1) Before observing, preferably well before dusk, make an observing plan for the night.
2) Create a new directory on the PC, or clean an old one, solely for tonight's data. Link to the camera and filter wheel (see page 26 of the CCDSoft Manual).
3) Be sure the camera is cool, and change the camera temperature if necessary (see page 27 of the CCDSoft Manual -- it should be running ~30C below ambient dome temperature).
4) Start a log of the night's observing. Front and back pages are available in the STScI folder in the warmroom. Be sure to document the sky conditions and dome temperature around dusk. Also record the Universal Time (UT), Local Sidereal Time (ST), and Local Time (LT) at the beginning of the night.
5) Before opening the dome, with the dome lights OFF, take 8 bias frames (also known as "zero" frames).
6) Open the telescope as usual, following the procedure sheet kept in the warmroom.
7) If there are no/few clouds outside, take twilight sky flats.
8) Focus the telescope.
9) Pick a star from your observing list, and observe the star.
10) Repeat till dawn, or you are tired, or it clouds up.
11) If it is dawn, perfectly clear, and you can stand it, take more sky flats.
12) Close up the telescope as usual.
13) With the dome lights OFF, take 8 more bias frames.
14) Email Dr. Andy Layden with the location of the nightly directory on the PC, rough number of stars/observations obtained, and number of hours spent.
NOTE: here is a page for Troubleshooting basic problems, and instructions for dealing with major problems and emergencies.
Updated 1999 June 09
This should help you plan your night of variable star observing.
1) Login on galileo (the Sun workstation in the observatory warmroom).
2) Copy the current list of stars to your account. It is in priority order, 1-N where 1 is highest and N is lowest prority. The RA and DEC (epoch 2000) are apparent. It may be handy to print out this file.
3) Run the planning command file.
4) The program begins by deleting all the files it generated the last time you ran it:
5) Then it asks you for tonight's date (specifically, the date on your watch at sunset, not UT date), in Month Day Year format (eg 6 19 1999).
6) And the length of observations (how long it will take to observe a typical star) in minutes: 15
7) And frequency of observations (how many observations of that star per hour): 0.67
8) And the maximum airmass: 1.6
9) And the filename to read: stars.prio
10) The program is done. You can plot up your plan using SuperMongo (aka, SM) and print it out:
11) The plot may look something like this.
12) You can print a hardcopy of the plot on the warmroom's laser writer.
13) HINT: while observing, after you finish the observation set for a star, place a mark on the plot at the ST. In the event that you deviate from your plan (lose time to clouds or equipment failure, or even if you observe faster or slower than the default 15 min per object) you can revise your plan "on the fly". Just revert to the Rule in Red ("In general, ...") in Step 11.
Updated 1999 June 09
:C\IMAGES is the main directory for storing images. I suggest making a subdirectory specifically for this night's work, to avoid filename confusion with images from other nights.
Create a new folder (directory), I suggest a name like 99jun03 for the 3rd of June 1999.
When you Save your first image to disk, be sure you navigate to this folder. Subsequent Saves will automatically go here.
Updated 1999 June 03
1) Be sure all light sources are OFF in the dome, including the eyepiece reticle lights. The dome slit should be closed, and the dome as dark as possible. The covers should be on the telescope.
2) On the CCD PC, take an exposure (flash camera icon). In the dialog box,
3) Once the image is read out and has appeared on the screen, save it to disk (floppy disk icon). For the first image of the night, you will have to change directories to the one you cleared/created earlier. Call the image "biasNN.fit", where NN is the running number 01, 02, 03, ... . If you take another set of biases later in the night, be sure not to repeat numbers -- you may want to call them 11, 12, 13 , ... .
4) Repeat Step2 seven times, giving 8 bias frames in the sequence.
NOTE: you should check the number of counts in your first bias (slide the pointer around the image and note the (X,Y,Counts) readout in the lower left, or do a histogram on a bias frame), you should see a peak around 974 counts. This is the usual bias level for our chip. If it is different by more than 10 or 20 counts, please report it to Dr. Layden or Dr. Laird.
DEFINITION: A "bias" or "zero" frame measures the amount of signal, or charge, on the frame due simply to the read-out. It has zero exposure time, and recieved no illumination. In addition to determining and correcting for the mean level of this signal, the bias can also be used to correct the object images for any pattern in this signal which is constant in time (often referred to as "fixed-pattern noise").
