OSMOS Multi-Object Observing

For the experienced OSMOS MOS observer, please click here.

This page is a basic guide to making observations using multi-object masks on OSMOS at the MDM 2.4m. More detailed information can be found at the links given within.

Table of contents:

Setting up

Telescope control

Biases, flats, comps, and offset skies

Focusing and pointing

Rotating, guiding, and alignment

The observing sequence

Shutting down

The MDM 2.4m Telescope Manual contains most of the information presented here, along with a lot more details than this overview provides. It is definitely worth reading and should be kept handy as a reference, but this document should be able to get you started. The manual can be found here:

http://mdm.kpno.noirlab.edu/24_manual/contents.html

Another worthy read is the Observer's Guide for the 2.4m. This is less a technical manual and more of a general guide with tips and tricks. It is especially useful for the startup and shutdown procedures; you should give this one a read through before the first night (it's not that long). It may be found here:

http://mdm.kpno.noirlab.edu/Manuals/mdmguide_current.html

There is also a good amount of reference information on OSMOS here, in the OSMOS User's Manual:

http://www.astronomy.ohio-state.edu/~martini/osmos/

1: Setting up

Relevant manual sections:

1: General Guidelines and Preliminary Information

2: What Every Observer Should Know

Insert your masks:

Before running the startup checklist, you'll want to insert your masks. The masks go in a hatch on the side of the MOS (it's the upper hatch on the box with the ohio state logo on it). The hatch is a bit stiff to open. Inside is the wheel where the masks and slits go. Even on the first night there may be some already /ed - longslits and whatnot from whoever had it before you - if so, these should be removed and placed in the wooden box where they are stored, which should be somewhere about the dome. On subsequent nights, of course, your masks from the previous night will be inserted. To remove, loosen the two screws with a blue allen wrench (size 2.5 usually located on the cart in the dome) and slide the mask/slit out. The screws are locked into the mask, so don't worry about them dropping when you loosen them. Your masks are inserted the same way with screws pointing down, make sure each mask has both of its screws. The masks are keyed so they only go in one way.

To load the masks, use the "Prospero" window to move the slit positioner wheel (it can be moved manually in the hatch, but make sure to return it to the position it was in when you opened it). In Prospero, type <slit N> (don't use brackets) and N is the slit number IN THE FOCAL PLANE. With this slit in, the slit opposite is visible through the hatch. The image below shows the mapping between the slits. For instance, is you type <slit 2>, then slit 2 is in the focal plane and slit 5 is visible in the hatch. Load the masks by cycling through the slits and noting which one goes into which slot. Then, in Prospero, use <tedit slit> to edit the slit parameter file with the mask name in each slit position.

Note that slot 6 must be left empty, so you can have a maximum of 5 masks installed at a time. Make a note of which mask you put in which slot (the mask numbers are in the .oms files). Leave the stairs next to the instrument package so you can open the dark hatch later.

Then, get out the startup checklist (which can be found here):

http://mdm.kpno.noirlab.edu/24_manual/hiltA.html

(1) Insert masks into OSMOS

(2) Take some mask images (use a filter and flat without disperser but with mask).

(3) Fill dewar (and note on whiteboard in control room).

It's a good idea to get the dewar filling early, since it takes a few minutes (you will have to do this once at the start of the night and once at the end, more or less). The liquid nitrogen in is the shorter silver container on the hand cart. The wand is located at the end of the rubber tubing and goes into the socket on the bottom of the dewar and screws into place - be sure you've got it all the way in and attached. The dewar is full when a steady stream of liquid nitrogen shoots out of the overflow tube. It is *not* full if there's just spatters or a flow that doesn't have any velocity behind it. After you turn off the nitrogen you'll have to wait a few minutes for the tube and fittings to unfreeze before you can remove it. Make sure that the dewar is full and the tank detached *before* you unlock the telescope drives. These are more reasons to get it started filling early so that it doesn't delay you. Once you fill it, note the time on the whiteboard on the door to the dome in the control room.

(4) Open vents out in dome (manually). If windy, also open the garage door because otherwise it causes a racket that isn't fun to listen to. The padlock combo is on the whiteboard in the control room.

(5) Open dome hatch (press left green button on wall out in dome one; when finished, hold green button on right to flip open hatch).

