\documentstyle[11pt,twoside]{report} \topmargin=0mm \oddsidemargin=1cm \evensidemargin 0.2cm \textheight=23cm \textwidth=15cm \parskip=3mm \parindent=0mm \pagestyle{plain} \begin{document} \begin{titlepage} \vspace*{3cm} \begin{center} \huge\bf EMMI \& SUSI \vspace*{1cm} The ESO Multi-Mode Instrument \\ and \\ The Superb Seeing Imager \vspace*{2cm} \large{\bf E. Giraud} \vspace*{0.1cm} \vspace*{3cm} \Large{\bf ESO OPERATING MANUAL No. 15} \vspace*{1cm} \large{\bf Version No. 2.8} \vspace*{0.5cm} {\large\bf August 1995} \end{center} \end{titlepage} \newpage {Acknowledgements:} A user manual is a collective work. The first version of the EMMI-SUSI users manual was written by J. Melnick, S. D'Odorico, H. Dekker. While the basics of the instrument have not changed, new cameras and detectors have been installed, changing the scales, the efficiencies, more powerful tools of quick reduction have been implemented, parameters have been tightened (e.q. shutter delays, image quality), a new echelle has been tested, the critical points in the use of the instrument are better understood, the parallel image analysis will be offered soon. It became worthwile to write a new version of the manual taking into account the experience gained in the last years. This new version was prepared with the input of P. Gitton, G. Mathys and J. Storm. \newpage \chapter{Introduction} \pagestyle{headings} \pagenumbering{arabic} This manual describes the operation of the ESO Multi-Mode Instrument (EMMI) at the New Technology Telescope (NTT) on La Silla. EMMI is a very flexible instrument which allows a wide range of observing modes, from wide-field imaging to high-dispersion echelle spectroscopy, including long-slit and multi-object spectroscopy. This manual also describes SUSI (SUperb Seeing Imager) which is mounted in the other Nasmyth focus of the NTT and complements the imaging capabilities of EMMI. A brief description of the active optics system of the NTT and its basic operational principles are also provided in this manual. The driving concepts in the instrument definition were the high image quality foreseen for the NTT, the need to complement or improve on the instrumentation on the 3.6m telescope, and the need to minimize instrument change-overs. The concept which was adopted is that of a dual-beam instrument, fully dioptric, and based on the white pupil principle. CCD detectors were foreseen for the two arms with the possibility to adapt to the geometric characteristics of future detectors by changing the cameras only. \ The main advantages of this type of design are the high efficiency in both channels and the easy conversion from wide field imaging to grism and grating spectroscopy. After the first observations with the NTT, it became clear that the telescope and the atmosphere at La~Silla could provide stellar images with diameters as good as $0.3^{\prime\prime}$. \ Images of this quality could not be sampled adequately with EMMI, which scale had also to be adapted to the spectroscopic modes of observation. Thus, SUSI was designed and built for the other Nasmyth arm of the NTT. The design is such that switching between EMMI and SUSI is done in a matter of minutes, so that they can be considered as different parts of the same instrument. \section{ How to read this manual} This manual gives an overview of the capabilities of the instrument and guidelines for the observations. You are likely to read it twice: first when applying for observing time, and second when preparing the actual obervations. When applying for observing time, you are advised to first read Chapter 2 which gives an overview of the different observing modes of EMMI and SUSI. This will help you to decide which mode of EMMI/SUSI is the one best suited for your observations. The chapter also describes the optical components and the CCDs. After this, go to the chapter describing the particular mode relevant to your observations. The modes are described in Chapters 5--10. In these chapters it is supposed that you are acquainted with the information in Chapter 2. Additional information can be found in the Appendices: these are referred to regularly in the manual. Chapter 11 gives additional information on anomalies, flexure, etc. Chapter 3 introduces the user interface used to control the instrument: this makes more sense after you have arrived on La Silla and it is recommended reading for the day before your observations. Finally, Chapter 4 gives information on the telescope: the active optics, focusing and tracking/guiding. This also should be read when preparing the actual observations, and will help you to get the best image quality out of the telescope. The user interface of the telescope control software is operated by the night assistant: it provides parameters on the dome and telescope status (windscreen, mirrors, hydraulic system, etc), the positions of the telescope and the rotator(s), the guide probe(s), and some external parameters ( e.g. wind speed). This interface is not documented in this manual: please ask your night assistant and introducing astronomer all questions you would like to know. The night assistants are also responsible for the operation of image analysis and are acquainted with the judgement of the parameters. It is recommended, however, that observers participate in operational decisions that will affect their data. Basic information on that subject is provided in Appendix . The NTT and EMMI/SUSI can be operated in remote control from Garching. Except for chapter 3, this manual equally applies to remote control. The user interface in Garching is different from the one on La Silla. It is described in the Remote Control manual (ref [1]) which is available from the Visiting Astronomer Section in Garching. This manual by necessity reflects the status of telescope and instrument at a particular moment (Summer 1995). Updated information is posted on the ESO Bulletin Board, and can be viewed by telnet to {\tt mc3.hq.eso.org} and using the login: {\tt esobb}. A WWW page under the ESO home page is being developed and will eventually replace the bulletin board. Important changes or technical improvements are regularly described in articles in The Messenger which appears every trimester. There is also a section "NTT bits and pixels" in the Messenger which contains short notes on changes in performance, configuration, and operation of the NTT and its subsystems. Information on the current technical activity at the telescope and the problems encountered during the previous nights and weeks is available on La Silla. By typing {\tt nttrep} on any workstation on La Silla, you will have access to the description of technical problems concerning the various subsystems of the telescope ordered by type or by date. The interface is self-explanatory. \newpage \section{Abbreviations and Acronyms} The following abbreviations and acronyms are used in this document: \begin{table} \label{abbrevia} \vspace*{0.2cm} {\small \begin{tabular}{ll} ADC & Atmospheric Disperion Corrector unit \\ AOS & Active Optics System \\ BIMG & Blue IMaGing mode of EMMI \\ BLMD & BLue Medium-Dispersion mode of EMMI \\ CCD & Charged-Coupled Device \\ DAT & Digital Audio Tape \\ DIMM & Differential Image Motion Monitor (seeing monitor) \\ DIMD & DIchroic Medium-Dispersion mode of EMMI \\ DO & Data Organizer \\ EFOSC & ESO Faint Object Spectrograph Camera \\ EMMI & ESO MultiMode Instrument \\ FITS & Flexible Image Transport System \\ ftp & file transfer protocol \\ GUI & Graphical User Interface \\ IA & Image Analysis \\ IHAP & Image Handling And Processing system \\ IRAF & Image Reduction and Analysis Facility \\ IRSPEC & InfraRed SPECtrograph \\ LU & Logical Unit \\ MIDAS & Munich Image Data Analysis System \\ MOS & MultiObject Spectroscopy \\ OBST & OBServing Task control program of EMMI \\ REMD & REd Medium-Dispersion mode of EMMI \\ RILD & Red Imaging and Low-Dispersion mode of EMMI \\ SUSI & SUperb Seeing Imager \\ TCS & Telescope Control Software \\ TRS & Technical Research Support department, La Silla \\ UIF & User InterFace \\ WS & Work Station \\ \end{tabular} \vspace*{0.3cm} } \end{table} \newpage \chapter{Instrument Overview} \section{Optical design} A detailed description of the optical design of EMMI can be found in Dekker et~al. (1986). Figure~\ref{schematic} shows the optical layout and identifies the main components of the instrument. The instrument has two 'arms' with separate detectors. One arm ('EMMI blue arm') is coated for high efficiency in the region 300 to 500 nm, the other ('EMMI red arm') for 400 to 1000~nm. Each arm supports several different observing modes. The capabilities of the blue and the red arm are not identical, e.g.\@ Multi-Object Spectroscopy (MOS) is only available in the red arm. \begin{figure} \vspace{5cm} \caption[Schematic layout of EMMI]{\it Schematic layout of EMMI showing locations of the main components.} \label{schematic} \end{figure} \newpage \section{Observing modes} Each of the two arms has two possible light paths: one is used for grating spectroscopy and the other for imaging and (only in the red arm) low-resolution spectroscopy using grisms. In grating spectroscopy, a dichroic can be inserted into the beam allowing the use of both arms at the same time. Each of the five possible light paths (two per arm plus the dichroic mode) is called an 'Observing Mode'. The five modes are called: \begin{description} \item[RILD] which stands for \underline{R}ed \underline{I}maging and \underline{L}ow-\underline{D}ispersion spectroscopy \item[REMD:] \underline{Re}d \underline{M}edium \underline{D}ispersion spectroscopy \item[BIMG:] \underline{B}lue \underline{Im}agin\underline{g} \item[BLMD:] \underline{Bl}ue \underline{M}edium \underline{D}ispersion spectroscopy \item[DIMD:] \underline{Di}chroic \underline{M}edium \underline{D}ispersion spectroscopy \end{description} Table~\ref{obsmodes} summarises the correspondence between type of observations and EMMI mode (and SUSI), and lists the main specifications. \begin{table}[hhh] \caption{\sl Types of observations possible in the various modes of EMMI and SUSI} \label{obsmodes} \vspace*{0.1cm} {\small \begin{center} \begin{tabular}{llllllll} \hline \bf{Observation type} & \bf{wavelength} & \bf{Mode} & {\bf resolution} & {\bf size} \\ \hline Wide-field imaging & $\lambda < 500\,$nm & BIMG & $0.37^{\prime\prime}$/pixel & $6.2^\prime \times 6.2^\prime$ &\\ & $\lambda > 400\,$nm & RILD & $0.27^{\prime\prime}$/pixel & $9.15^\prime \times 8.6^\prime$ \\ \\ High-resolution imaging & & SUSI & $0.13^{\prime\prime}$/pixel & $2.2^\prime \times 2.2^\prime$ &\\ \\ Low-dispersion long-slit& $\lambda > 400\,$nm & RILD & $ R=280$--$1670$ & slit $8^\prime$ \\ \quad spectroscopy \\ \\ Low-dispersion slitless & $\lambda > 400\,$nm & RILD & $ R=280$--$1670$ & $9.15^\prime \times 8.6^\prime$ \\ \quad spectroscopy \\ \\ Low-dispersion multi- & $\lambda > 400\,$nm & RILD & $ R=280$--$1670$ & $5^\prime \times 8.6^\prime$\\ object spectroscopy$^a$ \\ \\ Medium-dispersion long-slit & $\lambda < 500\,$nm & BLMD & $R=840$--$9000$ & slit $\le 6^\prime$ \\ \quad spectroscopy & $\lambda > 400\,$nm & REMD & $R=600$--$70000$ \\ & & DIMD$^b$ & BLMD+REMD \\ \\ echelle spectroscopy & $\lambda > 400\,$nm & REMD & $R=7700$--$70000$\\ \hline \multicolumn{5}{l}{$^a$ no. of slits $\approx 25$} \\ \multicolumn{5}{l}{$^b$ combined BLMD ($\lambda<410\,$nm) and REMD ($\lambda>410\,$nm)} \end{tabular} \end{center} } \end{table} In RILD the instrument works as a normal focal reducer with the possibility of doing imaging, low-dispersion (grism) spectroscopy, and MOS, and is thus very similar to the EFOSC instruments at the 3.6m and 2.2m telescopes on La Silla. In REMD, the light coming from the telescope is first diverted to the intermediate dispersion slit unit (Figure ~\ref{schematic}). A small prism below the slit sends the beam to the collimator and grating. After a second pass through the collimator, an intermediate spectrum is formed which is re-imaged by the focal reducer on the CCD (the upper folding mirror is inserted into the beam in this case). In the case of echelle spectroscopy, one of the grisms is used as a cross-disperser. The same optical principle is followed in the blue arm except that grism and echelle spectroscopy are not possible. Here EMMI offers only the possibility of direct imaging (BIMG) and medium-dispersion long-slit spectroscopy (BLMD). In DIMD, a dichroic beam-splitter below the intermediate dispersion slit sends the light to the red and blue arms simultaneously. The light paths in the various modes are illustrated more precisely in the corresponding chapters. It is recommendable, when at La Silla, to take a look at EMMI itself: the light path is nicely illustrated on its cover. \section{Cameras and Detectors} EMMI works with two scientific CCD cameras, one at the red arm and one in the blue arm. The image scale at the F/11 Nasmyth foci of the NTT is 5.36 arcsec/mm or 186 micron per $1^{\prime\prime}$. This is also the scale of the direct imaging with SUSI. The actual field size and scale depend on the detector and camera being used and are given in Table~\ref{ccdscales}. \begin{table}[bhb] \caption{\sl Image scale and field size for EMMI and SUSI} \label{ccdscales} \vspace*{0.2cm} {\small \begin{center} \begin{tabular}{||l|c|c|c|c|c||} \hline\hline & & & & & \\ {\bf Mode} & {\bf CCD} & {\bf CCD type} & {\bf Camera} & {\bf Pixel size} & {\bf Field} \\ & & & & {($\mu$m/arcsec)} & (arcmin) \\ & & & & & \\ \hline & & & & & \\ EMMI RILD & \#36 & TEK 2048 & F/5.2 & 24/0.27 & $9.15\times8.6^\ast$ \\ & & & & & \\ EMMI BIMG & \#31 & TEK 1024 & F/4 & 24/0.37 & $6.2\times6.2$ \\ & & & & & \\ SUSI & \#25 & TEK 1024 & (F/11) & 24/0.13 & $2.2\times2.2$ \\ & & & & & \\ \hline\hline \multicolumn{3}{l}{ $^\ast$ In MOS: $5^\prime \times 8.6^\prime$} \end{tabular} \vspace*{0.3cm} \end{center} } \end{table} An indication of the characteristics of the CCDs is listed in Table~\ref{ccdchar}. However, these characteristics are not constant and the numbers are indicative only. All CCDs on La Silla are regularly tested, and the results of these tests are posted on the ESO WWW pageS. We give in this manual information on the CCDs presently (August 1995) mounted on EMMI in Chapter 5 (RILD) for the red camera, Chapter 6 (BIMG) for the blue camera and Chapter 10 for SUSI. Updated information may be obtained from the ESO bulletin in Garching using: telnet {\tt mc3.hq.eso.org}, login: {\tt esobb}. Any technical problems encountered with the detectors are described in the NTT report facility which can be accessed by typing {\tt nttrep} on any La Silla workstation. \begin{table}[bhb] \caption{\sl CCD characteristics for EMMI and SUSI} \label{ccdchar} \vspace*{0.2cm} {\small \begin{center} \begin{tabular}{||l|c|c|c|c|c|c||} \hline\hline & & & & & & \\ {\bf CCD} & read-out & conversion & read-out & mean bias & dark & read-out \\ & mode & factor & noise & level & current & time \\ \hline & & (e$^-$/ADU) & (ADU) & (ADU) & (e$^-$/hr) & \\ \hline & & & & & & \\ \# 25 & slow & 3.4 & 1.8 (1.2)(*) & 185.4 & 2 & \\ (SUSI) & fast & 6.6 & 1.1 (1.3)(*) & 155.2 & 2 & \\ & & & & & & \\ \# 31 & slow & 1.7 & 2.5 & 280.2 & 8 & \\ (EMMI blue) & fast & 3.4 & 1.6 & 275.2 & 8 & \\ & & & & & & \\ \# 36 & slow & 1.34 & 3.3 & 279.2 & 3 & \\ (EMMI red) & fast & 5.7 & 1.3 & 229.2 & 3 & \\ & & & & & & \\ \hline\hline \multicolumn{5}{l}{ (*) in braces: when $2\times2$ binning used}\\ \end{tabular} \vspace*{0.3cm} \end{center} } \end{table} Saturation is in most cases defined by the ADU converter, at 65 kADU. The actual well depth is around $2\times10^6\,\rm e^-$ so that the linearity is good up to digital saturation. The exceptions are EMMMI red and SUSI when read out in fast mode. For EMMI red in the fast mode, exposure levels should be kept below 40 kADU, and for SUSI below 24kADU. Otherwise, the linearity of the CCDsis better than 0.5\%. The measured linearity curves can be found in the CCD test reports. The fast readout mode has as main advantage a reduced readout time. This becomes important on EMMI red where fast readout saves two minutes. It is much less important in SUSI. The disadvantage is increased readout noise and digitization noise, and sometimes increased electric interference. For broadband imaging and many spectroscopic programmes, the readout noise is not important compared to the photon background, and the fast readout mode would be recommended. Electronic interference could be larger in fast readout: when in doubt, it is worthwile to take a few biases in the afternoon to check on the readout noise and on the presence of pattern noise. Remember that, in general, calibration frames such as bias, flat fields, and darks taken with slow readout cannot be used for correcting fast exposures, and vice versa. The red CCD can be read in dual ouput readout mode: two output preamplifiers are connected to the CCD, the upper half of the chip is read through one of the amplifiers, the lower half through the other. The main problem with this mode has been cross-talk between the two amplifiers. However, this has recently been reduced to a level of 3 ADUs for a saturated star. The effect is that of a faint star at the mirror position of a bright one (i.e. mirrored across the dividing line between the two amplifiers). Using dual output mode results in a time gain of about 80s for the slow mode and 25s for the fast read-out mode: it may be worthwhile for time-critical observations. The selection of single or dual output is not done in the exposure form. Ask the NTT coordinator on how to proceed using the EMMI UIF. For CCD \#36, the second amplifier (normally A; for single-output amplifier D is used) has a cf of $\rm 1 ADU = 1.4 e^-$ and a ron of $\rm 6 e^-$ in slow read-out. In fast read-out the cf is $\rm 1 ADU = 6.1 e^-$ and the $\rm ron = 8.5 e^-$. The parameters of the CCDs may degrade. In particular, the readout noise of the CCD may increase by a factor of 2 depending on electric environment. For programs which are not sky limited it is worth checking the real noise before an observing run, and to check for pattern noise on the CCD. Spurious patterns may appear when an image is loaded with demagnification factors. Genuine noise patterns can be detected only with no demagnification applied. The readout noise of CCD No 36 slowly increases when it is read in 2~x~2 or 3~x~3 binning ( 8\% for 3~x~3 in {\tt SLOW } readout). CCDs have an electronic bias that can be evaluated by averaging several 0s dark exposures and substracted from the science images to take it out. By using these exposures the observer can check the CCD readout noise and possible pick-up patterns in the electronic background. At least one, but if possible more, long (at least 1 hour) dark exposures are important to monitor the dark current of the CCD (if possible, take a dark longer than the longest science exposure). \\ \\ The following section describes the optical components which are available in the different mode. You are encouraged to read this, before turning to the specific chapter describing your observing mode in more detail (Chapters 5 to 10). Chapter 11 gives additional information, mainly on known problems you may encounter---we recommend you reach this before the start of your observations. Detailed efficiency curves for the various modes are given in Appendix~\ref{efficiency}. \section{Optical components} \label{sisetup} \subsection{Instrument configuration} For each observing mode of EMMI there are a number of elements that can be installed, in order to configure the instrument to your specifications. Thus, there is a range of filters, grisms, slits, and gratings which can be mounted on the instrument. Not all of these components can be mounted together and therefore you must specify the instrumental configuration required for your observations. This has to be done one day before your observing run by filling out a special form available at the Astronomy Lounge on La Silla. If anything is unclear, talk to the NTT coordinator (beeper 50). Two of the echelle gratings (\#10 and \#14) require a large effort to mount and they cannot always be made available on short notice. Normally, proposals using these gratings will be scheduled as a block.. Please note that ESO is not committed to make available instrument configurations which deviate significantly from the one requested with the original application. \subsection{Filters} EMMI has four filter wheels: the blue and red imaging filter wheels, and the blue and the red below-slit wheels. The last two are only used for grating spectroscopy and usually contain neutral density filters only. Each of the two filter wheels used for imaging has 9 positions of which 8 are available for mounting filters and one is kept free. The R-filter is usually needed for focusing. Both red and blue filters have a free circular diameter of 80~mm and an outside diameter of 85~mm. Adapter trays are available for filters of other instruments (e.g. EFOSC) but use of smaller filters will produce vignetted images only useful in the centre of the CCD. In such cases it might be better to use SUSI. The SUSI filterwheel has 8 positions of which 7 are available for filters. In contrast to EMMI, SUSI uses 60-mm filters which are the same size as the ones used for EFOSC. EFOSC filters can therefore also be mounted in SUSI (if they are not required by the EFOSC observer!). There are also a large number of 60-mm filters which are not allocated to a particular instrument and which can be requested. All filters are permanently mounted in special cells which make replacement very easy. EMMI red-arm filters, which are inserted in a parallel beam, are mounted at $5^{\circ}$ inclination to avoid reflections between the CCD and the filter. Blue arm filters, used in the converging beam in front of the blue camera, do not show this effect and hence are mounted with no inclination. Using a red-arm filter in the blue will result in a slight change of the central wavelength and will cause some astigmatism. If a blue filter is used in the red arm, every object in the field produces a ghost due to the mentioned reflections, which is about 5 magnitudes fainter than the original object. Thus, although it is possible to use blue filters in the red and vice-versa (one might want to do this in the overlap region, 400 to 500 nm), filters should normally be used in the wheel they are intended for. Filters with very narrow bandwidths are not really suitable for the red arm. Because the filters are mounted in the parallel beam, the central wavelength will change across the field. In extreme cases, the central wavelength may be outside the filter response near the edge of the CCD. The tabulated wavelength corresponds to the centre of the CCD. The effect is further described in Chapter 11. As a guide line, avoid filters with $\Delta\lambda <5nm$ for wide-field imaging. This effect also affects wide-field photometry if using narrow filters. The ESO Image Quality Filters Catalogue (Gilliotte, 1992) contains a list of available filters and transmission curves. More recently (1995), a number of new filters have been acquired and all transmission curves re-measured. This new data can be viewed using the MIDAS graphical user interface (GUI) {\tt FILTERS}, available in MIDAS version 94NOV or later. The most recent version is always available in La Silla. Lists of standard EMMI filters are also given in Chapters 5 (RILD) and 6 (BIMG), respectively (Tables 5.1 and 6.1), and a list of standard SUSI filters in Chapter 11 (Table 11.1). \subsection{Grisms and Gratings} Grisms are a combination of a prism and a grating, used for low-dispersion spectroscopy. EMMI has a grism wheel in the red arm only. The reason is that grisms have poor efficiency in the blue: in EMMI they cut off just below 400~nm. The list of available grisms is given in Chapter 5 (RILD). The grism wheel of EMMI has nine positions of which one is kept free for direct imaging and one is used for the focus wedge. Two of the remaining positions are rotated by $90^\circ$ to be used as cross-dispersers for the echelle gratings. Thus, for low-reslotion spectroscopy, five grisms can be mounted for a single night. Changing the grisms during the night is not possible because of the need of aligning the dispersion direction with the CCD pixels. Except when used as cross-disperser, grisms are normally used with either a long slit or with multiple slitlets (mulit-object spectroscopy). The slit widths are fixed but several different widths are available. Grisms and filters can also be combined to obtain slitless spectra imaged on the CCD. This mode of operation can be useful in a number of survey programs. The use of filters in combination with a grism reduces the sky background intensity. It also selects the wavelength region of interest and limits the length of the spectrum. The spectral coverage within this wavelength region depends on the position of the object in the field. \begin{figure}[hph] \vspace{20cm} \caption[Resolution and wavelength coverage of EMMI gratings and grisms] {\it Resolution and wavelength coverage of EMMI gratings and grisms} \label{Resolution} \end{figure} EMMI contains two grating units, one in each arm, which are used for (long-slit) medium-resolution spectroscopy. The gratings are mounted in special housings, and one such housing is mounted on EMMI. The housings can contain two gratings back to back; the allocation of gratings to a given housing is permanent. Seven housings, four for the red arm and three for the blue, are presently available. The housings cannot be exchanged during the night. There are three echelle gratings available for the red arm. Two of these, \#10 and \#14 (the ones with the highest resolution) each have a special housing containing only one grating. Mounting these two housings takes extra effort and cannot be done on short notice. Exchanging between these two gratings and the other housings in an observing run is normally not possible. If you require these echelle gratings, please indicate this on the observing proposal. Figure provides a global view of wavelength coverage and resolution--slit products (Rs, nominal resolving power for a 1$^{\prime\prime}$ slit) of the grisms and the gratings. The characteristics are listed in more detail in the chapters corresponding to the various modes of the instrument. The properties of the gratings available in the blue and red arms are listed in Chapters 7 (REMD) and 8 (BLMD) respectively. The echelle spectra obtained using different cross-dispersers are also described in Chapter 7 (REMD). The grisms are tabulated in chapter 5 (RILD). The efficiencies are given in Appendix. Notice that, as is the case for some grisms, order-separating filters are required for some of the gratings in mode REMD. \subsection{Slits, slitlets and starplates} EMMI contains two slit units. In grating spectroscopy, a $6^\prime$ long slit is used, mounted before the beam splitter so that the same slit is used for both arms. The length of the slit can be limited using dekkers: this is necessary in echelle spectroscopy to match the interorder separation. The slit width is continuously adjustable between $0^{\prime\prime}$ and $9^{\prime\prime}$. An intensified TV camera provides images of the central field of view of the telescope as reflected off the slit jaws in grating spectroscopy. Under good circumstances, $18^{\rm th}$-magnitude stars can be seen on the slit viewer. The field is about $1.5^{\prime}$. The second unit is mounted in the red arm and is used for grism spectroscopy; it contains a wheel which has 5 positions. In four of the five positions, a so-called {\it starplate} can be put. The fifth position is kept free for direct imaging. A starplate, in the EMMI language, is a dismountable plate containing a slit. There are six fixed slits available, each with a length of $8^\prime$ with the present F/5.2 camera and the TEK 2048 CCD. The available widts are $0.5^{\prime\prime}$, $1.0^{\prime\prime}$, $1.5^{\prime\prime}$, $2.0^{\prime\prime}$, $5.0^{\prime\prime}$, and $10^{\prime\prime}$. Four of these can be mounted at any given time. There is no TV camera for this slit unit, but it is possible to take a direct image through the slit to check the centering of the object. It is also possible to make your own starplate. For this purpose, there is a punching machine mounted inside EMMI. Up to 4 of the regular long-slit plates can be replaced by starplate blanks. Any remaining regular long-slit plate should be protected by the software in order to prevent accidental punching: please check before punching. A special program running in MIDAS allows the user to define slit positions and lengths, based on a previously taken EMMI acquisition image. The punch heads of EMMI produce rectangular slitlets of the dimensions indicated in Table~\ref{mosa}. Long slits may be created by punching several adjacent slitlets. This mode is normally used for Multi-Object Spectroscopy. It is very similar to that of the EFOSC instruments described in Melnick et~al. (1989). \ Table~\ref{mosa} compares the capabilities of EMMI with EFOSC1 (3.6m telescope) (EFOSC2 on the 2.2m does not have a MOS mode). Unlike EFOSC, the punching machine of EMMI is incorporated in the instrument. Thus, slitlets are punched inside of EMMI and are automatically positioned in the focal plane of the instrument for MOS spectroscopy. \begin{table}[hhh] \caption{\sl Comparison of MOS modes in EMMI and EFOSC1} \label{mosa} \vspace*{-0.2cm} \begin{center} \begin{tabular}{||l|c|c||} \hline\hline & & \\ *[-0.2cm] & {\bf EMMI} & {\bf EFOSC1} \\ & & \\ *[-0.2cm] \hline & & \\ Wavelength range & $420-1000$nm & $360-1000$nm \\ & & \\ *[-0.2cm] Imaging field & $9.1^\prime \times 8.6^\prime$ & $5.1^\prime\times5.1^\prime$ \\ & (TEK2048 CCD) & (TEK512 CCD) \\ & & \\ *[-0.2cm] Punch field & $5^\prime\times8^\prime$ & $3.6^\prime\times4^\prime$ \\ & & \\ *[-0.2cm] Aperture shape & Slit & Circ. hole \\ & & \\ *[-0.2cm] Size & $$ & $2.1^{\prime\prime}$ diam. \\ & $$ & $3.6^{\prime\prime}$ diam. \\ & & \\ *[-0.2cm] No. objects per field & $10-30$ & $10-20$ \\ & & \\ *[-0.2cm] Punching machine & On line & Off line \\ & (on EMMI) & (control room) \\ *[-0.2cm] & & \\ \hline \hline\end{tabular} \end{center} \end{table} Punching MOS plates is most efficiently done during the afternoon if images are available. Starplates may also be prepared during long exposures, but the RILD mode may not be used during the actual punching procedure which may take about 15 minutes depending on the number of slits and their positions. Notice that there is some play in the starplate support so {\it a mask that is removed and mounted again may no longer be aligned with the image used to define the slit positions}. It is possible to submit requests for taking acquisition images for MOS observations. The NTT team will attempt this on a best-effort basis, during test time. The request should be submitted 2--3 months before the observations. These observations will be done during bright(ish) time and are limited to 30-minute exposure time through the R-filter and 1--2 exposures per run. They will not be very deep. If deep exposures are required, make sure that the time needed for these is included in your application for observing time. % is it possible to use images from other telescopes? It should be noted -- and is a source of confusion -- that the two slit units are rotated over $90^\circ$ with respect to each other. The grating slit unit is oriented East--West, whereas the starplate slits are oriented North--South. EMMI can be rotated as a whole to align the slit with a particular angle on the sky, but this rotation angle differs by $90^\circ$ between the two modes. \subsection{The Dichroic} A dichroic beamsplitter can be inserted in the beam to allow simultaneous grating spectroscopy in the red and blue arms. This beam splitter is permanently mounted in the instrument and cannot be exchanged. The efficiency curve of the presently available unit is shown in Appendix~\ref{efficiency}. The central wavelength is approximately 450~nm. Close to the central wavelength, the dichroic acts as a polarizer, sending one polarization to the red arm and the other to the blue. For this reason we do not recommend to use the dichroic in the range 410--490~nm. There is no dichroic available for the imaging modes. \subsection{ The Atmospheric Dispersion Corrector } EMMI has an Atmospheric Dispersion Corrector unit (ADC) located in front of the mode selection unit. Although all parts of the ADC unit are present in EMMI, the control software for this unit is incomplete, so the ADC unit is normally put in position ``free'' and disabled by the Operations staff during the observations. If it is not disabled, you should select position {\tt free} in the corresponding panel of the control software. \section{Calibration unit} There is a system of calibration lamps associated with the adapter/rotators at the NTT which can be used for most of the wavelength calibrations required for the EMMI data. The main component of the calibration system is an integrating sphere mounted on the side of the adapter. Light from the output aperture of the integrating sphere passes a lens and is reflected to the center of the focal plane by a $45^{\circ}$ mirror which is moved to the optical axis. On the integrating sphere He, Ar, and ThAr lamps are mounted, while the light of flatfield and other spectral lamps that are mounted in a rack on the floor is fed to the sphere through an optical fibre. The fiber induces some broad absorption features around 724$\mu$m and 880$\mu$m which do not occur in the scientific data. The angular size, location, and shape (including central obscuration) of the NTT exit pupil are approximately simulated. The illumination is homogeneous and unvignetted in a $3^\prime\times6^\prime$ field and is still usable in a field of $5^\prime\times8^\prime$ which is the maximum field size for MOS. \section{Format of the scientific data} The data from the EMMI and SUSI detectors are simultaneously transmitted to IHAP and MIDAS databases. MIDAS runs on a Unix workstation equipped with a DAT tape unit. IHAP uses standard 1/2~inch 2400~foot tapes at 6250 BPI with a total capacity of 45~$1024\times1024$ images in FITS format. ESO has the policy to archive all EMMI/SUSI data, and make them later available to the general users. The observer will get a tape from the archive containing all his/her observations. IHAP is in principle used to write the tape used for archiving. Thus, observers are requested to save their data on an archival tape, even if they also make a back-up for private purposes. The FITS headers of CCD files contain all the information necessary for the scientific use of the data, that is all the telescope, instrument, and detector parameters. Most of these parameters are stored in so-called hierarchical keywords. MIDAS can read these keywords, but some other packages may not since these are an extension of the FITS standard. If you are not using MIDAS, it is worth to check the actual FITS header for further information which may be useful. \newpage \chapter{Observing with EMMI} \section{Instrument control and User interface} The NTT is controlled by two HP1000/A900 computers, one for the telescope (called NTT) and one for the instruments (called NTI). The control software of EMMI is organized in such a way that EMMI is presented as five sub-instruments called RILD, REMD, BIMG, BLMD, and DIMD. Depending on the type of observations, the user selects one of these modes and the control software automatically moves the functions to be set for this mode. This leaves only the parameters of the particular type of observation to be defined. For instance, when setting up an exposure in RILD, the required mirror unit and the upper red folding mirror are automatically set. The observer must only specify the camera focus, the choice of slit, filter and/or grism, and exposure parameters (see section Getting started). The user interface (UIF) consists of a RAMTEK monitor where mouse driven menus and forms are displayed, and a CRT (LU:53) monitor where messages from the system about the instrument are given and commands may be entered. Parameters are entered by filling in forms on the RAMTEK screen. Once all optional optical elements are installed by the operation group, according to the observer's request, a setup form is produced. A printout of this form is left in the control room so that the observer can verify the setup and can use it as reference during the night. The positions in the wheels of filters, grisms,and slits, and the gratings in the housing will be displayed, on the RAMTEK UIF in sofar used in the chosen mode, whenever a setup in that mode is defined. An example of a typical EMMI setup form is reproduced in Figure~\ref{setup}. \begin{figure}[p] \vspace*{-0.5cm} {\footnotesize \begin{verbatim} =========================================================================== EMMI OPTICAL SETUP TEL: NTT OBS:_______________ SETUP BY:________________ DATE:__________ =========================================================================== ADC prisms: 0.5'' , 1.0'' , 1.5'' , 2.0'' , free =========================================================================== Blue imaging (BIMG) | Red Imaging and Low Dispersion (RILD) | Blue filter wheel (FILB) | Red filter wheel (FILR) Grism wheel (GRIS) | focus name identifier offset | name identifier offs. name identifier 1 B #603 -3 | 1 z #611 3 1 #1 #1 ALIGNED 2 U #602 40 | 2 H alph #654 5 2 #2 #2 ALIGNED 3 OII/5 #649 0 | 3 V #606 32 3 #3 #3 ALIGNED 4 HE II #588 0 | 4 R #608 0 4 #4 #4 ALIGNED 5 He I #587 0 | 5 I #610 3 5 #5 #5 6 FOCW+B #604 focwB 0 | 6 Bb #605 2 6 #6 #6 ALIGNED 7 #1 HARTMANN 0 | 7 HALPHA #596 11 7 CD FREE 8 #2 HARTMANN 0 | 8 BG 38 #543 0 8 FOCW ALIGNED 9 FREE FREE 0 | 9 FREE FREE 0 9 FREE FREE | | Starplate wheel (STAP) | | name identifier | 1 FREE FREE | 2 5'' X=555.75 | 3 1.5'' X=552.8 | 4 2.0 X=555.7 | 5 MOS RED000100 | focus = 0.0 + 20.0 * T | focus =7994.0 + 9.7 * T T in deg C =========================================================================== BLue Medium Dispersion (BLMD)| Red Medium Dispersion (REMD) | Blue grating unit (GRTB) | Red grating unit (GRTR) | name identifier | name identifier # 3 1200 g/mm | # 6 1200 g/mm # 5 158 g/mm | # 7 600 g/mm | Below slit filter wheel blue | Below slit filter wheel red (BSLB) | (BSLR) focus | focus name identifier offset | name identifier offset 1 FREE 0 | 1 FREE 0 2 nd 1.0 ND 663 0 | 2 nd 0.5 ND 661 0 3 nd 0.5 ND 662 0 | 3 nd 1.0 ND 664 0 | focus = 7613.0 + 39.0 * T | focus =7785.0 + 27.5 * T T in deg C =========================================================================== Dichroic Medium Dispersion (DIMD) Blue focus offset = -25.0 | Red focus offset = -250.0 \end{verbatim} } \vspace*{-0.5cm} \caption[Example of a typical printed setup form of EMMI]{\it Example of a typical printed setup form of EMMI. The form shows the optical elements that have been mounted on the instrument, and the temperature dependence of the instrument focus.} \label{setup} \end{figure} In addition to the list of optical components mounted in the instrument, this form gives the temperature dependence of the instrument focus, the positions of the starplate slits for the RILD mode, and the focus offsets of the filters. The optical elements are mounted and aligned with the CCD rows by the operations staff. Check that your selected grisms are labeled {\tt ALIGNED} and call the NTT coordinator (paging 93-50) should this not be the case. The requested filters are marked with a '$\star$'. \section{ Data processing and data saving} The data is send simultaneously to the A900 running IHAP and to a UNIX workstation (WS) running MIDAS. IHAP is running on terminal LU:65. The MIDAS implementation at the WS uses two monitors, one of which is exclusively dedicated to displaying the images. They arrive from the EMMI and SUSI CCDs in FITS format (.mt) and are automatically converted to .bdf format and displayed on the large display monitor. The other monitor (the left-hand one) contains a number of X11 windows. The light blue window at the upper right shows the status of the data transfer. MIDAS is running in the white window: this should be checked regularly. A number of often-used MIDAS commands are accesible via a number of 'buttons' on the display monitor, called the 'observing batches'. In this way they can be used quickly even if the observer has little prior knowledge of MIDAS. The most popular batches are {\tt seeing} (obtaining the seeing from the raw frames), {\tt statistics of region} (found under {\tt utils}) and {\tt trace}. Further batches calculate the focus setting from a focus exposure {\tt focus}, or the pointing offset needed to get a particular object on the slit {\tt point} Tape drives are available on both the IHAP computer and on the workstation. One of these is used to make automatic backups of all incoming images: this tape will go to the ESO archive and the observer will be sent a copy. Because the link between the IHAP computer and the workstation has not been fully reliable, so that occasionally images will not arrive at the workstation, in general the IHAP tape is used for this backup. However, this gives a 1-2 minutes extra overhead which is significant when taking many short exposures. If the observer is willing to take the risk, the DAT on the workstation can also be used for the archival copy. In addition to the archival tape, most observers also make a private backup using the DAT. \section{Getting started} The program that controls EMMI is called {\tt OBST} (OBServing Task). To run OBST, logon at the EMMI terminal (LU:53) ` {\tt RTE-A LOG ON : OBST} (ask for the password), and follow the instructions appearing on the screen. During the initialization process, a special form called {\it Assembly} will appear on the OBST screen. The form requests the name of the observer and the programme identification, and asks you to set the flags which enable/disable communication with the relevant nodes of the NTT and CCD controllers. For normal operation during the night, all connections to these nodes ({\tt EMMI, CCDR, CCDB, ADAPTB, TELNTT}) should be enabled (i.e. set to \fbox{\tt TRUE}). If you have to start the system during the day, when the NTT is stopped, always \rm set the \fbox{\tt TELNTT} flag to \fbox{\tt FALSE}. To exit OBST, use the softkey EXIT, and answer Y to do a shutdown. On the second terminal (LU:65), you may run IHAP by logging in with the username {\tt EMMIHAP}. \ After login, press the first softkey in the terminal ({\tt f1:} \fbox{$_{\tt EMMI}^{\tt IHAP}$}) to start IHAP. A number of softkeys will appear that allow you to manipulate IHAP in the standard way. Once OBST is running, EMMI is operated using mouse-driven menus on the RAMTEK User Interface (UIF). In order to operate the UIF, simply slide the mouse to the selected commands (that appear in bars on the right hand side of the RAMTEK monitor), and click at the desired command using the {\it middle} button of the mouse. In {\bf top menu}, the observer is confronted with five different menus, one for each mode of EMMI. Clicking these menus leads to more menus and forms to be filled out which together allow one to setup the instrument and define exposures. After choosing a mode in the {\tt top menu}, one clicks on {\tt setup instrument} to select the optical elements to be used, and then on {\tt define exposure} to define the exposure parameters (see below). Up to six instrument setups and eight exposures may be predefined for each mode. Much of this can be done during daytime or during integrations so that mixing observing modes during the night is a straightforward task. \section{Selecting the light path: the mode and the setup} During observations, the instrument configuration is done in two steps. First, the EMMI {\it mode} must be selected. This is done in the \fbox{\tt Top menu} of the user interface. (This is the menu that appears after startup.) After clicking \fbox{\tt Top menu} the five EMMI modes appear on the UIF menu bar. Click the desired option using the middle button. For example, for DIMD, click \fbox{\tt DIMD-Dichr.Med.D.