Users Manual

Click here for latest news!



Design of the instrument

MOSCA is installed at the RC focus of the 3.5m telescope. The design of the instrument itself is quite simple. Various focal plane apertures - mask, slit, or multislit - can be selected according to the observational mode. A collimator forms a parallel beam which passes one or two analysers (grism, filter, FPI, and/or polarizer) and is then focussed by a camera optics onto the detector. A second filter wheel is placed in the convergent beam. The calibration unit has three spectral lamps for wave length calibration and a continuum source.

The design is modular; the modules are put into slots in the instrument. Available modules are currently calibration unit, aperture unit, filterwheel 1, filterwheel 2, grismwheel, FPI. Note that the 2 filterwheels are not identical, each one has its own slot in the instrument, so both filterwheels can also be used simultaneously. Filterwheel 1 is in the parallel beam whereas filterwheel 2 is in the convergent beam. The FPI is put into the slot which normally takes up filterwheel 1.

The instrument is controlled with a gui. The data are stored on disk in FITS format. You can look at and analyse your data using MIDAS, IRAF or IDL. However, several utilities (e.g. computation of offsets, alignment of masks) are available in MIDAS only.

Image scales (approximate)

Starting MOSCA

Note: since the computer environment is permanently subject to changes, the actual start procedure may differ from the one described here!
  • log in workstation "ultra1" , for userid and password ask the local staff.
  • wait until a GUI opens, click CCD+MOSCA. This will open several xterm-windows.
  • enter: start_mosca in window labelled instrument. Wait until this is finished, and then
  • enter: start_ccd in the same window.
If you start your observing run make your own directory in the area given to you by the local staff, eg. by
mkdir mydata
and change to this directory by the command
cd mydata
Now start two parallel MIDAS sessions, one for running the receiver, and another one for quick look of the data:
  • to start the MIDAS quick look, change to your directory in the xterm labelled MIDAS quick look, enter
    inmidas -p 33
    Use this window to execute the prgs available for MOSCA
  • in the xterm labelled receiver, change to your directory, enter
    inmidas -p 31
    Start the MIDAS receiver with
    @@ getfits
    when the MIDAS session has started. This receiver converts new files to bdf format and flips or rotates the files to standard orientation. Binning is taken into account, and the use of windows on the CCD does not cause problems too.
  • a third MIDAS session for analysing data may be opened in the window labelled MIDAS data reduction. Do not forget to start this too in your directory!
Note that these MIDAS sessions are working parallel on the same area. When the MIDAS receiver has converted a frame, you may look at it in the quick-look session or data reduction session.

To get this manual on-line start netscape
netscape -install &
If the manual is not bookmarked go to

Now you are ready to go!


The GUI corresponds to the configuration chosen. The status of the instrument is always clearly visible in the fields next to the yellow buttons. All yellow fields are clickable with the right mouse button. The use of the GUI is self-explanatory; nevertheless we give here an overview over the most important functions:
  • calibration: this opens a sub-GUI to select the calibration lamps. Note that turning-on of a lamp automatically causes the mirror which reflects the light into the optical path to be moved in. With "exit" you close the sub-GUI.
  • aperture: opens a sub-GUI. You can select the aperture to use, change the decker position either by moving the grey arrow with the left mouse button or by clicking into the yellow field "masked range" with the right mouse button and then select "keyboard" in the menu which opens up, the number you enter is in arcsec. In the field "decker: fixed position" select the length of the slit. The width of the slit can be changed by moving the arrow with the left mouse button or by clicking into the yellow field, selecting "keyboard" and entering the desired slit width in arcsec. With "exit" you close the sub-GUI.
  • filter 1: control of filter wheel 1;its use is self-explanatory
  • filter 2: control of filter wheel 2
  • grism-wheel: control of grism- wheel
  • setup: to edit the filter/mask/grism lists or to print these lists.
  • config: shows the configuration of the instrument
  • quit : to leave the gui
  • restart: to restart, eg. after changing of a module
  • sleep: to set MOSCA in the position for storing the instrument
  • lambda-cal.: prepares wavelength configuration
  • flat :prepares internal flatfielding

Preparing MOSCA

The CA staff will put the desired filters in the modules, update the filter list, clean-up the disk space of the data disk and align the columns of the CCD parallel to the slit. The following steps should be done during daytime by the observer together with CA staff:
  • find the internal focus of MOSCA as described here
  • use the procedure @@ install to measure a reference position (hole) and store it to the data bank
  • measure the position of the slit on the detector (select the continuum lamp (60% intensity), slit 1-2 arcsec, grismwheel in free position, and take an exposure of 5 seconds.
  • find the focus of the TV guiding unit on the slit of MOSCA: calibration unit out, select the slit, TV guider in slit viewing mode, low lights in dome on, main mirror open. Move TV guider focus until the slit is in focus (77 or so). Note the focus.
The last step you have to do is best done during dawn.


Since MOSCA has a very fast f-ratio of f/2.7 it is vital to determine the focus of the instrument and the telescope carefully, especially under good seeing conditions!