(6) Make sure TCS computer window reads: "Dome at home".

Next, follow the rest of the start-up directions. Note that you should check the airbags (in the computer/rack room on left computer rack) for pressure.

Order of operations isn't hugely important for most of it, but there are a couple important points. First, don't try to rotate the instrument package until the mirror covers are open - it won't work, and if it did you'd damage things. Second, don't open the mirror covers until you're ready to observe - we don't want anything dropping onto the mirror. When you do open them, go out with a flashlight and check that all four panels are completely open - if they aren't, see the observer's guide for help. Most likely you will have to climb a ladder and unstick them with a broom handle. Doesn't that sound fun! The sliding portion of the dome cover is rather noisy while it's opening, but the banging doesn't seem to indicate anything wrong.

Note that the 'external computer' switch on the telescope controller is obsolete and not used - that is why the checklist doesn't mention it.

2: Telescope control

Relevant manual sections:

4: Telescope Control & Data Acquisition Computers

5: Telescope Control System (TCS)

6: The Multiple Instrument System (MIS)

The telescope and instruments are controlled by a variety of pieces of software running on Hiltner. The relevant windows are the Prospero terminal, the xtcs telescope control window, the xmis instrument package control window, the prospero status window, and the TCS telescope status window on the adjacent monitor.

The telescope position is controlled through the 'xtcs' window. To move the telescope, coordinates are entered into the RA, Dec, and equinox boxes, the 'send coords' button is used to send the coordinates to the telescope, and the 'go' button is used once the coordinates have been recieved to command the slew. There are details, of course.

You will want to create a target file with your observing targets in it to allow you to quickly and easily command the telescope to slew to them. This is a file in the following format:

HILTNER102 1 5 53.8 62 31 32.0 1950.0

My_star 1 46 28.5 -13 18 17.0 1992.6

HD60778 7 33 39.0 -0 1 49.0 1950.0

2207+8201 22 07 13.2 82 01 13.9 1975.0

The columns are name, RA (HH MM SS.S), Dec (DD MM SS.S), and epoch. The filename can be no more than 22 characters long, and the object name should be no more than 10, with no spaces. RA and dec *must* be in the above formats separated by spaces (not colons), *not* in decimal hours/degrees or degrees/degrees. The name doesn't need to be elaborate (especially since you only get 10 characters); it's only an identifier for the coordinate line in the file, so it just needs to be easily remembered and distinctive. This file should be placed in the visitor directory on hiltner (e.g., /lhome/obs24m) and then just enter the filename using /data/hiltner/filename into the xtcs.

The 'get coords' button/submenu can be used to avoid having to enter coordinates manually. The 'nearest bright' menu option enters the name and coordinates of a nearby bright star - useful for pointing and focusing. The 'named target' menu option reads the coordinates of the corresponding object from the target file you created earlier - you must enter the target file filename in the 'target file' box for this to work.

If you enter the coordinates of some other object manually (or read them from 'nearest bright' or whatnot), you can enter an object name and then click 'append' to add this object and its coordinates to your target file for later use.

To command the telescope to slew by a specified offset rather than to a particular target, enter the desired offset *in decimal arcseconds* into the RA and Dec boxes and hit 'send offset', followed by 'go' (this is useful for taking offset skies).

To reset where the TCS thinks the telescope is pointing, hit 'set encoders' in the 'setup' button/submenu. This resets the TCS's current telescope coordinates to the values currently displayed in the RA, Dec, and epoch fields (this is useful to calibrate the pointing).

The 'xmis' window controls various instrument package functions. It can be used to set the lamps, the retrocam prism, and move the guider probe.

What you'll be using most here are the retrocam and lamp controls. The retrocam prism slides in to (among other things) redirect the camera to the instrument package lamps rather than the sky. So, to take instrument package flats or calibrations, it must be in - just remember to slide it out again before you try to observe. To open the lamp control subpanel, click the 'lamps' button. There are four line calibration lamps available as well as a flat lamp, which are turned off and on simply by clicking on them (they are highlighted when on). Note that to get rid of the lamp panel, click 'dismiss' - if you click the 'x' in the corner, it will kill the entire xmis window.

