This page is designed to walk you through taking long-slit observations using the OSMOS instrument, as well as the MDM4k chip, on the MDM 2.4m Hiltner telescope. The point of this is to provide a reference to remind you of the various steps required to take your science frames, as well as your calibration frames. If you follow these steps, then you should be able to use my OSMOS Reduction Guide to reduce these data!

There are five basic images that we can take:

1. Darks

   Darks are images taken with the shutter closed for some period of time equal to the exposure times you will be using for your science and flat field data. The purpose of a dark is to remove the "dark current," which is the response of the chip even when it is not exposed to light. It will have to be subtracted out. For a good detector, it is pretty minimal. It can also be used to look at bad pixels on the chips, and to examine cosmic ray statistics. We'll cover taking darks in Step 1.

2. Biases

   Bias frames are like darks, but are taken for the minimum amount of time allowed for the chip. They help to remove the underlying noise level found in every frame. However, the MDM4k has four different quadrants which all have different bias levels, and it is much easier to use something called the "overscan region" to remove the bias in each quadrant. The overscan is a section in an image that is found on the sides of the chip (and can be specified, but for the purposes of this guide, we're using a 32 pixel region to the left and right side of the chip) designed to have counts at the bias for that quadrant. The observer then, can go without taking biases, and just use this free region to remove the bias from all of their images. If you want to take biases, you are able to, but I haven't had success using them as compared to just using the overscan region.

2. Flats

   Flats are images taken of an illuminated source to measure the response of the detector across the chip. You'll want to put the slit and the disperser in, and then provide some uniform-ish illumination source, such as:

     The Sky - At twilight, for a short period of time of around 10ish minutes, you can take images with the telescope tracking turned off and get enough counts to illuminate the chip equally. Taking these images will be pretty important, but it's stressful to race against the setting or rising sun to get good twilight flats. However, while twilight flats are excellent for fully illuminating the chip, they suffer from having absorption features from the sky, so you'll have to take other flats to correct for this. I'll discuss these in Step 3.

     The Dome - Inside the dome is a big screen, and if you point the telescope at this screen with some illumination source such as the lights inside the dome, or even the ambient light during the day time, you can get a pretty good illumination of the chip.

     The MIS Lamp - There is a lamp that is used specifically for flat-fielding called the MIS lamp. You control the lamp with the xmis controls, and you'll put in the finder mirror, then turn on the lamp, and take short exposures to get some measure of illumination. These flats don't illuminate the chip fully, but can be used in tandem with sky flats to create a good master flat.

3. Arcs

   Arcs are spectra taken of special gas lamps for wavelength calibration. Depending on the chip / slit / wavelength range you are using and / or care about, you'll want to use some of the built in lamps available. We'll be taking Xenon lamps for the setup in this guide. The lamp images will be dark except will have long vertical lines in them corresponding to the bright emission features in the lamp spectrum. I recommend taking arcs after every science target, at the same rotation angle, RA, and DEC, because various issues with the movement and flexure of the telescope can affect the wavelength calibration. Arc spectra will be discussed in Step 7.

4. Standard Star Spectra

   If you want to flux calibrate your science data, you'll want to get a spectrum of some sort of standard star for a short time. As with the arc images, I recommend taking a standard star spectrum for each set of science spectra, using a standard star near to your science target. I'll go over these observations in Step 5.

5. Science Data

   Finally, the science data is the most important data! It's why you're at the telescope. You'll want to take enough data so that you get enough signal to satisfy your needs as an astronomer. These will be discussed in Step 6.

You are going to want to take dark frames at some point in the afternoon, when you can't observe anything anyway. Dark frames take a lot of time, because you're going to want to take them with the same exposure time as your science / arc / standard frames, and are pretty good for taking in batches. I tend to start taking them right before I go to dinner in the afternoon, so you can take them while you're eating, but it's up to you.

You'll want to take the darks before you do any moving of the telescope, or opening any domes or mirror covers. You'll also need to make sure that the dark hatch is closed for this as well. The point of a dark is that there is no light reaching the detector!