Internal (MOSCA) focus

The internal focus, which means that the apertures are focused onto the detector, is achieved by moving the camera optics. Focus differences between filters are automatically corrected for if standard MOSCA filters are used; the focus offsets are contained in the filter table. To take a focus series,
  • select all filters/grisms free
  • Use the procedure @@ autofocus to take the focus series automatically and analyse it on the MIDAS quick-look with @@ focus. The internal focus in white light must be near 8300! If you are happy with the focus series, set the focus when @@ focus asks you to do so. Note that this is the focus zero point. When you use a filter the actual focus will differ from this value. Both focus values - focus zero point and actual focus - are displayed in the GUI.
The internal focus is very stable; nevertheless it is recommended to check it a few times during an observing run.

External (telecope) focus

  • select the slit, turn on the light in the dome and focus the TV Guider onto the slit (the focus is near 77) . Then move the telescope to a bright star, and change the telescope focus until the stellar image on the slit is sharpest. The telescope focus is near 31000.

Now you are already quite close to the best focus. If you do only spectroscopy, you are ready to start to observe. If you do direct imaging, improve upon this focus by one of the following steps:

  • focus series. Take a focus series of a field on the sky using @@ autofocus. The exposure time should not be less than 10 seconds. Change the telescope focus by about 100 units from exposure to exposure. Take 7 or more exposures. Load the image into the MIDAS display and evaluate the focus series with @@ focus and then set the telescope focus. If the seeing is less than 1.5 arcsec it may be good to repeat the procedure with smaller steps!
  • use of the focussing prism (if installed and initialized). Select the prism in the grism wheel. Move the telescope to a suitably bright star (M=10 requires a few sec integration time in a broad band filter) and take an exposure. Do not overexpose! The prism splits the stellar image into 2 images, the positions of which are used to determine the focus. Load the image into the MIDAS display and measure and apply the focus offset with @@ fpris.

Direct imaging


The focal ratio of MOSCA is f/2.7 which means that it is vital to determine the focus accurately! Proceed as described here

Due to the excellent pointing of the telescope it is usually not necessary to check the position of the telescope if you have good coordinates. If you want to check the pointing anyway, a short exposure of 10 seconds in a broad band filter goes as faint as 20 mag and is usually sufficient. Use @@ offset x,y to place an object into the desired position x,y on the CCD. The offset autoguider is recommended for exposure times longer than 1 minute.

The procedure @@ sequence allows to take a series of exposures with dithering of the telescope.


Flat fielding can be done in one of the following ways:
  • using the internal continuum source
  • illuminating the dome (domeflat)
  • taking dawn/dusk sky (skyflats)
  • taking a median of many object frames
Flats with the internal continuum source are useable for determining pixel to pixel variations, domeflats are ok for online quick-look. Better results can be obtained with skyflats, but best results are obtained by first correcting multiplicative terms with skyflats; then a median over at least 5 skyflats corrected dithered frames is taken pixelwise (command aver/window in MIDAS) to compute an additive term which is subtracted in a second step.

@@ autoflat is an automatic procedure to determine exposure times for flats and taking a series of dome or sky flats. For domeflats @@ expose will work too.

To take skyflats, make sure that the TVGuider is in the guiding position (otherwise there is vignetting). Move the telescope to E for evening and W for morning skyflats, tracking on. @@ autoflat is useful to take the flats. If you want to do it manually,
  • set the TVGuider is in the guiding position
  • select a small (eg.100x100pixel) area near the center of the field
  • expose for 1 sec, measure the light level. If the CCD is not saturated, calculate the exposure time for the desired level, set the CCD to full frame and expose. If the CCD is saturated, wait if you are taking evening flats, or go to bed if you are taking morning flats. The exposure times should be in the range 5-150 seconds.
  • the exposure time changes from flat to flat by a factor of about 2.3 due to the setting/rising of the sun. Do not forget to move the telescope by 10-20 arcsec in an arbitrary direction between the flats, so stars which may be in the flats can be taken out with a median over all skyflats.
Since taking of skyflats may not be possible because of weather conditions, it is a good idea to take several domeflats for all filters one intends to use in an observing run. @@ autoflat or @@ expose may be useful.
  • close all doors
  • turn off all lights
  • turn on the flatfield lamp
  • turn the tlelescope tracking off
  • point the telescope towards the flatfield screen
  • open the mirror and the Cass shutters
  • make sure that nobody enters the dome!
  • set the autoguider in guiding mode
The exposure times depend on the set-up of the photo lamp. Try the following exposure times, but make sure that you do not overexpose!

Filter U B BV V R I z
lamp 100% 50% 50% 50% 50% 50% 50%
exp.time 35 25 7 6 2 1 2

Internal flats
Approximate exposure times for internal flats are:

Filter U B BV V R I z
lamp 100% 60% 40% 40% 60% 100% 100%
exp.time 60 14 14 17 6 20 22

Flux calibration

For bright 10-11 mag standard stars exposure times of a few sec should be ok, but make sure that you do not overexpose, especially when the seeing is very good (defocussing the telescope may help in this case).

Limiting magnitudes

The limiting magnitudes depend of course on seeing, atmospheric conditions, filter and CCD. Here are some useful measured limiting mags:

Let me know some numbers!!!!

A preliminary version of an exposure time calculator can be found here !



If you do only spectroscopy, it is probably sufficient to focus the telescope on the slit using a bright star and the TVguider. If you have determined the telescope focus as described here you are ready to go.


