat_080504K.png, flat_080526J.png and flat_080612J.png. The following impact has been observed: Telluric standard stars can match the large size bad pixel areas. Telluric correction is not possible in this wavelength range.

On 2011-05-18, the SW Hawaii detector went blanc (no signal) in the Upper Left quadrant. See the quick report . After intervention the detector returned to regular operation in July 2011.