MES-SPM focus and focussing
18 September 2006
Michael Richer
Occasionally, there are queries about the focus of the MES-SPM spectrograph as well as the occasional request to refocus the spectrograph outside the 5000-6750Å range. Here, I explain a variety of issues relating to the focus of the MES-SPM spectrograph as well as how it should be focussed. I am indebted to John Meaburn and José Alberto López for their patient explanations of setting the spectrograph focus.
Normal focus
The MES-SPM has a slow ~f/8 camera. Normally, therefore, there is no need whatsoever to focus the spectrograph when it is put on the telescope. The focus is usually good from one observing run to another. This is in striking contrast to the Boller & Chivens and REOSC echelle spectrographs, but those spectrographs have much faster ~f/2 and f/1.4 cameras, respectively. In those cases, the small differences in the CCD position, which are the normal result of cooling to cryogenic temperatures, are sufficient to cause the CCD to not fall in the camera focal plane from one installation to another. Because the MES-SPM has a much slower camera, these small changes in CCD position have no significant effect because the camera is much more forgiving. Therefore, it is usually NOT necessary to focus the MES-SPM when it is put on the telescope.
The camera resolution
One effect that causes some doubt about the spectrograph focus is the high resolution of the MES-SPM’s camera. The camera is able to work at up to resolutions of nearly 100,000. As a result, it resolves the 70 and 150 micron slits usually used, i.e., the camera sees these slits as resolved objects, with a measurable width, not an infinitely thin line. Because of this, the line profiles are more box-shaped than Gaussian. In practice, this effect is not terribly noticeable because the CCD binning is usually chosen so as to sample the slit width using only 2.6-2.8 pixels. At this CCD sampling (note that this is different from the camera resolution), the slit will only be slightly boxy compared to a Gaussian profile. Fig. 1 shows the profile of the 150 micron slit with the SITe3 CCD binned 2x2 (slit width ~2.7 pixels FWHM). In this case, the profile is slightly wider at the mid-height than the Gaussian profile that is fit to it. Likewise, the line profile is flatter at the top than the Gaussian fit. Both of these small effects are the result of the camera resolving the slit. Therefore, such profiles are in fact evidence of good focus, and not a reason to worry.
Fig. 1: Line profile, 150 micron slit, CCD binned 2x2
That the camera easily resolves the line profile can be seen in Fig. 2, which shows the same spectral line observed through the same 150 micron slit, but with the CCD binned 1x1 instead of 2x2. In this case, the line profile is very clearly flat-topped. Having many pixels sample the slit profile is no advantage as regards data analysis, so the CCD is usually binned 1x1 with the 70 micron slit and 2x2 with the 150 micron slit. In principle, the 30 micron slit allows the camera’s resolution limit to be reached. However, the SITe3 CCD that is currently used with the spectrograph cannot correctly sample the resulting profiles (with 2.4 pixels FWHM).
Fig. 2: Line profile, 150 micron slit, CCD binned 1x1
Note that the line shape has no effect upon wavelength calibration, since this is unaffected by symmetric distortions about a Gaussian shape, i.e., there is no error in determining the line centre provided that the line profile is deformed in the same way on both sides of the line centre.
Finally, while one might worry that the non-Gaussian line shape might have some effect upon the kinematics measured for celestial objects, this is not the case either. Both seeing and guiding errors will assure that the object is not kept motionless within the slit, effectively smoothing the slit’s somewhat square profile into a very nearly Gaussian one.
Focussing the spectrograph
As noted earlier, it is not necessary to focus the spectrograph when working within the usual wavelength range used for most observations, 5000-6750Å. (The focus should be equally good within the design range, 3900-7000Å.) Normally, too, it is impossible to change the spectrograph focus, since the lenses are locked into position, which causes the graphical interface to ignore requests to move them (the lenses button is grayed out; see Fig. 3). If it is necessary to change the focus, the lens clamp must be loosened to unlock the lenses and allow movement. This also activates the lenses button on the graphical interface (it changes to a black colour), through which lens movements are effected. The spectrograph is focussed by taking images of the 70 micron slit in diffuse light (during the day). The lenses are moved so as to minimize the width of the slit in the resulting images. The User’s Manual provides more information regarding the graphical interface.
Fig. 3: MES-SPM graphical interface
The recipe for focussing is then
1. Unlock the lens clamp. (This is done on the observing floor. It is the bronze knob indicated in Fig. 4. It must be loosened considerably to unlock the lens clamp. There is a quiet “click” when the clamp unlocks. When the clamp unlocks, the lens setting in the graphical interface will turn from gray to black.
2. Position the filter for which you wish to focus. Insert the mirror. Position the 70 micron slit in the beam. Set the CCD binning to 1x1.
3. Take an image of the slit. (The telescope covers need not even be opened.)
4. Measure the slit width, using IRAF’s imexamine task, for instance.
5. Move the lens encoder by about 10 units.
6. Repeat steps 2-5 until the minimum slit width is encountered (2.6-2.8 pixels, FWHM).
7. When focussing filters to the blue of the usual wavelength range, the lens encoder should be driven to lower numbers.
8. Take an arc spectrum to check that it too is in focus.
9.
Tighten the lens clamp. Do not force the mechanism,
but tighten firmly. The spectrograph should NEVER be left with the
lenses unlocked.
Fig. 4: MES-SPM on the telescope. The lens
clamp knob is indicated.
It is necessary to compare the slit/line width in imaging and spectroscopy since there are some combinations of wavelength and focus when images of the slit and arc lamp spectra are not simultaneously in equally good focus. The [Oiii]5007 filter used with the MES-SPM provides an illustrative example. Following the above recipe, the “best focus” occurs with the lens encoder at about 40 units lower than the best focus at H alpha. However, spectra of the arc lamp are in better focus at the H alpha focus position. (The line widths of the arc lamp spectra are also very similar through the H alpha and [Oiii]5007 filters at the H alpha focus position.) Likewise, spectra of celestial objects using the [Oiii]5007 filter are in better focus at the H alpha focus position that at the nominal [Oiii]5007 focus position. In this case, images should be focussed using the H alpha filter, but spectra may be taken over the 5000-6750Å interval. The reason that these contradictory results may occur is that the “fold plane” is different with the mirror and grating used for imaging and spectroscopy, respectively (see Meaburn et al. (1984) or Meaburn et al. (2003) for schematic optical diagrams). If in doubt of which focus setting to use, a spectrum of a bright emission line source (e.g., Orion, Ring nebula, etc.) should resolve which provides the better true spectral resolution.
Fig. 5 shows the variation of focus in the case of images taken through a He ii 4686 filter. Note that both slit profiles were obtained with the 70 micron slit. The “out of focus” profile was obtained at the initial encoder setting (good focus at H alpha) while the “in focus” profile was obtained at an encoder setting about 50 units lower. In this case, the focus in imaging and with the arc lamp coincided.
Fig. 5: A comparison of slit profiles with the
spectrograph in and out of focus.