} (in yellow). After selecting the desired mode of operation, a {\it setup} has to be defined, specifying each optical element (filter and such) for that mode. In order to define a setup, click the corresponding setups bar (in green). Thus, to setup RILD for example, click \fbox{\tt RILD Setups }. A form will appear on the screen that allows you to predefine up to 6 instrumental configurations (for each mode). The form corresponding to \fbox{\tt RILD Setups} is illustrated in Figure~\ref{rildset} . The list of optical components mounted for that mode are displayed in the lower half of the RAMTEK screen. The forms are filled using the keyboard --~{\it not the mouse}~-- to move the cursor from one field to the next and typing the desired parameters. The TAB key may be used to skip a field. The arrow keys may be used to navigate through the form. The optical components (filters, grisms, gratings, slits) may be defined either by their position in the corresponding wheels (e.g. 1, 5, 9, etc.), or by their names as they appear in the optical setup form (e.g. \#1, $1.5''$, ESO567, etc.). The fields marked {\tt NOT ENAB.} in the setup screen should be ignored as the corresponding functions are not enabled. For any light path (setup) you have to focus the instrument according to the temperature and the selected optical components. The temperature dependence of the instrument focus is given in the setup forms of the various modes. The focus value corresponding to a given setup must be entered in the last field of that setup. When the form is complete, it can be send with the return key. There is an automatic mode of temperature correction of the camera focus, which can be set in the last field of the exposure definition (Sect.~\ref{sexpdef}). It is not stabilized, and suffers from problems with the link between the 2 computers (NTT and NTI). We do not recommend to use the automatic mode. \begin{figure}[p] \vspace{20cm} \caption[Example of a RILD setup form as it appears on the UIF screen]{\it An example of a RILD setup form as it appears on the user interface screen. The optical setups are defined by the observer by filling out the upper part of the form. The list of optical components mounted for that mode are displayed in the lower half of the screen.} \label{rildset} \end{figure} \section{Defining and executing exposures} \subsection{Exposure definition} \label{sexpdef} The exposure definition form is used to fully specify the exposure or exposure sequence. It contains the number or name of the setup to be used, and specifies exposure time, number of exposures, the CCD window, binning, and readout mode, possibly the calibration lamps, whether to save the exposure on IHAP tape etc. They are described below in the order in which they appear on the exposure definition form. To define exposures, simply click the {\it exposures} bar (in green) for a given mode. For example in BIMG mode, click the command \fbox{\tt BIMG exposures}. A form will appear on the screen where the exposure parameters must be entered. Up to eight different exposures may be defined (for each mode). The exposures form has two areas where system information is given (located at the top and bottom of the form), and three areas used by the observer to define the exposure parameters. The form used to define exposures in BIMG on the user interface screen is illustrated in Figure~\ref{bimgexp}. \subsubsection{First area} The first area defines the exposures to be executed, sets flags to enable or disable automatic transfer of data to IHAP tape at the end of the exposures, and records the EMMI temperature for the automatic camera focus mode. Single exposures, or sequences of exposures may be defined. Thus, if you want to execute exposures 4 through 7 you must enter \fbox{\tt 4} $-$ \fbox{\tt 7} at the top of this area. To execute only one exposure, its number must be entered twice. In mode DIMD, two exposures must be defined, one for the blue and one for the red CCD: both exposures will be started simultaneously. IHAP data are stored on standard 6250 BPI magnetic tapes using FITS format. To set IHAP tape recording, set the {\tt save to tape} flag to \fbox{\tt T} or \fbox{\tt F} (True or False). If set to \fbox{\tt T}, the exposure will be written to tape after the end of the exposure. However, the next exposure will only start after the tape-writing has finished since IHAP can only execute one thing at a time. For large frames, tape recording is time consuming and some users prefer to do the tape saving manually. This is done using the command {\tt WFITS} in IHAP (type: {\tt WFITS, $\#$ A, $\#$ B, NH} in IHAP if you want to save the files Nos A to B) during a long exposure. Notice that the full header information is not stored in the IHAP header, but in FITS keywords. Therefore, the softkey \fbox{\tt LIST FITS KEYS} must be used instead of {\tt DLIST} to display the information on the IHAP terminal. Only when the files are written to tape are data and header information merged. The temperature of EMMI is displayed after the tape control flag. It is followed by a {\tt sensor} message which indicates whether the focus of EMMI is automatically adjusted for temperature at the beginning of each exposure. This should be set to {\tt NOT ENABL} for reasons given above. \begin{figure}[p] \vspace{20cm} \caption[The form used on the UIF screen to define exposures in BIMG]{\it The form used on the user interface screen to define exposures in BIMG} \label{bimgexp} \end{figure} \subsubsection{Second area} The second user area of the exposure definition form consists of 8 lines with a series of fields where the actual exposure parameters must be entered. From left to right these parameters are: \vspace*{-0.3cm} \itemsep 0.3cm \begin{description} \item[Type:] The following types are defined: {\tt dk} for dark; {\tt sci} for scientific exposures; {\tt foc} for a through-focus sequence for telescope focus ({\it not} for a focus wedge exposure); {\tt foi} for instrument focus; {\tt cal} for exposures using the internal lamps (including white light for flat fielding) , {\tt cam} for multiple-calibration exposures (i.e. exposure with more than one lamp on the same frame), and {\tt ff} for flat-field exposures using external light. When the shutter should not open (dark or bias), the type {\it must } be set to {\tt dk}. When the internal calibration unitis to be used, the type should be set to {\tt cal}, with one exception only: if a \ calibration using multiple lamps is required (e.g. to get a He--Ar spectrum), set the type to {\tt cam} and define one exposure for each lamp. The IHAP identifier in any set of {\tt cam} exposures must be identical: only the exposure time and the lamp may differ. The software will turn on all lamps, but pause the exposure after the amount corresponding to the shortest exposure time. It than turns the corresponding lamp off and proceeds exposing for the remaining time. There are three different focus types: Instrument focus ({\tt foi}) is normally determined by the operations, but may have to be determined when using private filters. A focus sequence ({\tt foc}) is used in the EMMI blue arm and SUSI to focus the telescope. In the EMMI red arm there is a special prism (the focus wedge) to determine the telescope focus with one exposure: this should be defined as a {\tt sci} exposure. \item[Time:] the exposure time in seconds must be entered in this field. \item[\#n:] this is the number of times a given exposure is to be repeated (maximum 9). \item[IHAP identifier:] The exposure identification goes in this field. The identifiers in any set of type {\bf cam} exposures must be identical. \item[IHAP batch:] enter here the name of an IHAP batch to be executed at the end of the exposure. The most-used batch is {\tt CLEANIHAP}, which at the end of each exposure sequence will clean the IHAP data base, keeping only the 6 most recent files. The batch increases the dead time between exposures because it repacks the database after each exposure, but IHAP only has a small database and when it is full it will not allow a new exposure to start (it will hold approximately nine 2k$\times$2k frames). This will sometimes go unnoticed (e.g. during a exposure sequence) resulting in loss of time. The batch {\tt KDISPC} may be used to display the image on the IHAP Ramtek, if for some reason the images are not already displayed on the MIDAS workstation. \item[Setup:] in this field the number or the name of the instrument setup table must be entered. The previously defined instrument setup tables appear in the lower half of the screen, so you don't have to remember or write down what you specified in each of the six possible setups for each mode. \item[Calib. lamp:] enter in this field the name or the number of the calibration lamp you wish to use. The list of available lamps is displayed on the bottom-left of the TCS Ramtek (ask the night assistent). Only one lamp can be specified for each exposure. Multiple-lamp exposures are described above. \end{description} In mode DIMD, an extra field {\bf Path} appears in the exposure definition form that should be set to {\tt B} or {\tt R} according to which exposure is to be executed in the Blue and which in the Red arm. \subsubsection{Third area} The third user area of the exposure definition form is used to set the CCD readout parameters. {\tt Binning x,y} defines the number of pixels that will be combined in each direction. The default is \fbox{\tt 1},\fbox{\tt 1} where all the pixels of the CCD are read out independently. The next parameter that must be specified is the {\tt Fast readout mode} field: \fbox{\tt T} (true) stands for fast readout and \fbox{\tt F} (false) for slow(!). In fast readout ({\tt T}) the readout noise will be somewhat larger, but the readout time will be cut by approximately 40\%, or 2 minutes for the full red CCD. The conversion factor between electrons and ADUs is significantly lower for slow radout than for fast readout. Note that with the higher gain factor in fast readout, non-linearity and saturation are reached at correspondingly lower ADU values. The {\tt window readout} field specifies the geometry of the CCD. If set to \fbox{\tt F} (false) the values that appear in the {\tt XL,XR} (Left and Right borders), and {\tt YD,YU}(Down and Up borders) fields are ignored and the full CCD is read out, including the over-clocked columns and rows. The full size of the frame including the overclocked pixels is given between parentheses. IHAP can store and handle frames of at most 2048 x 2047 pixels (in any combination not exceeding the value of this product). This makes windowing always necessary for the red CCD, and the field should be set to \fbox{\tt T}. We recommend to use the window 1--2086 in X and 80--2046 in Y which includes the overscan but leaves out a part of the field which is vignetted. \begin{center} {\bf Note:} \\ {\it Remember that if you window the CCD you may loose the overclocked pixels and thus the possibility of checking the actual bias level of your science frames }. \end{center} The CCD parameters apply to all exposures defined in the form, and for all modes using this CCD: they should therefore be changed before executing an exposure if you wish to use different values for different exposures (e.g. for direct imaging and spectroscopy). In mode DIMD two CCD readout forms appear, one for the red and one for the blue CCDs. There are two more fields defining the number of bits used and the transfer to the workstation which should both be set to {\tt T}. The number of bits should {\bf always} be set to 16 (= {\tt T}): 15 bits is obsolete and may lead to serious problems. The transfer to the workstation is not strictly necessary (turning it off saves up to 45 seconds) but checking the quality of the data is much easier on the workstation than in IHAP. The last field of the form allows one to enable the automatic temperature correction of the instrument focus. If set to \fbox{\tt T} the EMMI camera focus will be automatically set to the value corresponding to the temperature displayed at the top of the form, taking the filter offsets into account. This feature is presently not stable and should be set it to \fbox{\tt F}. In that case, the focus value of the instrument corresponding to a given setup has to be entered in the last field of each setup. Exit the form by pressing RETURN or ENTER in the keyboard. Wait for a series of beeps and a status message at the bottom of the screen: {\tt Input form no longer active}. \subsection{Executing exposures} To execute an exposure click the \fbox{\tt Start exposures} (orange) command. If you are still in the exposure definition form, an error message will appear at the bottom of the Ramtek screen with an audible signal indicating that the mouse is disabled: type return to send the form. Now you may start the exposure. The status of the instrument and CCD will appear on the Ramtek screen after the exposure has started. The end of an exposure is signalled by a beep. Exposure status and integration time information will also appear at the upper corners of the screen (upper left for the red CCD; upper right for the blue). This information will appear in every menu so you may safely change forms during exposures and still have information of the remaining time and instrument status; it disappears when read-out starts. There are several other orange bars in the UIF that allow to pause or abort an exposure, and change the exposure time. In order to prevent accidental execution of these commands , OBST prompts for confirmation. The command \fbox{\tt EMMI \& CCD status} (blue) provides at any time full information about the status of EMMI and the CCDs. In DIMD, there are separate forms for EMMI ($\,$\fbox{\tt EMMI status}$\,$) and for the CCDs ($\,$\fbox{\tt CCDr \& CCDb status}$\,$). \begin{center} {\bf Note:} \\ {\it Exposure and setup definition of an {\rm ongoing} exposure cannot be changed, except for the exposure time}. \end{center} \section{Troubleshooting} \label{bugs} In case of problems, first ask your night assistant for help, since he will be able to solve most of the problems which may occur. If still nothing works, and it is before 11p.m., you should call the NTT coordinator (paging system 93-50); if it is later, call Operations (paging system 93-34). (Dial the number, wait for the beeps to end, say clearly your phone number two or three times and hang up.) Reading the next paragraphs might also help. Error messages are printed at the top and the bottom of the Ramtek UIF, and on the OBST console. A reserved area at the very top of the UIF gives error messages that need to be specifically cleared. For example, if OBST is assembled with the link to the NTT computer disabled, a message reading: \fbox{\tt Assembly defined with NO access to: TELNTT} will appear in yellow, preceded by the time at which this particular condition was verified. This does not necessarily imply a catastrophic failure. The messages at the {\bf bottom of the UIF} inform you whether the system expects input from the mouse or from the keyboard, and whether the parameters in the forms have been sent to the computer. Normally these conditions are cleared by simply sending the forms (using the RETURN or ENTER keys) before using the mouse. Sometimes, however, the terminal stays inactive and does not acknowledge input from either the keyboard or the mouse. In that case, hit RETURN several