In article <4004ec1b$1@news.barak.net.il>,
Gil Teva <gilteva@actcom.net.il> wrote:
It is a fact that for many years billions of dollars have been spent to try
to locate other civilizations, but without a success.
I think this is untrue. SETI is done on a shoe string. I'd be surprised
if more than US$ 50,000,000 had been spent on this (if you ignore the
opportunity cost of machines running S@H). It might be as little as
US$ 20,000,000.
1. The antenna is very much directional and receives or transmits
signal only to a very narrow angle in the sky. A 70 db antenna ha an angle
of only 1 part of 10 million of the sky. To detect signal from other
That's the penalty for sensitivity.
2. A large dish antenna used today transmits in a narrow bandwidth of
64 KHz using a carrier frequency of 8 GHz. The ratio between the two
Not true. Dishes have very wide bandwidths. The bandwidth is only
limited by the feed (which can be a dipole). The dish doesn't matter
provided it is at least a few wavelengths across and the wavelength is
more than about 10 times any surface profile errors.
frequencies is about 1:100,000. If an antenna from another civilization will
A DBS dish at 12GHz is going to have a bandwidth of around 10MHz per
channel and cover many channels. I don't have DBS, but I'd assume at
least 10 channels, so at least 100Mhz at 12GHz, which is 1:120. Arecibo
operates from 430MHz or lower to something around 4GHz; that seems more
like almost 2:1 on the centre frequency! S@H is doing 2.5MHz or 1:568,
but is actually taking part of the 100MHz bandwidth from SERENDIP (1:14)
and I think that is on one feed. The average Newtonian telescope also
has 1:1 or better coverage, and is also a dish antenna.
The antenna I thought of is a simple dipole antenna in space. This antenna
Anything in space is very expensive. The whole annual budget for
radio SETI is probably only about US$ 1 miillion. I think the annual
budget for the SETI Institute is about that, but they also do astro-
biology projects, and theoretical work.
transmits and receives to all directions and is not limited to a very narrow
This one is not new, but it last came up a few years ago.
Dipoles tend to have doughnut patterns. Isotropic patterns are physically
impossible. In particular, dipoles do not transmit or recieve in the
length direction of the wire.
frequency.
A resonant dipole has a bandwidth which depends on the wire diameter.
Practical wire diameters would result in extremely narrow band
antennas! Operating non-resonant would result in heavily losses in
the matching network. However, these would equally affect signal and
sky noise so might not matter, as this is not the real problem.
In order to make it very sensitive to signals I thought to make it as long
as possible.
Long antennas, especially when made with practical quantities of conductor,
have high resistive losses, which both remove wanted signal and tends
to introduce noise at an equivalent temperature to the wire (not that
thermal noise matters compared with other sources).
Antennas over about half a wavelength get their added sensitivity
by being directional, a characteristic you considered undesirable.
At frequency of 50 KHz and 1 AU from the sun the noise is not significant.
At 2 AU the frequency is 25 KHz and at 10 AU the frequency we can use is 5
KHz.
That's not your noise problem; the problem is galactic synchrotron noise,
which will, I think, be extremely strong at such frequencies (unless they
are two low). It is about equivalent to 10K temperature at 1.4GHz, and
goes up as the inverse 2.2 (approx) th power of frequency.
Your other problem is that the galaxy is
opaque at low frequencies. In fact, if I interpret
<http://www.gao.spb.ru/english/publ-s/viii-rfs/p099.ps.gz> correctly,
the opacity sets in at around 1MHz. (This also indicates that useful
measurements are being made by satellites at 250kHz, which indicates
that sky noise is dominant, even for simple antennas, at this frequency.)
Finally, it is almost impossible to work out the direction of the source,
and it is not practicable to achieve significant transmit antenna gains.
The effective area of the antenna will be 9 square kilometers. This is 100
times more than the largest antenna array on earth.
But only 9 times larger than planned arrays, which would have signal
to noise ratios many of times better. Incidentally, the
figure I've seen for the capture area of a half wave dipole is about
half a wave by a quarter of a wave, which would give 4.5, not 9.
Even using your figure of 100, which is a 20dB improvement, one
has to offset this against the reduction from over 70dB [1] of gain for
Arecibo to only about 2dB for a dipole transmitter, assuming optimal
orientation of both transmitter and reciver. Arecibo has a feed point
power of 1MW when transmitting. Overall, that is about a 50dB penalty.
With your assumptions, you can get back about 10db for being able to
coherently integrate for about 1,000 seconds rather than about 10,
which still leaves you about 40dB short. This is before accounting
for attenuation by the interstellar medium and natural noise sources.
40dB is a 100 penalty on range and a 1,000,000 penalty on volume searched.
(This is one of the reasons that optical SETI is attractive - you can
use reasonable powers with quite small transmit and receive apertures).
In order to detect a signal from outer intelligence, there is a need to
detect not more than 100 bits of information. The time for this can be for
Not sure where this figure came from.
example 1 day or 100,000 second. The bit-rate will be about 0.001 bits per
second. The noise level of an amplifier for that antenna is e-21 watt per
Hz. The noise level for that need is e-24 watt.
Pre-amp noise is totally irrelevant at such frequencies as the sky noise
will completely dominate. In fact, with modern amplifiers, most of
VHF and everything from shortwave down is dominated by sky noise. (It
would have made these figures easier to check if you had stated the
noise temperature you are assuming for the reader, and shown the derivation
of formulae.)
[ ASCII art can only safely assume monospaced fonts - fixed ]
Distance^2 Wavelength^2 36
----------------- = ----------------------- = ------ =~ e26
P transmitter P receiver e-24
I'm not sure about this formula, but I suspect that you are using a
detection threshold of signal = mean noise power. That will give a
very high bit error rate. If you use the easier to calculate case
of slicing at 1.5 times mean noise power, 22% of zeroes will read as
ones. Getting the number of ones that read as zeroes is more difficult,
which is why it is non-trivial to work out the optimum slicing level.
You also have to consider that you must search many frequencies
(and to a lesser extent baud rates) which will increase your false
positive rate. There is also likely to be quite a lot of
human noise at these frequencies, although maybe the frequency is
low enough not to get through the ionosphere.