Copyright © 1997 by Dr. H. Paul Shuch, N6TXIntroduction to Amateur SETI
Executive Director, The SETI League, Inc.
PO Box 555, Little Ferry NJ 07643
email n6tx @ setileague.org
Presented at the 1997 Central States VHF Conference, Hot Springs AR, 26 July 1997
The recently privatized Search for Extra-Terrestrial Intelligence (SETI), with its strong amateur component, has resulted in the deployment of dozens (soon hundreds, and eventually thousands) of small amateur radio telescopes worldwide.  Generally operating in the 1.3 to 1.7 GHz range, these receiving systems have been loosely based upon the best amateur practice for 1296 MHz EME ("moonbounce") communications. A typical example of such a system, built by SETI League member Daniel Fox, KF9ET, is seen in Figure 1 below.
The Spectral Dimension
Amateur SETI stations have shown themselves sufficiently sensitive to detect the strongest likely signals from the nearest stars.  However, a significant limitation of these amateur radio telescopes is their narrow instantaneous spectral coverage. With digital signal processing being done at audio frequencies in multi-media personal computers, the receivers employed are restricted to bandwidths of not more than tens of kiloHertz (a 12.5 kHz instantaneous bandwidth being typical). But the frequency region of greatest interest (the so-called Water Hole) extends from 1420 to 1660 MHz, a total spectrum of 240 MHz. Scanning each 12.5 kHz channel for just 24 hours in meridian-transit (drift-scan) mode, one station could cover the entire Water Hole at a single declination in just under 53 years. I submit that it will prove difficult to recruit volunteers for a sustained search of such scope.
The so-called "professional" SETI programs (of which the now defunct NASA SETI office is a prime example) devoted much of their resources to the development of Multi-Channel Spectrum Analyzers (MCSAs) allowing the simultaneous scanning of tens of MHz of spectrum, to resolutions typically in the tens of Hertz. Simple dimensional analysis reveals that such receivers must typically operate on perhaps a million simultaneous channels.
Outstanding MCSAs have now been deployed by the University of California, Berkeley (Project SERENDIP), Harvard University (Projects BETA and META), Argentine Institute of Radio Astronomy (META II), Ohio State University (Big Ear), the SETI Institute (Project Phoenix), and others. Yet such broad spectral performance has been, and probably remains, beyond the amateur state of the art. The present project contemplates changing all that.
The META Model
The SETI MCSA which is probably best known to the ham radio community is the Mega-Channel Extraterrestrial Assay. META was developed by Harvard University's Prof. Paul Horowitz, W1HFA, and went on the air in 1985. Although hardly an amateur effort (the development of the META-1 receiver was funded by a $100,000 grant to the Planetary Society from filmmaker Steven Speilberg), Dr. Horowitz' amateur radio background makes this an attractive benchmark for future ham efforts. The receiver, occupying several six-foot high equipment racks, scans over 8 million simultaneous channels, each 50 milliHertz wide, for a total instantaneous bandwidth of 400 kHz. Paired with its twin in Argentina, Project META conducted one of the most exhaustive searches ever for microwave signals of possible intelligent extra-terrestrial origin. 
Just ten years after the first META receiver went on its air, its technological grandchild, BETA, was activated. BETA's 250 million channel MCSA operates at 0.5 Hz resolution. Essentially a supercomputer running at 40,000 MIPS, the ambitiously named Billion Channel Extraterrestrial Assay scans the entire Water Hole in eight bands, each 40 MHz wide, at a rate of two seconds per band. Although BETA represents Paul Horowitz' greatest technological triumph to date, we expect even better performance from him in the years ahead, as computer technology continues to advance.
STAR-1, the SETI League's first Spectral and Temporal Analysis Receiver (see Figure 2 below) is a work in progress. Far from completion and not really intended for duplication when finally operational, it is a test-bed for developing low-cost MCSA technology for use by the world's radio amateurs and amateur radio astronomers.
STAR-1's frequency coverage, and much of its basic design, were dictated by an accident of electromagnetic interference (EMI). It has been noted that the prime radio astronomy spectrum extends between roughly 1420 MHz (the precession frequency of neutral hydrogen atoms in space) and 1660 MHz (one of the resonances of interstellar hydroxyl ions). Smack in the middle of this prime real estate, at 1575 MHz, is an angry swarm of navigation satellites. The US's Global Positioning System (GPS) and its Russian counterpart, GLONAS, make the center of the Water Hole all but unusable for terrestrial SETI.  (See Figure 3, below).
