Focussing CCDs on ISIS

or

How to avoid astigmatism

Introduction

There has been a history of astigmatic images being seen on the red arm of ISIS particularly when using the 1200g/mm grating. Recently it has also been reported when the 600g/mm grating has been used. ISIS is set up to have the fine focus control by moving the collimator mirror, there is about an 11:1 demagnification between the collimator movement and the focus shift at the camera so that a 50mm movement of the collimator gives a 4.5mm shift of the camera focal plane. However the collimator cannot be used to correct over a large range of focus because, as it is moved, it no longer delivers a collimated (parallel) beam to the diffraction grating. When a diffraction grating is illuminated by a non-parallel beam astigmatic images are formed with the size of the aberration dependent both on the amount of defocus and the ruling frequency of the grating.

What is the size of the problem?

With the ISIS collimators set to deliver parallel light to the grating the camera focus lies 26 mm (27.39 mm for red camera only, new value after ISIS overhaul May 2002) from the camera interface plate (this is made up of 22 (23.39) mm of air and 4mm of fused silica for the cryostat window). The shutter assembly and field lens mount protrudes by 10mm from the interface plate with the spring-loaded seal fully compressed. This leaves 16 mm (17.39mm) for the mechanical distance from the front of the CCD cryostat to the CCD, plus whatever additional light seals are required, plus any leeway required for tilt adjustments. If greater clearance is required and the collimator is adjusted in focus to achieve this then there will be a consequent deterioration in the image quality. Because the ISIS Hartmann test is set up to minimise the image spread in the dispersion direction the full astigmatic line spread will appear in the spatial direction along the slit.

The spot diagrams on the following page shows the line spread for different amounts of decollimation i.e. departure of the collimator from ideal focus. The same focussing criterion has been used; that is to minimise the image spread in the dispersion direction. The grating for the first diagram is 1200 g/mm and the on axis wavelength is 650nm. The problem will be worse for longer wavelengths.

The second spot diagram if for the same central wavelength, 650nm, and a 600g/mm grating. As might be expected the size of the astigmatic line formed for a given amount of collimator defocus is halved with respect to that obtained with the 1200 g/mm grating.

Line images for 1200g/mm grating operating in red

Line images for 600g/mm grating operating in red

The offset distances of 2,5 and 10 mm at the collimator correspond to camera focal plane shifts of 0.18, 0.45 and 0.9 mm when working at low spectral resolution. The astigmatism, generated when using higher ruling frequency gratings, changes the effective focus shift at the camera when the spectrograph is focussed so that the line focus lies along the slit; it reduces the camera focal plane shift to 70% of the low resolution value for the 600g/mm grating and to 50% for the 1200g/mm grating.

Shift of collimator	Movement of camera focal plane
	Low resolution	600g/mm grating	1200g/mm grating
2mm	0.18mm	0.14mm	0.10mm
5mm	0.45mm	0.34mm	0.24mm
10mm	0.90mm	0.67mm	0.46mm

Focussing process to avoid astigmatism from the grating

In order to avoid astigmatism from decollimation at the grating the following process should be followed.

When focussing a detector on the blue arm set the blue collimator focus to 5600. For the red arm remove any dichroic and set the red collimator focus to 9100. Reduce the Hartmann shift to about 10 to 15 mm by using the capstans on the CCD then the final focus can be done with the collimator. When a dichroic is introduced into the beam the collimator focus in the red arm must be increased by 2300 (~11400). The dichroic is on a 4mm glass substrate that is tilted at 45° to the transmitted red beam. This tilted plate will introduce some astigmatism; this amounts to a 0.41 arcsecond spread along the slit. The focus increment given above puts the line image along the spatial direction of the slit.