Project Cyclops, A Design... - Department of Earth and Planetary ...

sec,therebyobtaining100samplesof thecomplete
powerspectrumwith0.1-Hzresolution. These are then
added in perhaps 400 different ways to synchronize with
the assumed drift rate, but since additions with almost
the same slope are correlated, there are only about 100
independent signals obtained from the adder, each
having 2X109 independent Nyquist intervals per polarization.
Thus, a total of 4×10 _ tests is made per
observation, and to keep the probability of a false alarm
to 20% for the entire observation, the probability of a
false alarm per test must be about 5X 10-_ 3
With the threshold set so that the probability of the
noise alone exceeding it in any one Nyquist interval is
5× 10-l 3, the probability of missing the signal is 50% if
the received signal-to-noise ratio is unity (0 dB) in the
0.1 Hz band or -93 dB for the 200-MHz total band.
l'or only a 1% chance of missing the signal an additional
2 dB of signal power is needed. We know of no other
method of processing the signal that can even approach
this performance. The technique is believed to be an
optimum one, though of course other and better means
of implementing the operations may be found.
IMAGING THE RADIO SKY
The proposed Cyclops array is orders of magnitude
more powerful than any existing fully steerable radio
telescope. As such it would be a magnificent tool for
finding and studying distant weak radio sources and for
mapping the structure of known sources. Because all the
IF signals are returned independently to the central
headquarters, they can be combined to form more than
one simultaneous beam. In fact, with n elements we can
form n independent beams, or more than n correlated
beams. By arranging the beams in a closely packed array
and portraying the sky brightness measured by each
beam as the brightness of a corresponding point on a
screen, we can form a real-time high-resolution map of a
portion of sky within the beamwidth of the antenna
elements. In this way the time required to map the sky
or to search for new sources is only one nth as great as
with a single beam.
Several ways of doing this imaging were studied, none
of which is ideal in all respects. The proposed method
involves reradiating the signals from the antenna elements
as electromagnetic waves from antennas in a
scaled down model of the receiving array. If the scale
factor is o, the angular magnification of the telescope is
l/o. The imaging, if broadband, must be done at the
original received frequency; otherwise, the angular magnification
is frequency dependent and changes appreciably
across the received band.
72
The model, or signal, array focuses the radiation on a
second array whose receiving antennas pick up samples
of the signal in the image plane. The energy collected by
each receiver is then detected and converted to a
proportional brightness of an element of a display
screen.
The useful angular field of the image is inversely
proportional to frequency. Thus, to keep the number of
points in the image constant, the separation between the
two arrays must be proportional to the RF center
frequency. In the proposed system, the signal and image
arrays are 20-m in diameter and their separation varies
from 40 to 120 m over the tuning range from ! to 3
GHz. lm_,ging is not attempted from 0.5 to 1 GHz.
The size of the proposed imager is awkwardly large.
To prevent coupling to the receiving array it must be
housed in a completely shielded space, which, in turn,
must be lined with microwave-absorbing material to
prevent undesired reflections. The dimensions of the
anechoic chamber needed are about 75 ft wide, 75 ft
high, and 550 ft long, a building roughly the size of the
turbine house
of a large power plant.
The building volume needed is directly proportional
to the number of antennas in the array, to the ratio of
the high to the low frequency limit of operation, and to
the cube of the wavelength at the low frequency limit. If
we were content to do the imaging over the band from
1400 to 1700 MHz, which includes the H and OH lines,
a chamber 60 by 60 by 125 ft would suffice. However,
even the cost of the proposed imager is small compared
the total system cost, so we have proposed the large
version, leaving possible compromises to the judgment of
future study teams.
The shielding requirements, though severe, involve
only a few mils of copper. Most of the problems are
associated with bringing leads into and out of the
chamber. All shielding problems are greatly reduced if
the imaging frequency is offset from the receiver band
by the IF bandwidth. This introduces an appreciable, but
tolerable, amount of lateral chromatic aberration.
The proposed imager can image both polarizations
simultaneously. By combining the intensities, we can cut
the integration times in half. If the selected polarizations
can be varied, then subtraction of the intensities would
reveal polarized sources. This might be a powerful search
tool for unusual types of emissions in the universe. In
any event, the imaging capability would reduce by about
three orders of magnitude the time needed to make
certain kinds of radio astronomy studies, and, for the
first time, would place radio astronomy on a more
nearly equal footing with optical astronomy so far as
data collection rates are concerned. Who can say what
major discoveries might ensue.'?