OSMOS is controlled through Prospero, which will mostly mean that you will be typing commands into the prospero controller terminal window. The set of commands you can issue can be found here. There are some important things you'll want to set for your dark. First, you'll want to tell the instrument that you're taking a dark. To do this, in Prospero, you'll type:

   pr> dark

And you'll be presented with a response, saying something like:

       DARK Name [dark] ?

You should type in the name you want to be put in the header for this object. You could also, in this example, just press enter, since prospero is already set up to call them a dark. It's important that you run this step, since by telling Prospero that you want to take a dark, it won't open the shutter for the images.

Next, you'll want to change the filename:

   pr> filename dark.0001

Here, we are going to name the actual file that is produced from this observation "dark.0001.fits". You can name the files anything you want, but I like to name them in a pretty specific way. First, I put the type of exposure at the start. If I'm taking an arc, I'll call it "arc.000X" or whatever, and if I'm taking a science target, I'll call it "j0641.000X", so that when I'm going through the data in the future, I can quickly see what each exposure was targeting. After the dot, you'll see a number. This is a running number throughout the night, and it's also important, since you'll want to not repeat this number, or skip numbers. Prospero is pretty clever, and will increment this number if you are taking multiple exposures without changing the filename. I am also a proponent of changing the first digit of this number for each night. So, on the first night, you'd start with dark.0001.fits, and then the next night would be dark.1001.fits. In this way, you can really easily see when you took a particular image. You can do what you want, but hey, this works for me.

Now, you'll want to set your exposure time:

   pr> exp 1800

This is in seconds. Remember, you want to take darks for each of the exposure times used in future taking science / standard / arc images.

Finally, time to take your exposures!

   pr> go

You could also type go X, where "X" is a number of images you want to take in a row. This should work well, but sometimes, Prospero will have trouble incrementing the file name, so you should be careful and watch out for this. It will just write the file to a dummy name, and you'll have to manually switch it.

So, now you can just do this and change the exposure time and keep taking all of the darks that you need.

Once you've taken your darks, and you're ready to get everything set up, you'll want to run the startup sequence. There are guides in the control room, and you can read all about the startup here, and you should follow this EXACTLY AS WRITTEN. I'll mostly focus on how to take the various types of exposures, as well as what you should be doing during the course of the evening. But keep referring back to the Hiltner manual when you need to, because it is quite well written, and very helpful.

When you're done, if the weather is right, you should have an open dome, and the telescope should be pointing to zenith, and the mirror covers and dark hatch should be open. Make sure to use xmis to send the guide probe to the origin, so it's out of the way entirely, although this isn't necessarily that important for spectroscopic observing. Now, you wait for astronomical twilight.

For twilight flats, you should keep the telescope pointed straight up, looking at the sky. Then, you should go to JSkycalc. Keep it updating, and keep your eye out for the ZTwilight field, which shows you a number corresponding to how bright the sky is at the current time. I can't just tell you when is a good time to observe, and for how long, but I can tell you the ZTwilight value that worked for me for taking good spectra.

The MDM4k chip saturates at about 60k counts, and so you're going to want to continually examine your flats so that you get ones with counts at around 10 - 20k. At ZTwilight = 13, and an exposure time of 30 seconds, I was able to get 15k counts for my spectroscopic flat. You're going to want to take the flats in the short period of time when ZTwilight goes from about 14 to about 11, and you'll just keep upping your exposure times.

To take the flats, you'll want to set everything up in Prospero first:

   pr> flat

And you'll be presented with a response:

       FLAT Name [dark] ?

You should type in something like "twiflat" or "flat".

Next, you'll want to change the filename:

   pr> filename flat.XXXX

Where the XXXX is the number corresponding to the next file number in the sequence.

   pr> exp 30

This will take a 30 second exposure, but you may want to take shorter ones if it's earlier.

Now, you'll want to put the slit into place:

   pr> slit 5

I've put "5" here, because for my most recent run, this was where on the filter wheel the slit was that I cared about (the 1.2" center slit). You should find out where the longslit is that you care about, and put that filter wheel value in here.

Next, put in the disperser:

   pr> disp 1

I think that the VPH Grism, the disperser often used for the long-slit observations, should always be in the wheel position 1, but you should check.