A set of 7 grisms with dispersions from 1.3 to 6 Angstroem per pixel is available:

grism            lambda   lambda_central  dispersion
                  [nm]        [nm]        [A/pixel]
green_250        330-1000     570             6.0  
red_500          540-1000     850             2.9
blue_500         330-650      480             2.9
green_500        430-820      550             2.9
blue_1000        330-510      400             1.3
green_1000       420-650      530             1.3
red_1000         570-820      680             1.3

Note: the mean dispersion is given for a pixelsize of 15 microns

Total efficiency curves are given here.

Placing an object into the slit

There are two ways to place an object into the entrance slit:
  • if the object is bright enough, use the TV guider in the slit-viewing mode. Under good conditions, 19-20 mag objects are visible.
  • if the object is too faint for this, take a short direct image (select aperture free, grismwheel free). Measure the position of the object on the detector and offset the telescope using the procedure @@ offset. If you want to rotate the instrument, measure the position angle with @@ pa (note that the slit is oriented in EW direction, i.e. at pos.angle 90!), then rotate the instrument, take a short exposure and use @@ offset to move the telescope. Select the desired grism, the slit and take the spectrum.

Only the southern slit jaw moves.

Wavelength calibration

The slit position is reproducible within 1/7 arcsec at various telescope positions. So it is probably sufficient to take only a few calibration spectra during the whole observing run and correct small shifts using the night sky lines. The command @@ calspec is very useful for this. Exposure times given below are for the LORAL 11i:

grism green250 red500green500 blue500red1000green1000blue1000
Ar 25 25 40 3000 45 3000 3000
HgAr 20 20 2015 45 30 200
Ne 15 15 1520 25 30 3000
Ar+HgAr+Ne 15 15 2020 30 30 2000
recommendation HgAr,Ar HgAr,Ar,Ne HgAr,NeHgArNe,Ar HgAr,Ne HgAr

Note the long integration times required for the blue1000 grism.

Please turn off the calibration lamps when not in use!

A good wavelength calibration is accurate to about 1/10 of the slit width used. It is a good idea to check the wavelength calibration by rebinning a comparison light file and measure some lines in it.

I have made very good experiences with the MIDAS spectra reduction software which is enabled with the command crea/gui long. It works very well and is easy to use. Work through the buttons from left to right; first search lines and then identify a few lines in the spectra; plots with lines marked are listed here as well as a MIDAS table with all lines; note this table has to be activated as line catalog in the long-slit gui. A polynome of 3rd degree is ok.


For best results, especially in the near IR where fringing is important, it is recommended to take flats at the actual telescope position. Domeflats are better than internal flats. Exposure times for the internal flats are:

grism green250 red500 green500 blue500red1000green1000blue1000
cont 100% 20 60 40 50 100 200 200

Flux calibration

For the usually used standards which have V=10 or so, a 1 minute exposure time should be ok.

Small wavelength shifts of the spectra

It may be useful to shift the whole spectrum by small amounts, e.g. if a spectral line just falls out of the spectrum or on a bright night sky line. The command @@ aperoffset npix offsets the aperture (slit) by npix pixels. A positive number shifts the spectra to the blue, i.e. will increase the spectral coverage in the red. Go back to the original position by @@ aperoffset 0



Note: Preparation of the masks is the responsibility of the user, especially the astrometry of the objects of interest. Since we have developed quite a fast and easy procedure for the critical alignment of mask and telescope, we strongly recommend that users make use of our experience. To help all observers we offer to make the masks in our mechanical shop. Note, however, that the observer is responsible for his/her masks.

To align mask and sky we use relatively bright stars (called reference stars hereafter) the positions of which can be measured precisely even on short exposures. To this end the masks contain holes (10 arcsec diameter) at the positions of the reference stars. The positions of these reference stars measured on short exposures are used to correct Cassegrain flange angle and telescope position to improve the alignment of mask and sky. Two reference stars are generally sufficient, but the prgs allow use of more reference stars; more than 3 are not required. Alignment of sky and telescope is done with 2 reference stars only; so at least on the telescope you have to decide which ones to use. An image of the sky taken through a mask is shown here.

Since the positions of 10 arcsec diameter holes can not be precisely measured on images, we put reference holes with 2 arcsec diameter at known offset from the large holes in the mask. The positions of these reference holes can be measured precisely on exposures of the sky. An example for a mask is given here showing the mask from below (i.e. the upper surface is the focal plane). One can see easily the slits, the large holes and the small holes. Orientation of the mask in the instrument is unique.

To make a mask the astronomer has to determine

  • the positions (RA, DEC) of the objects of interest to < 0.3 arcsec accuracy (good astrometry is essential!)
  • the positions (RA, DEC) of 2 or 3 bright stars in the field which are used to align mask and sky. These stars should be bright enough to be well visible on a 10 sec exposure but not too bright; a good range is V=17-19.
  • RA, DEC for the desired center of the field

Note that all these coordinates must refer to the same equinox! RA and DEC must be given in decimal degrees.

The procedures which are used in the following are in MIDAS. You can download a tarball with sample files and a semi-automatically running example( to start it: @@ example)

To compute the positions of the slits and reference stars on the mask the program @@ compmask has to be used. This program projects the celestial coordinates of the objects onto a plane centered on the given center of the field, and maximizes the lenghts of the slits. Compmask wants to read a table with the columns :ad,:dd, and dd are RA and DEC in decimal degrees, and type is used to distinguish between reference stars and objects: type=0 for objects, type=1 for ref.stars.