times until the condition is cleared. If the terminal continues to be blocked, go to another terminal of the NTI computer (not NTT) and from CI type {\tt CI> SETP,lu} where {\tt lu} is the logical unit of the terminal you wish to unblock (usually written on the terminal itself). If this does not work, ask your night assistant for help, and eventually call the NTT coordinator or Operations, depending on the time of night. The third list of messages appears on the {\bf upper half of the OBST console}. These messages contain information about the current activity of the system, as well as error messages. They are normally not ordered, so you must check the time that is displayed together with the message to find the most recent ones. The error messages usually inform of some failure that prevents an exposure from being started (for example if a wrong parameter is given in one of the forms or if there is an ongoing exposure). A common error message occurs when the IHAP database is full. Before the exposure starts, the message \fbox{\tt IHAP database full} appears. In that case, the IHAP commands {\tt PURGE} and {\tt PACK} must be used to remove files and clear the disk space. Make sure that the files have been written to tape before using the {\tt PACK} command. Files accidentally purged may be recovered using the {\tt FRESTORE} command, but this is no longer possible once {\tt PACK} is used. Sometimes the frames arrive at the MIDAS workstation without the full list of descriptors (headers). This is due to a bug in the HP computers where a program called {\tt AOIB} stops. This happens randomly and it is therefore recommended to check every afternoon if all the descriptors are present (use the MIDAS command {\tt READ/DESC filename *}). If only a short list of descriptors is present, ask the night assistant to restart {\tt AOIB}. Sometimes MIDAS gets into a funny state where it is impossible to get a proper cursor on the display window. The quickest solution is normally to log-off and log-in again. If you do this while an exposure arrives, it will not be converted to a .bdf file: you will have to use the {\tt INDISK/FITS} program to read the corresponding .mt file. \section{The NTT report facility} \label{obsrep} A computer-based system has been installed for reporting technical problems encountered by the astronomers using the NTT, and follow-up by maintenance staff. This NTT report facility is used as a source for tracking anomalies, actions which have been taken, solutions, recipes for future problems and NTT upgrading. {\bf It replaces the printed night reports.} The observers have access to the reporting system by typing {\tt nttrep} on the workstation. A window interface appears that is self-explanatory. Astronomers can read previous reports on the telescope and instruments, and submit reports as they used to do with the night reports on paper. \section{The NTT daytime activities calendar} \label{calendar} Daytime activities scheduled at the NTT are recorded in a computer-based calendar. To access this calendar one can type {\tt nttcal} on the workstation. \newpage \section{Check list} \label{checkl} \begin{enumerate} \item Have a look at the notice board in the control room . You will get a first view on the technical activities or changes at the telescope. Should you wish to know about the problems encountered with the instruments, detectors, adapters, or the telescope itself in the previous days or weeks, type {\tt nttrep} on the workstation and follow the instructions. The NTT report facility is self-explanatory. \item To inform yourself about daytime work at the NTT type {\tt nttcal} which will display a mouse-driven calendar. \item Compare the optical setup form with your request. \item When starting OBST on EMMI terminal (LU: 53) check the connections to the nodes ({\tt EMMI, CCDR, CCDB, ADAPTB, TELNTT}). During daytime {\tt TELNTT} must be set to {\tt FALSE}. \item If you are not familiar with the user interface, prepare as an exercise, for example, a calibration exposure of 1~sec with a grism, a He lamp that will be read in fast readout and start the exposure. It takes some time to send the information from the interface to the instrument. It takes additional time to rotate the wheels, move the mirrors, heat the lamps etc. Thus after you fill a form, press RETURN and wait for the audible signal before using again the keyboard or the mouse. \item Check the readout noise of the detector(s) on bias exposures. Check for any gradient in the bias. \item By running exposures with various setups, you may find that one of the functions cannot be initialized (a message appears on EMMI terminal LU: 53). Call the NTT coordinator using 93-50 (paging system). \item In the afternoon the alignment of the slits with the CCD can be checked by using calibration exposures. \item It is very advisable to begin the preparation of MOS tables as early as possible, whihc wil generally be at 2p.m. \item The first exposures of the night, including focus exposure and quick-look exposure of the field, may be defined in advance. \item After starting an exposure, check the instrument status display. Make sure that the correct slit, filter or order sorting filter, grism or grating, cross-disperser grism and echelle etc. are in. \item Check the CCD readout mode and binning. \item Did the exposure (count down of exposure time) actually start? \item After starting an exposure, look for any error message on the EMMI terminal (LU: 53). Check the monitor regularly because messages may scroll off the page and, therefore, escape notice. \item Check EMMI, dome and CCD temperatures from time to time. \item Check the seeing and focus regularly. The seeing measured by the seeing monitor can be obtained from the meteomonitor panel on the left workstation (Sect.~\ref{seeingm}). \item Is the target on the slit? \item Is the telescope guiding? \item Is the remaining rotator-angle range sufficient for your long exposure? \item Do not forget to consider seeing effects when estimating exposure times for narrow-slit observations. \item Check the parallactic angle of the slit. \item Is the central wavelength of the grating correctly set? \item Are the slit width and heigth adequate to the required resolution, leaving enough inter-order space? \item If the seeing becomes excellent ($\rm < 0.7''$), try to make an image analysis close to the observed field, if possible in parallel mode. \end{enumerate} \newpage \chapter{Image Quality, Focus and Pointing} \section{Image analysis} Confusion sometimes is found about the difference between active optics and adaptive optics. Adaptive optics can correct for turbulence in the atmosphere by means of very fast corrections to the optics. Active optics only corrects the mirror for slow changes in the telescope structure, e.g.due to different pointings. Thus, whereas adaptive optics can reach the diffraction limit of the telescope, active optics only allows the telescope to reach the ambient seeing. The NTT has active optics and can therefore not improve on the ambient seeing. In fact the good image quality of the NTT is due as much to the design of the enclosure as to the active optics. The median seeing at La Silla is $0.8^{\prime\prime}$--$1.0^{\prime\prime}$ and comparable image quality is routinely achieved at the NTT. Both the primary and the secondary mirrors of the NTT have active optics. The active support of the primary (M1) unit consists of 75 actuators and three fixed point supports. The force applied to each of the 75 actuators can be adjusted and thus the shape of the M1 unit can be modified. The M2 support can be moved in X,Y,Z, where the X,Y motion of M2 is used to correct for decentering coma and motions in Z to control the focus. The procedure to calculate the proper settings of the active supports is the so-called {\it image analysis}. The image-analysis sytems, located inside the instrument adapter/rotators, consist of a Shack-Hartmann grid and a CCD to record the image of the telescope pupil through the grid. Software running on the NTT computer analyses thes image, determines the telescope aberrations, and calculates the differential forces to be applied to the active support in order to correct these aberrations (Wilson et al. 1991 ). The night assistants are acquainted with the system and are responsible for its operation. It is useful, however, that observers are aware of the basic operational principles since they may be requested to take operational decisions during the night according to the seeing conditions. The NTT Active Optics System (AOS) is initialized every afternoon by the night assistant. In practice, this means setting the forces of the M1 support, and the position of the M2 unit to the default values which have been calibrated for zenith position. This procedure alone can already be sufficient to operate the telescope under average to poor seeing conditions (FWHM$>1''$). A full image analysis will be done at the beginning of the night, when it has become sufficiently dark. This will be immediately after the taking of twilight sky flat fields. This analysis is done not only to improve your images, but also to monitor the telescope and detect possible problems. The first measurement takes about 15 minutes. The observers may decide to shift these measurements to later in the night if they conflict with urgent observations, but the night asistents are instructed to do this test every night. A record of all active-optics measurements is kept in the NTT control room. For seeing conditions around $1^{\prime\prime}$ there is no need to further check the AOS unless the images become severely elongated. The active optics control automatically corrects the position of the M2 unit as a function of zenith distance, while the primary mirror is sufficiently stiff to retain its shape without need for large corrections. There is an automatic correction possible which uses a look-up table to determine the optimal distribution of forces for the corresponding telescope position, but this has not been fully stable and also adds an overhead of several minutes every time the telescope is pointed. However, if the automatic mode is used, it should be used for every preset because the corrections are differential relative to the previous values. Thus, if the seeing conditions deteriorate, the telescope should be pointed close to the zenith before disabling the automatic correction mode. As a better altenative, it is now possible to run the image analysis software in parallel with your exposure, with the option to apply the calculated corrections at the end of the exposure. If your guide star is sufficiently bright, a dichroic can be inserted in the guide probe which deflects most of the light to the Shack-Hartmann grid. After the exposure is finished you can decide whether to apply the corrections. This cuts the overhead to a few minutes. It can also be used to monitor the telescope focus. However, the parallel mode only works with zero rotator offset. In imaging mode this will in most cases be true. but in spectroscopy it may be necessary to apply a rotator offset to orient the slit along the parallactic angle. If the parallel mode is not used, and the seeing conditions become exceptional (e.g. $0.5''$ or better) or if the conditions are good (e.g. $<1''$) {\em and} the telescope is pointed to zenith distances larger than $30^{\circ}$, it may be best to do a full image analysis using a bright star near the position of your target field. The image analysis does not always succeed. Sometimes the uncertainties on the corrections are so large that the solutions are not implemented. The two main reasons for this are either very poor seeing or very little wind: in both cases the tubulence in the tube distorts the pupil. If two consecutive image analyses fail, it is better to postpone it to when the conditions improve. A nice aspect of the NTT is that if anything is wrong, you are likely to obtain elongated images. If the elongation is more than about 15\% (to check use the MIDAS command {\tt CENTER/IQE}), you should consider to check the focus and possibly redo the image analysis. You should measure the focus every time corrections to the mirror are applied. \section{Focusing the telescope} The focus of the telescope can be determined either using a through-focus exposure sequence, or using a focus wedge similar to EFOSC. The focus wedge is the fastest method, and can be used for any of the EMMI modes except BIMG for which a through-focus sequence is required for each filter. However, with the B-filter in the blue arm, the focus will be approximately the same as in the red arm with no filter or the R-filter. The reference point for the telescope focus is either the long slit or the starplate wheel. However, there is no convenient reference position in BIMG and the camera focus and telescope focus cannot be individually determined. This is the reason that the focus wedge will only give an approximate solution for BIMG. For SUSI, the camera cannot move and the focus is purely set by the telescope focus. The temperature dependence of the NTT focus is $$\rm \Delta F/\Delta T = 0.0764~mm/^{\circ}C. $$ \noindent The night assistent will monitor the temperature and if necessary adjust the telescope focus. It is recommended to check the focus every time the optics are corrected (cf. Appendix~F). The observer should check the EMMI temperature and if necessary adjust the instrument focus. \subsection{Focussing with the FOCUS wedge} \label{sfocwed} To focus the telescope with the focus wedge, select the RILD mode and prepare a setup with the focus wedge in the grism wheel and preferably the R-filter. The R-filter is best because it does not introduce an extra offset while minimising atmospheric dispersion. Thus, click \fbox{\tt RILD}, then \fbox{\tt RILD Setups}. Focus the camera according to the temperature, using the formula on the setup form. When this is ready prepare an exposure by clicking the command \fbox{\tt RILD exposures}. The exposure type should be {\tt sci} (not {\tt foc} which is used for through-focus). In order to average out seeing effects, do not use exposures of $\rm (>20 sec)$ for focussing. The use of fast readout and windowing is recommended to reduce the overheads, and the exposure should not be saved to IHAP tape. Now start the exposure. On the resulting image, you will see two images for every object in the field. Select the MIDAS batch {\tt FOCUS} on top of the display, with the option {\tt FOCUS WEDGE}. This program, described in detail in Appendix~\ref{batches}, gives the focus offset to be applied to the telescope. The focus wedge divides the pupil into two halves and separates the two images horizontally by a fixed amount. The vertical separation between the centroids of the two images depends on the defocusing and has been calibrated empirically. For a fair range of focus this dependence is linear, but it becomes more complicated if defocusing is severe. For that reason, if the focus offset is large (say more than 0.10mm) it is recommended to repeat the procedure. Normally, two exposures suffice to determine the focus to better than 0.010 mm, which is the accuracy required under good seeing conditions. The focus wedge works in a parallel beam and, therefore, the focus is in principle the same for all optical elements in the beam. In practice, however, some filters have optical power and therefore introduce focus offsets. The focus wedge is calibrated with no filter or the R-filter in the beam. Offsets introduced by the filters are listed together with the lists of available EMMI filters (Table~\ref{emmir-filter}). These offsets are applied to the camera focus. The focus wedge is calibrated by the operations staff using the Hartmann masks. If necessary, the calibration can be checked by taking an exposure of the pinhole mask mounted in the aperture wheel and illuminated by the He or Ar lamps. \subsection{Focusing with a through-focus sequence} \label{sfocthr} With a through-focus sequence, you take a series of exposures with different settings of the telescope focus. To gain time, the CCD is not read out in between: instead the exposure is paused, and the telescope is offset by a not-too-large value so that you get one frame with several images for each star. This procedure is fully automated. Through-focus is needed in the EMMI blue arm and in SUSI; in the EMMI red arm the focus wedge is faster. In EMMI, click the focus parameters bar of the BIMG mode i.e. click \fbox{\tt BIMG Focus param.