Now, you're ready to go. If you are feeling nervous about capturing the exact time when you get enough counts, you could only read out the central portion of the chip, by typing:

   pr> call roi1k

This will save a lot on readout time, since the full chip takes ~2 minutes to read out. Just remember to change it back to the full 4k when you're ready to take the real data:

   pr> call roi4k

If you are planning on only taking spectra, and you want to keep the spectra near the center of the chip, you could also go the route where you only read out a cutout section that doesn't include the top and bottom of the slit:

   pr> call roi4x1k

You should figure this all out before your run, and stick with it. I am going to go the route where you just use the full chip, even though it takes a while to read out.

So, with everything ready, you should take images and look at them quickly after they read out to see if they're exposed properly. If not, take a longer exposure. You're going to want to get 3 or 5 good twilight flats for your observing run. You'll want to take flats for each observational set-up you will be using. If you're using another slit, you'll want to put that in and take flats with that, too. It's good to keep taking flats until you'd have to spend 5 minutes to take one with enough counts, because then it's too dark, and it's time to observe.

Before you take your science data, you are going to want to make sure that the telescope is pointing well, and that it is somewhat focused. This is a pretty simple procedure. The first thing you want to do is use the xtcs window to select a nearby bright star. The xtcs window has a button labeled Get Coords click on this and then select Nearest Bright from the pull-down menu. This will load the Yale Bright Star catalog entry nearest to where you're pointing the telescope, which in this case, should be zenith. You will want to click Send Coords, and then Go, and the telescope will slew to the bright star. Now, you should turn on tracking to just roughly follow the star in the sky, and then you're going to want to put the telescope into a mode with quick readout, so change the exposure time and the ROI to be something like 1 second, and 1k, respectively:

   pr> disp 6

I've moved the disperser and the slit to the empty filter wheel slots. Notice that I haven't changed the file name, or the object name. You can do this if you want, but movie mode doesn't save images, rather, it just reads them out to the screen such that you can see quickly where the telescope is pointed. When you type movie, it will take a 1s image, and read out the image, and repeat until you type stopmovie. Hopefully, unless the telescope is pointed terribly wrong, you should see a really bright star near the center of the movie image. Like, a really, very bright star. Saturated in the 1s exposures. Grab the big paddle, and your job is to move the telescope such that the star is right in the center. You're going to have to figure out how holding N, S, E, or W moves the telescope, but it shouldn't be too hard. The paddle requires you to hold the button down for a little while for you to really observe the motion in the movie images, but you'll see it. Get it right in the middle. Then stop the movie mode.

Now, with the telescope all centered, you are going to want to reset the RA/DEC encoders, and essentially tell the telescope where it's really pointing. First, look at the TCS display monitor, which tells you where the telescope is pointing, and make sure that the "Next Object" field has the RA and DEC for the bright star. Then on the xtcs window, click and hold down Setup, and then let it go on Set RA/Dec Encoders. Good, you've pointed. If you ever lose pointing in the night, like, completely lose pointing, you're going to have to follow these steps to restore pointing. It's tedious, so just be careful. Over the course of the night, you can reset your pointing any time you want with this method, and it's sometimes a good idea to do it by slewing to your target, then slewing to the nearest bright star, fixing the pointing, and then coming back to your target.

Now, it's time to focus. You probably don't want to focus on the bright star you've pointed, so find a standard star field, and load up the coordinates of the star. To do this, you should just place your coordinates list file (which is in Jskycalc input format) into the main directory on hiltner, and then if you type the name of the file into the xtcs window, and then the name of the target, and press enter, it'll load the coordinates. Then, you can click Send Coords, and then Go, and the telescope will slew to this field.

