Observations

Flats and darks

Make them in the usual way. But take flatfields often. MUCH more often than in imaging work. The flats must be acquired very close in time to the observations. Do not hesitate to take a break in the observations and make flat fields even if the seeing is good. Without flats the spectra are worthless.

Geometric distortions

To rectify the spectra, the spectral lines can be used in the spatial direction. For the other direction direction, we have a device that places a grid of 0.5-mm spaced lines in front of the slit giving rise to a set of lines following the dispersion direction on the spectral frame. This is called the gg device. Place it at the slit and make a flatfield exposure.

NOW: The current gg device is provisional. The glass plate with the etched lines is glued to a piece of pink plastic that can be slid on the three rods around the slit in a similar way to the focus device. As always be careful when working close to the slit.

Reductions

Flat fields are constructed by computing a mean spectrum for each flatfield exposure, then dividing each spectral row with this mean. If the spectra are curved and not orthogonal, this becomes complicated. We lack a good set of routines for flatfielding with the current image distortions (smile and keystone). This may be made available later.

NOW: An old set of routines (history: dan $\rightarrow$ rouppe $\rightarrow$ dettori $\rightarrow$ dan) has been brushed up and is available on royac17. Use flat_setup as usual. then give the command spectflat which has the same interface as flat. Be careful to define the image area to include only the illuminated part of the CCD. The spectflat command is defined by an alias in the .cshrc of obs. Code is in /home/obs/spec.

Use this for a quick check. It is recommended to look at reduced data as soon as possible - preferably the very same day. Are there heavy, time-dependent fringes? Are there enough flatfields? Have there been wavelength shifts?

Correlation tracker and adaptive optics

The correlation tracker hardware is set up on the small table in front of the spectrograph as shown in Fig. 8. Except for an extra folding cube everything is like the imaging setup. Operations are identical except for some coordinate changes. This should not concern general observers, but here they are.

The calibration procedure takes care of the tracker orientation. What must be manually set is the signs for the drift corrections that are being sent to the turret computer. Under the menu: system - drift correction, set sign X to 1 and sign Y to 1. Remember to press save and restart the program. On the image table the sign of the x coordinate changes.

Figure 8: Wavefront sensor and correlation tracker.

The wavefront sensor is also set up on the small table. Operations are identical to those on the imaging table except that the orientations of the electrode response wavefronts that are displayed after a control matrix run are different.

The illumination levels will be different from those on the imaging table.

Scanning

The slit is narrow (0.11 arcsec). Seeing movements, image rotation, and, when two separate wavelengths are simultaneously observed, differential refraction all calls for some scanning procedure unless purely statistical studies are enough.

The scanning method available involves sending offsets to the tracker system and having it introduce the appropriate tilt on the mirror.

To use scanning simply enable the ``Spectrograph Step'' in the options menu. The following values are read from the configuration file defining operation. (An easier to use interface is on the way)

SCAN_HOST

The correlation tracker computer. royac15 at the time of writing.

SCAN_PORT

The port, this should always be 11126

SCAN_STEPX

The step size in the x direction (arcseconds).

SCAN_STEPY

The step size in the y direction (arcseconds).

SCAN_STEPS

The number of steps to make.

When running in all data taking modes the system will scan make a single step in x and y at the end of eachh data taking period. It will repeat this SCAN_STEPS times before going back to the starting position.

Note the following:

• Set the 'repeat' counter to 0 for frame selection.

• Scanning will change the AO target.

• There is an upper limit to the scanning range.

Fine-tuning slit position.

Depending on how precise you want to be, it can be quite hard to get the slit exactly where you want it using the telescope handpaddle. To allow fine tuning of the slit position an interface is available on all camera computers for adjusting the slit position using the same mechanism as scanning. To start the interface (shown in figure 9) simply click on the slit icon on toolbar (when logged in as obs.)

To use the system one should first choose a good correlation tracker target and lock on it. Then by using this interface you can use the tracker (which is much more precis than the turret handpaddle) to fine tune the slit position. The smallest step size is 0.11'' (the slit width) and 0.5'', 1'' for medium and large respectively. Note that the AO target will change as you do this.

Figure 9: Interface for fine tuning slit position).