Problems can be related to

• the instrument itself, resulting in low performance raw frames
• to the pipeline processing, resulting in imperfect pipeline products
• to a not optimal observing strategy, which may impact the scientific expectation

Here we report on problems related to the instrument alone.

Start of operations is on April 01 2015.

From 2015-04-01 to 2015-07: IRDIS optical distortion calibrations using the pinhole mask are often blurred resulting in a lower contrast between the pinhole peaks and the background. This can lead to a failure of the pipeline recipe (V 0.15.0) . In some cases are more tolerant threshold parameter helps. The cause of the low image quality is not yet fully understood. Turbulence inside the instrument and a non-optimal shape of the deformable mirror are discussed. (WH 2015-06-30)

IRDIS optical distortion raw frame (a bad frame), showing the part of the detector illuminated by a blurred and contrastless pinhole image. The corresponding good raw frame is given here .

Blurred pinhole mask image :

Since the pinhole mask for IFS and IRDIS is the same, IFS optical distortion calibrations can be blurred in the same way as described for IRDIS (WH 2015-11-09).

IFS optical distortion raw frame (a bad frame), showing the slitlets illuminated by a blurred and contrastless pinhole image. The corresponding good raw frame is given here .

Persistence:

From 2015-04-01 until 2017-05, IFS dark frames with DIT=1.650726 (and other of higher DIT) show persistence, in particular the first low-DIT dark exposure of the night is affected, others acquired later during the day-time calibration sequence are not affected. The pattern looks like an imprint of the IFU flat. The persistence is monitored in a dedicated HC plot. It is recommented to use IFS background calibrations for the mentioned period. The following image shows an affected DIT=1.650726 dark, the unaffected DARK with the same setup has been subtracted.

Arc-shaped optical ghost:

Since 2016-10-01 all IFS detector flats of the YJH prism show a large prominent optical ghost in the upper part of the imaging flat (~5% more counts). The ifs detector flat pipeline recipe generates four products: the differential flat field (DFF), the large scale flat field, the pre-amplifier map and the bad pixel map (see here for examples), of which only the large scale flat and the pre-amplier flat are contaminated by the ghost but these products are not used by the pipeline. The DFF product used in further steps of the calibration cascade does not contain the optical ghost. After an alignment on 2016-10-11 the ghost vanished (WH 2016-10-11)

This figure shows the ratio between the large scale product from 2016-09-08 and that from 2016-10-02. The prominent pattern has canceled out, the remaining structure (fringe-like ringes and lines) is below 0.5% and the bright ghost shows an intensity of about ~5% of the flat field intensity.

Imaging flat format move:

Since 2016-09-15 un-illuminated areas in the lower left and right corner of the IFS detector flat (all setups) continuously increase, meaning the vignetted area becomes larger. The fixed pattern (prominent features) does not move as shown in the product ratio image above. After an alignmernt on 2016-10-11 the format was adjusted (WH 2016-10-11)

dark contamination

From 2015-04-01 to 2015-11-01: The ZIMPOL CCDs accumulate counts when not being read out since a longer time (several hours). The first ZIMPOL calibration in the day-time calibration batch is the imaging dark, of which the first of two frames is contaminated, while the second one is clean and corresponds to a proper measurement of the dark current. In case the calibration is acquired in cube mode, only the first layer of the data cube is affected, while all further subintegrations are not affected. The clean dark current corresponds to a value of ~3 ADU/100sec, the contaminated and useless value is about ~10 ADU/100sec. The issue is solved since 2015-12-01. (WH 2016-01-10)

shutter problem

Since 2015-09-09 some but not all ZIMPOL lamp flats as well as ZIMPOL modemeff calibrations and optical distortion pinhole images calibrations suffer from the following effect:
The first or more planes (the first detector read outs) of the raw frame data cube suffer from having not lamp counts or considerably less counts than the remaining readouts in the data cube.

The Fig. shows the median counts of the V-band imaging flats, resolved per plane of the raw frame data cube. The first four planes are not illiminated with a bias level of ~1000 ADU. Only planes 6 to 10 are fully illuminated. Blue points are from channel 1 (CAM1, Callas) and the red points are from channel 2 (CAM2, Bartoli).

Since the number of unilluminated planes varies it is not known how the data reduction recipe behaves in this situation. Master flats and modemeff frames are only processed for reasons to monitor the behaviour via the qc_plane_rms QC parameter, but products are discarded thereafter again. The corresponding health check plots are taken off the main page and are available here (WH 2016-04-14): flats imaging mode

The problem is solved in 2016-06.