The limitations of the observatories of amateur radio astronomers are analyzed as well as the reasons for these limitations. A solution is proposed to enhance the capabilities of the observatory of a serious radio astronomer by the addition of one of the modern, state-of-the-art personal computers that have appear in the market at a very reasonable price. Some of the possible applications are analyzed to prove the point. None of these applications were possible before, with a computer under thousand dollars.
Building an amateur radio astronomical observatory is not a trivial project [1]. It requires skills, dedication and funds. In most cases, the achievable results are limited, not by the skill or abilities of the amateur, but by conditions outside his or her control. Some of these limitations can lead to frustration of an amateur that could otherwise produce valuable results. The purpose of this paper is to show that with an additional moderate investment in computer equipment, a serious amateur can produce valuable results with normal, off-the-shelf equipment. For this to be true, the amateur should select a computer that can offer some improvement in the capabilities of the station, otherwise the investment will only complicate the operation of the station with little gain in performance.
One of the most important problems of amateur radio astronomy is the lack of adequate equipment to make meaningful observations. The results presented in an accompanying article [1], make me very proud, but they are nothing compared with what is obtained in professional observatories. This is not true, for example, with visual astronomy where comets and novae are routinely discovered by amateurs rather than professional astronomers; the same is true with radio amateurs, who always perform tasks in the forefront of communications technology.
The possible results of any observation by an amateur radio astronomer are limited by physical laws that it seems as if they cannot be changed or circumvented. Because of these laws, the possible results of an observation are, in general, limited by the aperture of the antenna as a function of the wave length of the signals, its temperature, and the length of time of the observation. To obtain better results, it is necessary to increase the aperture of the antenna, reduce its temperature and/or increase the length of the observation.
Increasing the frequency of operation (to reduce the wave length) requires the use of more complicated techniques, more expensive equipment and, what is more important, requires more specialized knowledge. Losses in the cables and connectors increase very much with frequency and it is much more difficult to obtain the same gain as in a lower frequency. Doubling the frequency of operation, although it increases the relative aperture of the antenna by a factor of two, increases the cost of the equipment by a larger factor.
With very few exceptions, the aperture of the antenna becomes limited by two very important factors: space and cost. As the aperture of the antenna increases, the requirements of space increase proportionally, if not at a larger rate. The mechanical problems associated with the antenna increase at a much larger rate, easily with the cube of the diameter, and the cost goes up in proportion. Further more, large aperture antennas have a narrower beam, which is good, but they require a better and more accurate positioning system, and a more rigid structure.
A very much related problem is that, as the aperture of the antenna increases, the requirements on the receiver increase. A receiver with very low temperature is needed to use the possibilities of a large aperture antenna.
It seems as if the length of the observation is the only element under the control of the amateur, because the amateur has time to devote to the hobby, but this concept is very misleading. A long observation time following a source, requires a very precise driving mechanism that, for any antenna, is expensive and difficult to maintain in proper operation. A long observation time following the source also requires a very rigid structure and good adjustment. From this point of view, the amateur is mostly limited to transit observations.
So, it seems as all three basic parameters of the observation conspire to prevent the amateur radio astronomer from getting results.
During the last few years, and specially during the last one, a very important change has taken place, that if the amateur takes advantage of it, can increase the possibilities of performing valuable observations. This refers to the appearance in the market of powerful and economical personal computers.
The personal computer has been with us for some years but up to last year, they were small and limited units; the few with a certain power were too expensive. This is the important change that has taken place: last year, a personal computer was put in the market with a price below $800, with capabilities that were not available before for less than fifty thousand dollars.
Without entering into technical details, the Atari 520ST, or its big brother the 1040ST, have the best processor available today, the Motorola 68000. They have an architecture using all the state-of-the-art advances to produce fast and powerful computers. Both computers have a high resolution graphics display. They also have advanced software that makes them easy to work with. Powerful languages are available for these computers, from Forth to Pascal, Basic, Logo, Fortran, etc. The 520ST has one half of a megabyte of read/write memory and the 1040ST has one megabyte of read/write memory, which make possible large programs, or large data bases to be processed very fast. Both computers have their main programs in permanent memory, and most of the read/write memory is free for the applications. Both computers have many already installed ports which permit their use in radio astronomy.