There are generally two situations: (1) the astrometry of objects and 2 reference stars is known from some source (2) the astrometry has to be determined from an own exposure, eg. an image taken with MOSCA.

The astrometric data are known:

Check that there is no offset between different astrometrical sources if your input data come from more than 1 source!

If your astrometry is an ASCII file, then the required table can be easily created with the command crea/tab. Example:
crea/tab ngc6240 3 100 ngc6240.asc

This will create a table ngc6240 from the ASCII data file ngc6240.asc. Name the columns with RA and DEC (must be in decimal degrees) properly, eg. if RA and DEC are in columns 1 and 2:
name/col ngc6240 #1 :ad R12.7
name/col ngc6240 #2 :dd S12.6
The formats R12.7 and S12.6 display the RA and DEC values in HH:MM:SS and DD:MM:SS (although they are given as degrees). Using eg.F16.8 instead of R12.7 or S12.6 would display the values in desimal degrees. Additional columns are allowed and are simply copied.

Then create an additional column

crea/col ngc6240 :type I1

To set the new column type to the proper value: write/tab ngc6240 :type 0 all sets all entries of type to 0, set type in the proper rows with the data for the 2 reference stars to 1 with write/tab ngc6240 :type 1 @row
where row is the row number.

The astrometric data are unknown

If the coordinates of objects and reference stars are not known, e.g. if one has an astrometrically uncalibrated CCD frame, one has to proceed in three steps:
  • determine the astrometric positions of 10 or so astrometric standard stars on the CCD frame
  • determine the 'plate solution' for the CCD image with these stars
  • apply this 'plate solution' to calculate the astrometric positions of the objects of interest
You could proceed as follows:
  • use eg. skycat to get coordinates of >10 objects which you want to use as astrometric standards. Make a MIDAS table with at least 3 columns :PPM,:ad,:dd as described above. You can also use @@ std to measure the positions of the standards on DSS images and create a proper outputtable
  • 3. load the CCD image into the display: loa/ima CCD
  • 4. use @@ obj to measure the same standards on the CCD frame in the same order! Input 2 when @@ obj asks for type (2=standard). When you have entered the last object, click on it again and enter 3 to finish @@ obj (this is important, because the astrometry looses 1 row (not my fault), so the last object has to be entered 2x).
  • 5. Proceed in the same way to measure the positions of the objects for which you want to get spectra (enter 0 for type) and the 2 reference stars (enter 1 for type),and again enter 3 to finish @@ obj.
  • 6. merge the tables. First make sure that there is no selection active: sel/tab stand all and sel/tab obj all, then merge these 2 tables: merge/tab stand obj input
  • 7. Enable the astrometry package set/context astromet
  • 8. @@ astro stand input mask to compute and apply the astrometry and supply output for further use. The astrometry works only if the astrometric reference stars are in the same order in both tables (eg. output from std and obj)
The output of @@ astro contains all columns required.

Computing the masks

When you have prepared the tables as described you are ready to compute the masks. The command compmask (or compmask2)
  • checks that the columns of the input table and the reference star data exist
  • projects the RA,DEC data onto a plane (generalized coordinates) relative to the plate center
  • rotates the coordinates to the position angle given
  • computes the lengths of the slits. The slits end in the middle between 2 objects
  • computes the positions of the 2 10 arcsec diameter holes for the reference stars
  • computes the positions of the reference holes of 2 arcsec diameter
  • computes spectral range covered (approximately)

@@ compmask ngc6240 16,52,55.5 2,23,42 -90. green_500

will compute the slit positions from the input table ngc6240 using RA = 16h 52 min 5.5sec and DEC = 2 deg 23 min 42 sec as field center and a position angle of -90 degrees. The spectral range covered by grism green_500 is computed. The output is stored in table ngc6240_mask. Note that @@ compmask help gives help information.

To plot the results use

@@ plotmask ngc6240

You should get a decent plot (click here for an example) showing the whole mask with all slits and a dashed square indicating the field of MOSCA. If not, you have probably created a highly rectangular graphics window; try crea/grap 0 500,500

The position angle PA is defined in the standard way, i.e. 0 degrees, which is normally used for direct imaging, gives direct images in the standard orientation. Note that the slits are in EW direction if PA=0.

When you have computed and plotted a mask, check the slit lengths. They should not be less than 10-20 arcsec because then the sky subtraction is difficult if not impossible. Play with the position angle to get optimal slit lengths.

Compmask has been used many times and we know the slit centers are calculated very accurately. The slit lengths are determined such that the slits end in the middle between two objects. This may not always be optimal. You can modify the slit lengths by writing into the table or editing it (the relevant entries are :x_s and :x_e for the starting and end coordinates of the slit). Note also that shifts of the field center in NS direction result in wavelength shifts: a shift of the field center to S (if PA=0) results in a shift of the spectra to the blue, i.e. gives a better coverage of the red.

Note: negative DECs do not cause a problem; however if you have decs like -0 10 30.5 you must enter as field center -0,-10,-30.5. If your field center is -1 10 30.5 enter -1,10,30.5.