}. A similar bar is available in SUSI. A form will appear on the RAMTEK display, requesting as input the number of focus steps (typically 7 to 9), the focus increment ({\tt step size}, typically $10-30 \mu$m), and the telescope offset in R.A. or Dec. (or both; typically $10''$). The parameters for the instrument focus are also specified in this form. Here you must give the number of steps and the camera focus offset. Define an exposure with as type {\tt foc}. In order to average seeing effects, use individual exposures of $\rm (> 20 sec)$ for the focus sequences. The procedure takes the current position of the telescope or camera focus as the value of the middle step of the sequence. Thus, if you want to start at any given focus value, you must position the focus to the value mid-way of the sequence: e.g., if you want to start at $-3.000$~mm, the focus of the telescope should be positioned to $-2.960$ ~mm for 9 focus steps of $10\mu$m. Use the MIDAS batch {\tt FOCUS} (on the top of the MIDAS display window) with as option {\tt focus sequence}. The use of that procedure is described in Appendix. \section{Focusing the EMMI cameras} The focus of the instrument is normally checked by the operations staff. The optical set-up form gives the value of the instrument focus for the various modes and their temperature dependence. Most filters require a correction to be applied to the instrument focus. In the BLMD mode the temperature is critical and should be closely monitored. As the temperature of the EMMI room is kept approximately constant by the air-conditioning system, adjustments to the instrument focus should not be required during the night except for mode BLMD. The EMMI temperature as well as the telescope tube and mirror temperatures are displayed by the Telescope Control System (TCS) UIF screen (a RAMTEK monitor located next to the EMMI display). The focus of any mode can be moved manually from the instrument console using the commands {\tt EMMI} $>${\tt FOCB,value} and {\tt EMMI}$>${\tt FOCR,value} for the blue and red arms respectively. The instrument focus must be entered in the last field of each of the six instrument setup tables. These values will be applied at the beginning of each exposure. The temperature equations are also stored in the system. If the automatic mode is selected, (not recommended) the program checks the EMMI temperature, calculates the focus, and applies the filter offset. The EMMI focus is usually determined by the operation staff, but should you want to check the EMMI focus in a given mode, the procedures described in the Appendix can be followed. The dichroic beam splitter used in DIMD introduces a focus offset in both arms. The values of these offsets are given at the bottom of the instrument setup form (cf. Figure 3.1). \section{The seeing from the seeing monitor} \label{seeingm} The seeing monitor is a program which is running on the workstation, under the blue button ({\tt meteo}). If it is not running, it can be started by typing {\tt meteomonitor} on the workstation. A window appears on which you can type {\tt h} for help on the commands, {\tt x} to create a new window in which the temperature, humidity, wind speed and seeing measurements of the last 24 hours period are displayed. Experience has shown that under normal circumstances NTT images should get to within 10\% of the displayed value; better values than given by the seeing monitor are also possible. There are some exceptions: when there is very little wind dome seeing may be a factor. In strong winds the seeing may also detoriate. Finally, if M1 is warmer than the tube, extra turbulence is also generated. The temperature of M1 is displayed together with the tube temperature and the EMMI temperature on the TCS display. If none of these conditions apply but your images are considerably worse than the seeing monitor indicates, consider an image analysis or a focus check. \section{Tracking and autoguiding} The pointing model of the NTT is normally good to $1^{\prime\prime}$ rms over most of the sky; close to the zenith and close to the horizon the model is not as good. The NTT being an alt--az telescope, it is not possible to observe too close to the zenith. However, the pointing model presently gets noticeably worse at zenith distances where observations are still well possible. It has also been noticed that the pointing model can detoriate in time, so new pointing models are regularly established. If you notice that the pointing is off by a significant amount, while being well away from zenith or horizon, please include this in your night report. The pointing model is defined with respect to the axis of rotation. In the case of SUSI, this does not quite coincide with the center of the array but is about 15$^{\prime\prime}$--20$^{\prime\prime}$ displaced. With the small pixel size of SUSI, one should be aware of this when windowing the CCD. The tracking makes use of the same software as the pointing and should therefore also be quite good over most of the sky. Without guiding, one can already expect images close to $1^{\prime\prime}$ in not-too-long exposures (10--15 minutes) away from zenith or horizon. Guiding the telescope during an exposure is usually done setting on a star using one of two guide probes located in the rotator/adapter. and using the autoguider. This operation is normally carried out by the night assistant. The resolution of the guide probes is $0.05^{\prime\prime}$. However, the autoguiding is fairly slow (several seconds) and may have problems correcting when the stellar image dances in and out of the box. In poor seeing or strong wind, autoguiding may make the images worse since the corrections always lag behind actual conditions. The NTT is also fairly sensitive to windshake which can lead to similar problems. As a rule of thumb, if the seeing is significantly worse than $1^{\prime\prime}$, your exposures are short (10 minutes or less) and you are not near the zenith or the horizon, consider not to use the autoguiding. \newpage \chapter{Observing in RILD} \newpage \chapter{Observing in BIMG} \section{Optical configuration} In the Blue Imaging mode (BIMG) the instrument works as a focal reducer at F/4. The optics in the blue channel is coated for high efficiency transmission in the region $\rm 300~nm $ to $\rm 500~nm $. A filter wheel is mounted in the converging beam in front of the camera for doing imaging. The light path in this mode is illustrated in Figure~\ref{pathbimg}. Spectroscopy in the blue arm is possible by using gratings. It requires a different configuration of the instrument which is described in Chapter 8: Observing in BLMD. \section{Instrument setup} The instrument setup is simply performed by selecting one position of the filter wheel. This is done in the user interface by going first to \fbox{\tt Top menu} then choosing the BIMG mode by clicking \fbox{\tt BIMG}. After selecting the mode, the setup can be done by clicking the setup command \fbox{\tt BIMG Setups}. A form appears on the screen that allows you to predefine instrumental configurations. Do not forget to focus the camera according to the EMMI temperature and the chosen filters. To define exposures click the command \fbox{\tt BIMG exposures} and fill the form as described in Section 3.5: Defining and executing exposures. \subsection{Filters} \label{filbimg} Up to 8 filters can be mounted in the filter wheel. Blue filters are mounted in cells perpendicular to the optical axis. Information on the transmission curves and quality of the filters can be obtained by using the MIDAS GUI {\tt FILTERS}. A list of standard EMMI blue filters is given in Table~\ref{emmib-filter}. Observers who wish to use other filters are reminded of the image quality requirements. U and B photometry are normally be done in the blue arm of EMMI, the red optics having a sharp cut-on at $\rm 390~nm $ due to the coating. The Bb filter is a blue filter intended for use in the red arm of EMMI. It is open towards the UV so its effective colour equation depends on EMMI and camera transmissions and the CCD QE at $\rm \sim 400~nm $. A red filter used in the blue will cause some astigmatism. If a blue filter is used in the red arm, every object in the field produces a ghost which is about 5 magnitudes lower in brightness. Image anomalies are described in Chapter 11. \section{Observing} \subsection{Focusing using through--focus sequences} \label{sfocthr} Click the focus parameters bar of the BIMG mode i.e. click \fbox{\tt BIMG Focus param.}. \ A form will appear on the RAMTEKdisplay. \ The parameters required are the number of focus steps (typically 7 to 9), the focus increment ({\tt step size}, typically $10-30 \mu$m), and the telescope offset in R.A. or Dec. (or both; typically $10''$). \ The parameters for the instrument focus are also specified in this form. \ Here you must give the number of steps and the camera focus offset. {\tt foc} type exposures use the current position of the telescope or camera focus as the value of the middle step of the sequence. \ Thus, if you want to start at any given focus value, you must position the focus to the value mid-way of the sequence. \ For example, if you want to start at $-3.000$~mm, the focus of the telescope should be positioned to $-2.960$ ~mm for 9 focus steps of $10\mu m$. The optimal focus may be derived using the procedure {\tt FOCUS} (in the MIDAS display window). \ The use of that procedure is described in Appendix~\ref{batches}. \ In order to average seeing effects do not use excessively short individual exposures $\rm (< 20 sec)$ for the focus sequences. \begin{table}[hhh] \caption{\sl EMMI filters for the blue arm} \label{emmib-filter} \vspace*{0.3cm} \begin{center} {\scriptsize \begin{tabular}{||c|l|c|c|c|c||} \hline\hline & & & & & \\ {\bf ESO} & \multicolumn{1}{c|}{\bf Filter} & $\lambda_0/\Delta\lambda$ (nm) & {\bf Peak efficiency} & {\bf Cell type} & {\bf Focus offsets} \\ & & FWHM & {\bf ($\%$)} & {\bf (arm)} & \\ & & & & & \\ \hline\hline & & & & & \\ 588 & He II & 469.0/6.6 & 71 & B & $-$97 \\ \hline 602 & U & 354.2/54.2 & 67 & B & $-$55 \\ \hline 603 & B & 422.3/94.1 & 66 & B & 0 \\ \hline 644 & GG375 3mm & $>$369.2/lwp & 99 & B & $-$10 \\ \hline 647 & Ne V & 342.2/8.3 & 39 & B & $-$80 \\ \hline 648 & O II / 0 & 372.5/6.9 & 35 & B & 45 \\ \hline 649 & O II / 5000 & 379.5/6.7 & 44 & B & $-$70 \\ \hline 650 & O II / 10000 & 385.3/7.0 & 43 & B & 35 \\ \hline 651 & O II / 15000 & 392.7/7.8 & 41 & B & $-$95 \\ \hline 658 & EUV (UG11/5) & $<$366.1/swp & 70 & B & $-$95 \\ \hline 671 & Spec. & 468.0/15.2 & 57 & B & $-$110 \\ \hline 723 & Spec. & 394.9/3.5 & 44 & R & $-$40 \\ \hline & & & & & \\ 587 & He I & 448.0/4.8 & 53 & R & $-$10 \\ \hline 652 & He II & 469.3/7.3 & 62 & R & $-$52 \\ \hline 589 & O III / 0 & 501.4/5.6 & 64 & R & $-$30 \\ \hline 590 & O III / 3000 & 505.7/6.4 & 52 & R & $-$10 \\ \hline 591 & O III / 6000 & 511.2/6.1 & 69 & R & $-$43 \\ \hline 592 & O III / 9000 & 516.0/6.3 & 68 & R & 10 \\ \hline 593 & O III / 12000 & 521.1/6.7 & 66 & R & $-$30 \\ \hline 594 & O III / 15000 & 526.0/6.6 & 65 & R & $-$40 \\ \hline 643 & BG38 2mm & 481.9/276.9 & 96 & R & 0 \\ \hline 645 & OG530 3mm & $>$530.0/lwp & 95 & R & $-$15 \\ \hline 769 & BG39 & 472.2/237.2 & 86 & R & $-$85 \\ \hline \\ \hline\hline \end{tabular} } \end{center} \end{table} The focus in RILD with no filter or with the R-filter is approximately the same as the BIMG focus with the B filter. \subsection{Checking the seeing} The seeing can be checked on images taken in BIMG by clicking the field \fbox{\tt seeing} that is in the menu on top of the MIDAS display window (Sect.~\ref{seeing}, and Appendix A). The image scale is read in the header of the displayed image and does not need to be specified. The seeing measured outside the telescope by the seeing monitor can be displayed on the workstation by typing {\tt meteomonitor} (Sect.