Sbig St-i User Manual

ST-i Spectrograph

With ST-i Spectro Software

SBIG Imaging Systems, A Division of Diffraction Limited©
59 Grenfell Crescent, Unit B, Ottawa, ON Canada, K2G0G3

Tel: [phone] | Fax: [phone]|

E-mail: [e-mail] | www.sbig.com©

Contents

Introduction ........................................................................................................... 2

Section 1: Setting up the ST-i Spectrograph ........................................................... 6

Section 2: Using the ST-i Spectrograph .................................................................. 8

Section 3: Installing and Using ST-i Spectro Software ............................................. 8

Section 4: Some Suggested Observations ............................................................ 10

Appendix A: HyperSpectral Imaging ..................................................................... 11

Appendix B: Calibration Data ............................................................................... 15

1

Introduction

Entrance Slit
Location

CCD Sensor
SF11 Prism

SBIG’s new ST-i spectrograph was designed specifically for our ST-i camera. This
unit is intended to enable an amateur to characterize his/her skies, flat field
sources, filter passbands, and other light sources. Its main purpose is to allow an
amateur to measure his sky spectrum, and optimize his flat field sources to better
match his conditions. The reason why this is important can be found in an article
by Alan Holmes titled Flat Fields – The Ugly Truth found at the SBIG web site:

http://www.sbig.com/about-us/blog/flat-fields-the-ugly-truth/

This unit also provides a good way for an amateur to compare his light pollution
situation to users at other sites. The cold reality of the world is that good paying
jobs are associated with cities and light pollution, and light pollution is fact of life
for most of us!

The optical design of the spectrograph is shown in Figure One. A prism is used as
the dispersing element instead of a grating. The reasons for this are two-fold.
Most important, the optical efficiency is high across the spectrum, roughly twice
that of a grating on average. This is valuable since the night sky is not that bright.
Secondly, the spectral range without confusion from multi-order light is from 430
to 1000 nm (4300 to 10000 Angstroms), encompassing the full sensitivity range of
SBIG CCD products. The skyglow in the near infrared (700 to 1000 nm) is
significant to the CCD camera, even though it is invisible to your eye. This near
infrared response is important in detecting faint stars and distant galaxies, but
also is not well baffled by many commercial telescope optical systems, so it needs
to be understood.

Figure One: ST-I Spectrograph Optical Design

Location

2

The design is simple: light enters the spectrograph through a 25 micron entrance

555.7 nm

slit and is collimated by an achromatic lens. It then passes through the Schott
SF11 glass prism, where blue light is bent through a greater angle than red
wavelengths. A second achromat focuses the light onto the CCD, with an
additional plano-convex lens to shorten the focal length and increase the
photographic speed of the system to F/3.66. The speed is important when trying
to capture the sky background. The plane of the CCD is actually tilted a little bit
relative to the angle of incidence of the light, as shown, to reduce the
contribution of chromatic aberration to the optical blur. The spectrum of a neon
gas discharge tube captured with this system is shown in Figure Two. Note that
the spectral lines are straight, top to bottom, even though the slit is seen to be
curved when inspected visually. The slit curvature was added to correct the
natural tendency of the off axis rays from the slit being bent by a slightly lesser
amount by the prism than on-axis rays. Curving the slit straightens out the lines,
enabling greater vertical binning of the CCD for faint targets. In these figures,
blue wavelengths are on the left, and red on the right.

Figure Two: Spectrum of Neon Discharge Tube

Figure Three illustrates the sky background from the author’s back yard, which

th

has about 5

magnitude skies. To capture this, the spectrograph is simply
pointed straight up. As a result, background starlight is also included in this
spectrum and is visible as the semi-uniform continuum baseline. The resolution
around the natural airglow line at 557.7 nm (marked) is adequate to separate it
from the pervasive mercury line at 540.6 nm, and the sodium line at 568.8 nm
(both from streetlights).

