[Lf] Re: lf digest, Vol 1 #147 - 3 msgs
Vittorio De Tomasi
vdetomasi at tiscalinet.it
Tue Jun 27 18:56:44 CDT 2000
Frank,
I think that 14 USD are too much for a bandpass filter: a DSP solution
is really a free lunch!
You need about 2N(log2(2N) + 1) operations to do the whole filtering
process with FFT. Let us say you sample at 11,025 samples/s, and you use
a buffer of 1024 samples. It follows you need about 25,000 operations
for each buffer to do the filtering, i.e. you can go real time with a
system that is able to deliver 250 kFLOPs (if I'm not wrong).
Since 1002.2 Hz is a bit odd sampling rate, you can interpolate down to
1000 Hz using nearest-neighbour interpolation (this usually results in
some THD, but it needs only a look-up operation) or linear interpolation
spending about 3 FLOP for each emitted sample.
However I have a question: how do you think to transmit data ? Remember
that if you send 16 bit samples, you need to find a way to synch them,
otherwise you could put together the wrong sequence on the other side...
probably it is better to use some transport layer like TCP/IP: anybody
there willing to hack KA9Q code ?!?
Vittorio
--
*************************************************************************
Vittorio De Tomasi ik2czl at amsat.org
Home page: http://space.tin.it/scienza/vdetomas
My DSP page: http://www.freeyellow.com/members/padan
"Wir muessen wissen; wir werden wissen" (David Hilbert)
More information about the lf
mailing list