The other major set of controls here are the 'x', 'y', 'dx', and 'dy' boxes; these control the guider probe. More on this in the guiding section.

The prospero terminal is where you'll actually control the camera and wheels from. A brief listing of useful commands follows - format is <insert command here>, meaning to enter exactly what I type, minus the < and > on either side. Any use of a capital X is standing in for a number.

<runinit> - this should be entered at the start of the night to set things up. Mostly self-explanatory, although note that it doesn't seem to actually let you change the save path even though it says it does.

Image type setting:

Use these to set the image type, with 'name' replaced with something descriptive (keep the ' marks in) - i.e. <comp 'Argon comp'> or <bias 'Bias'>. The bias command also helpfully sets the exposure time to zero seconds for you.

Wheel controls:

These set the filter wheels, slit wheel, or disperser wheel to position X (note that position 6 is always supposed to be empty). Typing any of these with no argument will return the current position. If any of them appear to be lost or not returning the correct position, use 'reset' as the argument - for example, type <slit reset> to reset the slit wheel position. Note that you will need to reset the slit wheel if you've moved it by hand when installing masks.

<print [wheel]> - shows what is installed in each position on [wheel], where [wheel] is one of the above, i.e. <print filter1>.

<tedit slit> - allows you to edit the slit data table to let the software know what is installed in each position of the slit wheel. Use this after you've installed your masks to enter their names or numbers so that you can keep track of them (this is the information that is displayed when you type <print slit>.

Sets the camera readout region of interest to either the central 1k of the field or to the entire 4x field. The 1k field is useful for focusing/pointing/aligning/etc because it reads out in ~15s rather than 2+ minutes, but of course you should remember to switch back to 4k afterward.

Exposure commands:

<exp X> - sets exposure time to X seconds.

<go> - commands it to take an exposure with the current exposure time

<mgo X> - takes X exposures in sequence

<movie> - takes exposures continuously, displaying them on the monitor but *not* saving them to disk (useful for focusing and pointing).

<stopmovie> - aborts movie mode.

Also useful:

where X =1 or 2 to get 1x1 or 2x2 binning on output.

3: Biases, flats, comps, and offset skies

There is no need to take biases. There is a script that performs all the basic, essential reduction steps (e.g. removing cross-talk between quadrants) on the MDM4K page. It performs bias subtraction by removing the mean level in the over-scan region of each row.

Flats can be taken either using the internal flat lamp or using the twilight sky. The flat lamp can be done anytime, but has the problem that it doesn't evenly illuminate the field - which makes them pretty bad for the actual purpose of a flat image. This won't kill you if all you care about is the position of lines, but if you want intensity you might have problems. The alternative is twilight flats, which actually will illuminate evenly, but must be taken during a very limited window of time during dusk.

To take internal flats, slide the retrocam mirror in and turn on the flat lamp (making sure to turn off any of the others). You will need one for each new mask, and should use the filter (if any) you plan to use for the actual observations. An exposure time of ~30s seems to work for internal flats, and we were taking sequences of 3.

Twilight flats obviously need the retrocam mirror out, and exposure times will vary based on sky brightness (though they will be short). The problem here is that you'll still need a set for each mask - given the readout time, it's not clear that there's enough time to get 5 masks worth during twilight (even a sequence of 3 images for each will be over 30 minutes worth of readout time). The flats will be vastly superior quality if you can manage it, though.

You will need spectroscopic calibrations (otherwise known as 'arcs' or 'comps') for each mask. These should be taken with the retrocam mirror in, one of the calibration lamps on (and the other lamps off), the disperser in, no filter, and the current mask set. Argon and Xenon comps can be about 60s, neon much shorter (5-10s), and we were again taking sequences of 3.

If you have sky slits on your masks, congratulations, you just saved yourself a ton of time on each mask. If you don't, you'll need to take offset skies. This is as simple as entering an offset of 10 arcminutes or so into the tcs and exposing with the retrocam mirror out, lamps off, mask in, filters in, and disperser in, as if you were still looking at your objects. Sky exposures need to have a reasonable total exposure time, but our original 'half of the object exposure time' is probably rather overkill. Again you'll need more than one; we had 2, but 3 would probably be better.

