On Sat, 6 Feb 2016, gareth wrote:

"Brian Howie" wrote in message
...
In message , gareth
writes
Presumably those SDR rigs which do not work on the IF but
directly from antennae must have, separately from the DSP processor,
some semblance of a DDS generator (but without the final DAC) to
act as the equivalent of the VFO, for I cannot perceive that a
fractional-Hz
tuning rate could be achieved with machine code running in the DSP
processor?
I'm not an expert ,but I think what you're asking is " how is the local
oscillator generated" in a direct conversion SDR and "what determines its
resolution"
There is an example here, :-
http://www.radioelementi.it/public/saqrx.pdf
The "c" source code is here,which I can just about understand ( My
software background is FORTRAN and Matlab) :-
https://sites.google.com/site/sm6lkm/saqrx/
Softies shouldn't have a problem with it although I was able to mess about
with it and recompile it successfully
In this case the spectrum is dc to 22050Hz in 512 steps. It's not the LO
precision ( it's floating point in this one) that limits it but the size
of the FFT , the sample rate and thus the record length, that sets the
minimum FFT bin width . This one tunes in lumps of about 43Hz

Thank-you Brian, but what you have URLed is already at baseband, being VLF.

I thought that was the norm, not much doing A/D at signal frequency.
Initially, it was too fast for the hardware to handle, but there are
probably some good reasons still to downconvert.

Michael




When moving from FORTRAN to C, the major difference (apart from the
nitty-gritty
of statement syntax) is that in FORTRAN, variables are always passed by
reference
(at least in FORTRAN '66 which I did 47 years ago) and in C you have the
choice of passing
by value or by reference.