What follows is loosely based on a few pages which float
around with the manuals sometimes. I do not know who wrote the
original. If the project you have in mind requires PRECISE values
for any of the quantities listed here, plan on measuring them
yourself.
In future of this page, I would like to include a sketch of
the Mark III, pointing out what knobs to turn, and I would like to
know more about what the grisms do, and what order-blocking filters
are available. Most importantly, if you have a question, some
information, or a suggestion, please communicate with me.
-Guy Worthey worthey@astro.lsa.umich.edu
Mark III Spectrographs
The first Mark III spectrograph was designed and constructed
in 1982. The image forming optical parts consist of a 381 mm focal
length lens for a collimator and either a Canon FD 135 mm f/2 or a
Conon FD 200 mm f/2.8 for a camera lens. In addition there is a field
lens at the slit. It is a plano-convex lens with a focal length of 500
mm and a diameter of 45 mm. A grism is used for the dispersing
element.
A second, twin, spectrograph, the Mark IIIB, was built in 1986
with a removable slit assembly to permit dual use as a
spectrograph/imager. An alternative field lens is used for imaging to
reduce vignetting at the edges of the chip. (The point of this sort of
thing is to look at a wider field of view - something which is
becoming less of an issue now that our CCDs are larger. I know of
nobody who has actually used this option in the recent past. Contact
Gary Wegner of Dartmouth.)
Limitations:
1. The off-the-shelf optical elements are not very good for blue
spectroscopy. Throughput begins to fade below 4400 A. On the bright
side, the spectrograph does a good job of staying in focus down to
perhaps 3900 A, at least with the 600/4600 grism.
2. When used as an imaging camera, the Mark IIIB has internal
reflections that can seriously disrupt photometry in the vicinity
of bright stars. At the 1.3m telescope, the Mark IIIB produces very
undersampled images.
Grisms:
Table of ballpark dispersions in Angstroms per pixel
Grism:
lines per mm / Wilbur Charlotte/Templeton Nellie
blaze wavelength 2048 pixels 1024 pixels 2048 pixels
150/7300 ? ? ?
300/5400 3.3 5.4 4.7
300/6400 ? ? ?
300/8000 3.0 5.0 4.4
600/4600 1.4 2.3 2.0
600/5800 1.4 2.3 2.0
Slit Widths (for 135 lens)
SLIT Actual/Projected 2.4m projected 1.3m projected
POSITION Slit Width
1 0.071/0.025 mm .824 " 1.52 "
2 0.107/0.036 mm 1.17 " 2.17 "
3 0.145/0.051 mm 1.68 " 3.12 "
4 0.203/0.071 mm 2.36 " 4.36 "
5 0.287/0.102 mm 3.34 " 6.17 "
6 0.406/0.144 mm 4.72 " 8.73 "
7 0.574/0.203 mm 6.67 " 12.34 "
8 0.812/0.287 mm 9.44 " 17.46 "
9 1.15 /0.407 mm 13.4 " 24.7 "
10 +see note 0.076/0.027 mm .884 " 1.63 "
* The length of the slit is 27.94 mm, which corresponds to 5.4 arcmin
on the 2.4m or 9.7 arcmin on the 1.3m.
* Multiply numbers by 1.48 if you are using the 200 mm lens.
+ position number 10 is a series of square holes instead of a continuous
slit. This position is EXTREMELY useful when calibrating long-slit
spectroscopy, since it gives the entire spatial mapping in one exposure.
The camera lenses have magnifications of 0.354 and 0.525 for the 135mm
and 200mm lenses, respectively. The linear dispersion with the 135
mm camera is 100 A/mm.
Spatial Scale:
1. Assuming the numbers in the above table are correct, the spatial
scale at the 2.4m/f7.5 is 32.84 arcsec per mm of detector area. At the
1.3m/f7.6, it is 60.69 arcsec/mm. (135mm camera)
2. On the other hand, if you use the magnification for the 135mm
camera (0.354) and the numbers for spatial scale in the ccd page, you
get 32.56 arcsec/mm for the 2.4m and 58.53 arcsec/mm for the 1.3m.
Adopting the latter, and scale factors from the ccd page, we get the
following spatial scales in arcsec/pixel:
2.4m: f7.5 1.3m: f7.6
Charlotte 0.781 1.41
Wilbur 0.488 0.878
Nellie 0.684 1.23
Projected length of entire 27.94mm slit, in pixels:
2.4m: f7.5 1.3m: f7.6
5.37 arcmin 9.65 arcmin
Charlotte 412 pix
Wilbur 660 pix
Nellie 471 pix
I measured 416 pixels for the slit projected on Charlotte (from a flat
lamp exposure; I took the place where the intensity dropped to 0.5 as
the "edge" of the slit.) In another measurement, I ran a star back and
forth along the slit, took a focus frame, and recorded the declination
from the telescope monitor. This gave a scale of 0.77 arcsec/pixel for
Charlotte on the 2.4m. These numbers imply a slit length of 5.3 arcmin
with an error of around 1% to 1.5%.
User notes:
In case you ever have occasion to wonder, NORTH is toward HIGHER
column numbers, the way things are set up now & with a zero rotator
angle.
At the 1.3m, strung cables limit the free motion of the instrument
rotator. It can go all the way to +90 degrees on the positive side,
but only goes to about -28 degrees the other way. This will hopefully
change in the future (Bob is working on it.)
Each grism setting requires adjustments for the spectrograph focus,
tilt, and translation. The translation determines the range of
wavelengths projected on the CCD. The focus setting "focuses" the
dispersion at the center of the chip, while the tilt angle balances
the amount of defocusing for wavelengths at the edges of the chip. For
the 300-line grisms, the sharpness of a lamp line (at the smallest
slit width) varies by a ratio of 1.4 edge/center (this is an old
number from the old, smaller chips - has anybody measured this
lately?). The 600-line grisms are roughly a factor of 2 better
behaved: I get variations of only 25% across the 2400 Angstroms
covered by Charlotte, and my setup is probably not optimal.
Bob Barr customarily provides the initial setup.
If you decide to change grisms yourself, remember to tighten up the
thumbwheel clamps before undertaking a series of focus tests, since
the tightening itself causes a very significant change in the
positions of the grism and the CCD.
The Mark III spectrographs have a slot for filters (76.2 mm square)
above the grism slot. Thickness may be as large as 7 mm. For red
spectra, an order-separation filter is necessary so that, e.g. 4500-A
light does not show up on your 9000-A spectrum. For bluer spectra,
such a filter is not necessary since virtually no light bluer than
3500 A will get through the spectrograph lenses.
Some more apocryphal stuff regarding imaging with the Mark IIIB: The
filter slot just described is useful for imaging with the Mark
IIIB. Note, however, that the observatory does not have a set of 3x3
inch standard filters for this purpose. The 2x2 inch filters that
could be positioned in the MIS filter wheel will not permit imaging
without vignetting.
As customary for spectral observations, the guiding TV camera is
carefully focused at the slit prior to observations by using the MIS
flat lamp and pulling out a special mirror ("slit viewing optics" in
the 2.4m manual) attached to an adjustment rod above the spectrograph
in the MIS box. The mirror reflects an image of the slit (itself
mirrored so that the sky image reflects, except for what passes down
the jaws of the slit) into the center of the guider TV when the guider
probe is set to the "slit" position.
worthey@astro.lsa.umich.edu
last updated 27 December 1995