Figure Three: Airglow from my Backyard (Magnitude 5 Skies)

3

A program for a PC is included with the spectrograph that allows the user to
control the ST-i and acquire spectra, to create a wavelength calibration for the
spectrograph, output text files of the data for processing with Excel or another
program, and re-bin the data into a format with uniform sized wavelength bins for
simpler comparison with grating data or from other sources. The intent is to
make this functionality easy to use, and to provide a window into the interesting
spectral properties of light sources around us. Figure Four illustrates a screen
shot from that program. A graph of the spectrum is shown, as well as a graph of
the spectrum binned into spectral intervals of equal width to correct for the non­linearity of the prism design.

Appendix A describes how to use this spectrograph for hyperspectral imaging.
Our software support for this interesting new capability for the amateur is a bit
simplistic at present, but we think it will intrigue many users and enable the
development of some new techniques. An example is posted in Figure 5 to
illustrate the power of this technique.

Appendix B graphs the nominal calibration data for the spectrograph, as well as
the dispersion in terms of Angstroms per pixel across the CCD. As you can see,
the dispersion is much lower in the near infrared.

We hope this simple-to-use spectrograph will become an essential part of
the amateur astronomer’s toolkit. It enables the amateur to visualize the
spectrum of the light around him, and to easily measure the transmission of
filters, the reflectivity of diffuse surfaces, and even atmospheric transmission with
a bit more time and setup. This technique will be described in a future paper.

4

Figure Four: ST-I Spectroscopy Program

Figure Five: Hyper-Spectral Image

5

Section 1: Setting up the ST-i Spectrograph

This section assumes you are familiar with use of the ST-i with CCDOPS. To begin,
insert the ST-i into the spectrometer block as shown in Figure Six. Note that the
USB connector is to the right of the ST-4 style tracking connector in the middle of
the camera. It is important to get this orientation right.

Figure Six: Correct Rotation Orientation of ST-i in the Spectrograph

Next, under fluorescent light illumination or even subdued daylight, point the
combined ST-i and spectrograph at a white surface. In CCDOPS, establish a link to
the camera, select a resolution of 1xN under CAMERA-SETUP, with a vertical
binning of 4, and then start focus mode with about a 0.1 second exposure. You
will get a blurry pattern with tilted, badly out-of-focus spectral lines as shown in
Figure Seven.

Figure Seven: Initial Out-Of-Focus Rotated Pattern – Fluorescent Lit Room

Make sure the small setscrews that secure the ST-i in the spectrometer body are
loose, and using your thumb and forefinger, pull the ST-i out a few mm until the
sloping lines in the image become sharp. At the same time (and this isn’t very
easy), rotate the ST-i to make the lines vertical. It is easier to see a slight error if

6

you hit PAUSE to halt the focus process, and resize the image window to put one

5461

7650

edge near a line. Hit RESUME to continue. The ST-i may rotate a little bit when
you tighten down the setscrews, so this can a little time to get perfect. Do not
over-tighten the setscrews. This can dent the ST-i and impede future small
adjustments. When this is done the spectrum is probably not centered perfectly
on the ST-i CCD, so one next needs to translate the nosepiece slightly to achieve
perfect centering. One technique is to grab an image and display it on the screen
at the same size and position as for focus mode. Use the SBIG crosshairs mode to
place the cursor on the desired pixel. Then, stick a Post-It note on the screen with
one corner marking the pixel location to which you will try to move a spectral
feature. An easy line to see and recognize is 5461 Angstroms (for fluorescent
lights). You should move this line to image column 349 (+/- 2 pixels). Another one
is the atmospheric oxygen absorption feature at 7650 Angstroms. It is easily
recognized as a dark band. Move this to pixel 558. When translating the
nosepiece the lines may twist away from vertical you might have to repeat these
steps until it is set. Once it is set, you don’t need to touch it again.

These alignment lines are identified in Figure Eight. Note that the fluorescent
light spectrum may not look exactly like this since the bulbs have different color
temperatures and phosphors in them. However, they all have the 5461 line.

