PF-QHY Camera

The PF-QHY camera (centre-right) mounted on-axis at the prime focus of the WHT during pre-commissioning tests of the new corrector for WEAVE. The second QHY camera (seen off-axis at left) is not offered for imaging observations.

UPDATE: 33 service programs were submitted from which 18 programs were completed (55%) during two months between 10 May and 12 July 2021 (alternative with some tests).

During the integration of WEAVE at the WHT, a second call of opportunity has arisen to offer the community some service-mode imaging time with a camera. This PF-QHY camera is a test imaging camera mounted behind the new prime-focus corrector for WEAVE. It is available for service-mode science observations during a short period in 2021. The PF-QHY has been characterised and tested by ING, but users should understand that it is not a fully-commissioned common-user instrument.

The QHY camera (model QHY600L) is based on a back-illuminated CMOS detector (Sony IMX455), with 9576×6388 3.8-μm pixels, giving a field of view of 10.7'×7.1', at a scale of 0.067"/pixel. We recommend binning 4x4 (0.267"/pixel).

At sky position angle PA = 0, North is left and East is down, when the image is displayed in the usual way in DS9, thus the DS9 needs rotation +270 deg counter-clockwise to match the normal sky orientation.

The PF-QHY field is extremely flat across its whole field, thanks to the excellent optics of the new prime field corrector. This fact could be probed at 20210531 on sky and is demonstrated in the following Scamp distortion plot (sky PA=0, DS9 rotate 270) which shows scale variations less than 0.0001 "/pixel from centre to corners of the TWFC2 camera. In sky position angle PA=0, the long axis of the camera is along the DEC direction to within an alignment error of about 0.5 deg.

The filter wheel holds up to seven 50-mm circular filters. Initially, the filters will be: Sloan u, Sloan g, Sloan r, Sloan i, Sloan z, the broad-band ('clear') luminance filter L-1, and a blank (used for darks). Enquiries about the filters should be addressed to the instrument specialist.

The QHY600 quantum efficiency is close to 90% around 550 nm. The net throughput of the PF-QHY camera is probably not very different from that of the old PFIP prime-focus imager with the EEV detector, so a rough idea of the expected SNR can be obtained by running the ING's SIGNAL exposure time calculator for the previous WHT prime focus camera (PFIP) whose pixel size and characteristics are similar to PF-QHY binned 4x4.

The PF-QHY readout noise is 11 electrons for binning 4x4 and 6 electrons for binning 2x2.

The detector can be windowed, and binning options 4x4 and 2x2 are available (binning 1x1 and 3x3 are not offered). For most science aims observed in average or good seeing (0.5-1.0"), we recommend binning 4x4 (0.267 "/pixel). Requests for 2x2 binning (0.134 "/pixel) should be justified in the service proposal. The camera saturates around 60k counts but is linear to 55k counts.

The following plot describe the linearity of the PF-QHY camera, based on recent ING WFC2 dome flat tests corrected by bias, using gain 0.37 electrons/ADU (ampgain 30 as listed by the company), binning 4x4 and a stable lamp. PF-QHY is linear below 55k counts where the camera shows deviations up to 0.3%. The QHYCCD company linearity plot made for gain=0 looks also linear up to 70-80 ke (units not directly comparable with our ING plot).

Based on recent images using no filter, L1, Sloan r, i and Sloan z filters, the PF-QHY camera does not show any fringing patterns.

Darks are strongly recommended to be used for image reduction. Dark images must be taken at night using the blank filter, with the dome closed, no lights, and petals closed. Darks taken during the day with dome closed and lights off show lots of counts and look similar with flats fields - samples: 10s dark taken during day with Sloan i in front of the camera, compared with 20s dark taken at night facing R filter. For longer exposures (above 5 minutes), the dark could generate saturated pixels. To correct them, the PF-QHY deputy Richard Ashley created a dark mask for binning 4x4 where hot pixels are marked with a '1' other regular pixels with a '0'. This mask was generated from a set of 3x600s exposures, flagging all pixels that showed 10xsigma above the median ADU pixel value in all three exposures.

The full-frame readout time when binning 4x4 is 9 sec. If a multiple exposure is requested, the readout time for exposures after the first is 3.4 sec. For 2x2 binning, the times are slightly larger: 10 sec for the first exposure, and 6 sec for subsequent exposures. Windowing does not (currently) save much readout time - e.g. if the detector is windowed 700x675, the readout time with 4x4 binning reduces by only 1 sec (readout 2.4 sec). In future, it's possible that faster readout modes will be commissioned, but this depends on how long the camera is available, and on how much interest in faster readout is expressed by potential observers.

Autoguiding of observations through the Sloan filters is provided by a standard ING autoguider head mounted at a radius 0.9 deg from the optical axis. While autoguiding is possible for the Sloan u, g, r, i and z filters, due to the fixed focus position of the autoguiding camera we cannot guarantee that all runs can be supported with autoguiding, particularly in sparse regions at heigher galactic latitudes. Observations made through the L1 filter cannot be autoguided. Please also note that, due to differential atmospheric refraction, the large on sky separation of the science camera and the autoguider camera has implications for the accuracy of autoguiding over long runs.

Non-sidereal tracking (but not autoguiding) is also available with good results up to 3 min exposures.

Saturated stars could persist as star-ghosts in the following images, as probed by a test reproduced in the following mosaic. The saturated star in the upper-left image (over a 20 pixel diameter disk at its centre) is followed by 3 successive biases which show the persisting star-ghost (slowly diminishing its brightness in time, by around 100 counts between successive biases). This behavior is normal for most CMOS cameras.

We thank Bernard Karpinski of Modern Astronomy for his help with characterising the cameras and filters.