WHY A BIAS? The bias raises the background level of counts on the chip. Imagine if there was no bias applied. If you took a frame with a very short exposure time or in very dark conditions, the counts due to the sky background would be about 0. However, there is noise associated with reading out the chip, so some pixels would want to have negative counts. Since the chip only records integer counts between 0 and 16384 (2^14 -- we have a 14-bit chip), these pixels would get the value 0, and the statistics of the sky noise would be corrupted. Putting a bias on the chip ensures that all pixels will have a positive value, and the statistics will be computed correctly.
Updated 1999 June 03
Twilight Sky Flats can be taken at dusk (evening) or dawn (morning). Doing them at BOTH is preferable, if you can manage to stay awake in the morning! The procedures for the two are subtly different, so explicit instructions for dusk flats and dawn flats are given separately. After that, there is some wisdom on when to start taking your sky flats.
NOTE: Sky flats are best taken when the sky is perfectly clear, so the sky is uniformly bright. Uniform clouds are ok, but you should try to get more images per filter (10 or more, if possible) to help average out any non-uniformities due to the clouds.
GOALS: ideally, we would like 5 or more images in each filter you will be using that night. An ideal exposure would have 9000 ± 2000 counts in each image, though any sky flat with 2000 to 13000 is worth saving. All of the images must have exposure times longer than 10 sec. It is also important to move the telescope between images, so the stars change position on the chip from one image to the next.
DEFINITION: Sky flats measure the relative sensitivity of each pixel. Assuming the sky is uniformly bright, the pixels with lower sensitivities will have lower numbers of counts. As part of the "data processing" procedure, we will later correct for the pixel-to-pixel sensitivity variations by DIVIDING each object (star) frame by a combined sky flat frame. The pixel sensitivities may depend on the color of the light they are seeing, so we must take flats through each filter we will use that night. We need to get 5 or more flats per filter to reject "cosmic rays" and star images that appear on the individual frames. We move the telescope between images so the stars do not appear in the same (X,Y) location on the chip.
WHY THE 10 SECOND RULE? One should never take sky flat or object (star) exposures shorter than 10 sec with our CCD. The shutter in front of the camera takes a finite amount of time to open and close. Tests have shown that for exposure times less than 10 sec, this "shutter delay time" results in a true exposure time that is different from the requested time by more than ~1%. What's more, the shutter is an iris (like in your 35mm camera), so the center opens before and closes after the corners, resulting in extra exposure in the center relative to the corners. In theory, one can map out this effect (create a "shutter correction image") and apply the correction to all your images, but it is a pain. Easier to just avoid the short exposures altogether!
1) If you are taking dusk flats, point the telescope about 1 hour East of the Zenith, and rotate the dome so the slit is facing East (away from the setting Sun). Of course, be sure the dome is not occulting the telescope!
2) Be sure the telescope tracking and autodome are ON.
3) Which filters will you be observing with tonight? For example, BVI. Do the bluest filter first, when they sky is brightest, and the reddest filter last, when the sky is darkest. In our example, we are taking evening sky flats, so we begin with the B filter -- hit filter button 2 (1=U 2=B 3=V 4=R 5=I 6=Clear).
4) Take a 0.1 sec exposure (flash camera icon),
Once the image has read out and appeared on the screen, move the cursor around near the image center (brightest region) and see roughly how many counts there are using the (X,Y,Counts) readout at lower-left. A well-exposed skyflat has between 7000 and 11,000 counts.
5) If the image is saturated (all pixels uniformly 16383), you will have to wait a while and try again when the sky is darker (maybe 5 minutes). If this image is not saturated, but has more or less than the optimal number of counts, estimate your next exposure time with
though remember that while you are doing this, the sky is changing brightness outside. Experience will enable you to guess a "seat of the pants" correction to your computed exposure time.
6) Take another test image with your improved exposure time estimate. When your exposure times finally get longer than 10 sec, start saving the images to disk.
7) After each good sky flat, move the telescope East with the handpaddle (about an arcmin is good). This is to ensure that star images don't land atop each other.
8) When you have 5 good flats in this filter, switch to the next filter (V in our example -- hit filter button 3).
9) Take your first flat with an exposure about 1/2 that of your last flat (in B), since the redder filters in general have a higher throughput. Adjust subsequent exposure times as in Step 5.
