Using the FLI and Andor Cameras for Guiding and Slit Viewing
John Thorstensen, Dartmouth College
2011 February 1
Until 2010, the standard acquisition/guiding cameras at MDM
were based on three-stage image tubes. These were deployed
in the mid-1980s, and served well, but had numerous shortcomings.
Early in 2010 we took delivery of three Andor Ikon DU 937-N
cameras, to replace the image tube systems.
These are mostly aimed at the laboratory microscopy
market, which is much larger than astronomy in dollar
terms, so they represent a lot of expensive research
and development. They are also designed with laboratory
scientists in mind -- scientists who have other things to
worry about than tweaking their CCD systems -- so they're
basically 'turnkey' systems.
Salient features of the cameras are as follows:
The Andor cameras are vastly more capable than the
image tubes. In addition, they are controlled
entirely by a PC, so they may be operated remotely.
- Thinned, backside-illuminated CCD sensors, with
over 60 percent QE from 4000 to 8500 A, and over
90 percent between 5000 and 7000 A.
- 512 pixel square active area; 13-micron square pixels,
6.6 millimeter square imaging area.
- Frame-transfer (i.e., data are shifted from the active
region of the chip to a covered 512-square area for readout,
eliminating the need for a shutter).
- Up to 2.5 MHz readout speed, resulting in 8 frames per second;
something like 10 e- read noise at 2.5 MHz, down to 2.5 e-
read noise at 50 kHz (which is too slow for most purposes).
- Thermoelectrically cooled (they can go much colder than we
As of this writing, one of the Andor cameras has replaced
the guide camera on the 2.4 m, and the image tubes
have not been used at this port since the beginning of
the 2010B semester. Also, in the summer of 2010,
Yorke Brown (a consultant with ties to Dartmouth and
the Sloan telescope) designed and implemented a simple
slit-viewing port for the Modspec and Mark III spectrographs.
Another Andor has been dedicated to this port, so that
an observer can use one camera to view the slit and
another to handle guiding.
We have at present two different software options
to run the cameras. Both run under Windows.
They have complementary strengths,
but unfortunately neither one "does it all".
- Solis. This is Andor's own package to
run the cameras. It is quick and convenient to use,
and takes advantage of the very high speed these
cameras are capable of. This is what you want for
quick centering, focusing, and the like, and it is
the software of choice for slit viewing with
the Brown optics (or with CCDS). There
are no astronomy-specific functions, and in particular
there is no autoguiding option.
- Maxim DL. by the Canadian concern
Diffraction Limited. This package is aimed at the amatuer
astronomy market -- you can see their ads in Sky and
Telescope. Maxim DL is not built to handle the
Andors' high frame rate, but it does have an autoguiding
function. This saw first use in 2010 June and
has been in routine use ever since.
Note that you'll need
the observer password to get onto the
PC that controls the camera. This is not the
same as the password to the Linux boxes, so --
be sure to get it from the staff!
Autoguiding with Maxim DL
Maxim DL is a huge, feature-rich program aimed at
amateur astronomers. We use only a small subset of the
As noted earlier, the autoguiding capability of
Maxim DL is its great advantage over the Solis software.
To start the Maxim DL software, double-click on its
icon. Unlike the Solis software, it doesn't
look for the camera right away, but simply pops
up a large grey window with menu bars at the top.
Find the Camera Control icon -- it resembles
a little power plug, but it's supposed to represent a
camera with a cable coming out of it. It's about the
sixth icon from the left. When you click it, the
Camera Control window appears.
The Camera Control window has three tabs at the
top -- Expose, Guide, and Setup.
Start with the Setup tab. It has panels for
Camera 1 and Camera 2, which for us are
one and the same.
- Just to be cautious, verify that the shutter
is being held permanently open (or else it'll
wear out pretty quickly). To do this, click on
Setup Camera in either Camera 1 or Camera 2; in
the resulting window, hit the Advanced button;
there's a check box at the bottom of that window
(for which we paid extra!) that holds the shutter
open. See to it that it's selected.
Note that this needs to be done before the camera is
connected (next step).
- Click on Connect -- the program should
take over control of the camera. If it doesn't, either
another program (e.g. Solis) is running the camera, or
the camera needs to be power-cycled (do this as
a last resort).