It's important to start guiding, now, so follow the procedure here, which is very straightforward. Once you're guiding, it's good to take an actual exposure, so you're going to need to change some fields, first:

   pr> object standard

The instrument should still be in roi1k mode, so it'll read out an image. Ideally, if your pointing went well, the standard star should be very near the center. Now, take a look at the TCS display screen, and find the value for the focus. There's a script called focus4k which will take a series of five images and shift the focus between them. You're going to take these images, and then use IRAF and ds9 to measure the seeing for each images. If you want to stray away from the current focus, I would maybe take the value that the focus currently is at (which is probably a great place to start), and subtract 50. So if you read 4255, then run:

   pr> call focus4k 4205

While the images are reading out, display the resulting images in iraf, and then run imaxamine, which will make the cursor, when hovered over ds9, into a little target/circle blinking cursor, and if you put it over a star, and type "a," it will fit the star with a Gaussian and print some information to the screen. Do this for a series of stars in the image:

# COL LINE COORDINATES R MAG FLUX SKY PEAK E PA ENCLOSED GAUSSIAN DIRECT

The number to look at is the second to last number, which is the Gaussian FWHM, in pixels. To convert to arcseconds, just multiply by the plate scale (0.273 arcseconds per pixel). So, in the example above, the FWHM is at around an arcsecond, which is pretty good! Repeat this process on each image until you reach the minimum FWHM, which means the telescope is focused, and you now know the seeing. At this point, input the focus ("XXXX") into prospero to set the correct focus:

   pr> telfocus XXXX

Now your telescope is focused and pointed. You may have to repeat one or both of these procedures over the course of the night as necessary.

So, now you should be set up on the standard star, and the telescope is pointed. We're going to need to set up the star on the slit, and then take a spectrum. To do this, we'll use osctrtask, which requires you to take an image of the field (10s or thereabouts, depending on how bright the standard is), and then an image of the slit (5ish seconds, put the slit in, but not the disperser), and then osctrtask will provide the dx and dy values to put into the xmis window to move the guide probe to set the star up on the slit. You should keep things set with roi1k, since these steps would be lengthy with the full 4k chip reading out. The procedure is fairly straightforward.

IMPORTANT NOTE: When you are guiding with the telescope, and when autoguiding is turned on, you've created a feedback loop such that whenever the guide star moves with respect to the guide window, the guider will send a signal to the telescope to move the star back to the center of that window. As such, when you want to actually move the telescope, you have a few ways to do it. If you're not guiding, you can use the paddle, and just kind of slew in a dumb way. This is all right for pointing, when you want to reset the RA/DEC Encoders, but it shouldn't be used for most other situations, especially when you have to put the object on the slit. To do this, you're going to move the guide probe, which will cause the guider to have to move the star, and thus moves the telescope. It's a clever way to do this, but it's slightly confusing! It's important that this concept makes sense to you, because it is the primary way in which you'll make fine adjustments to the telescope.

When you put in the dx and dy values on the xmis window, and move the guide probe, the star will jump in the guider window. The window only allows for steps of ~100 before the star leaves the boundary, and then it's pretty tough to find the star! Thus, it's probably best to not just input a 267 step offset immediately, since you'll lose your guide star. Instead, split it into little bite sized chunks.

Also, there are a few ways you can input the dx and dy values. You can turn off autoguiding, input a value, and then use the paddle to recenter the star. You can also input the dx and dy values with the autoguider turned on, and let the autoguiding do the work for you, providing the star stays in the guide window.

So, once you've put the star on the slit, you can take another image through the slit, and hopefully the star will be a bright spot in the position you wanted it to be on the slit. At this point, it's time to get going on your first standard star spectrum.

   pr> call roi4k

The XX should be adjusted to the file number, exposure time, and slit number, of course, and it's important to double-check the prospero window to make sure you don't start an exposure before everything is ready. But, once it is, take your exposure, and wait for it to start reading out! Hopefully you'll see your spectrum where you think it should be, and it should be bright, and have the counts you are hoping for. You can take multiple spectra if you want, but generally, standard star spectra are at a low enough exposure time that cosmic rays aren't really an issue. I tend to just take one, so that I can get to the science images.

Now you will presumably be getting antsy to start taking some data, so you should get the coordinates for your first science target, send those coordinates, turn off the guiding (both on the guider computer and also by toggling the autoguiding switch on the TCS control box) and finally, go to the object. Hopefully your star (as well as the standard) are sufficiently close to the zenith so that the pointing is good.