Note that the status of the lamp and the mirror fail to update sometimes. We issued the command 'lamp' in prospero to force the values in the GUI to be updated.

4: Focusing and pointing

Before you can observe, you'll need to make sure that the telescope pointing is accurate and the focus is good.

Use 'nearest bright' to slew to a bright star. Set roi1k for quick readouts and enter movie mode. Hopefully you should see a bright star in the field of view - if not, the pointing is drastically off, and you'll probably have to go through the manual zenith reset procedure (section 5.4.7: Manually restoring coordinates using the tilt meters). If the star is already somewhere near the middle (doesn't need to be exact, we'll refine this after focusing) you can go on, otherwise roughly center the star using the handpaddle next to the control computer. Use 'guide' speed (i.e. just press the directional buttons, not any of the speed buttons) - if the instrument package isn't rotated, E moves the star up, W moves it down, S moves it left, and N moves it right. If the instrument package *is* rotated, you'll have to experiment.

Stop the movie and take a 1s exposure. Go to the desktop workspace with the iraf window. If you need to open iraf then you will need to cd to the iraf directory in the home directory before returning to the data directory. Close the ds9 window on this desktop and reopen it by typing <!ds9 &> into iraf - iraf frequently gets confused if any other ds9 windows have been active since the last time you used iraf, so it's best to just reopen each time you focus to be on the safe side. Type <imexam [image]>, where [image] is the filename of the image you just took (without extension), i.e. <imexam object.0001> if your image is named object.0001.fits. Note that it seems to take several seconds (up to ~10 or so) for the data to be written to disk completely, even after the monitor says the readout is done. If you try to open it in iraf too soon, the file will be incomplete and iraf will complain, so wait a few seconds. On occasions when this happened we had to log out of iraf and log back in. If it opens correctly, the image will pop up in the ds9 window. Find a star in that window to examine - a smaller, dimmer star is best, though you can use the central bright star if there aren't any others. Hover the mouse over it and press 'r' to bring up a radial brightness profile - the bottom-rightmost number in the corner is the FWHM. You can also press 'e' to bring up a plot of the brightness contours. Press 'q' once you're done examining this image.

You will be trying to minimize the FWHM (and, optionally, make the contours as circular as possible). Take repeated exposures at different focuses (focus controls are on the handpaddle and the current focus is displayed on the TCS telescope status monitor) and make a note of what the FWHM of the star is for each one. If you're *very* far off, you can move the focus by more than 20, but otherwise it's best to go in steps of 10 or so to start with. Once you find the value with the lowest FWHM, it's worth it to search around in increments of 5 and even 2 or 3 to fine tune it. Make a note of this number once you find it.

After you focus, we'll refine the pointing (this is done second since if you do it first, it'll probably drift again before you're done focusing; it drifts significantly in only a few minutes). Go back to movie mode and use the handpaddle to center the star - note you have to wait an entire readout cycle to see your last adjustment, so it's easy to overshoot if you forget about this. Once the star is centered, reset the encoders (making sure that the star's coordinates are still entered in the xtcs window), and move on to starting the guider before the star drifts too much.

5: Rotating, guiding, and alignment

Rotating:

Each mask will have a rotation angle defined in its .oms file. You will need to rotate the instrument package to match. This should be the first thing you do for each mask (so before you do focusing and pointing). Info on the rotator can be found here:

http://www.astronomy.ohio-state.edu/~martini/osmos/rotator.html

The control is outside in the dome, hanging from the telescope fork. The current rotation angle is displayed on the TCS telescope status monitor. There is a repeater monitor on a cart out in the dome so you can check the angle without going back inside, just turn it on with the red switch. Watch carefully - the minus sign for a negative angle is kinda blurry and could be missed. Rotating the instrument package is best done with two people - one to rotate, and one to watch the monitor and call out directions - but it can be done with one, it'll just involve a lot more running back and forth. Make sure to watch the package continuously as it rotates in case any of the cables snag.

You don't need to get the rotation closer than about half a degree, as there is a fine adjustment available on hiltner. If the rotator gui isn't open, it's in the Applications-Telescope Control folder at the top. You can send offsets in degrees, arcmin, or arcsec, but *only* up to one degree (you'll be using this later for alignment, too).

