One of the challenges in building a large CCD mosaic is focusing the detectors simultaneously when cooled. The CCD42-90 package is ideally suited for this purpose. The CCD is epoxied to an Invar block. Three precision shims are attached to the bottom of this block, lapped by the manufacturer so they define a plane 14 mm from the CCD surface. We then faced the task of providing a flat surface on which to bolt them. We chose to mount them on a 6 mm thick Invar plate, suspended around the edge by six titanium flexures from a warm support ring (see Fig. 1). These flexures provide the necessary thermal isolation. The tilt and axial position of the cold plate was set by adjusting the location of the warm ring with a set of spring loaded screws. Once the location was set, precision spacers were machined and bolted into place. The mounted plate was measured to be flat to ~10^m P-V before any CCDs were installed. Note that our original design used a molybdenum plate, which provides much better thermal conductivity. However, this plate arrived from the vendor broken in two pieces, so we chose to replace it with the less brittle material.
To verify the focal plane flatness we relied on focus measurements on the sky. Our current measurements indicate that that there is roughly 100 ^m P-V of deviation from the best focus across the plate. This is larger than expected, but has a completely negligible effect on the image quality. An image taken in 0.4 arcsec seeing shows essentially no variation in FWHM across the entire Megacam field. There are no discontinuities from one CCD to the next, showing that the CCD packages were manufactured well.
Was this article helpful?