BackgroundThe receive system I have been using during recent years for weak signal detection has one serious weakness, and that is poor tolerance to interference from very strong local stations. Otherwise the performance is quite good, and EME contest results show the benifits of the methods with adaptive polarisation and adaptive filtering described at this site.
There are two problems:
The DSP, a TMS320C25, operates with a 12 bit A/D converter and 16 bit arithmetics, so the dynamic range of the DSP system is not very high. Consequently I have to reduce the bandwidth in the analog system now and then to avoid interference caused by overloading. The normal bandwidth is 3.5kHz. Such a bandwidth is desireable because it is not uncommon to find stations calling at a frequency offset of more than 1kHz. When I have to reduce the bandwidth to 500Hz, I loose the possibillity to hear stations more than a few hundred Hz away, but such a bandwidth reduction is nessecary now and then because many EME stations are active in the SM5/SM0 region and it may be hard to find a free frequency without coming close to a neighbour.
A more serious problem is the noise blanker. The TMS320 system uses a first blanker operating after a gaussian shaped filter with 500kHz bandwidth. This blanker is very efficient, even against electrostatic rain, but very strong local stations destroy the operation. When the wideband blanker is disabled, a second noise blanker operating at about 15 kHz bandwidth, also after a gausian filter takes over the job. The second blanker is less efficient, but it is good enough as long as the problem is not electrostatic rain. The problem is that it is much more sensitive to interference from local stations within the 15 kHz bandwidth.
During EME contests it is very difficult to find workable frequencies when a 15kHz window is useless due to a local fellow EME-er. (weak signal operation without noise blanker is often meaningless) One objective of the PC receiver project presented at this site is to overcome this problem.
The Pentium MMX has more computing capacity than the TMS320C25 by orders of magnitude, so it seems natural to use the standard PC as the platform when trying to improve the receive system.
Right now the Soundblaster 16 is the natural choice for tha A/D conversion. Rumours tell that professionals (CD industry, radio stations...) soon will have a 24 bit standard at about 90kHz sampling frequency. I am told that this new standard will be in a separate box that fits to the PC, and that the cost will be about 1000 USD(?).
The Pentium MMX at 200MHz easily handles the 16 bit stereo data sampled at 44kHz by a SB16, but the computing capacity is not good enough to fully use much more bandwidth, so for the time being I have not put any effort in trying to find a better A/D board.
Once the PC receiver is fully operational, it is a trivial thing to increase bandwidth and dynamic range by use of faster processors and better A/D conversion sub-units when they becoma available at a reasonable price.
More bandwidth will greatly improve the noise blanker function, and more bits with better linearity will improve the dynamic range. Already with a 200MHz Pentium MMX and a SB16 the performance is impressive and I am not aware of anything that even comes close when it comes to receiving weak signals in the presence of strong signals and man made noise.