The observatory described in a previous paper [1] is equipped with an Atari 520ST. Some of the possible uses of the computer will now be analyzed:
The computer has a parallel port intended to be used with a printer and a program port intended to be used with cartridges, among several other ports. Either one of these ports could be used for connecting an analog to digital converter to sample the signals coming from the receiver. In the observatory described, the cartridge port is used, to leave the printer connected all the time. This permits printing messages at any time. There are today very fast and inexpensive analog to digital converters [2] that permit sampling the signals at a rate well above the Nyquist limit in order to avoid problems with aliasing. The data collected in this form can be stored in memory to perform any analysis that will be desirable.
As an example, the way of operation of the observatory mentioned above could be as follows:
The transit of an object takes on the order of a quarter of an hour, so the system is set up to store values for half an hour and put the data in memory. The analog to digital converter gives 8 bits of resolution. Since the 520ST has more than 300,000 bytes of free memory after loading the programs, 180,000 of them will be used as storage buffer. This permits the computer to store 1000 values per second, or one every millisecond. The timing is done with the internal clock of the computer.
At the end of the process, the program stores the data on disk and returns control to the user. While the computer is sampling data points, it displays the collected values on real time, as a graph, to give feedback to the user. Since the screen has 400 horizontal lines of dots, the screen would fill in 4/10th of a second, which is too fast. The program displays only one out of every 256 samples, fills the display in 20 seconds, and then scrolls as any pen recorder. The horizontal resolution of the display is 620 points, but only 512 are used in this program.
More interesting than the raw collection of data, is the possibility of sampling the signals at a very high rate, say every few microseconds, accumulating the results in the digital equivalent of integration. This could be done with this computer since the instruction time is on the order of one microsecond, and only a few instructions are required. The 68000 computer works with 32 bits, consequently more than 16 million values can be added before getting into problems with overflow. If we work with 16 bits, 256 samples can be added together, without any possibility of overflow.
As an example of operation consider the following: The analog to digital converter gives only 8 bits of resolution, 256 samples from the analog to digital converter are accumulated by adding them together, forming a 16 bit number which is stored in memory at the end of the process. 36,000 of these values constitute a record, with the values stored at 20 values per second, or one every 50 milliseconds. This leaves time for the computer to do other tasks, like displaying the values or printing summaries. Another possibility would be collecting samples every 40 microseconds, or one average every 10 milliseconds, or 100 samples per second. In this way, noise can be removed and only the quasi-stationary characteristics of the signal are preserved.
With a computer, it is possible to do something that is difficult to do in the analog word; that is, we could maintain several integration periods simultaneously to permit more complete analyses of the signal.
Much more complex operations and manipulations of the data could be done with this computer, like filtering, averaging, smoothing, etc. since the time between samples is quite long compared with the capabilities of the computer.
With a computer of the capabilities of the Atari, it is possible to accumulate in a single record, the observations made of an object over a very long period of time. For example, the observations made of an object during transit every day over a period of several months, can be combined into a single record. The computer finds the proper way of correlating each individual record into the combined one. Several strategies can be tested and the results compared, until finding the best one. The combined record will be equivalent in resolution, to an observation made with a very large telescope. See Dr. Kraus' book [3] for a description of this procedure and the impressive results he obtained with the Andromeda Galaxy. In this case, the only requirement of mechanical rigidity is that the antenna is not moved too much by the wind so as to make the records unreliable. If this happens, such a record must be disregarded. Since the antenna is fixed in one position, it can be very tightly guyed and secured to assure stability.
Another interesting possibility for the use of a computer, as powerful as the one used in this observatory, is the correction of the observations for systematic errors. A systematic error is one that depends only on the system and that repeats itself every time an observation is made.
Consider the plot shown on Fig. No. 1. Imagine this is not a plot of Cygnus A as it is, but a record from a source that is strong and with no other source close to it. The two peaks at the sides would be, then, artifacts of the equipment we wish to remove, side lobes. For this purpose, we obtain many measurements like this, of strong, isolated sources. The computer averages all these measurements to remove noise and irregularities, coming up with the pattern for the antenna. This pattern can be used, at any level of complexity, to remove the side lobes from the observations.
A similar procedure can be used to remove the effect of the drift during observations. This drift produces a tilting of the base line that distorts the observations. Unless the conditions are very pathological, this tilting of the base line is linear and can be removed with the computer if careful observations are made before and after the record. It is to be noted that this effect is different for different observations. Dr. Kraus [3] also gives an example of this procedure and its results.
SETI workers are very much interested in the detection of signals that appear of artificial origin. The most common method of work is the spectrum analysis. The cost of one installation can run into several hundred thousand dollars.