- Turn the Coolers On with the
button provided. The other Cooler buttons
in the Camera 1 and Camera 2 areas let you set the
temperature; -40 (that's minus 40) seems to
be a good set point.
Now that the camera is connected, you can move on to
the Expose tab. This is what you'll be using
for acquiring objects. On the right side is a
column of three buttons; you want Continuous,
and be sure Autosave is off. Toward
the bottom, you'll want to set X binning
to 2, and Y binning to Same. Near the
top, set the exposure time to something short,
like 0.7 seconds.
If you hit the Start button, the camera
will start taking pictures continuously. If it
doesn't, it might have hung up -- try disconnecting
and reconnecting the camera in the Setup tab.
You'll see the image appear in a window. Here
are some things you can do to change the display:
- To change the magnification, put the mouse in the
image area and use the scroll wheel.
- To change greyscale stretch, you have
two options: (1) Right mouse button, "Screen Stretch"
item; useful choices are "low", "medium", "high", and
"range"; or (2) Hold down the Shift key and
the left mouse button simultaneously, and
move the mouse in the image area.
Note that the image updates rather slowly -- a couple
of seconds -- even if your exposure time is short.
This is somewhat inconvenient for fast centering,
but it's not too bad. (This is why you want
Solis if you don't need the autoguider, e.g. for
the dedicated slit-viewer).
Note - you can of course display the Guide tab (next step)
at any time, but the guider can't do anything until
you stop taking continuous exposures in the
Finally, we get down to business with the Guide Tab.
Here we have three important options in the buttons on the
right - Expose, Calibrate, and Track.
Some notes about the user interface and nomenclature:
- The "Expose" radiobutton inside the Guide tab
is entirely different from the "Expose" tab. I feel
that it was a design error on Maxim DL's part to give them
the same name.
- MaximDL uses the word "track" to mean "issue guiding
corrections to the telescope". The "track" switch on the
2.4m control panel means "drive the telescope in HA
to follow the stars."
- Note that the radiobuttons do not initiate the
action you want -- they only select what will
happen when you push Start.
If you select the Expose radiobutton under the Guide
and hit Start, then:
- A single full-frame picture is taken and displayed
in a separate window;
- The software automatically finds the brightest
star in the picture and enters its coordinates in the
Guide star X and Y
boxes. If you don't like the star it selected, you
can click on another star with the mouse, and its
coordinates will be entered into the boxes. (Note:
I believe the integer pixel you click is selected --
it doesn't centroid the image, unfortunately.)
- Note that the Expose action under Guide resets the
fiducial XY location of the guide star. This
will be necessary when you set up on your first night,
but you probably do not want to do this more
than once -- there is much to be gained by keeping the
guide star fixed.
In the Options menu (right side of the box)
there's a control for Track box size.
64 pixels seems to be a good option. The track
box needs to be contained entirely within the image, so
the guide star can't be close to the edge of the field.
Now you can move on to the Calibrate
- First, go into Settings; under
X Axis and Y Axis, set the
Cal time settings both to 5.
- Hit Start and sit back -- After a while
a little red line appears by your guide star, the guide
star moves over, then it moves back, and then it does
the same thing in a perpendicular direction. The
program has mashed the guide buttons and watched the
star, then figured out the guide speeds and
directions! If you now look in Settings, in the
manual calibration section, you'll see
numbers for X speed, Y speed, and
angle -- the program has just set all these
for you. If you rotate the instrument you
manually change the number under Angle.
It appears to work to make this simply equal to the
rotator angle, but your mileage may vary.
(If in doubt, you can simply recalibrate at the
new angle, until you're sure of your procedure).
- Once you have the calibrations, you'll want to
write them down. Unless you change the guide rates
on the TCS, or rotate the instrument, they should
be good for the rest of your run. I have found
that with 2x2 binning and the offset guider,
the X and Y rates were both around minus 13,
and the rotation was close to the rotator angle.
You're finally ready to guide!
- Select the Track radio button.
- For Aggressiveness, enter something like
5 in both axes -- the guider will then try to correct
50 percent of the image decentering.
- With the guide star somewhere near its fiducial position,
hit Start. A little postage stamp
around the guide star will be displayed in the
guider image window (you can blow this up and
stretch it to your heart's content); the guide
star should walk to the middle of the box, and
you're off and running.
Here's a strategy for guiding and acquisition.
