Previous Article
Next Article
Table of Contents

Risley Prisms at the 4-m: Better Than Nothing (1Dec92)
(from KPNO, NOAO Newsletter No. 32, 1 December 1992)

Although we all learned about differential atmospheric refraction back
in Astronomy 101, the current generation of spectroscopists owes a debt
to Filippenko (1982, PASP, 94, 715) for reminding us exactly how severe
the implications can be for spectrophotometry. With the proliferation
of CCDs on Cassegrain spectrographs, it is not uncommon for an
astronomer to believe she/he is doing spectrophotometry with a 1.5-2.0
arcsec slit, when in fact much of the blue light is falling outside the
slit. (Even at a "modest" airmass of 1.5, the light at 4000 A will be
displaced by 0.7 arcsec relative to the light at 5000 A.) Mark Phillips
recently gave an excellent illustration of this in the December 1991
NOAO Newsletter No. 28, p. 17.

If the spectrograph slit is aligned with the direction of refraction
(i.e., the line from the zenith to the horizon), then there is little
wavelength-dependent light loss. However, aligning the slit with this
"parallactic angle" can be a mild pain with most telescopes. At the
2.1-m, for instance, the telescope must be moved to the zenith and the
spectrograph rotator adjusted by hand from the platform. Although the
rotator at the 4-m can be controlled remotely from the control room,
the telescope must be first slewed near the zenith in order to protect
the rotator's bearings from too much strain. An alternative to rotating
the slit is to use optics to compensate for refraction. These
"atmospheric dispersion correctors" (ADCs) are an intrinsic part of the
design of most modern telescopes, due in large part to the desire to
use multi-object fiber feeds, for which alignment with the parallactic
angle is meaningless since fibers ends tend to be round.

Many users of the 4-m telescope may be surprised to learn that in fact
there are a pair of prisms available designed to act as ADCs. These
"Risley prisms" can be inserted into the beam remotely using software
commands, and are automatically rotated to compensate for atmospheric
dispersion at a given telescope position, in principle. (Their field
size is too small to be used with Hydra, however.) During two recent
runs with the R-C Spectrograph, two staff observing teams opted to use
the Risleys.

Our experiments on standard stars suggest that the Risleys do help, but
they do not do the whole job. The plot below shows the observed flux of
the standard star Feige 110, calibrated by observations of a standard
star near the zenith, observed both with and without the Risleys. The
observed flux of Feige 110, obtained at an airmass of 2.2, is a
reasonable match to that expected when the slit is rotated to the
parallactic angle. (The two spectra at the top show the observations at
the parallactic angle both with and without the Risleys; the small
differences between these are due to small differences in the fit of
the sensitivities curves.)

		        [figure not included]

The bottom-most curve shows what would happen if we attempted to
determine the flux of Feige 110 with the slit rotated 90 degrees to the
parallactic angle. This is about the worst case imaginable, but would
actually be what you are doing if you attempted to observe an object at
a declination of -30 degrees on the meridian with the slit oriented
east-west. (Our slit width was 1.8 arcsec for this experiment.)
Filippenko's Table I reminds us that the light at 3500 A was displaced
by 2.1 arcsec relative to that at 5000 A; we were lucky the seeing was
poor, or we might not have gotten any UV light into the slit at all!

The middle spectrum shows how effective the Risleys were in this
"horrible-case scenario." Clearly they help, and yet, it is also clear
that they do not quite do the whole job, either. Note that the glass
cannot match the atmospheric dispersion at all wavelengths, and so the
prisms represent a compromise by the designer over some spectral
range.  Furthermore, the prisms introduce a small (~ 7%) transmission
loss above 4000 A which worsens to a 30% loss by 3700 A.

We are planning to conduct further tests this spring, and substantiate
that the Risleys are being aligned correctly. As we approach the era of
the ADCs, it behooves us to understand the ones we already have. In the
meantime, we recommend rotating to the parallactic angle to really "get
it right."

Phil Massey, Todd Boroson,
George Jacoby, Richard Green

Previous Article
Next Article
Table of Contents