~\ref{seeingm}). \subsection{Pointing and guiding} The pointing of the NTT is better than 1.5 arcsec rms. Guiding the telescope is normally done by the night assistant by centering a star on one of the two guide probes in the adapter and starting the autoguider. \subsection{Direct imaging} The camera on the blue arm of EMMI is at F/4 and the detector presently used is a TEK CCD of $\rm 1024^2$ pixels, $\rm 24 \mu m$ in size (ESO No 31). This gives a scale of 0.37 arcsec/pixel and a field of view of $\rm 6.2 \times 6.2$ arcmin. \section{Calibration exposures} \subsection{Bias and darks} \label{biasbimg} It is not safe to assume the bias to be always a scalar and therefore it is recommended to take many bias exposures. It has proven to be extremely difficult to isolate the CCD electronics from electrical interference from components in the NTT adapters/rotators. Therefore to some extent, the EMMI CCDs show pick-up patterns in the electronic background (the bias). This noise is minimized in {\tt SLOW} readout mode, but may be rather strong in {\tt FAST} readout frames. The patterns are not stable, but change from one exposure to the next, so it is difficult to remove them completely by substracting bias frames. However, some reduction can be achieved and, therefore, it is recommended to take a good number of bias frames throughout the observing run. Should strong patterns (i.e. more than a few ADUs) appear on {\tt SLOW} readout bias frames, call the NTT coordinator (93-50). Note that spurious patterns are introduced if images are displayed with demagnification factors. At least one, but preferably more, long (1 hour) dark exposures should be taken to monitor the dark current and any exposure dependent features. The normal values for the TEK1024 CCD (ESO No 31) are as follows. Conversion factor: $\rm 1 ADU = 3.3 e^-$. Read-out noise $\rm \sim 7 e^-$ rms. \ The dark current is $\rm \sim 8 e^-/pixel/hr$ at 166 K. \ There are approximately 8 small traps extending over less than 20 pixels. \subsection{Flat fields exposures} The linearity of CCD No 31 is very good up to $\rm 160 \, 000~ e^-/pixel$. \ The most accurate results for flat fields in broad band imaging are obtained using sky flats. This may be achieved by median filtering of science images, if they are not too densely populated with stars and do not contain very extended objects, or by doing multiple exposures of sparsely populated fields, using spatial offsets. A list of such fields is available in the control room. Another approach is to use morning and evening twilight ($\sim$ half an hour when the sun's elevation is $\sim \rm -12^o$). Tyson (ref []) sequences are often used to obtain flats of similar flux level . There are also dome lamps that can be used to take flat fields in the dome. The introducing astronomer can assist the visitor in their use. \subsection{Absolute flux calibration} Lists of standard stars for calibrations of broad band images with CCDs are available in the control room of the telescope. \section{Instrument performance} \subsection{Shutter timing} A time delay of 0.80 seconds has been measured for the shutter in the blue F/4 camera. Because of the location of the shutter in the optical path, the exposure time over the field is constant and equal to the chosen time plus the average shutter delay. If critical for your application, it is recommended that you check the shutter timing by either taking exposures of increasing duration on a star, or using one of the internal lamps and a pinhole in the apperture wheel. \subsection{Typical count rates} In blue imaging the efficiency of EMMI is the product of the transmission of the atmosphere, three reflections in the telescope, the transmission of the blue coated optics of mode BIMG, filter, and quantum efficiency of the CCD. The efficiency in B and U was checked for the F/4 camera and TEK CCD No 31 in 1993. The count rates in $\rm e^-/sec$ deduced for a 15th magnitude A star with approximately zero colour are U: 2 200, $\rm B:~16\,900$ at unit airmass. \subsection{Colour equation} An approximate colour equation beween u and b has been obtained from standard stars. The data for CCD No 31 give the following relation: $$\rm U~ -~ u = +0.01~(u~ -~ b) +~ 23.34 $$ \subsection{Check list and remarks} A general check list for observations can be found in Sect.~\ref{checkl}. Please make use of it: it is intended to help in tracking errors and saving observing time. Remember that you can get information on current technical problems by typing {\tt nttrep} on the workstation (Sect.~\ref{obsrep}). This NTT report facility will have to be used to report on any technical problem during your observing run. The operation staff will also use it to report on technical solutions. To get information on the seeing and other meteorological local parameters type {\tt meteomonitor} . \newpage \chapter{Observing in REMD} \newpage \chapter{Observing in BLMD} \newpage \chapter{Observing in DIMD} \newpage \chapter{Observing with SUSI} The Superb Seeing Imager (SUSI) is physically distinct from EMMI but complements its observing capabilities. A supporting plate is mounted on the adapter of the Nasmyth A focus of the NTT. On the plate, a mirror with 3 positions is mounted. The first position sends the light to SUSI, the second feeds an IR camera (not available), and the third position is free for the operation of IRSPEC. A CAD view of the assembly is presented in Figure~\ref{susi}. Between the diagonal mirror and the CCD is a filter wheel with 8 positions. In the control software, SUSI is also referred to as DIFA: the Direct Imaging FAcility. \begin{figure}[hhh] \vspace*{10cm} \caption[\rm CAD drawing of SUSI identifying its major components] {\em CAD drawing of SUSI identifying its major components. The second dewar shown in the figure corresponds to an IR array camera.} \label{susi} \end{figure} \newpage SUSI uses a TEK CCD (ESO \#25) with $1024\times1024$ pixels of $24\mu$m corresponding to $0.13''$ on the sky. The field of view is thus $2.2^\prime \times2.2^\prime$. More details about the CCD can be found in the ESO CCD manual, or on the ESO WWW pages. \section{Should I use SUSI?} SUSI has two advantages over EMMI. First, the small pixel scale which gives a much better sampling of the point-spread function. Second, the lack of optics which gives better throughput. It also has two disadvantages: the CCD is much less efficient in the blue than the one in th blue arm of EMMI, and the field is small. SUSI is therefore well suited for either periods of good to very good seeing ($<0.8^{\prime\prime}$) or deep imaging in the red. For seeing $>0.8^{\prime\prime}$, the red of arm of EMMI will in most cases give adequate sampling, and the blue arm for seeing $>1.0^{\prime\prime}$. \section{SUSI filters} The filter wheel of SUSI accepts up to 7 filters 60mm in diameter (one position is kept free) There is a basic set of filters for SUSI which are listed in Table~\ref{susifilters}. However, there are many more available which can be found in the ESO filter catalogue. In particular, the EFOSC filters fit SUSI directly without the need of special adapters. The filter catalogue can be accesed within MIDAS (with the command CREATE/GUI FILTERS), or from the WWW ESO pages. \begin{table}[hhh] \caption[\sl SUSI filters: \ basic set]{\sl SUSI filters: basic set} \label{susifilters} \vspace*{-0.2cm} \begin{center} \begin{tabular}{||c|c|c|c||} \hline\hline {\bf ESO \#} & {\bf Filter} & $\lambda_0/\delta\lambda$ (nm) & Peak efficiency \\ & & (FWHM) & (\%) \\ \hline 640 & U & 354.2/53.1 & 69 \\ 639 & B & 434.0/101.1 & 58 \\ 641 & V & 547.2/113.2 & 80 \\ 642 & R & 643.8/166.7 & 85 \\ 703 & Gunn g & 515.2/75.1 & 82 \\ 621 & Gunn r & 678.5/81.5 & 84 \\ 705 & Gunn i & 803.3/150.8 & 97 \\ 623 & Gunn z & 837.0/lwp & 97 \\ 724 & Tys. B & 444.5/183.8 & 82 \\ 707 & OIII & 500.9/6.51 & 82 \\ 708 & H$\alpha$ & 656.8/6.92 & 90 \\ 709 & H$\alpha$r & 664.5/7.12 & 88 \\ 742 & H$\beta$ & 486.1/7.20 & 83 \\ 743 & H$\beta_c$ & 477.1/7.19 & 81 \\ 700 & SII & 673.0/6.05 & 60 \\ \hline\hline \end{tabular} \end{center} \end{table} \section{SUSI control software} SUSI is nowadays run from the same terminals as EMMI. To start the SUSI control programme, log-on with username {\tt SUSI} (no password required) and follow the instructions that will appear on the screen. As EMMI, SUSI uses a mouse driven graphical interface on a Ramtek monitor. After SUSI starts, the Ramtek UIF displays a number of bars on the right hand side. Click \fbox{\tt SUSI observations}. \ A number of new commands will appear. The command \fbox{\tt Define Exposures} allows you to define up to 8 exposures in the same way as for EMMI. Only 4 types of exposures, however, exist for SUSI: {\tt dk} (dark), {\tt sci} (scientific), {\tt ff} (flat field), and {\tt foc} (focus exposures). The list of filters mounted on the filter wheel (up to 8) appears on the form, and can be selected by number or by name. \ The \fbox{\tt Start exposures} command is used to start the sequence defined in the form. Switching from EMMI to SUSI only takes a few minutes. However, it is necessary to move the telescope to point at the zenith: only in that position can M3 be turned, to move the beam from on Nasmyth platform to the other. The EMMI control software should be terminated, and SUSI started up as above. You will have to redetermine the focus, and if you are switching because of significant improvements in the seeing, a new image analysis should also be done to take advantage of the new situation. \section{Focusing the NTT with SUSI} Telescope focus is critical to obtain good images. The slow angle of the NTT beam at the Nasmyth focus (F/11) facilitates focusing the telescope with SUSI, which must be done using through-focus sequences. The parameters for these sequences are entered in the \fbox{\tt SUSI Exposures} form. Tests done on reasonably good seeing ($0.75''-1.0''$) showed that the optimal focus step is 30 microns. Step the telescope by $\sim10''$ in the most convenient direction depending on the field. $7-9$ focus exposures give the best results. The MIDAS observing batch {\tt FOCUS} (Appendix A) can be used to analyse the focus sequences. The following table gives approximate focus values for the most used SUSI filters. These values should only be used as initial guesses for your focus sequences. The NTT focus changes with temperature as $\Delta F/\Delta T = 0.0764$~mm/$^{\circ}$C. If the main mirror is activated, it is recommended to check the focus every time a correction is made (cf. Appendix F). To make the most of a period with sensational seeing, the focus should be checked often. Slightly elongated images are a good indication of a focus drift (or of gravitational arcs): use the MIDAS command CENTER/IQE to check. Images of $0.34^{\prime\prime}$ have recently been obtained in 10 minute integrations, so it is possible! \begin{table}[hhh] \caption[\sl Indicative SUSI telescope focus]{\sl Indicative SUSI telescope focus for $T=10^{\circ}$C} \vspace*{-0.2cm} \begin{center} \begin{tabular}{||c|c||} \hline\hline & \\ *[-0.2cm] {\bf Filter} & {\bf Telescope focus (\bf mm)} \\ & \\ *[-0.2cm] \hline & \\ *[-0.2cm] U & $-3.48$ \\ B & $-3.47$ \\ V & $-3.50$ \\ R & $-3.47$ \\ I & $-3.43$ \\ Z & $-3.43$ \\ *[-0.2cm] & \\ \hline\hline \end{tabular} \end{center} \end{table} To speed up focusing of SUSI the following initial guesses may also be useful: \begin{center} \begin{minipage}{12cm} focus[SUSI(V)] = focus[RILD(R)] $-$ 0.26 \\ focus[SUSI(V)] = focus[Image analysis side A] $-$ 0.10 \end{minipage} \end{center} \section{Calibrations} As with all CCDs, the afternoon should be used to check that the CCD is in good working condition. Because SUSI is not used as often as EMMI, faults can go unnoticed for some time and the night is not the time to find out about those! Take a few biases and check the read-out noise and check for the presence of patterns. The CCD is normally used in the slow mode, especially when used with binning: the time gained in reading out in fast mode is 25sec in slow mode and 5sec with $2\times2$ binning. The CCD is linear up to 150000 e$^-$/pixel. Dome flats have been found to flatten images to about 3\%. Sky flats are generally more accurate: SUSI shows some vignetting which is not well removed with dome flats. For twilight flats Tyson sequences have been used with good succes: the program to calculate these is available on the off-line computersystem (lw0--lw10) under the directory {\tt /astro/progr1/twilight}. Twilight flats should preferably be taken in the evening. If the background in your science images is bright enough, you can also create flats out of these. A photometric system for SUSI has not been established. The normal calibrations using the Landholt standards or other CCD standards can be used, but due to the small field size it is difficult to get more than one star on the CCD. If you have determined colour terms for SUSI, please let us know! \newpage \chapter{Additional Information about EMMI} \section {Ghosts and image anomalies} \subsection {Imaging (RILD and BIMG)} Most ghosts in imaging are due to a reflection between the CCD and a lens surface, and as such, they depend on the reflectivity of the CCD surface and the efficiency of the antireflection coatings. \ The antireflection coatings on EMMI have a low reflectivity (1\% in the red and 0.4\% in the blue, compared to ~2\% for EFOSC). A sky concentration already familiar from EFOSC is also present in EMMI. \ This type of effect is very difficult to avoid in any focal reducer design. \ It is due to light ---~from the sky background and from stars~--- that is reflected back into the camera by the CCD and returned by some optical surfaces. The effect is not very noticeable in normal flatfield exposures. An on-axis star has a faint halo around it with an intensity level dropping to $10^{-4}$ at 5~pixels from the parent star (RILD). \ This was measured without any filter in the beam. In RILD, filters are tilted so reflections between the CCD surface and the filter are excluded. However, multiple reflections inside the filter may in principle lead to satellite images close to the parent if the filter has a small wedge. \ No such ghosts with a level greater than $10^{-3}$ were found for filters 587 to 645. \ In BIMG, filters are in the converging beam and no nearby in-focus ghosts are expected. \subsection {Spectroscopy (RILD, REMD and BLMD)} Image ghosts and anomalies may originate from both the grating (or grism) and the spectrograph optics. Grating ghosts are caused by periodic variations in the position of the grooves. \ Long-periodic errors of the grating engine produce line satellite (Rowland) ghosts which perturb the line shape and so may change the width and shape of the point spread function. \ The observer should be very careful in looking for faint or unidentified lines, especially when observing objects with strong emission lines, because these might be the more dangerous Lyman ghosts. \ These are far away from the parent line and are caused by short-periodic errors of the ruling engine. \ According to manufacturers' specifications, all EMMI gratings have Lyman ghosts $< 10^{-4}$ except No.~6 ($3.8\times10^{-4}$), No.~7 ($1.8\times10^{-1}$) and No.