It is very important to take a good direct image of the field with the reference stars clearly marked to the telescope!

Getting the masks machined

Machining of the masks is done in the mechanical shop in Heidelberg. Create the code for the CNC machine by the command @@ mechmask. Example:

@@ mechmask ngc0815 1234

Note that a 4 digit number - in the example 1234 - has to be used as outputfile ! (This has to do with the CNC machine program). Use a different one for each mask. The output of mechmask is an ASCII file which outside users have to put on our ftp server in the directory pub/fried/ . Send me an email ( and I will take care of the rest (outside users only!). Allow at least 2 weeks for machining and transporting the masks to Calar Alto, and more in case of holidays /vacation. It might be a good idea to check dates with me as soon as you know the schedule of your observing run!

The slit width has to be specified and communicated to us; we use mostly 1.5 arcsec slits, but 2 arcsec slits can be made as well.

Observing with masks

To observe with a mask
  • put it into the aperture unit. There are 2 places for putting masks which are accessible by removing the cover with yellow label (mask1+2). A snap-in mechanism allows quick installation in a unique orientation. The masks are accessible only if the aperture unit is in position "free". Eit the names of the masks in the GUI (setup)
  • set the Cassegrain flange to the desired position angle
  • move the telescope to the required center position of the mask
  • enter @@ alimask. This command will
    • 1. take an exposure of the mask with the build in lamp to measure the positions of the reference holes
    • 2. take a short exposure of the whole field to measure the 2 bright stars and improve upon telescope position and Cass.flange angle
    • 3. take 2 short exposures to measure the positions of the stars in 2 large holes in the mask to compute flange angle and telescope position corrections. This step can be iterated.
    • take an exposure through the slits to verify that all objects are in their slits. This step is optional. An example is this image
If things work well, this procedure will take 15 minutes. If you have already used alimask (e.g. the night before and you know telescope position and cass.flange angle) and want to continue, you can skip step 2 in the command @@ alimask which then reads the positions of the reference holes from the aquisition exposure already taken and skips step 2 and thus saves time. Note that it works only if telescope position and cass.flange angle are already known well enough so that the 2 reference stars fall into the 2 holes; so if you want to observe with the same mask several times write down accurate position of the telescope and the file name of the exposure taken during step 2 (something like AQ0037a.bdf, the program tells you the file name).

Do not forget to take flats and wavelength calibration exposures!

Up to now it is not possible to set the Cassegrain flange angle from the program. Instead of changing it at the Cassegrain focus one can also change it from the TECS computer by entering the command CASFL. Then enter the new position angle and start to change it with CTRL A (CTRL Z = abort). This operation is possible only if the telescope control is set to 'PULT'. Use this only for small corrections of the position angle!




Settings on CS100:

Before turning on the CS100 check the settings:
    fine     coarse  quad.bal.
x    276       -1     404
y    862       -2     389
z    562       -2     378
With this z setting the Ne 6929.5 line should be near the maximum when the electronics is turned on!

Turning on control electronics CS100:

If the FPI goes into overload when turning the CS 100 on, then
  • turn off CS100, turn on again. Switch to BALANCE and then back to OPERATE.
  • the cables are defect (or the connectors not porperly put into place). WARNING: Do not disconnect cables when the CS100 is turned on!
  • the QUADRATURE BALANCE knobs may be set wrong. Check the settings by switching from OPERATE to BALANCE and from OFFSET to QUADRATURE ERROR. Adjust the QUADRATURE BALANCE knobs until all needle readings are zero. Switch back to OPERATE and readjust the QUADRATURE BALANCE knobs until all needle readings are zero again.

Wavelength calibration :

The Ne 6929.5 line is used for the calibration in the 24th order. @@ fpi will configure MOSCA like this:
  • calibration unit in, Ne lamp on
  • aperture unit: select slit, 10 arcsec width
  • FPI in beam
  • grism free, filter1 free
@@ fpi asks for the order sorting filter 685/20. If you have just started the calibration procedure, wait at this point for 1 min until the lamp has warmed up and then continue. @@ fpi performs a scan in z and analyses the result file automatically. Enter -20,2,18 for z start, z step and # steps. If you are happy with the result (FWHM about 14.0 units), store it when the routine asks you to do so. Now you are ready to control the FPI using the GUI.

When the etalon is properly aligned the FWHM should not be larger than 14.5. Otherwise the alignment has to be checked by doing several scans with X and Y values variing independently by steps of about 50 units, thus finding the optimum X/Y values (the variation will not only lead to a change of the resolution but also to a small shift of zmax). If the FWHM of < 14.5 is not reached then contact Hans Hippelein (or JWF, but only if you are very desperate!)

If the zmax value of the calibration is off from zero by more than 50 units, check the z setting of the CS100 controller.

Do not forget to turn off the Ne lamp when the calibration is done!

Operation of the FPI

To set the FPI to a certain wavelength, enter it in the GUI. Note that it has to be entered in nanometers. The program will then give you a list of orders and corresponding resolutions. Clicking on one of the orders will set the FPI to the desired wavelength and order.

The reset button in the GUI resets the CS100 control electronics and sets the z value to zero, the x and y values will be as initialized. Use this button after an overload has occurred.