Figure Eight: Alignment Lines

Fluorescent Lights

Angstroms

Daylight

Angstroms

7

Section 2: Using the ST-i Spectrograph

It is easy to use the ST-i spectrograph to capture the spectrum of any diffuse
source. For night sky light pollution, for example, just mount the unit to a tripod
and point the slit straight up. No optics are necessary since the sky fills the field
of view. The acceptance angle of the spectrograph is F/4.3 (13.3 degree full
width), so you need a clear view of the sky with a greater angular width than that.
If you average 3 to 7 images with exposures of two to five minutes each you can
get a pretty good measurement of the night sky. When capturing such images,
always bin 4:1 vertically to improve the SNR and so the data can be accepted by
the analysis program ST-i Spectro later. The program will read both FITS and SBIG
format images, but only saves in SBIG format. CCDOPS can convert between the
two, as well as can many other astronomical image processing programs.

Section 3: Installing and Using ST-i Spectro Software

The ST-i Spectro program can be used to plot the spectrum, and assign a
wavelength to the values. It operation is described in this section.

To install the software, download the folder labeled ST-i Spectro Installer and put
it in a directory on your PC. Next, open the folder and double-click SETUP.EXE.
This should install the program. Find the new program in the program list
(selected by clicking the START button in the lower left corner of your PC’s
screen), and select it to start the program. You can also right click it to create a
shortcut on your desktop.

To become familiar with use of the program, start it up, and go to the installation
folder and go to the FILE I/O menu, select LOAD SBIG IMAGE, and find the image
file SkyGlow.SBIG in the installation directory. Select it. It will load and show on
the screen. The image is shown at the top left. Below the image is a graph of the
average counts between the two green crop bars lines in the image, minus the
background crop region. The program assumes the image has been dark
subtracted and has a “no light” pedestal level of 100, so it subtracts this from the
image data in preparing the graph. If a background crop area is selected it will
subtract the average of that region from the image data. The idea behind having
a cropped region is that when a star is focused on the slit it produces a spectrum
that is a streak only a few pixels tall.

By moving the cursor around on the image data, you will see a line move below it
on the graph, as well as the pixel location and value displayed in the text box

8

below the graph. The wavelength is also shown. Note that the wavelength value
initially will be off, and requires a simple calibration step to improve it.

The graph to the right of the image graph shows the counts per 100 nm wide
wavelength interval. With a prism system the wavelength is not a linear function
of the pixel location, so some math has to be performed to “linearize” the data.
This is done here, and is useful for estimating the total energy in a wavelength
interval. Above it is a visual representation of the energy distribution, color coded
according to the appearance of that wavelength to the eye.

Under display you will find a menu item labeled “ADJUST CONTRAST”. A new
background and range value can be entered to change the display of the image on
the screen. Under TOOLS you will find numerous items. The SHOW CROP DIALOG
item allows you to set the cropped regions to optimize the data shown to the
portion of the image desired. The calibration of the spectrograph can also be
adjusted here. When the SHOW CROP DIALOG item is not shown, slide the cursor
around on an image that shows mercury (fluorescent) emission lines. Find the
pixel corresponding to the peak of the line, and write it down. Enter that value in
the appropriate box that pops up when you select SHOW CROP DIALOG. It is
unlikely you will need to adjust the scale factor for wavelength calibration. You
can determine if there is a significant error by looking at an emission line near the
edge of the spectrometer’s range. The mercury line at 4358 Angstroms is a
convenient one to check. Note that the program does not centroid to find the
exact wavelength of the center of the line, so errors of a few angstroms should
not be cause for concern. The intent of this program is more focused on energy
distributions rather than spectral line identification or shifts.

TOGGLE CROP LINES merely turns the green lines on and off. The SMALL
SPECTRAL ROTATION allows you to rotate a near horizontal spectrum to be
horizontal. This is not actually a rotation, but rather a vertical shift that varies
across the array. It is not uncommon for the horizontal dispersion direction to be
a few pixels off the perpendicular to the spectral lines. This is not a great cause for
concern. The dispersion direction is set by the prism, while the orientation of the
spectral line can be changed by slight rotations of the entrance slit. They are not
necessarily square to each other because of the physics of the spectrograph.
Spectral rotations up to +/-8 pixels can be used.

