Dear Leif Tnanks for the answers. To be complete: my sdr has a front end and a backend.
The front end's main function is to allow selection of a subband, mixing with a software defined LO and decimating. The backend therefore operates on a lower frequency, currently analog modes, rtty (many parameters), psk (b, q, b63), navtech are implemented. The backend frequencies are selectable, for rtty, psk and navtext I prefer 6000 Hz, while most AM requires 9600 Hz. Furthermore mfsk is working fairly well, I recoded an implementation from pawel jarochi. I am looking into an olivia implementation now. CW is still a problem though selecting a few parameters allows software to generate reasonably often "cq, cq de ..." RTTY and psk are my favorites in listening using the elektor may 2007 card, both are working fairly well, even for relatively small signals. This summer I was playing with a variety of filters and I included in my software a selector for chebyshev, butterworth, FIR with 255, 511 elements as well as the fft convolution implementation of the fir. >From a software point of view, I am reasonably satisfied with the implementation, although it showed that the handling of the soundcard is less portable than I thought. Based on recent experiences. I am working on that. I created a simple frontend for MW, experimenting a little with tuned-loop for selected sw bands. My next steps are a. continuing to increase the portability of the implementation b. performance improvements (now it takes app 50% cpu of a 3 Mhz amd K6/7 laptop c. improving the frontend technology (dynamic range, noise) best, and again, thanks jan 2009/9/25 Leif Asbrink <[email protected]> > > > Hi Jan, > > > > My experiences are limited. For AM reception > > I can vary the band (e.g. more lower side band, less upper > > side band or vv) around the virtual center, that helps a lot in handling > > noisy signals nearby the frequency received. > Yes, with a conventional AM receiver you have two parameters: Center > frequency and bandwidth. That is the same in analog and digital systems:-) > > > I think I mean improving S/N ratio, esp for smaller signals, > > receiving rtty (b)psk is my favorite. > With an SDR you might remove various kinds of non-random noise by > use of clever software but otherwise there is in principle no difference > between analog and digital receivers. > > To receive RTTY you need a linear receiver (SSB mode.) The detect process > would be in some external software that would work exactly the same > regardless of whether the linear receiver is an SDR or an old conventional > receiver. > > > My background in information theory is limited. > > In the series of articles in QEX on software radio for the masses > > (that inspired me quite a lot) there were some mysterious > > comments in the description of the front end signal handling where > > he spoke about "improving the signal" without elaborating on it. > Improvements in the front end could be better rejection of image > frequencies and other spurious responses as well as lower noise floor > (higher sensitivity) In principle it is trivial. Just add a preselector > with narrow bandwidth and low noise figure. With high enough gain > any radio would get ideal performance for those parameters. (As long > as dynamic range is not a problem. With high enough selectivity it > would nearly always be OK.) > > There is no particular article that comes to my mind where you could > read about those very basic aspects of radio. Just read "here and there" > until you find something specific that you do not understand. Then you > would be able to formulate a specific question that you might post > on this or some other mailing list:-) > > The theory of an ideal receiver is quite simple. Most of what you can > find in the literature is about non-ideal behaviour (dynamic range.) > > Very often receivers are good enough to be considered ideal. The signals > we hear are the real ones and not spurs, images or artefacts caused by > poor dynamic range. Then the radio is just a linear process that > shifts the frequency and removes signals outside some selected filter > bandwidth. > > The linear radio is always followed by a non-linear detection process > which could be inside a human brain (CW, SSB) or in a computer (RTTY, > JT65,...) or a combination (AM, FM.) The theory of detection is the > exciting and difficult part which is right now under development. > > 73 > > Leif / SM5BSZ > > > -- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355