Guiding:

Detailed guiding instructions can be found here:

http://mdm.kpno.noirlab.edu/Manuals/guiderdoc/autoguiding.html

However, this really goes into too much detail and is rather overcomplicated for our purposes. You probably should read it at some point, but the steps here are good enough to start with.

Step 0: Set the rotator, slew to your cluster, use 'nearest bright' to go to a bright star close to the cluster, and do the focusing and pointing alignment.

Step 1:

If it's not already, open up jskycalc on hiltner. Not that it may be in one of the other open spaces. You can load up your target file to make the entering of coordinates easier.

Click on the 'guide stars' button to open up the main guide window, if it's not already open.

Select your target from the object file to center the window at those coordinates.

If applicable, enter the rotator angle in the appropriate box.

Click around until you find a reasonably bright (11-14th mag, ish) star, then click 'move guide probe' to center on that star.

Step 2:

Move over to the guider computer. Load up the maxim pro software and click on the 'camera control' button (looks like a weird plug) to open up the main camera window, if these haven't been done already.

In the Setup tab, hit 'connect' to establish the link. Turn the coolers on, and under camera 1 hit 'coolers' to open a dialog box and check the setpoint (-35C is reasonable).

In the Guide tab, options menu, camera settings - make sure that the third button on the right says 'close shutter' (if not, click it), this ensures that the shutter stays open the whole time so it doesn't wear out.

This options menu also has the track box size, set it to 64x64. You might need to increase it to 128x128.

In the main guide tab, aggressiveness should be 5 in both axes

On the expose tab, do the exact same thing as on the guide tab regarding the 'close shutter' button. Also, in the options menu, make sure that 'no calibration' is checked.

Go to Guide->Settings and set X-speed = -16, Y-speed = 16 and Angle = -90 - rotation angle of the mask. Use these manual settings. If you use the Calibrate function, these will change and they can go bad. Don't calibrate.

Step 2 is all setup stuff that you will only have to do once (and you don't need to turn this off at the end of the day, so once at the start of the observing run only, unless it crashes or something). Each subsequent time you will only have to do step 1 and the following.

Step 3:

Go to the guide tab, select the 'expose' *radiobutton* (not tab) and hit 'start'. A starfield should pop up, and a fairly bright star should be obvious. If not, check your pointing. If yes, continue. and it will automatically pick out the brightest star as your guide star.

Select the 'calibrate' radiobutton and hit start. Sit back for a few seconds as it will slew the telescope around a bit, watch which way the star goes, and generally figure out how to properly move things. Note that you can skip this step if you haven't changed the telescope rotation since the last time you ran the calibration. (You don't need to recalibrate if the only rotation change is one of the very small adjustments from the alignment process) Do not Calibrate. Instead, make sure to set the movement manually. Go to Guide-> Settings and set Xspeed and Yspeed and Angle according to the numbers -16, 16, -90. Note that this differs from the official documentation, and was supplied by Jules Halpern. This may change in the future.

Finally, select the 'guide' radiobutton and hit start. The starfield will change to a small window just around the target star, and it will commence guiding on that star. Now you are ready for alignment.

Note: sometimes maxim pro gets confused and refuses to stop using that small window even when you're trying to do the large exposures. We fixed this by going to the Guide tab, hitting 'settings', and in the bottom right box (exposure settings) hitting the 'reset' button. You'll need to go through all the steps (including step 2) again if you have to do this.

Alignment:

The 'official' reference for this task is here:

http://www.astronomy.ohio-state.edu/~martini/osmos/oalign.html

Again, worth a read, but we're not going to be following it exactly. This is one of the more time consuming parts of each mask, which is why it helps if your alignment stars are all close enough to the center of the mask that they fall into the 1k roi - unfortunately, this isn't always feasible depending on what you're looking at.

step 1:

First, we're not going to use oalign.pro. It doesn't actually do anything other than take a couple images for you, and it's not flexible enough to do the job right. We're just going to take the input images for oalign_bin2.py manually.

You need to take two images: one with the mask in with the disperser out, g band filter in, mirror in, and flat lamp on, and one with it out (no filters or disperser in either case). The 'mask in' image should have an exposure time of 1s or so. The 'mask out' image should be 5s or so.

