Previous: A Basic Description
Up: General Principles of Observing with CCDs
Next: Limitations, Extra Observations, and the First Stages of Analysis
Previous Page: Saturation
Next Page: Limitations, Extra Observations, and the First Stages of Analysis

Operational requirements

CCDs require electronics generating electrode biases, clocking waveforms, low-noise amplification of weak signals, A/D conversion, buffering and digital storage. They also require temperature-controlled environments mountable at different telescope foci for direct imaging, and at spectrograph foci for use as spectroscopic detectors. The details of how operational systems meet these demands differ, but the major elements are similar. To illustrate these, Fig 4 shows the outline for RGO-built systems now in use at the Anglo-Australian Telescope and at telescopes of the Isaac Newton Group, La Palma.

The cryostat or dewar provides the cooled environment for the CCD. A liquid-nitrogen cryostat designed at RGO can be used in either downward or upward-looking modes, for examples in direct imaging at prime or Cassegrain foci. It has a hold time of some 12 hours so that a single filling of N lasts through the observing night. The CCD itself is mounted behind an anti-reflection coated quartz window and on a copper block with which it makes good thermal contact and which has the temperature- sensing element attached to it. Resistive heating controlled by a feedback loop maintains the temperature to +0.05 about the optimum operating temperature of 150 . This circuitry is mounted on the dewar, as is circuitry to sample and pre-amplify the output signal.

A local driver-box sits within 1 metre of the cryostat, holding the remaining circuitry for which proximity to the CCD is important for optimum performance: the A/D converter and buffer, clock driver circuitry, and telemetry circuitry to monitor the status of the CCD, its environment and its electronics.

The camera controller is a purpose-built microcomputer which determines the basic sequencing of all camera operations. It provides the timing for the clock-driver circuitry, and generates commands for signal sampling. This provides for a system of maximum flexibility - reprogramming the microprocessor allows different CCDs to be used, for instance to accommodate the new generation of large chips, and it also allows different ways of reading out the chip, involving on-chip binning. Before considering this, first consider the standard readout sequence which the microprocessor controls. It runs as follows: the observer requests a run of N seconds to be initiated, the controller:

1. Calls for two erase cycles of the chip,
2. Opens the shutter,
3. Closes the shutter after N seconds,
4. Clocks out, samples, and stores the individual pixel responses as described in section 1.

The erase cycles consist of clocking out the (unexposed) CCD to remove any charge which has built up due to dark current (section 1) or cosmic rays (section 3). This standard sequence reads the CCD out pixel-by-pixel. With the camera controller, it is easy to change the clocking routines via software to provide on-chip binning in which the summed charge from several adjacent pixels is read out as a single charge signal. Spatial resolution is correspondingly reduced, but this may not matter, for instance in direct imaging with heavy oversampling, or in spectroscopy if spatial resolution perpendicular to the slit can be sacrificed. On-chip binning causes the summed charge signal from the adjacent pixels to be measured with the readout noise corresponding to a single sample, and if the signal is dominated by readout noise rather than shot (photon-statistic) noise, the improvement in signal-to-noise is a factor of n, for the binning of n pixels. Of course post-readout binning can be done via software at the analysis stage; but signal-to-noise only improves by n in this instance. A further feature offered by the camera-controlling microprocessor is the ability to opimize chip performance through software changes, timing and the form of clocking signals being set by the microprocessor. All these changes are implemented by different versions of the microprocessor program, which is loaded or downloaded through the interface shown in Fig 4.

Figure 4 CCD camera systems built at the RGO for the Anglo-Australian Telescope and for the telescopes of the Isaac Newton group on La Palma. The block diagram indicates that the camera may be placed at Prime or Cassegrain focus, for imaging at the former and for imaging or as a spectrograph detector at the latter.

This interface and the other modules of Fig 4 are of no particular interest here. They are standard, and demonstrate that the system is built to be used at observatories where telescope/instrument control and data storage media already exist. Stand-alone (portable) systems may have camera control which is essentially hard-wired, dedicated interfaces, little monitoring circuitry, and control and data-acquisition via a single microprocessor. Two features do remain essential - the accurate temperature control for the CCD itself, and a large storage medium if on-line monitoring of images is required. In this regard, each image must be digitized to > 12 bits to maintain the dynamic range for which CCDs are so valuable. Each frame from the small (400 x 500 format) CCDs thus occupies some 0.3 megabytes in a 16-bit medium. Large CCDs will produce images each of 5 megabytes or more.