First, set up as follows:
- Select an unmistakable target, such as a
bright star or anything else you're sure you can
- Use JSkyCalc24mGS to
drive the guide probe into the predicted position
of a nice bright guide star (like 10-12 mag).
- In the Expose tab, take continuous short
- Center the telescope "science instrument"
accurately on the target, e.g. by taking test
exposures and ensuring you're centered, or by
simply centering it in the spectrograph slit.
- Look for the guide star in the Maxim DL
image. If the center of
your instrument is offset significantly from the center of
rotation of the guide star XY stage, the guide
star may appear
close to the edge of the image, or it might be
off the image altogether. If so, move the
guide probe to approximately center the guide
star. [Note: if there are other stars near
the guide star, the pattern appears flipped
vertically compared to the field map in JSkyCalc24mGS.]
- Once you're happy with the guide star,
stop the continuous exposures.
- Select the guide tab.
- Select the Expose radiobutton
- Hit Start to record the position of the guide
- Write down the XY coordinates that the program found
for the guide star; they should be near the field
- Compare the true guide probe position in xmis --
they're shown in white letters on a blue background --
to the nominal JSkyCalc24mGS guide star position.
Figure out (and write down)
the delta-x and delta-y that, when applied to the
JSkyCalc24mGS coordinates, will give you the coordinates
you have. [Be sure to get the true coordinates from
the xmis window, since the coordinates returned
to JSkyCalc24mGS when you hit "Read Guide Probe" are
- Calibrate the guider and set up the
aggressiveness as described earlier.
You're now set up -- for subsequent targets,
all you need to do is
- Use JSkyCalc24mGS to select a guide star and
move the probe to its expected location. Be sure
to include any non-zero instrument rotator position,
and if you're using the inner slit on OSMOS be sure
to select the red slit in JSkyCalc24mGS.
- In xmis, apply the "dx" and "dy" that you
figured out earlier, by entering
the values into the appropriate boxes and executing
the moves with a carriage return (with the mouse
pointer in the box). (This accounts for any misalignment
between the instrument center and the center of the
MIS stage. The dx and dy corrections are independent
of rotator angle.)
- Select the Expose tab and start taking
- Identify the guide star. If the telescope
pointing is off, you may need to move a little to
find it. If you have a slit-viewer running, and
can identify the field,
put your object near the slit and
that should bring your guide star into view.
- Move the telescope to put the guide star
where it should be in the guider field of view;
anywhere within the tracking box will do fine.
(This is why you want a not-too-small track box).
- Stop the continuous exposures.
- Select the Guide tab.
- Be sure the Track radiobutton is selected.
(There is no need to recalibrate the guider
- Click on the Start button to being guiding.
- Once the guider settles, your instrument should
be tracking within
a couple of arcsec of the target coordinates.
- If you need better centering than this -- e.g.
to drop your object into a
spectrograph slit -- apply small
dx and dy motions in the xmis
(This is much quicker than re-setting the guider
box, and offers finer control.)
Note: for Mark III and Modspec: if the
slit-viewer has the orientation correct (N near
top and E near left), then if the target appears to the
left of the slit, you apply positive dy to
bring it into the slit If the target pokes out on the
right, and vice versa. One arcsec is
12.8 guider steps. [The directions for guide
are independent of the instrument rotator, since
the guide stage co-rotates with the rest of
These procedures seem very complicated
when written out explicitly like this,
but in practice they go pretty quickly.
The procedure leaves you
exactly centered, and guiding, in one go.
Viewing the Slit Using Solis
The staff focuses the slit-viewing camera on the slit jaws
as part of the instrument change, using
using the focus ring on the lens attached to the camera.
The focus ring has a locking screw on it, so the slit-viewing
camera should remain in focus once they've set it up --
you shouldn't have to touch it.
Once the slit-viewing camera is in focus, one can focus
the telescope by acquiring a star, putting it near the
slit (the jaws are tilted with respect to the focal
plane, so the focus varies slightly across the field),
and focusing the telescope until the star appears as
sharp as possible. This obviously requires a star and
can't be done until after sunset; there are instructions
further down as to how to set up the program for focusing.
To start the cameras, simply log in to the control PC as
observer and start up the Andor Solis control/analysis
software by double-clicking on its icon. This will look
for the camera and will not start up unless the camera is found.