Step 2:

Run oalign_bin2.py *in a regular terminal, not prospero* as the following:

You will need to copy the .oms file and the images into the directory. Again, *wait* several seconds (~10 or so) after the image readout is done before you try to copy them.

It will pop up a ds9 window with the 'mask in' image, with regions helpfully culled from the .oms file placed where it *thinks* the alignment boxes are. So far, it's always placed these in positions reflected top/bottom from where the alignment boxes actually are, so you'll have to go hunting for where the alignment boxes *actually* are. Move the mouse over the center of each alignment box (*not* the regions, the actual boxes on the mask) in the image and hit 'x', then hit 'q'. It will load up the 'no mask' image, and place regions around where you clicked on the alignment boxes in the previous image. If you're very lucky, each of these boxes will have a star in it already, but probably not. You should be able to tell fairly easily which are the alignment stars, though - they're the brightish stars near the boxes, presumably all offset from the boxes by roughly the same vector. If you can't find them, check your pointing. Similar to last image, move your mouse over the center of each alignment star (in the *same order* as you did for the boxes) and hit 'a', then hit 'q' once you've done them all. It should pop up a pink 'x' where the solution it calculates will put each star; hopefully these are near the center of the alignment boxes. Close the ds9 window (or minimize it, but in that case make sure to close it once you're done).

Step 3:

In the terminal, it will have output the suggested guider and rotation offsets. These are the 'gdx', 'gdy', and 'dPA' values.

Bring up the 'rotator GUI' if it isn't already open on the main workstation. Over on the guider computer, hit 'stop' to temporarily suspend guiding and quickly enter the dPA into the rotator box and hit send, and once it's done, turn the guider back on by hitting 'start'. As long as the rotation was less than ~0.5degrees, the guider should be able to handle it; if it can't and the star's gone outside the view, send the reverse rotation to bring it back, then apply the rotation in multiple smaller chunks, allowing the guider to catch up in between.

This is the only correction you will make the first time. It seems clear that the rotator angle needs to be adjusted before dx and dy. With you correction made, and the guide star back in the center of the guider window (always wait for it to come back to the center after every correction), take another alignment image as before and run oalign_bin2.py with this as the last entry. Now we will adjust the dx and dy. First, enter in the negative of the supplied gdx into the dx box!!! Note that you cannot shift it more than about 100 guider units in one go, or the star will leave the field and the guider will get lost. If the supplied offset is greater than 100, break it into multiple chunks of 100 or less and let the guider catch up between each one. If the guider loses the star, apply backwards offsets until it finds it again, then continue more slowly. After you have successfully applied the dx offset, now apply the dy offset. This is not the negative like the dx-case but exactly what the output of the program supplies.

Step 4:

Repeat this alignment procedure until the offsets in dx and dy are less than +/-~7.

Finally, you are aligned. You can now finally go on to taking science images.

6: The observing sequence

Roughly, observing a science target can be broken down into the following steps, each of which should have been explained in the previous sections:

Slew to nearby bright star

(Check focus and adjust if necessary - check a couple times during the night, as it may change due to changing temperature)

Adjust pointing (this is done *right* before alignment so that it doesn't have a chance to drift)

(Set rotator angle, if the current mask has a different one than the last mask)

Slew to target

Begin guiding

Take comps

(Take flats, if you didn't do twilight flats)

Do mask alignment

Take science images (we were doing two half hour exposures)

(If taking offset skies:)

(Take science images)

(Apply offset)

(Take offset sky images)

7: Shutting down

The shutdown checklist can be found here:

http://mdm.kpno.noirlab.edu/24_manual/hiltB.html

You can rotate the instrument package back to zero if you want, though you don't necessarily *have* to (unless it's the last night; if so, you should probably put it at zero and take out your masks). If you do, remember to do it before you close the mirror covers. The mirror covers should be closed before you start the dome closing.

Note that the dome must be at home position to be closed. The checklist tells you to flip all the switches at once, but after you flip 'dome home/free' to down (home) you have to wait for the dome to actually get there before you flip 'auto dome' to down (off).

Remember to fill the dewar and record the fill time on the door whiteboard. It's probably a good idea to take your night's data and move it out of the main directory into a subdirectory, and also to copy it over to whatever you're using to bring it home with)