The last two items under the TOOLS menu allow you to save either the spectral
data or the binned data (into 100 Angstrom increments) to a text file, along with
the wavelength, for graphing using Microsoft Excel or some other spreadsheet
program.

9

ST-i Spectro can also control an ST-i camera and provide real time spectral data.
Under the ST-i Camera menu item, after you have attached a camera to your
computer, select ESTABLISH LINK to connect to the ST-i. Then you can take a
single exposure using TAKE EXPOSURE, or select CONTINUOUS UPDATE to
continuously take new exposures and refresh the screen. Both choices
automatically take dark frames at the start. You can click the red ABORT button
that pops up in CONTINUOUS UPDATE mode to terminate the process, after
which you can save an image or view the data. Note – if you are capturing
skyglow spectra with long exposures you will be better off using AUTOGRAB in
CCDOPS for acquisition since it will save multiple images.

Under the FILE I/O menu you will find commands to load an image in SBIG or FITS
format, or save it in SBIG format. The EXIT command will not only close the
program, but it will also shutdown any attached SBIG camera.

Section 4: Some Suggested Observations

As mentioned earlier, this device is superb for capturing the spectrum of the night
sky. No other low cost device has the sensitivity to accomplish this observation.
At the same time it is a good idea to measure the spectrum of the twilight sky. A
lot of users use twilight flats as flat fields when doing CCD imaging, but spectrally
they are a very poor match to the dark sky, as you will find. You can also measure
any flat field sources you might have, at which point you will discover that they
are not great either. White LEDs are very blue, and incandescent bulbs are very
red. You can measure filter transmissions by pointing the unit at a white screen
and collecting some data, and then holding a filter in front of the slit and
repeating the exact exposure. You can then calculate a ratio of the two readings
using Excel or some other program, and determine a transmission curve. And, as I
mentioned previously, it is possible to measure your absolute atmospheric
transmission with this device and a cardboard box and some white paint. I will
describe this technique in an Application Note in the future. For those who are
interested in trying it now please look up “Langley plots” on the internet. This
technique can also produce an absolute radiometric calibration for your device
since solar irradiance is known to a percent.

10

Appendix A: HyperSpectral Imaging

The ST-i Spectrometer can be used for hyperspectral imaging. This is a technique
where an image can be taken of an object at many wavelengths simultaneously.
It is useful in agriculture and mineral exploration where the spectrum of individual
points can be examined for the signature of chlorophyll, drought stress, and other
attributes. The technique was actually used in astronomy to determine that the
“canals” and darkish patches on Mars were not caused by green plants. Green
plants become very reflective at wavelengths longer than 700nm, a feature not
present in the spectrum of the Martian regions. Anyway, it is possible to
experiment with this technique with this spectrograph. If your optical system is a
long focal length telescope, your targets are limited, with the moon being the
obvious one. SBIG sells a C-mount lens attachment for the spectrograph, though,
similar to that shown in Figure Nine. The way to collect a hyperspectral image is
to mount the ST-i Spectrograph and C-mount lens on a rotary turntable and point
it at the desired scene. One then collects 648 frames of data while the turntable
rotates. If it rotates at the sky’s sidereal rate (15 arcseconds per second) then you
need to collect an image every 140/FL seconds, with FL being the focal length of
your optics in mm. A pixel appears to be 10.34 microns wide at the slit, for this
calculation. SBIG has HyperSpectral Analysis software that will then pick the same
pixel from each image and reconstruct an image from the data set.

Figure Nine – C-Mount Lens Attachment for ST-I Spectrograph

This brief explanation is an introduction. Below I will show a specific example.
Figure Ten shows the ST-i Spectrograph and 35 mm C-mount lens mounted on an

11

AstroTrac mount for astrophotography with camera lens, turned so the motion is
in the horizontal plane.