In view of the above interference, it was felt reasonable to develop a receiver with two spectral response bands, one in the vicinity of the hydrogen line, and the other encompassing the hydroxyl line, with a gap in between. Since the two frequencies are 240 MHz apart, they can be images of one another if heterodyne downconverted to a common 120 MHz IF. The required local oscillator frequency would then fall halfway between the two signal bands, or roughly in the already spectrally polluted GPS band!
An added bonus of excluding the middle of the Water Hole from analysis is that we can now develop a precise frequency reference, based upon reception of GPS navigation signals. This possibility is explored in the Section which follows.
As seen in the block diagram, STAR-1's second conversion is done directly to baseband. By utilizing image-recovery technology, it is possible to extract two 6 MHz wide outputs for each of the two input channels, for a total of 24 MHz real-time frequency coverage. Signal analysis is then done in four advanced Digital Signal Processors, with ultimate resolution limited primarily by the performance of the DSP engines.
Tom Clark's Clock
If we can't receive SETI signals in the GPS band, is there a way to capitalize on the interference? It turns out there is: we can use GPS as our primary time and frequency reference.
There are number of interesting things which a SETI enthusiast can do with a high precision time and frequency standard. We can use them to calibrate or control our receiver frequency coverage, determine accurate local sidereal time (and hence right ascension of our drift-scan radiotelescopes), and to provide a common time reference for correlating long-baseline interferometry observations. In the STAR-1 MCSA, such a reference would facilitate the design of frequency synthesized local oscillators. Unfortunately, the technologies used in the past for accomplishing these goals, Cesium or Rubidium frequency standards and Hydrogen masers, are priced beyond the reach of most experimenters. Thanks to a ham (of course!) that is no longer the case.
Dr. Thomas A. Clark (W3IWI), the NASA scientist whose name is virtually synonymous with very long baseline interferometry, has been grappling with the time and frequency standard problem for his geodesy experiments. By deriving timing information from the atomic clock-controlled Global Positioning Satellites, he is able to produce a relatively low cost time and frequency standard with long term accuracy rivaling a Hydrogen maser.
Tom Clark's Totally Accurate Clock, or TAC (the acronym happens to be his initials) produces a 1 PPS output exhibiting 30 - 50 nSec precision and 30 nSec rms accuracy, even with Selective Availability implemented in the GPS system. At 1-day averaging, the unit produces a standard frequency accurate to one part in ten to the twelfth. This is equivalent to Cesium or Rubidium standards, at a cost of about $800. Better still, Tom has made hardware and software details available in the public domain. And as a special bonus, Tom has written SHOWTIME.EXE, a nifty program for displaying GMT time, sidereal time, and user location on a PC screen. The program, which is driven from Tom's GPS-based clock, is especially useful for radio astronomy applications, and is being distributed free of charge.
Those SETI enthusiasts interested in assembling a Totally Accurate Clock can obtain full hardware and software details via anonymous FTP from the following directory: ftp://aleph.gsfc.nasa.gov/GPS/totally.accurate.clock/. Note that GPS is upper case. Also check the Tucson Amateur Packet Radio (TAPR) web site at http://www.tapr.org for news concerning the availability of the TAC-2 in kit form. As seen in Figure 2, the first and second local oscillator frequencies for STAR-1 are synthesized in a phase-locked loop derived from a precision 10 MHz temperature compensated crystal oscillator (TCXO). The multiplication chain for the first LO is derived from an existing no-tune transverter board from Down East Microwave.
Since all LOs derive from the same TCXO, accuracy and frequency stability are limited by the performance of a single stage. W3IWI is currently working on an add-on for the TAC-2 to facilitate synchronization of just such a TCXO to GPS time. While we await availability of that product (probably through TAPR), the STAR-1 prototype is employing a surplus PTS commercial frequency synthesizer for its local oscillators.
The Front Ends
The low noise amplifier (LNA) shown in Figure 2 is a room-temperature pseudomorphic high electron mobility transistor (PHEMT) stage. System design objectives call for a 50 Kelvin noise temperature, at 20 dB gain, with the LNA mounted directly at the antenna feed. It remains to be seen whether we can achieve the desired bandwidth from a single LNA (or for that matter, from a single antenna feed). It is expected that the final configuration will employ separate feedhorn probes, and separate LNAs, for the hydrogen and hydroxyl bands.
