In recent years there has been a tremendous increase in the number of Slow Scan Television (SSTV) stations operating on HF, VHF, and UHF bands. In large part, this increased activity reflects the development of computer-based approaches to receiving and displaying pictures. At present, there are three basic approaches to implementing SSTV on an IBM-compatible PC: The use of simple one or two stage op-amp interfaces connected to and powered by the PC serial port. More complex, dedicated SSTV interface systems, implemented as ISA bus cards or designed to interface with a PC parallel port. SSTV implementations using the PC's multi-media sound-card as the SSTV interface. All of these approaches can do an excellent job under good-signal conditions and none are particularly expensive to implement, providing one already has a computer in the shack.
In practice, with less-than-optimum conditions, the various systems do not perform in a comparable fashion. The problem is, while there are standardized and widely accepted approaches to evaluating the performance of most of our equipment, SSTV reception performance has not been evaluated in any systematic way and standardized methods have not been developed. How different systems perform under weak-signal conditions is one case in point. When it comes to HF DX work, for example, stations must exchange pictures that include the callsign of the other station, a signal report, and, typically, ancillary information such as the operators name and location. If the information is readable at both ends, a valid contact can be logged. This seemingly simple approach is critical, for, with modern equipment, it is possible to receive useful SSTV signals under conditions where the other station's voice signal (typically with a much lower duty cycle) is buried in noise. If you are going to chase DX, being able to select an optimum combination of hardware and software to extract pictures from noise has obvious advantages.
Unfortunately, you cannot make such an optimum choice by simply using several SSTV systems in normal operation. There are simply too many uncontrolled variables for any kind of precise analysis. The opinions of other operators, while well-meaning, are based on precisely this type of approach and thus are no more useful than comparing all the available systems on your own! With respect to noise performance, any valid system for objective comparison must incorporate the following elements:
What follows is a very simple approach that can realize these test constraints and thus produce objective results. While the method will be illustrated in terms of display capabilities in the presence of noise, precisely the same approach can be used to evaluate resistance to interference, the relationship of image content to readability under various conditions, and even the effectiveness of specific SSTV image modes under varying conditions.
Recording
The key element in this particular method is to use precision recordings of a specific image in the performance analysis. Ordinary audio recordings will not suffice for several reasons. First, tape imperfections and other varions during playback can produce artifacts in the recorded signal which can complicate comparative analysis. Secondly, and even more important, the most widely used SSTV modes today are synchronous, and are transmitted and displayed by reference to very precise timing standards. Speed variations in recording and playback are a major problem with even high-end analog recording systems and are sufficient to completely destroy proper image synchronization. A digital audio tape (DAT) recording system would be ideal, but these are not widely available and they are quite costly. A cost effective alternative is the use of a quality video tape recorder (VCR). The various VCR systems vary considerably in terms of how the audio channel is modulated and recorded. To maximize the fidelity of the recordings, the use of an 8 mm recording system is suggested. A Soney EV-C3 8mm recording deck was used in these tests. To assure image synchronization, the SSTV audio signal (with noise) was applied to the audio line input while a composite NTSB video signal was applied to the video line input. On playback, the SSTV signal was derived from the audio line output, with the recorded video signal providing the needed reference for locking the recorders servo drive system to maintain a constant playback speed.
Image and Noise Source
The source image for these tests was a 320 x 240 color image of my wife, produced by a high-quality flat-bed color scanner and saved in .TGA format. It is very important that all the image files used in such testing be uncompressed. Image compression formats such as JPEG almost always introduce artifacts, especially in the presence of noise.
The noise source was primitive but effective. The station transceiver (an IC-707) was set on upper sideband, the receive preamp was switched in, the system was set up on 10 meters into the station dummy load. The noise spectrum was thus primarily shaped by the transceiver IF filters and thus provided a reasonable thermal noise profile that would be encountered in actual use. A sequence of six recordings was made of the test image in the Scottie 1 mode using a Pasokon Classic ISA board with version 3.1 of the Pasokon Classic software. Each picture in the sequence had progressively higher levels of noise mixed with the SSTV signal:
No attempt was made to precisely quantify the SNR in each case, since the same recorded image/noise mix would be used in each test. An audio spectrum analyzer could easily be used to set a specific SNR for each image, or the instrument could be used to measure the resulting SNR from the recorded tape.
Systems Tested
A total of three representative systems were tested to evaluate the method:
All systems were tested on a 75 MHz Pentium system with a fast hard-drive and 32 megabytes of RAM. In the case of the sound-card systems, signal input levels were set between 40 and 50%.
I have put the results on another page because the material is very heavy on graphics. If you are to be able to evaluate the test pictures, you must turn off any image compression on your browser. This means the pictures will take a lot longer to download, but if you view them with compression on, you are wasting your time!
Finally, I want to emphasize that I set up these tests to find out, for myself, which was the best system to use for weak signal work. Remember, at each noise level, each package is displaying exactly the same image, right down to the last noise pulse! I have no axe to grind with respect to any system. I use them all and will continue to do so. I have recommended each of them to others and I will keep on doing so. They are all excellent products, but, as you shall see, each package has its own strengths and weaknesses.