If the software fails to start and immediately complains
of a timeout, you may be able to solve the problem
by power-cycling the camera. Unfortunately, it looks
as if this requires unplugging the camera's power brick out
at the telescope.
Once the program comes up, turn on the cooling (unless
it's horribly humid and you're just playing around). This
is in the Hardware menu, under Temperature. Minus-40
is a good operating temperature. A box at the lower left
of the screen reports the temperature; it's red and
turns blue when the operating temperature is reached.
There's no reason not to take images when the camera is
warm, but there will be substantial dark current (and
Now open the Acquisition menu "Setup Acquisition" item.
This brings up a window with lots of tabs, most of which
you don't need. Important -- the tabs don't all fit
at once, so you scroll left and right with little arrows
at the upper right of the window.
The leftmost tab is "Setup CCD", which is obviously
important. On this one:
- Check the box that says "Frame Transfer"
- Acquisition mode can be "single" or "kinetic"
- Triggering is "internal" (default)
- Note the "Horizontal Pixel Shift". The chooser there
lets you select "2.5 MHz at 16 bit", "1 MHz at 16 bit",
or "50 kHz at 16 bit". 1 MHz seems to be a good all-round
choice, but 50 kHz is useful for when you want to go as
deep as possible, with some penalty in read time. In that
case read a subarray.
The next tab is "Binning". The slit viewer gives 0.24 arcsec
per pixel unbinned. 2x2 binning is adequate for most
slit-viewing purposes; for the initial focus on the slit
jaws you'll want 1x1 binning. (Also, very bright objects
don't saturate as quickly with 1x1 binning). Binning reduces read time
and read noise at the expense
of resolution; it only takes a few seconds to change
it "on the fly". Note that the Solis program still displays absolute
pixel coordinates (512 square for whole frame) even
if the pixels are binned.
You can also select a sub-image (measured in original pixels).
If you do this you can move it around on the tiny display
with the mouse. There's seldom a need for this.
The "Image orientation" tab: For the Yorke Brown Modspec/MkIII slit viewer
- no rotation
- Select both Flip Horizontally and Flip Vertically.
The cartoon will show an "R" standing on its head. This
choice aligns the Brown slit viewer with North near the top
and East near the left when the instrument rotator is
at zero. The field is rotated by 18 degrees, and
the field is about 122 arcsec square at the 2.4m.
Once you're set up, you can start looking at stuff.
In the bar of icons just under the menu bar, there are
little red and green circles -- these are the start/stop
buttons. You want the one that says "RT" -- it starts
the camera taking data continuously and reading it out.
The red button stops this; you need to stop to reset the
The exposure time is set by a control box that comes up
when you click on the "run-time" icon, which looks like a
little tv remote held diagonally. The Run time box
that pops up has a slider that says "Exposure (seconds)".
It starts at the fastest possible exposure, then you slide
up and when you get to the top it trips over to a new
range, and so on up -- when you start over, it reverts
to the shortest time, which is a little awkward but not
hard to get used to. You can also simply type your
desired exposure in to the entry box at the bottom.
Note that you can make the picture bigger by dragging
out the corners of the Solis window.
Much of the power of the Solis software comes from the
speed and convenience with which you can control the
display to optimize your view. These controls are
- Farther along the icon bar is one that has a red arrow
pointing to the lower left and a blue one to the upper
right -- this is the "Reset" (reset view) icon. It
does two things: (1) restores the display to show the whole field being read
(2) autoscales to min/max values.
You'll use this a lot.
- Two icons to the right is a little thing that looks like two
bold blue arrowheads, up and down. This toggles autoscaling;
if it's on, the greyscale is adjusted image-by-image (which
makes it look pretty crazy but can be good if you're moving
onto a bright star) If it's off, the greyscale stays stable.
- There's a cursor on the image display. Clicking on the image
moves the cursor to the spot you click on . X and Y cuts of
the image through the cursor's location appear to the left and bottom
of the image.
- Here's something really cool -- if you click on the image
and hold the mouse down, you can
drag out a rectangle. When you release the mouse, the
rectangle pops out to fill the whole display.
This lets you instantly blow up any region -- like the
slit, or a star you're using for focus. The "reset"
button described above (diagonal blue and red arrows)
restores the whole area.
At the top of the image is a slider bar for the greyscale.