10) You are finished when you have run out of filters to do (you may want to go back and get more good images per filter) or the exposure times in your reddest filter are longer than 200-300 sec.
1). If you are taking dawn flats, point the telescope about 1 hour West of the Zenith, and rotate the dome so the slit is facing West (away from the rising Sun). Of course, be sure the dome is not occulting the telescope!
2) Be sure the telescope tracking and autodome are ON.
3) Which filters did you observe with tonight? For example, BVI. Do the reddest filter first, when they sky is darkest, and the bluest filter last, when the sky is brightest. In our example, we are taking dawn sky flats, so we begin with the I filter -- hit filter button 5 (1=U 2=B 3=V 4=R 5=I 6=Clear).
4) When the sky levels in your object (star) exposures start reaching around 5000 counts as the dawn is beginning to brighten the sky, you will want to stop taking stars and start with dawn sky flats (if your star exposure times are shorter than 50-100 sec, you may want to start taking skyflats when the sky level reaches 3000 counts). From you last star image, calculate the exposure time needed to give you 7000 counts in your skyflat (say its 300 sec), and take an exposure (flash camera icon),
Once the image has read out and appeared on the screen, move the cursor around near the image center (brightest region) and see roughly how many counts there are using the (X,Y,Counts) readout at lower-left. A well-exposed skyflat has between 7000 and 11,000 counts. If the image is good, save it to disk and compute your next exposure length as in Step5.
5) If the image was saturated (all pixels uniformly 16383), you cut your exposure time a lot (you waited too long to start sky flats): try 50 sec. If the image was not saturated, but had more or less than the optimal number of counts, estimate your next exposure time with
though remember that while you are doing this, the sky is changing brightness outside. Experience will enable you to guess a "seat of the pants" correction to your computed exposure time.
6) After each good sky flat, move the telescope East with the handpaddle (about an arcmin is good). This is to ensure that star images don't land atop each other.
7) Take another image with your updated exposure time. When your exposure times finally get shorter than ~50 sec, or when you have obtained 5 or more good sky flats, switch to your next-bluest filter. In our example, this is V, so hit filter button 3.
8) Take your first flat with an exposure about equal that of your last flat (in I), since the sky is brightening. Adjust subsequent exposure times as in Step 5.
10) You are finished when you have run out of filters to do (you may want to go back and get more good images per filter) or your exposure times in your bluest filter are shorter than 10 sec.
This depends on whether you are taking dawn or dusk sky flats, and what filters you are using. The following are vary rough estimates for when you should start taking 1 sec test exposures. Of course, only flats with exposure times longer than 10 sec should be used, but starting with 1 sec test exposures allows you to increase exposure times gradually
Updated 1999 June 14
GENERAL: we will determine the best telescope focus by obtaining a sequence of short exposures of a bright star, stepping the focus slightly from one exposure to the next. CCDSoft has a handy tool which allows us to determine which focus setting gives the narrowest, roundest images, with the highest peak intensity, ie, the best focus. We then set the telescope focus to that value. You can read more about this procedure in the CCDSoft Manual.
NOTE: the focus is set using the buttons on the handpaddle (not with the keyboard).
1) point the telescope at a bright star ( V ~ 9.0) near the zenith. You might pick a star from this list of candidates.
2) set a starting focus. Below are some final focus values at different dome temparatures, to give you a first guess at where your final focus might be. Take the estimated final focus, and subtract 5 focus units, to get your starting focus. Drive the focus 10-15 units below this value, then approach the starting value from below. There may be some "backlash" in the focus screws, so always aproach focus from below. For example, it is 15 C in the dome, from the list below, we expect a final focus around 2345, so our starting focus will be 5 lower, or 2340.
2) start a sequence of 12 sec focus exposures (flash camera icon). In the dialog box,
this will take one 12 sec exposure and write it to the screen, then pause.
3) find your bright star on the image. Place a box around it by putting the cursor to the lower-left of the star, holding down the mouse button, and dragging the cursor to the upper right. This defines a region inside which the focus tool in CCDSoft will look for a stellar image (be sure your bright star is the only star in the box). Hit continue and the CCD will take another 12 sec image and display its radial profile and peak counts in a special box.
