Detector Specifications

ShARCS detector is an H2RG engineering grade detector read out with a Teledyne SIDECAR ASIC. All rows above row 1457 are bad, as is amplifier number 1 (numbering from zero). The ShARCS imaging area is well outside these regions. All other parameters for the detector pass the requirements for a science-grade H2RG.

The detector can be read out with 1, 4, or 32 amplifiers. The default setting for ShARCS is 32 amplifiers at 100 kHz per pixel.

Minimum readout time for the full detector:
N amps	Time (sec)
1	43
4	11
32	1.455

All exposures are multiples of the readout time. Hence, if you request a 2.0 second exposure with 32 amplifiers, the actual exposure time will be 1.455 seconds. If a 3.0 second exposure is requested, the actual exposure time will be 2.91 seconds. The actual exposure time is recorded in the FITS file header and frames are written in units of DN per coadd.

There is also a window mode that reads out a sub-region of the detector. This mode can only be used with a single amplifier, so in order for this mode to give shorter readout time than reading the full detector with 32 amplifiers, the window size must be smaller than about 360 pixels on a side. The window need not be square.

The first full frame exposure after a windowed frame has an artifact in the bias, and should be discarded. Also, the minimum exposure times are calculated such that they over-predict the time for short window-mode exposures by a few hundredths of a second.

Read Noise

The detector may be read out with Fowler sampling (default) or in up-the-ramp mode. The more reads of the chip reduces the noise, but increases the minimum exposure time. Below is a table of the minimum exposure time and read noise for each number of reads.

N reads	Time (sec)	Read Noise (e-)
1	1.46	21.2
2	2.91	15.0
4	5.82	10.9
8	11.64	8.0
16	23.28	6.4
32	46.55	5.7

Bad Pixels

Bad pixels are 1.6% of the imaging area. A bad pixel map can be found on covert (the data-taking computer) in /data/LabData/badpix_g9_slopes.fits. (A link to the current bad pixel map will be added in the future...)

Crosstalk and Darks

There is about 1% crosstalk between pixels, which is normal for these detectors. Even if bad pixels are flagged and not used in stacked images, it helps to subtract a dark to take out the high pixels next to very bright pixels. Because the hot pixels can saturate in long exposures, it doesn't always work to scale darks. For this reason, observers should take darks that are the same as the exposure times for the science frames. One may also want to break very long exposures into shorter exposures with coadds to synthesize a longer exposure to prevent the hot pixels from saturating and adversely affecting the neighboring pixels.

Linearity

With the default gain setting, the data are linear to about 30,000 DN (about 70,500 e-).

Detector Specifications Summary

Array Size:	1024x1024 (full detector), Imaging area ~600x600 pixels
Pixel Pitch:	18 microns
Pixel Scale:	0.033 arcsec/pixel
FOV:	~20 arcsec
Readnoise:	21 e- RMS with CDS
Readnoise:	6.4 e- RMS with 16 Reads
Full well:	100,000 e-
QE:	85 % over wavelength range 1.0 - 2.5 µm
Wavelength Range:	0.9 - 2.5-microns (with IR dichroic in ShaneAO)
Gain:	2.35 e-/DN