home \ High Speed Imaging

High Speed Imaging

High Speed Imaging with "Zipper Mode" Readout

The occultation events we are searching for have typical durations of about 200 milliseconds, while the CCDs used in the survey (2048 x 2052 e2V CCD42-40 chips) have readout times of 2.3 seconds. If we want to resolve the occultation events our exposures need to be 200 milliseconds or shorter, and given the readout time of the camera we would have over 90% dead time if we read out in the standard way. We have thus developed a novel method to read out the CCD camera at a higher speed, which we call zipper mode.

In zipper mode readout, we open the camera shutter to start exposing, but it remains open throughout the duration of the run. After an exposure time of 105 milliseconds, a subsection of 76 rows is readout. This corresponds to 1/27th of the imaging area of the chip. Any stars imaged on the bottom 76 rows of the image are read out, while the electrons for all of the other stars in the image are shifted down by 76 rows. This is followed by another 105 millisecond exposure and another readout of 76 rows, and the process is repeated continuously, for up to 90 minutes at a time. The readout operation takes a total of 95 illiseconds, so the exposure time plus readout time is a total of 200 milliseconds, giving a readout cadence of 5 Hz.

This readout mode is illustrated in the animation below. Four stars are in the field of view, and they are color coded for easy identification. After the first readout step, photoelectrons from the black star are readout and digitized, while the photoelectrons for the other stars are shifted into the next rowblock subsection. Then another exposure is taken, and more photoelectrons are collected from each star. Note that after the first time we read out photoelectrons from the blue star, every subsequent image contains an image of all four stars. However, the light from each star was collected at adifferent epoch.

A section of a rowblock is shown in the image above. Note the streaks evident in the columns containing bright stars. These are due to the finite time of 1.25 milliseconds required to readout a single row. While the main exposure time is 1.05 milliseconds, each time a single row is read out we are collecting photons for the 1.25 milliseconds it takes to read it out. The presence of streaks in the images is one of the disadvantages of zipper mode readout. The other disadvantage is that sky background is collected all of the time. In the animation above, note that both sky background and photons from the blue star are collected during the first exposure. After the first readout, sky background is collected in the subsequent image subsections before the image containing the blue star is read out. In this example, we are thus collecting four 200-millisecond exposures of sky background in every rowblock, but only one 105-millisecond exposure for each of the stars. In practice, since the rowblock size is 1/27th of the size of the image, we are collecting twenty-seven 200-millisecond exposures of sky background for every single 105-millisecond exposure of a star. This reduces the signal-to-noise ratio we can achieve with the TAOS system by a factor of more than 50. Nevertheless, this method does in fact allow high speed readout of every star in the image, and while we do lose out in signal-to-noise, we are still able to achieve sufficient photometric accuracy on a large sample of stars and are thus sensitive to the occultation events for which we are searching.

Valid XHTML 1.0 Transitional| Valid CSS!