[hpsdr] TeamSpeak audio 2014/Nov/22

[Warren C. Pratt]

Edson,

Glad you enjoyed the TeamSpeak discussion.

The article you reference gives a very good and succinct tutorial. "Fast 
convolution" has been around for a very long time; however, it is 
apparently not well-known / well-understood.  Doing down-conversion at 
the same time is just an extension of the concept.  I'd recommend this 
article for anyone (1) interested in learning more about practical ways 
of implementing FIR filters in DSP or (2) looking at various DSP 
down-conversion techniques.

There are quite a few ways to do down-conversion in DSP.  Each seems to 
have its "place":  (1) Perhaps the easiest to understand is simply 
multiplying the IQ signal by a complex oscillator. Multiplication by a 
complex oscillator in the time-domain "rotates" the spectrum in the 
frequency domain (this is the frequency-shifting property of the DFT).  
(2) Using "fast convolution" with decimation done in the frequency 
domain via aliasing (discussed above).  (3) Google for "polyphase 
channelizer" and you'll find a technique that provides many fixed, 
down-converted "channels".  (4) One can use FFTs with windowing of the 
signal and selection of bins. ....  These are all useful in various 
situations.

For some of the down-conversions we've been discussing recently in 
HPSDR, we are starting with high-bandwidth data ... many Mhz of 
spectrum.  So, the down-conversion algorithms must be "drinking from a 
firehose" as Phil likes to say.  If FFTs are used in that process, that 
implies that we must execute FFTs very rapidly.  There are a couple 
other DSP concepts that I believe are also not well-known or 
understood:  (1) using a complex FFT to more efficiently compute the FFT 
of real data, and (2) using multiple complex FFTs, running in parallel, 
to compute the FFT of a complex sample stream.  For those interested in 
exploring DSP algorithms, these are also useful for the toolbox.

73,
Warren  NR0V

On 11/22/2014 8:47 AM, Edson W. R. Pereira wrote:
> ***** High Performance Software Defined Radio Discussion List *****
>
>
>
>
> I've enjoyed the discussions this week. Specially Warren's explanation 
> about frequency translation when using the overlap-and-save algorithm. 
> I've found this article that depicts Warren's explanations very nicely.
>
> http://www.3db-labs.com/01598092_MultibandFilterbank.pdf
>
> 73, Edson PY2SDR
>
>
> ---
> - We humans have the capability to do amazing things if we work together.
> - Nós seres humanos temos a capacidade de fazer coisas incríveis se 
> trabalharmos juntos.
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20141122/6b86faa6/attachment-0003.htm>