~9 ($17\times10^{-4}$). The efficiency curves of some gratings show irregularities called Wood's anomalies. \ These anomalies depend strongly on the polarization of the incident light and are most prominent when the polarization is perpendicular to the grooves. \ Wood's anomalies affect gratings numbers 3, 7, 8 and 11. The curved cemented surface inside one of the doublets of the red collimator happens to be exactly perpendicular to the incoming diverging beam. \ It is effectively behaving like a plane parallel glass plate in front of the grating (however with a very low reflectivity of about $4\times10^{-4}$). \ The white light ghost image of the slit generated by this surface was moved outside the CCD field by applying a small tilt to the collimator. \ A complementary ghost is formed by light that is dispersed by the grating, reflected by the cemented surface and dispersed a second time. \ This appears on the CCD as ghosts of a few times $10^{-4}$ of bright emission lines that are in the right--hand half of the CCD image, located approximately twice as far from the right--hand edge of the CCD as their parents. \ So far we only noticed ghosts in calibration echelle spectra, where some of the bright Ar lines of the ThAr calibration lamp produce ghosts of this type. Interorder stray light was measured with grism \#3 as cross--disperser and the red flatfield lamp. \ The absolute level is highest near 7000\AA\ and reached 1.9\% of the continuum with echelle grating \#9. \ With \#10 the stray light level was 6\% of the continuum at 7500\AA. \ On request, a stray light mask can be mounted in the intermediate focal plane (between the collimator and the focal reducer). \ This reduces the stray light level to 1.2\% and 3 \%, respectively. A stray light gradient which does not depend on the continuum intensity is present when combining echelle grating \#10 with grism \#3 as cross--disperser. \ The stray light is strongest near the top of the CCD with a level of about 50~e$^{-}$ in a 1~hour exposure. \ The effect was only observed with grism \#3 and was not noticed with grism \#4 as cross--disperser. \ The stray light seems to be generated inside the transfer collimator and its cause will be further investigated. The blue mirror train shows a 10\% reflectivity dip with a width of about 50\AA\ in one polarization at around 370~nm, which could easily be mistaken for an absorption feature. \ The antireflection coatings of the red lenses also show some dips that are however not polarization-dependent and several 100\AA\ wide. The dichroic prism shows large variations in the efficiency in the polarization perpendicular to the slit up to 30\%. \ The region between 400 and 480~nm is heavily polarized and it is not recommended to use the dichroic prism on polarized objects. Efficiency variations normally flatfield out for nonpolarized objects. \ The observer must take extra precautions when studying strongly polarized objects. \section{Image quality, scale and distortion} \subsection {Imaging} The measured scales determined using stars in an astrometric field are $0.268\pm 0.004$~arcsec/pix in the red, and $0.\pm 0.$~arcsec/pix in the blue. \ For a pixel size of $\rm 24 \mu$m, this corresponds to and $ \mu$m/arcsec in red and blue, respectively. The above coefficients are but the first terms in the polynomial that describes the transformation from sky to pixel coordinates. \ Higher order terms must take into account lateral colour (chromatic variation of focal length) which goes with the first power of the radial distance, and 3rd and 5th order distortions. \ The total contribution of these terms is less than $ \mu$m at ~mm from the centre of the CCD. \begin{figure} \vspace{20cm} \caption[Spot diagram showing theoretical image quality in RILD with the F/5.2 camera]{\em Spot diagram showing theoretical image quality in RILD with the F/5.2 camera, at a single focus setting at various wavelengths and radial field points.} \label{spot2} \end{figure} \begin{figure} \vspace{20cm} \caption[Spot diagram showing theoretical image quality in BIMG with the F/4 camera]{\em Spot diagram showing theoretical image quality in BIMG with the F/4 camera, at a single focus setting at various wavelengths and radial field points.} \label{spot3} \end{figure} Spot diagrams in RILD and BIMG cameras are given in Figures~\ref{spot2} and~\ref{spot3}. \ These were calculated assuming a perfect telescope and atmosphere. \ In RILD the image quality (80\% energy concentration) is better than $10 \mu$m in most of the field and better than $20\mu$m out to the corners. \ In BIMG the image quality is better than $20 \mu$m in most of the field. The on-axis image quality that was actually achieved, was measured by imaging a starplate with pinholes of different sizes in the telescope focal plane onto the CCDs and measuring the FWHM of the image. \ The instrument PSFs found (FWHM of 0.~arcsec (~pix) in the red and 0.37~arcsec (1.3~pix) in the blue) are clearly dominated by the pixel size.\ In the case of blue wavelengths, photon emission from the coating does also degrade the image quality. \ Thus, a $0.7''$ image delivered by the telescope to the Nasmyth focus is e.g. degraded by EMMI to $0.''$ in the red and $0.79''$ in the blue. The spot diagrams show that optical aberrations and lateral colour produce image elongations in the field that cannot be neglected. \ Note that the elongation axis direction varies with field position. \ Other possible factors contributing to image elongation are: sampling effects in the CCD, filter image quality, atmospheric dispersion and telescope tracking including wind buffeting effects and field rotation. \ Most of these produce an elongation that is constant in the field. \ Image size differences of $0.1''$ in X and Y on $1''$ images are normal and do not necessarily point to telescope tracking errors. \subsection {Spectroscopy} The image scales given for imaging also apply to EMMI used in spectroscopy. \ High-angle grisms and gratings produce anamorphosis which results in a slit image that is narrowed (the usual case) or widened by a certain factor. \ The scale along the slit is not affected. \ At the CCD centre, the anamorphic factor A is equal to: \[ A = \frac{\cos(\theta + \frac{\phi}{2})}{\cos(\theta-\frac{\phi}{2} + \psi)}\, , \] where $\phi$ is the angle between incident and diffracted beams at the field centre, $\theta$ the grating angle and $\psi$ the field angle. \ For gratings we have: \[ \theta = \arcsin\frac{(n \times m \times \lambda)}{2 \times 10^{+7} \times \cos \frac{\phi}{2}}\, , \] where $n$ is the number of grooves/mm, $m$ the order, $\lambda$ the wavelength in \AA. \ $\phi =5.5^{\circ}$ in EMMI while $\psi=arctan\frac{x}{120 \times F\#}$, where $x$ is the distance in the CCD focal plane from the field centre in mm and F\# is the camera speed. Gratings \#9 and \#10 have $\theta=28.7^{\circ}$ and $63.5^{\circ}$; the anamorphism is 0.96 and 0.82, respectively. Grisms have A = 1 at the central wavelength, A $>$ 1 towards the red and A $<$ 1 towards the blue. \ Here we have: \itemsep 0.3cm \begin{itemize} \item $\theta = 40^{\circ}$ for grism \#6 and $26^{\circ}$ for grism \#5 (less for other grisms) \item $\phi = 0^{\circ}$ \item $\psi = arctan$ ($\frac{x}{50 \times F\#}$), where $x$ is the distance from the field centre in mm. \end{itemize} In the case of grism \#6, A = 0.94 at the red end of the spectrum and A = 1.07 in the blue. \ Grism \#5 has less anamorphosis with A = 0.97 and 1.04, respectively. \ The anamorphosis for other grisms is less. If grisms are used as cross--dispersers in echelle spectroscopy, the scale in the slit direction will be affected: in the case of grism \#6 for instance, the height of the spectrum is 14\% smaller in the red orders than in the blue. With high-angle grisms and gratings, long slit spectra will show spectral line curvature. \ Spectral lines have a parabolic shape with shifts towards the blue at the upper and lower edges of the slit of up to several pixels depending on the grating used and the wavelength settings. \ The long-slit spectral reduction package LONG in MIDAS automatically corrects this distortion. \section {Filter properties} All filters are permanently mounted in their cells. \ Although it is possible to use blue filters in the red and vice versa (for example in the overlap region; 400 to 500~nm), filters should normally be used on the wheel they are intended for. \ Blue filters are mounted at 0~degrees in their cells while red filters are at 5 degrees to avoid reflections between the CCD and the filter. \ Below slit filters are glued in a rectangular cell that only permits to use them on one of the two below slit wheels of EMMI. As a general rule, all filters are blocked to better than $10^{-4}$ to $1.2 \mu$m. \ Note that the optices of the blue arm of EMMI provides some additional blocking to red light: 10\% beyond 600~nm\ and less than 2\% between 800~nm\ and $1.2 \mu$m. The spectral properties of colour filters are to a great extent independent of the angle of incidence, $\rm \phi$, and can be assumed to be constant within the field, both in blue and red imaging. \ The central wavelength of interference filters shifts to the blue if this angle deviates from 0~degrees. \ The following formula applies: $$ \lambda = \lambda_{0} \sqrt{\frac{1-\sin^2\phi} {n}}\, , $$ where $\phi$ is the angle of incidence and n the effective refractive index of the material forming the spacer cavity. \ The EMMI blue filters are placed in the diverging F/11 beam in front of the blue camera, so the effective filter curve (averaged and weighted for all incidence angles in the F/11 cone) will be somewhat broadened and blue-shifted compared to the filter curves measured at 0~degrees incidence. \ The effects are likely to be on the order of only a fraction of nm in the blue and may be neglected with the bandpasses used in the current set ($> 5$~nm\ FWHM) of filters. The EMMI red filters are placed in the parallel beam between the collimator and the camera. \ The angle of the beam with the optical axis depends on the field position of the object, and is for instance 4.5~degrees for a field point on the edge of the Thomson chip (F/2.5 camera, F/125 mm). \ However, red filters must be tilted 5~degrees in order to avoid reflections between the filter and the CCD, and so the angle of incidence is 0.5 degrees at the lower edge of the chip and 9.5 degrees at the upper edge. \ The effect on the bandpass of filter \#601 has been measured at incidence angles between 0 and 15~degrees, and a value of $n=2.06$ has been found. \ The wavelength shift is significant: about 0.3\% of the central wavelength (30\% of the FWHM: 2.2~nm) at 9.5~degrees in the case of \#601. Observations are affected in two ways. \ First, observers should remember that filter central wavelengths, and to a lesser extent also FWHMs, are field-dependent and take account of this, for instance by positioning critical objects nearer to the lower edge of the chip. \ Secondly, bright sky emission lines that are not in the filter passband at edge of the field may be just in at the other edge, giving rise to a background slope which then requires additional attention in flatfielding. U and B photometry should normally be done in the blue arm of EMMI. \ The Bb (blue B) filter is a B filter intended for use in the red arm of EMMI, and is open towards the UV in order to approximate the regular B as well as possible. \ The red optics of EMMI have a sharp cut--on at 390~nm due to the coatings, and so the effective central wavelength/bandpass of the Bb filter is 429.5/79~nm\ as opposed to 423/94~nm\ for the B filter used in the blue arm of EMMI. \ We expect that better results (efficiency and colour transformation coefficients) will be obtained with the B filter, but no systematic study has been made so far. Since the blue and below slit filters are used in diverging beams, they affect the focus. \ The red filters are used in parallel beam and introduce no significant defocus in most cases. \ The below slit filters are normally used with a wide slit for flux calibration on standards, where it is not important to achieve the optimum spectral resolution, so no focus correction is necessary. The image quality of some filters was investigated by imaging a $200 \mu$m ($1.07''$) pinhole in a starplate. \ The geometrical size of the image on the CCD corresponds to 3.8~pix in the blue and 2.4~pix in the red. \ There is some evidence of image degradation by filters \#588 and \#606. \ Some improvement may be expected by refocusing, as the focus was set at the optimum found for the B and R filters. \ No nearby ghosts could be identified down to a level of $10^{-3}$. \section{Image stability and flexure} EMMI rotates with the adapter to follow field rotation. \ Depending on the duration of the exposure and the location of the object, instrument rotations of 180~degrees and more may result. \ The instantaneous angle is displayed by the instrument control program. Instrument flexure has been measured in RILD and BLMD with a $1''$ decker on the slit and a test mirror instead of the grating. \ The position of the image on the CCDs was measured at various adapter angles. In RILD the peak to peak difference is 0.7~pixels. In BLMD, the peak to peak excursion in the dispersion direction is 0.35~pixels and 1.9~pixels along the slit. In REMD it is 0.9~pixels peak to peak in the dispersion direction. The blue and red grating units position the gratings with a reproducibility of better than 0.1~pixel in the dispersion direction, so it is possible to save observing time by performing the calibrations in the morning after the observing. \section {Instrumental polarization in EMMI and SUSI} A systematic investigation of the polarization induced by mirror 3 in the NTT and the mode-selecting mirrors in EMMI and SUSI has not been carried out yet. \begin{table}[hhh] \caption{\sl Instrumental polarization of EMMI} \label{emmipol} \begin{center} \begin{tabular}{||c|c||} \hline\hline & \\ *[-0.2cm] {\bf PA slit} & {\bf Flux (\rm $ \rm e^-/s/pix$ at 700~nm)} \\ & \\ *[-0.2cm] \hline & \\ *[-0.2cm] 135 & $29.7 \pm 0.2$ \\ & \\ *[-0.2cm] 180 & $30.0 \pm 0.2$ \\ & \\ *[-0.2cm] 225 & $30.2 \pm 0.2$ \\ & \\ *[-0.2cm] 270 & $30.4 \pm 0.2$ \\ *[-0.2cm] & \\ \hline\hline \end{tabular} \end{center} \end{table} A single measurement was obtained in the following way: a standard star (FEI 110) was observed with $9''$ slit and grating \#6, centered around 700~nm. \ The exposure time was 300sec; the different exposures were taken at different P.A. of the rotator. \ The efficiency of this grating at 700~nm in the two polarizations is 45\% and 83\% respectively. \ If the 3rd mirror of the NTT introduces significant polarization, the efficiency is expected to change with rotator angle. \ The measurements gave the values shown in Table~\ref{emmipol}. These preliminary measurements, therefore, indicate that the polarization induced by the NTT and EMMI may affect the accuracy of polarization measurements at the $\rm 1 \%$ level. \section{Bibliography} \begin{description} \item[[1]] Wallander A.: \ 1993, {\em Remote Control of the 3.5m NTT User Guide}, ESO Operating Manual No 17. \item[[2]] D'Odorico, S., \ Ghigo, M., \ Ponz, D.: \ 1987, {\em An atlas of the Thorium-Argon Spectrum for CASPEC in the 3400--9000\AA\ region}, ESO Scientific Report No. 6 \item[[3]] Dekker. H., \ Delabre, B.: \ 1987, Applied Optics, {\bf 26}, 8, 1375 \item[[4]] Dekker, H., \ Delabre, B., \ D'Odorico, S.: \ 1986, {\em SPIE}, {\bf 627}, 339 \item[[5]] Gilliotte, A.: \ 1992, {\em Image Quality Filters Catalogue}, {Internal ESO publication} \item[[6]] Melnick, J., \ Dekker, H., \ D'Odorico, S.: \ 1989, ESO Operating Manual {\bf \#4} \item[[7]] Prieur, J.-L., \ Rupprecht, G.: \, 1990, {\em Efficiencies of EMMI}, ESO internal report \item[[8]] Wilson, R. N., Franza F., Noethe L., Andreoni G.: \ 1991, {\em Journal of Modern Optics}, {\bf 38}, 219. \end{description} \end{document}