Figure Ten: AstroTrac Rotary Platform Turned Horizontal

This platform is a bit slow for short focal length lenses, but it is what was available
to the author at the time of testing. By sighting down the edge of the
spectrometer block, the unit is lined up. Then, 650 samples are collected, one
every four seconds, while the AstroTrac rotated. Each image looked similar to
what is shown in Figure Eleven

Figure Eleven: Individual Frame with C-Mount lens Mounted

The broad glow at the top of the frame is blue sky near the horizon, and a few

rd

white surfaces causes the horizontal streaks 1/3

of the way up from the bottom.

The glow on the right around the oxygen is the bright chlorophyll feature

12

mentioned earlier, from green plants. After installing the HyperSpectral Imaging
Utilities Program, run it. First, load the first image of your sequence by selecting
LOAD SBIG ST-i IMAGE under the file menu. The image will then load and display.
Next, under the TOOLS menu, select the SET SMALL SPECTRAL ROTATION
command and enter a value that you have determined to be correct for rotation
the dispersion into perpendicularity with the lines. If you get this value wrong
then your final images will appear to be shifted vertically with respect to each
other. This parameter must be set each time you run the program – it is not
remembered in this software version. Once this is set, select SET BANDS to enter
new pixel values you wish to have binned together for each of the eight bands the
program will reconstruct. These values will be saved in the directory with the
images. After these two adjustments are made then go back to FILE I/O and
select PROCESS FOLDER. The software will automatically go through all of the
files and extract the band data, and construct the eight images. It will show an
image building up every 50 pixels. In version 1.0 of the software the contrast is set
to be whatever the first spectral image was displayed with. However, when the
process finishes, you can look at several of the bands using the menu items under
the DISPLAY menu. The intermediated image that builds up on the screen is band

5. To save the resulting images select SAVE HYPERSPECTRAL COMPONENTS under
the FILE I/O menu. The files will be saved in the image directory where the
spectral files were read. The resulting image from my session is shown below in
Figure Twelve.

Figure Twelve: Looking Across a canyon toward Santa Barbara

This image illustrates a problem you might have if you have a slow turntable like I
did. It took 40 minutes to collect this data and during that time the thin clouds

13

near the sun shifted around, causing the slightly tilted dark bands in the
foreground. The total time spent in data collection here was about 45 minutes.
The ST-i can collect image data (for short exposures) at about 3 frames a second,
so with a more optimal turntable the collection could be done in as little as 4
minutes and reduce the likelihood of this annoying problem.

One can also use CCDOPS to process the final data in interesting ways. It can
create a color image by using the RGB Combine menu item under the UTILITY
menu. It also is interesting to flat field one image with another, which yields an
image that is a ratio of two bands.

Remember that curved entrance slit on the spectrometer? It causes the data to
be sampled along an arc instead of with rectilinear coordinates. We intend to add
a resampling routine in a later version to straighten these out (updates will be
available for free from the SBIG web site), but note that it is present in both of the
images in this instruction manual, so it does not prevent experimentation. Also,
the dark band at the bottom of Figures Five and Twelve is due to the light from
the lens used at F/16 not making it through the entire system. C-mount lenses
will vary from one another. At F/5.6 the problem is reduced. It represents a
vignetting effect that we also hope to provide a means to correct in a later
update. Flat fielding the band images (using CCDOPS) with a band image of a
uniform white scene like a white card or integrating sphere can correct the effect.

The fields that refer to red, green and blue colors on the program’s startup
screen, and their associated slider controls, are for a future version and should
not be used yet in version 1.0.

The HyperSpectral program is a work in progress, so please alert us to any major
bugs that are found. Also, if you find a particularly interesting result, please send
us some images so we can see what you are doing. This is a new arena for
amateur imaging, so many uses might exist of which we are not aware.

14

Appendix B: Calibration Data

Figure Thirteen: Nominal Wavelength as a function of Pixel Location

10000

9000

8000

7000

6000

Wavelength in Angstroms

5000

4000

0 50 100 150 200 250 300 350 400 450 500 550 600 650

Pixel Location

Figure Fourteen: Dispersion (Angstroms per Pixel) as a function of Wavelength

40

35

30

25

20

15

Angstroms per Pixel

10

5

0

4000 5000 6000 7000 8000 9000 10000

Wavelength in Angstroms

15