GLAS requires two wavefront sensor heads: one to observe the Natural Guide Star (NGS WFS, see Fig. 8) and one to observe the Laser Guide Star via a Shack Hartmann lenslet array (LGS WFS, see Fig. 9). The LGS will be quite bright (Mv9-10) and the current CCD39 WFS will serve well. It will, however, need to be physically moved and placed downstream of the Pockels shutter. Conveniently, the current NAOMI optical bench, which is located in a temperature controlled Nasmyth laboratory, has a spare corner where the LGS arm optics can be situated. The laser light is intercepted and redirected to this area of the bench using a narrowband notch filter tuned precisely to the laser. The wavelength range close to the laser line will not be available for science. The requirements of the second wavefront sensor needed by GLAS, the NGS WFS, are rather demanding and it seemed marginal that a conventional detector could reach the required Mv17 magnitude limit. A new head has therefore been built using a Peltier cooled L3 CCD60 and in May 2005 was mounted in the position previously occupied by the CCD39 destined to become the LGS WFS. This new head has already been tested on-sky in June 2005, not in its final role as a tip-tilt sensor but as a SH sensor, effectively emulating the behaviour of the original
CCD39. The SH image included in Fig. 1. was taken with this head It is hoped that, prior to the GLAS implementation, we can gain advantages from the L3 head in this role, and models predict a 1-2 magnitude improvement in guide star limit. The optics inside the WFS unit allow beam-switching between the master and slave WFS ports, so changing between the CCD39 and the CCD60 will take only a few minutes. This adds a useful level of redundancy if problems are encountered with the L3 head. The CCD60 only has a single output and this acts as a bottleneck preventing the full CCD39 (quadrant readout) pixel bandwidth being achieved. Using it with GLAS the CCD60 will be much faster since it only needs to read a single small window around the NGS. Three readout modes will be offered: 8^8, 16x16 and 32x32 pixel windows with a plate scale of 0.25" per pixel. The respective readout speeds through an SDSUII controller are 300, 200 and 100Hz. To achieve these speeds it was necessary to use the Modulo addressing mode of the controller DSP. Here, the waveform tables are first located at page boundaries. The DSP Rn register, which acts as a pointer to the next waveform-state word is then set to the start of the table and the corresponding Mn register set to the length of the table-1. When the post-incremented Rn pointer reaches the end of the waveform table it automatically loops back to the start to clock the next pixel. This gives a considerable time saving over the more conventional approach of JSR'ing to a subroutine for each pixel read or skipped.
A custom clock card has been designed and built as a PCB to generate the multiplication clock voltage. It fits into a vacant slot within the SDSU and connects via an edge connector to the standard SDSU clock and video cards. The clock amplitude can be programmed using one of the video board DACS for precise tuning of L3 gain.
Figure 9. The slave head to be used as the LGS WFS.
Was this article helpful?