[hpsdr] sampling

[Tony Langdon]

At 04:02 AM 4/23/2007, Jason A. Beens wrote:

>I think the "Why" question relates to "why sample at a rate that can
>recreate frequencies outside the range of human hearing, when the
>sampling system is marketed as a sound card?"

My understanding that I read a long time ago (it was in the context 
of using oversampling in CD players, where the source material had 
already been sampled at 44.1 kHz).

The idea was that when the original audio was sampled, there was a 
certain amount of quantization (+ dither) noise created in the 
process.  By interpolating the extra samples and digitally filtering 
the signal (assuming we use a DSP with greater resolution than 16 
bits), and also by using a higher resolution DAC, the original 
quantization noise gets spread across 4x the bandwidth (assuming 
192.6 kHz as the oversampling rate).  As a bonus, the reconstruction 
filter can be made with a less sharp cutoff and a flat phase response 
in the 20 - 20000 Hz region.

Of course, sampling at 192.4 kHz and using a higher resolution ADC 
initially means you can also use an anti aliasing filter with a flat 
phase response on the A/D side as well, then do the rest of the 
filtering digitally.  I strongly suspect this is done in practice in 
recording studios.

73 de VK3JED
http://vkradio.com


 1177278136.0