The most useful standard MIDAS commands are (for more details use help command):
  • load/image to load an image into the display. To load an image with cuts 100,200 and scale of 2 you may enter load ima cuts=100,200 sc=2 or shorter @@ ld ima 100,200 2
  • plot/col or plot/row to do plots
  • center/gauss to determine the position of an object
  • comp/ima z = a+b*3*c to do image arithmetic
The following special utilities are available:
  *** for installation *** 
@@ install      to measure and save installation parameters
@@ fprisini     initialize focussing prism
@@ fpi          to calibrate the FPI 
@@ rotcen       to measure rotation centre of cass flange

 *** for looking at the data ***
@@ autoload     load frame, determine cuts automatically
@@ backdet      load image with cuts near background
@@ zoom fac     zoom image by fac centered on marked  object

 *** simple but useful  *** 
@@ seeing       measure seeing for many stars
@@ pa           measure position angle between two objects
@@ pixel        convert measured to physical pixels
@@ offset x,y   measure offset to a position x,y and move telecope
@@ aperoffset   offset slit in direction of dispersion

 *** take series of exposures ***
@@ autofocus    do a focus series (MOSCA or telescope)
@@ autoflat     do automatically flats
@@ expose       configure MOSCA, take series of exposures
@@ focus        evaluate focus series
@@ sequence     sequence of exposures, dither telescope

 *** correction of data *** 
@@ fcor         flatfield correction
@@ spex         simple extraction of spectrum and skysubtraction

 *** working with masks ***
@@ alimask      to align mask and telescope

Detailed help information is given when entering help as first parameter in a command, eg. @@ seeing help. A list of all available special commands is obtained with @@ mosca

The following commands are short forms of MIDAS commands:

  • cg : center/gauss
  • cco : clear/chan over
  • stc : stat/cursor
  • rk keyname: read/key keyname
  • wk keyname: write/key keyname
  • rd read/desc
  • ld ima cutlow,cuthigh scalefactor
show/comm gives you an overview of all user defined commands

Saving the data

To save the data on DAT in tar-format use
mt rew
tar cv *
Be careful not to overwrite data; it is best to use a new DAT for each save! List the tape with
mt rew
tar tv > dat.list

To save the data as genuine FITS-files use the CA-script
and to list the DAT use
You may also use these commands in easy to use GUI versions (fitscopyx and ccdlstx). To activate these scripts enter the command getastro caha

Trouble shooting

The GUI does not start/work properly

  • if Netscape is running on the same terminal, close it, restart the GUI and then open Netscape again. Always start the MOSCA GUI before Netscape and CCD GUI! Note that Netscape should be started like this: netscape -install &!
  • quit the GUI, turn off the electronics of MOSCA, and restart the GUI (start_mosca)
  • if nothing helps, logout completely and restart properly.

The receiver does not convert the files

This happens if the receiver is working on a wrong directory; this can not happen if you start the software according with the macro start_mosca. If it happens by some reason, either set the proper directory manually, or logout and restart.

Communication problems

The computer environment is not as stable as it should be; this causes sometimes loss of communication to MOSCA or the CCD, so if you enter a command, it will not be executed properly and MOSCA or the CCD is in a confused state. If this happens, quit the GUIs of MOSCA and the CCD, turn off/on the electronics of MOSCA, reboot the EPICS system, wait for 2 minutes and start MOSCA and CCD. Up to now this has helped always. In tough cases, a reboot of the workstation (ask the night assistents to do it, there is a special procedure required!) may be necessary and helpful (the CA personnel will deny this, but it is my experience that it may be necessary sometimes).

Computer environment

Many mosca prgs communicate with the CCD camera and MOSCA. This requires that the UNIX variable path must contain the right entries and that some environment variables must be set properly. This is done automatically in the xterminal windows which open when selecting MOSCA in the instrument select GUI. However it happened in the past that someone modified the environment which caused some UNIX environment variables not to be set properly. As a consequence of this the communication between MIDAS and MOSCA does not exist and one gets error messages like " MOSCA_DIR/.../mosca_filter1 script not found" or so. If this happens, quit MIDAS and set the environment variables by entering the command set_mosca. If this does not work use source /disk-a/staff/MOSCA/scripts/.mosca

When you have to start via another computer, the following procedure may be successful: login the computer, eg. castor. Then enter the command

xhost +

connect to ultra1 via telnet, and then enter set_mosca

error message: final position not reached

If this does appear just repeat the command. It is caused by too small tolerances in the control software and not a real problem.

The calibration unit does not work

This has happened a few times after installation, never during the observing run. Take the module out and move the mirror in the beam ( a few mm are sufficient) by hand by turning the motor and start again. There was an electronical problem which has been fixed, and up to now this problem has not occured again.

No sound

Some MOSCA MIDAS commands are producing sound output. If you do not get sound then either the volume of the speaker may be turned off, or the speaker is not activated. In this case, start the soundtool and select headphones as output.