4) increase the telescope focus by one unit (to 2341 in our example) and hit continue to take a new exposure at this focus setting. CCDSoft will add the new stellar profile and peak count value to its box. Carefully record the focus value associated with each image, for example on a sheet of scrap paper... its easy to forget which focus value corresponds to which profile on the focus plot!
5) repeat Step 4. If all goes well, you will see that the first exposures in your sequence had low, broad profiles and few counts at peak, but as the focus increased, the images get taller and narrower. At some point, the images began to broaden and flatten again. This means that you passed through best focus, and it is time to stop your sequence. Look back and decide at which focus setting you got the tallest and narrowest profiles. This is your final focus value.
6) set the telescope focus to the final value, but drive the focus 10-15 units below this value and approach it from below to avoid "backlash" effects.
7) record the final focus value and dome temperature on the observing log.
NOTE: the telescope focus is probably a function of temperature, but we haven't yet determined the nature of the function. You may have to refocus during the night if the temperature changes much. We have also seen evidence that the focus changes when the telescope moves across the Meridian, probably due to the mirror shifting a tiny bit as its weight is redistributed. Be aware that you might need to refocus after a Meridian crossing, and please record clear events of this phenomenon so we can confirm, quantify, and attempt to correct it.
NOTE: the focus exposures should always be 10-15 sec each. Stellar scintillation (the variation of the star image in brightness and position with time) happens rapidly, as you know from watching stars twinkle. However, if you take a short exposure (a few sec or less), you are apt to "freeze" some of this motion, and get a deceptively good or bad estimate of the image shape and height. Observing the star for 10-15 sec enables these variations to average out to a characteristic value. Observing much longer than 20 sec is probably a waste of your time -- chose a brighter star to focus on.
Updated 1999 June 04
In most cases, we will obtain several frames of the same star, at slightly different positions on the chip and in different filters, to create an "observation set". For example, if your observing plan calls for observing the star SW And at 10:30pm, you might obtain 3 V and 3 I frames of the star in close succession, and we would refer to these as a "set".
1) chose a star to observe (see Planning Your Observations) and get out its finder chart.
2) point the telescope to the star's coordinates. They may be preset in the telescopes coordinate table. If not, feel free to place them there, but record on the finder which number the coords are in the table.
3) take a 10sec exposure (flash camera icon) to be sure you are pointed correctly. In the dialog box,
Can you identify the variable and comparison stars relative to the finder chart? Is the variable star near the center of the chip? Are all the comparison stars in the field of view? If not, move the telescope using OFFSET -- for scale, remember, the CCD has a 20x20 arcmin field of view, ie, 1200x1200 arcsec).
4) once you are happy with the pointing, run into the dome and ensure that the slit is not partially or fully blocking the telescope. If, after some experience, you find the AUTODOME is working correctly, you can skip this step.
5) On the finder chart, there are instructions on how to take an observation set for this particular star. For example,
6) Set the filter wheel to V (filter button 3). Take the first observation (flash camera icon).
7) Check the sky level; if it is bright (>6000 counts), you may want to shorten the exposure times and take more exposures through each filter (or, dawn may be coming on, or you may be pointed near a bright moon... check in the dome if the sky level seems odd).
8) Save the image to disk (floppy disk icon).
NOTE: For consistency and to avoid confusion, we will name our variable star files with the name listed on the finder chart. In this case, the star "SW And" appears as "sw_ and2v1", where the 2v1 indicates that it is the first observation through the V filter in the second set (eg, we obtained six frames of SW And a few hours ago, now we are obaining a second set of frames on this night). The bias frame would be called "sw_and2z1".
9) Go to Step6 and take the next frame in the sequence, or, if required, move the telescope using the OFFSET command on the telescope PC. If you have finished the sequence, go to Step10.
10) Record comments in your log : Starname, UT/Local Time, Sidereal Time, CCD Temperature, Exposure/Filters, Comments (eg, moon & sky conditions, problems, errors in file naming, etc). Put a cross on your observing plan sheet for this star at the time of observation, so you don't accidentally re-observe it. Go to the next star in your observing plan.
IMAGE ANOMALIES: sometimes something goes wrong, and your images look funny. Here is a nice web page devoted to describing anomalous images, identifying what went wrong, and how it can be corrected. Please inform Dr. Layden or Dr. Laird if you cannot correct the problem.
Updated 1999 June 03