With a computer like the one described, it is possible to sample the signals from a fixed antenna at a fast rate. The computer accumulates say 16384 samples in one second. Then, it performs an FFT analysis with bins at one hertz, and looks for any frequency where the output is above the threshold. This threshold can be set, for example, at twice the average of all the bins, to take care of the noise background. If the user so wishes, these results can be compared with successive samples to eliminate more of the influence of the noise. Since the computer used here has a very large memory space, many of these samples can be stored in a circular buffer, and when something interesting is obtained, these past samples can be put on disk for further analysis. Another possibility, nonexclusive, is to print some information at every sampling period to maintain some form of a continuous record. Possible values to print are the time, average value, maximum value, frequency of the maximum, etc.
If a larger precision is required in the analysis, say 65536 values can be sampled and stored in four seconds, and analyzed with the FFT program. This produces bins of one quarter of a Hertz. It is clear that something like this was not possible before with a computer under thousand dollars!
One of the projects that has captivated the imagination of our members is the detection of high energy pulses, believed to originate outside the Earth. No prove has been made one way or another. The way to prove the pulses are from outside the solar system is by detecting one or more of these pulses simultaneously in at least two frequencies and measure their dispersion [4].
The set up in my observatory is very well suited for this type of work. The computer can control a dual frequency converter like the one presented on the Journal several months ago [5], or like the Advanced Receiver Research converter used in my observatory. The computer sends a signal that sets the converter local oscillator to one of the frequencies and makes one or several samples at a very high rate, microseconds. It switches the frequency and takes one or more samples at the other frequency. Then the computer compares these two values with the previously stored values and, if they are similar, continues in the loop, having always in storage several thousand sets of past values in a circular file. As soon as a high energy pulse is detected in one of the frequencies, the computer alerts the operator and proceeds to store measurements at both frequencies until the pulse appears and disappears in the other frequency. Then, the computer puts all the information on disk and returns control to the operator.
If each set of values, one at each frequency takes one millisecond, the computer could store five minutes of data in its memory, which is clearly too much. With 100 microseconds between sample pairs, the storage capacity of memory is 30 seconds, which is still too much. Since the expected dispersion is in the order of 100 milliseconds [4], the computer can be continuously sampling as fast as possible to have maximum resolution.
A more accurate procedure could be used if two complete receivers are available with the antennas pointing to the same place of the sky, or with a single front end and two separate intermediate frequency strips, decoder and analog to digital converter. The computer sends the convert signal to both analog to digital converters and reads simultaneously the two values that were measured simultaneously. Note that the 68000 works with 16 bits. In this way, a resolution between 5 and 10 microseconds is possible. Higher resolution will require hardware to do the sampling, the comparison, etc. while the 68000 stores the values.
It is easy to see that the possibilities are many and varied and, although they were theoretically possible some time ago, the capabilities of the personal computers have increase so much and the price has become so low, that its use has become practical. It is clear that each one of you can think of many other interesting applications but those presented here suffice to prove the point.
The software is one of the most difficult parts of the development of one of these stations, although not impossible. The problem is simply that this software is very specialized for our application, and it does not have a market to attract any professional programmer. On the positive side, SARA has a number of highly qualified programmers among its members that, if they become interested in this project, can produce software of high quality for the use of everybody. The author is developing software for his own use that will be put in the public domain [6], as soon as it is ready.
The idea of enhancing the capabilities of the observatory of a serious amateur radio astronomer by the addition of a computer has been explored and it seems perfectly possible that any amateur can produce interesting and valuable work if his observatory is enhanced with the addition of one of the modern, state-of-the-art, and economical personal computer that have recently appear in the market. It is also clear that this is not possible unless the computer is one of these powerful new computers.
[1] M. E. Valdez, A Radioastronomical Observatory, SARA Conference, Green Bank, WV. June 1986.
[2] C. Drentea, A Simple Analog to Digital Interface to the VIC 20 Computer, Radio Astronomy, May 1986, page 12
[3] John D. Kraus, Radio Astronomy, page 226. McGraw Hill Book Co. New York, 1966.
[4] Gene Greneker, Determining the Origin of the High Energy Pulse using Automated Dispersion Receiving Techniques, Radio Astronomy, November 1984, pages 4, 12.
[5] Varactor Tuner Switcher for Simultaneous Two Frequency Observation, R. Sickels, Radio Astronomy, September 1984, pages 22, 23.
[6] Computer Programs for Radio Astronomy, M. Valdez, Editor, May 1986.