The MOSCA team

  • construction + mechanics: Bellemann, Benesch, Franke, Muench
  • electronics: Salm, Grimm
  • CCD: Marien
  • software: Zimmermann, Bruege
  • tests: von Kuhlmann
  • PI: Fried
  • responsible on Calar Alto: Aguirre, Thiele

send an email if you have questions to


Changing of modules and filters


To change a module proceed like follows:

  • quit the MOSCA GUI. This MUST be done first.
  • move telecope to zenith
  • turn off the electronics of MOSCA
  • change the module
  • turn on the electronics of MOSCA
  • reboot the epics computer
  • start_mosca, start_ccd

Changing the filter list

To change the filter list, enter the command


in the xterm labelled instrument. A GUI will appear, which allows to click the filters from the general Calar Alto list. Select the proper position in the corresponding wheel by clicking into it; the selected position will turn light blue. Then select the filter by scrolling in the list, and double click on the chosen filter. The values stored in the list will be taken over. If you want to set a position to free click onto the free button. Terminate by clicking the save+exit button and restart the MOSCA GUI.

Using own filters/installation of new filters

To find the focus offset for a new filter, install it in the desired wheel, enter its name in the filter list. Determine the focus without filter, then with the new one as described, then enter the focus offset (= focus with filter - focus without filter) in the filter list, restart the GUI and the focus offset will be automatically corrected for whenever you use this filter.

Calibration unit

4 lightsources are selectable :
  • 1 tungsten lamp + BG38 filter as continuum source
  • 1 Ne spectral lamp
  • 1 Ar spectral lamp
  • 1 Hg/Ar spectral lamp
Spectra and wavelengths of the lamps are here.

The light is deflected into the instrument by a plane mirror which is automatically moved in/out of beam.

Aperture unit

5 apertures are selectable :
  • 2 masks which can be easily exchanged even if the aperture unit is installed. The standard configuration is:
    • mask 1: pattern of holes, used for testing the distortion
    • mask 2: black. Useable as additional shutter (e.g. when storing the instrument)
    • free. Used for direct imaging
  • 1 longslit. Its width is adjustable from 0.3 to 15 arcsec. The length of the slit can be limited with a decker and the center part can be masked
  • 1 hole with 200 micron diameter. Used for determining the internal focus.


The following filters are standard MOSCA filters; if one of these is used, any focus offset due to chromatic effects in the optics of MOSCA or non-flatness of the filters is automatically corrected for.
glasses: UG1/1mm + BG 39/2mm
color: black , offset 46 in wheel 1
glasses: BG 37/1mm + BG 39/2mm
color: blue , offset 50 in wheel 1
glasses: BG 39/3mm
color: turquoise , offset -15 in wheel 1
glasses: BG 18/1mm + GG 495/2mm
color: green ,offset 0 in wheel 1
glasses: OG 570/2mm + Calflex X
color: red , offset 150 in wheel 1
glasses: RG 780/3mm
color: red , offset 40 in wheel 1
glasses: RG 830/3mm
color: black , offset 63 in wheel 1


Name    Offset (wheel 2)

398/10    -1735
455/110   -1679
522/16    -1779
529/30    -1787
586/32    -1685
612/18    -1585
628/15    -1653
685/20    -1628
703/22    -1663
817/24    -1705
916/32    -1666


There are 7 grisms available. Click onto the grism name to get the efficiency curves. These curves give the quotient of incident photons to registered electrons measured for a spectrophotometric standard star under perfect conditions with the LORAL 11i CCD. (The red part of the curve of the green250 grism may be contaminated by 2.order light; the true efficiency in the red part will be somewhat lower therefore) The efficiency curve for the red500 grism has to be deteremined

Note that grism green250 has no order separation filter. If the region of interest is in the red part of the spectrum, a GG400 or GG455 or GG495 filter may be used in filterwheel I to block the blue light.

Wavelength calibration

Calibration sample spectra with some line ids marked:

sample spectrum blue 1000

sample spectrum green1000

sample spectrum red1000

sample spectrum red500

sample spectrum green500

sample spectrum blue500

sample spectrum green250

Spectral line catalog in MIDAS table and fits format. To copy one of these files to your area click on the file and then use the save as button of netscape!

Image scales and distortion

Mean overall scale
The mean magnification factor of MOSCA is 0.2755 which gives an image scale of 1/(0.1697*0.2755) = 21.3893 arcsec/mm or 0.3208 arcsec/pixel of 15 micro meters.
Exact scale
The mean magnification factor is a good approximation over a large part of the field. The radial image distortion is < 1 arcsec over large parts of the field and reaches a few arcsec in the corners. The optical center is located 0.345mm E and 0.42 mm S of the position of the hole on the CCD. The true distances (R,A) from the optical center can be computed from the measured ones (r,a) by one of the following polynomials:

R = r + 8.0139*10**-4*r - 1.49886*10**-6*r**2 - 2.51185*10**-12*r**4   [pixel 15mue] 

A = a + 8.0139*10**-4*a - 4.66498*10**-6*r**2 - 7.57288*10**-11*a**4   [arcsec]

After these corrections the magnification factor = 0.27507 (green dots in the figure).

Flexure of MOSCA

The flexure of the instrument has been determined in February 1997 to be 1/7 arcsec (0.44 pixel) including reproducibility of the position of the aperture unit.

Shutter uniformity

The shutter is of the sickel type. Nonuniformities as derived from flats are maximal
  • 3% for 1 sec exposures
  • 1.9% for 2 sec exposures
  • 0.74% for 5 sec exposures
  • 0.35% for 7.5 sec exposures
  • 0.24% for 10 sec exposures
These values were determined in April 1998.

