There are three focus values to consider: the internal focus of ESI collimator onto the slit or focal plane, the telescope focus controlled by the z motion of the Keck II secondary, and the focus of the guide camera onto the slit or focal plane.

In general the collimator focus is set by imaging the slit or a grid of pinholes through the instrument. Dome illumination is better than internal lamp illumination for this purpose. Thedefault collimator focus values should suffice for most purposes, but individual observers may want to tweak the focus to their own taste.

The guider is focused on the slits or pinhole grid in a similar fashion. Normally this is rather tedious for the observer, requiring the OA to save frames at different focus values. The nominal values are probably adequate for most guiding and acquisition.

There are two ways to focus the telescope. One is called "autofoc" and is done using a sequence of guider images. The secondary is run through a set of z positions, the image size calculated, and a hyperbola fit to the resulting data. Best focus is the minimum of the fitted curve. Note that this relies on the TV camera being accurately focused onto the slit or focal plane!

The other technique, called MAlign ("em-a-line"), is less seeing-dependent than autofoc and can also measure higher order terms than simply the best z-position of the secondary. In particular it can calculate and eliminate coma by tilting the secondary mirror appropriately. However, it requires some discussion...

8.1. ESI Coma Control

The "best focus" can mean different things in different modes. Partly because ESI is an off-axis instrument, it has some inherent coma. The telescope, which is a true Ritchey-Chretien at f/15, theoretically has zero coma. At a given rotator angle it is possible to tilt the secondary mirror to provide telescope coma equal and opposite (to first order) to the ESI instrumental coma, cancelling it out. This would give better images in imaging mode, at that one rotator angle.

However, you would probably not want to do this for spectroscopy. For that you would like to get the best possible image quality at the slit, which means that the telescope should have zero coma. This will maximize slit throughput. You are then left with the internal coma slightly affecting the instrumental image quality.

These subtleties are not addressed in ESI, partly because the behaviour of the secondary depends on the instrument's observing mode, and this information is not passed between the two systems. There is a separate Coma Control Tool that the OAs can use to configure between imaging and spectroscopic tilts, or change from the imaging tilt at one rotator angle to the imaging tilt at a second angle.

In any case, MAlign is generally used only for imaging. Select the filter and collimator focus you want, and the OA will select a star and start the process. The mirror segments are spread apart a known amount to produce thirty-six individual images of the star. After an image is taken the segments are spread out the other way, producing a similar-looking pattern. Another image is taken. The difference in radial offsets of the segment images in the two images translates into a distance from the optimum focus. In addition, the deviation of the three "rings" of segments from perfect concentricity gives a measure of coma.

Either autofoc or MAlign takes about five or six minutes.

8.2. Collimator Focus

The best internal focus values have been determined and are fairly stable with time. However, they have been known to change. We try to keep the relevant Web documentation up to date, but the most up-to-date source of the proper focus values is the setup files available through the ESI Dashboard GUI. Note that there is a separate Web page on determing the focus of new filters.

Table 3. Recommended Focus Values

Mode/Filter Focus
Echellette 61600
Low-D 76994
B imaging 104800
V imaging 98254
R imaging 111581
Ellis-R imaging 116779
I imaging 108798