Previous: Signal levels, noise and dynamic range
Up: Introduction to the CCD system
Previous Page: Signal levels, noise and dynamic range
Next Page: Set-up

## Preflashing

Some chips currently in use suffer from a threshold effect such that signal charge below a certain level cannot be transferred within the CCD and therefore cannot be measured. In order to enable weak signals to be measured the CCD is subjected to a brief flash of illumination prior to each exposure. It has been empirically determined that a normal preflash of 40 units is sufficient to give good results for the GEC4 chip (this chip has already been replaced with GEC6). A lesser preflash would not allow weak sources to be correctly detected and a larger preflash increases the Poisson shot noise without significant signal gain.

The normal preflash gives an average signal level of 25 adu above the bias'' level; this corresponds to 25 e, with an associated shot noise 5 e. Thus the total effective readout'' noise (with preflash) is , this is measured as an rms noise of 8 adu.

The preflash illumination is not uniform and increases towards the edges of each row - see Figure 4.5. The level of preflash is however reasonably uniform down the centre of the CCD where a dispersed spectrum would normally lie.

Figure 4.6 shows several 1000 second long-slit exposures of NGC 3310 (a galaxy with a circumnuclear ring of star formation) with GEC4 covering 3650-9700Å . This example shows the various charge traps as dark columns. Cosmic ray events appear as bright spots. The dispersed spectrum of the object is seen falling-off towards the red wavelength end. Some night sky emission lines are seen. These frames have not had the preflash subtracted and so this modulation of the background is also visible. Figure 4.7 shows the corresponding line-plots.

Wed Apr 6 22:56:39 BST 1994