Filterwheel 1

  • 10 positions
  • filterdiameter 70 mm
At the position of this wheel, the beam has a diameter of 59 mm.
Position 1 must be free!
Used for the standard braod band filters. Since this wheel is in the parallel beam filters may cause reflections. The standard filters are inclined for this reason.

Filterwheel 2

  • 10 positions
  • filterdiameter 70 mm
At the position of this wheel, the beam has a diameter of xx mm.
Position 1 must be free!


Position 1 must be free! The focussing prism has been taken out.

Focussing prism

The focussing prism allows you to find the focus quickly. The principle is like a normal Hartmann test, where you take 2 images of a point source in 2 half beams. In focus the 2 images coincide. The focussing prism covers one half of the beam and deflects the image of a point source by a few arcsec, so one obtains the 2 Hartmann images simultaneously. There is a linear relation between focus and displacement.
Preparing the focussing prism
  • determine the focus of MOSCA by a focus series
  • select the tungsten lamp, the whole in the aperture unit, the focussing prism in the grismwheel, no filter.
  • take an exposure, measure the positions with MIDAS: @@ fprisini. Do what the command tells you to do. The measured positions are stored (table fprisini.tbl).
Using the focussing prism
Note: once initialized, the focussing prism can be used to focus both MOSCA and the telescope!
  • prepare the focussing prism as decribed above
  • take an exposure of a star or of the whole in the aperture unit
  • find the focus offset with the MIDAS command @@ fpris

Orientation, physical pixels

To get the files in standard orientation (N top, E left) the files are x-flipped in the getfits procedure (for the LORAL 11 and the SITE16a CCD) and for some good reasons the start is set to 2300-startold,1. Therefore the pixels you read off the frames (e.g. with get/cursor) are not the true physical pixels of the CCD. All MOSCA prgs work perfectly with these start values. However if you want to set a window on the CCD the pixels must be converted to the true physical ones which have to be entered in the CCD GUI:

xphysical = 2300-xmeasured

yphysical = ymeasured

The command @@ pixel converts the pixels for you, and it works also on binned frames. We are working on this nuisance (the solution is not as trivial as it looks).

Focussing unit

the focussing unit moves the camera optics. Maximum range is 2800 micro meters. Large numbers mean that the camera is near the CCD. Rings (0.1, 0.2, 0.3, 0.5mm or 1mm thick) can be placed between CCD and the mounting flange to get into the focus range. The distance between CCD and the dewar window must be 3-4 mm.
Changing the focus
The focus of MOSCA may be moved by the command

mosca_afocus value

where value is the desired focus value, e.g. mosca_afocus 8200. Note that value must be in the range 6500-8900. This focus will be lost when a filter or grism is selected, since upon this action MOSCA sets the focus corresponding to the focus zero point and offset for the filter chosen.

Installation Notes

Mounting MOSCA to the telescope:
  • set the position angle of the Cassegrain flange to 0
  • the TV guider in standard orientation, the slit-viewing optics of TWIN installed
  • the handle of the carriage of MOSCA must point towards N (the telecope pier). The calibration module points to N.
  • TV-guider has to be set in spectroscopy mode (s), f/10

Astronomical orientation

To check the orientation of CCD on the sky, put the mask labelled "orientation mask" in the aperture module, select the mask, the continuum lamp (60 % intensity) and take a 5 sec exposure. The image shows a star-like pattern, the shorter ray points roughly towards NNE for 0 degree position angle of the Cassegrain flange. The MIDAS frames are automatically flipped and/or rotated in the receiver so the MIDAS images are always in standard orientation, but not the FITS images!

SITe 16a CCD (normally used)

  • the label on the dewar point towards N, the mounting rails to S
  • no distance rings
  • internal focus near 8350 in white light
  • the files must be flipped in x to get standard orientation
  • the image scale is 0.3208 arcsec/pix
  • the chip has little fringing


  • the label N on the dewar points towards N, the mounting rails to S
  • distance rings 1.7 mm
  • internal focus near 8050 in white light
  • the files must be flipped in x to get standard orientation
  • the image scale is 0.3208 arcsec/pix
  • the chip has strong fringing

SITE 18b12 24 micron pixel

  • the label N on the dewar points towards N
  • distance ring 2mm
  • internal focus ?
  • the files must be rotated ccw by 90 deg to get standard orientation
  • the image scale is 0.5158 arcsec/pix
  • this CCD has benn used only once up to now

EEV l2 15 micron pixel

  • the label N on the dewar points towards N
  • distance ring 4mm
  • internal focus 8300
  • the files must be flipped in x to get standard orientation
  • the image scale is 0.3208 arcsec/pix
  • the chip has strong fringing

Software adaptation to CCD

The directory /disk-a/staff/MOSCA/mosca_prg must contain all prgs which are used by the software. Some of the prgs are chip-dependent: alimask autoflat detexp expose install fpi pixel These are stored in directories
/disk-a/staff/MOSCA/mosca_prg_SITe18b12 and
and must be copied from there to mosca_prg or the local user directory. For the EEV chip the SITE16a prgs should work (not tested). (Files in mosca_prg_SITe18b12 have been supplied by HJ Roeser)