The amplifier and bandpass filter stages which follow the LNA were built by modifying two existing Down East Microwave 1296 MHz no-tune boards (see Figure 4). It was necessary to raise the resonant frequency of the hairpin bandpass filters by trimming 0.1 inch and 0.3 inch lengths, respectively, off each end of each filter pole, for the hydrogen line and hydroxyl line frequencies. This modification was done with an approxo knife (there's nothing exact about it!)
The resulting gain curves, using three stages of GaAs MMIC amplification between and behind filter poles, are seen in Figures 5 and 6. A Mini-Circuits mixer and multiple-MMIC second IF amplifier are tacked on to the outputs of the two front end strips, with the bulk of the system gain designed into these VHF stages. Since radio telescopes operate best without automatic gain control (AGC), a step attenuator has been designed into each IF strip to optimize system gain.
The Back End
So far, there's nothing particularly exotic about the receiver we've described. The RF portion of STAR-1 represents a logical application of current amateur practice to a specific set of design objectives. What makes or breaks any MCSA design is its digital signal processing (DSP) hardware and software. As it happens, it is in the DSP area that the design of STAR-1 is most tentative. Did I remember to say that this is a work in progress? In that spirit, much of the DSP area is also vapor-ware at this writing.
It happens that our IF, though 72 MHz wide, is centered roughly on the 2 meter band. It seemed logical, then, to process the VHF IF down to baseband in an image-reject (or phasing type) direct conversion receiver, similar to that which Rick Campbell, KK7B, has described for 2 meter use.  However, whereas Rick did his image-reject conversion in two Mini-Circuits mixers, a Toko power splitter, and discrete LO phase-shift networks, we were fortunate to find all these components in a single package. Shown as a mixer with two outputs in the block diagram (Figure 2), that package is a commercial quadrature (or I-Q) demodulator.
The detector selected, Mini-Circuits MIQC-176D-111, sells for $55 in single quantities. Though a moderately costly component, when presented with IF signals and ample LO injection the I-Q demodulator selected produces two broadband (DC to 6 MHz) outputs exactly in phase quadrature, with a maximum phase imbalance of plus and minus 3 degrees, and amplitude balance within 0.4 dB.
Since we desire to analyze not kHz, but rather of MHz of spectrum at a time, we need to detect the VHF IF not down to audio, but rather to baseband signals which more closely resemble video. What better device to follow the IQ demodulator than a standard video amplifier module, with its DC to 6 MHz bandwidth? Both the I and Q channels are amplified, then digitized for input to a dedicated DSP chip. Note that since we are simultaneously processing both the hydrogen line and the hydroxyl line band, a total of four 12-bit A/D converters, each sampling at 12 MSPS, as well as four separate DSP engines, will be required. An ordinary personal computer is used not for signal analysis, but rather to control the four DSP engines, and for data storage and display.
High performance commercial DSP chips are evolving at a dizzying rate. This is a far cry from the days of NASA SETI, when custom DSP engines had to be designed and manufactured (at taxpayer expense). When this project was started, we contemplated employing perhaps four Graychip EV-4014 evaluation boards, which were priced at $6750 each. Fortunately, before we had set soldering iron to circuit board, Texas Instruments announced their new TMS320C6x, a 1600 MIPS dedicated DSP chip which was said to be able to perform a 1024 point Fast Fourier Transform (FFT) in 70 microseconds, and run at speeds up to 200 MHz. Price was announced at $96 in 25,000 quantity, with a developer's kit available for just (just?) $2995. Since new market entries occur almost weekly, we're still not sure what DSP chip we'll ultimately use. But at this writing (June 1997), the 'C6x (see http://www.ti.com/sc/C6x) is the odds-on favorite.
Note that if we analyze 6 MHz of baseband with a 1024 point FFT, our bin resolution will be on the order of 6 kHz. To achieve our goal of 10 Hz resolution would seem to require either analyzing only one thousandth of the baseband spectrum at a time, or running 1,000 parallel DSP engines, in each of the four output channels. At this point, some combination of the two approaches is contemplated. We feel this problem will solve itself as even more powerful DSP chips become available, and the cost of the existing ones continues to decrease.
In KK7B's direct conversion receiver designs, much attention was paid to developing phase-stable 90 degree audio phase shift networks to allow the upper or lower sidebands to be extracted from a pair of mixers fed in phase quadrature. You'll notice no such phase shift networks following the video amplifiers in Figure 2. This is because I could come up with no reliable way to achieve exactly 90 degrees of phase shift over a DC to 6 MHz bandwidth. A technique widely used in monopulse radar systems is to digitize the two phase-quadrature components, and then apply a Hilbert transform  to them in software.