It only makes sense to use this with autoscaling off.
Click-and-hold on the arrows on the left and right and you'll see
how you can manipulate the greyscale (unfortunately,
these move the greyscale range rather slowly). Also, if you
click-and-drag the greyscale bar, it drag the interval
up and down while keeping the separation of the black
and white levels constant.
If you right-click the mouse, it brings up a menu
that includes "Palettes" -- you can choose various color
schemes for the display. I find two of them to be useful:
"Grey.pal" and "false1.pal". It defaults to grey.
The false color can be helpful for seeing faint features,
if you choose your levels carefully.
- The Display menu item at the top has a 'preferences'
tag which you can use to set the color of the cursor to
some contrasting color. Cyan works nicely on greyscale.
The power of the Solis software is evident when you're
trying to focus the telescope on a star.
Here's a procedure for this:
You get a rapidly dancing star, and the seeing is
really dramatic. Note the starting telescope focus
value, and focus the telescope on the dancing star.
You don't have to worry about afterimages or gain like
you would with the old TV.
- Slide exposure to very short (like 0.3 sec)
(you can do this "on the fly" but it waits til the
current exposure is done to take effect).
- Set the greyscale. This is best done by autoscaling,
then turning off autoscaling (which gets you close), and
then using the sliders to fine-tune. You don't want
autoscaling on during focusing, because you want to
see the peak level change rather than having it normalized
- draw a small box around the star to blow it up.
For acquisition, you'll want to see the whole
field, and set the exposure to something reasonable.
Note that if you have faint targets, but also have
some brighter stars in the field, it's best to use
short exposures to get near the target (because the
feedback is quick), and then set the exposure long
to see the actual target.
Once your target is in or near the slit, you can
use the drag-and-expand feature to get it dead-center,
and so you can keep an eye on it
without having to squint.
Fast Photometry with the Andor
In 2011 January I had cause to use one of the standard
Solis features to do fast photometry - I'd discovered a
system with a white dwarf eclipse, which basically made the
star disappear in a couple of seconds, and reappear in a
similarly short interval some 7 minutes later. I was
delighted to find that the Andor could record this. Here
are some notes on how to do this.
Timebase. The windows control computer is supposedly
on NTP, but it synchronizes very infrequently. If absolute
timing is important, work with the staff to get it synched during the
day. If you don't have a chance to do this, collapse the
Solis window with the little "underscore-like" window control
button, bring up the control panel, look at date and time,
and figure out the offset by comparing the computer clock
to something more reliable (e.g. the Linux system clocks,
which are carefully synched via real NTP).
In the instructions below, recall that the tabs in Acquistion
are accessed using the left-right arrows toward the upper right.
Setup. In Acquisition, under the
Setup CCD tab,
- Be sure that the Frame Transfer checkbox is selected
- Set the exposure time to the desired value, like 0.5 sec
- Leave the number of accumulations at 1
- Set the Kinetic Series Length to the number of
exposures you want; for example, 900 seconds at 1/2 second
would be 1800.
Now turn to the Binning tab. To avoid huge data set and
keep the frame rate high,
you may want to bin 2x2 and specify a subarray. To do this,
- Select 2x2 binning using the radiobuttons
- Click on Custom, and then fill in values for
Left, Right, Bottom, and Top; these are in unbinned
- There's an button labeled Draw >> that draws your
custom subarray on the display. Use it to check that your
boundaries are correct, and then accept it.
Now select the Auto-save tab. In here,
- Click to enable auto-save;
- Select file type Fits (most likely)
- Give a file stem
- Give a directory for your data. I was using C:\Users\Observer\thorimages,
which I created; feel free to write in it.
- You can elaborate on this by automatically appending numbers, and so on;
note that each time series will give only one file, so there's
no need to have an elaborate numbering scheme.
You should be ready to go now. To take signal, go to the Acquisition
menu and simply Take Signal. The images will be displayed
as they come in, and you can play with the stretch, etc.
Once the operation finished, you'll have a 3-dimensional FITS file in
the directory you specify. The file is time-stamped to the nearest
second (only), and the time between frames is given in the header
to five digits or so. To retrieve the FITS file(s) from the Windows
machine, it's easiest (at this point) to use a USB memory stick.
If you display one of these datacubes in DS9, it will let you step
through the images and see what happens. For further reduction,
you're on your own!