Focus again looks reasonable throughout the night. There are individual images which appear around 100 microns out of focus, and a period at the very end of the night
with a more systematic defocus.
The focus looks good during the period of the DES mini 1a survey. During community time there were two images burned to allow the focus to settle, indicated by the blue arrows. These two looked to have reasonable focus already. However, there were three other images, circles, taken after large slews where the focus was indeed off by 50-100 microns. I conclude that there may be good reason to burn an image after large slews. Lastly, I mark the set of images taken for the program "Fast Cadence Imaging", with one second z band images with slews between every image. The mean focus is about right, but there do seem to be larger deviations here.
A histogram of the donut measured focus is shown here, for the 53 images taken for the Mini 1a survey. The RMS of the distribution is 12microns, essentially equal to the requirement. This demonstrates that the necessary focus control to meet the SV requirement is possible, although it should be noted that this is for a series of images having no large slews.
Most of the images on this night were out of focus. The exact cause is still under investigation, but seems to be due to the tweaks (ie. the values the AOS determines for how much the hexapod should move) not being implemented by the hexapod.
The focus generally looks fine, except after 150330 where the values are generally out of focus by 100 microns. Conceivably this is due to the same thing that caused problems on 11/12.
The focus looks reasonable, outside of a few images. The blue arrows point to burned images for focus. The focus is a little worse for these images, and there are also several other images which are a bit of focus, presumably from large slews.
The focus looks reasonable. Some of the community images rotated quickly through different filters, and the focus does seem to change here by a small fixed amount.