Since we already intended to digitize our I and Q components, SETI League president Richard Factor (WA2IKL) recommended we employ the Hilbert transform approach in STAR-1. The result is simultaneous USB and LSB reception, which allows us to cover downconverted hydrogen line and hydroxyl line components which extend from DC to both 6 MHz above and 6 MHz below the second LO frequency. When the second LO is programmed for 120 MHz, for example, the receiver is simultaneously processing 1414 to 1426 MHz (a 12 MHz slice of spectrum centered on the hydrogen line) and 1654 to 1666 MHz (a similar chunk centered on the hydroxyl line). By tuning the second LO in 10 MHz steps, this instantaneous 24 MHz of bandwidth allows us to tune a total of 144 MHz of prime SETI spectrum with slight overlap.
If the four square-law detectors shown in Figure 2 appear to have been added as an afterthought, it's because they were. Their purpose is to allow STAR-1 to function as a total-power receiver, for conventional broadband radio astronomy observations. (The required amplifiers and integrators may be accomplished in either analog circuitry or software, yet to be defined). For years, such projects as Suitcase SETI, Sentinel, and SERENDIP (operated by Harvard University and the University of California, Berkeley) have enabled SETI scientists to tap in to the outputs of the world's great research-grade radio telescopes, in order to perform what has become known as parasitic SETI. It seems only appropriate that a receiver specifically designed for advanced SETI should incorporate the capability of parasitic radio astronomy. Turnabout is indeed fair play.
Progress To Date
As of June 1997, the basic STAR-1 topology is relatively well defined. The front ends, IF strips, I-Q demodulator and video amplifier circuits have all been breadboarded and tested. The multiplier chain for the first local oscillator is operational, but since the 110 MHz GPS-locked PLL has not yet been developed, it is being driven by a 110 MHz crystal oscillator for test purposes. At present the second LO is a GPIB-programmable synthesized laboratory signal generator. And the A to D converters and DSP engines, the true heart of any MCSA, are still a distant dream. The SETI League welcomes the assistance of any and all interested radio amateurs. To participate in an email discussion group dealing with SETI system design, send an email to "Majordomo@sni.net" with the message "subscribe seti" in the body.
Conclusion: What's In A Name?
While preliminary development of the STAR-1 receiver was underway, the scientific community lost its most eloquent spokesman. The untimely death of Carl Sagan shook the SETI community especially, since he had been intimately involved in the Search for Extra-Terrestrial Intelligence since 1961. It is not generally known that it was Sagan who coined the name Mega-channel Extra-Terrestrial Assay, and the acronym META, for Prof. Paul Horowitz' landmark 1985 MCSA design. After consulting with Horowitz, with officials of the Planetary Society (of which Dr. Sagan was president), and especially with Sagan's widow Anne Druyan, The SETI League has decided to rename the STAR-1 receiver project "mini-META" in his honor, and to dedicate this effort to the memory of Carl Sagan. The name reflects NASA administrator Dan Goldin's watchword for future space missions in the present economic climate: "smaller, faster, cheaper." Amateurs may never best Horowitz' and Sagan's efforts. But two out of three ain't bad.
 Shuch, H. Paul, Project Argus and the challenge of real-time all-sky SETI. In Astronomical and Biochemical Origins and the Search for Life in the Universe, 693 -700, IAU, January 1997.
 Shuch, H. Paul, SETI sensitivity: calibrating on a Wow! signal. Proceedings of the AMSAT-NA Fourteenth Space Symposium: 130-141, American Radio Relay League, November 1996.
 Lemarchand, Guillermo, Project META reports 56 'alerts.' Bioastronomy News 6(4):1-2, Fall 1994.
 This is the reason that SETI scientists are seriously contemplating construction of a radio observatory on the far side of the moon, for deployment early in the next century. See http://www.setileague.org/press/pres9707.htm for further details.
 Campbell, Rick, High-performance, single-signal direct-conversion receivers. QST, January 1993, 32-40.
 Hahn, Stefan L., Hilbert Transforms, Chapter 7 in The Transforms and Applications Handbook, Alexander D. Poularikas, Editor. 1995, CRC Press, Boca Raton FL. ISBN 0- 8493-8342-0
entire website copyright © The SETI League, Inc.
this page last updated 28 December 2002
Top of Page