When you've picked your star and slewed the telescope to it, you're going to need to think about rotating the telescope. If you want to line up your slit at some angle on the target, or the parallactic angle, you want to do this before you start guiding. Don't slew the telescope or rotate the telescope while guiding! Then, you have to go into the dome, hopefully with someone else, turn on the TCS display screen outside (sometimes this screen rolls and is unreadable, and so you'll have to fiddle with the knobs on the bottom), and rotate the telescope until it reaches a value to within a degree or so of your target rotation angle. You can correct this in the control room. Make sure that you watch while the rotation is happening so the cords don't get caught on anything, and NEVER ROTATE WITH THE MIRROR COVERS CLOSED. EVER. When you leave the dome, make sure the display is turned off, as well as the dome lights. When you go back into the control room, there is a little rotator window on hiltner where you can put in finer rotation commands to line everything up the way you want.

With the telescope rotated to how you want it, you can start guiding. Make sure to input the correct rotator angle in JSkyCalc24mGS! Once you're guiding, it's time to take the disperser and slit out, and take an image of your field to see where the object is in relation to the slit.

   pr> call roi1k

Replace TARGET with the correct target name, and then I use a shorter version for the filename. So, if my target is J0641.5+3252, I'd type:

   pr> object J0641.5+3252

But do what you want! You'll have to run osctrtask, again, where you'll take an image of the field, and then an image through the slit, and you'll move the telescope to line everything up. Once you've input the dx and dy values, you can then take another image of the field and an image through the slit to make sure everything is lined up (they're short images, and with the readout set to roi1k, the overhead is pretty short). Then, you can put the slit and disperser in and take a spectrum!

   pr> call roi4k

Again, double check to make sure that everything is set up correctly in Prospero. When you're ready, start your exposure!

At this point, depending on your exposure time, you'll have a little bit of time. As part of a good observing run, you should use any time you have wisely. There are a few things you should be thinking about throughout the course of your observing run:

How long should I observe this target in total?

   When you get your first spectrum read out, it's important you look at it to see how it looks. Are the appropriate lines visible? Does it have the S/N you were expecting? You don't want to waste any time you have over the course of the night, so if you are satisfied with your data, you'll want to know as soon as you can.

What calibrations should I be taking over the course of the night?

   Depending on how sensitive you are to time, and how demanding your science goals are, you may want to take arc lamp exposures and standard star exposures right after each set of science frames. However, maybe you trust the arcs you have taken, or the standard star exposures from earlier in the evening, and want to quickly move to your next science target. This is important to know!

What is my next target?

   It's really important that you have a good plan for your evening, and that you know what you want to look at next.

As such, you should just be ready for when you gain control of the instrument after readout for your next move. While OSMOS reads out, you are free to move the telescope, or move the finder in (remember to turn off guiding when the finder is in!) in preparation for calibration lamp spectra, or even just get set for quick reduction and another spectrum! Just remember: every second of darkness is important! Never waste time!

Arc lamps are vital for running wavelength calibration on your data. The types of arcs and recommended exposure times for arc lamp data for OSMOS are detailed here, and I've found that the recommendations on that site are great. In order to get set up, you are going to want to do a few things. First, turn off autoguiding. You are about to put a little mirror into the mirror plane, and this can block the guide probe. You don't need fine guiding for this procedure, anyway. Next, use the xmis window to turn on the arc lamp you care about, by clicking on the lamp name. You're going to want to wait a bit for the lamp to warm up, so put the finder mirror into position, which is a slow process. While it's moving in, check to make sure that the slit and disperser are in, and set everything up in Prospero:

   pr> comp ARC

And, when the mirror is in and the lamp is on, take your exposure.

   pr> go

In the end, your exposure will read out, and you'll see a set of emission lines that will span the entire image in the spatial direction.

The Prospero display window is not showing anything. What do I do?

   If you bounce over to another window on the computer, and then come back, you'll find that Prospero is just blank. If you change the disperser or the slit, then some of it will reappear, but not everything. Just type:

   pr> startup

And everything will come back.

My images have four different levels in the quadrants!

   The MDM4k chip has four different bias levels in each of the four quadrants around the center. If you want to take a good look at your data, you're going to want to run a quick bias subtraction. I outline bias subtraction here on my guide for data reduction, and it's quite easy to quickly run the procedure on your data after it's read out.