[hpsdr] new architecture - FFT front-end

[Alberto I2PHD]

On 9/10/2014 1:41 AM, John Miles wrote:

> /Alberto's on the right track with regard to using Matlab's FFT and IFFT
> function to digitize a chunk of data, but without windowing there will be
> massive transient artifacts in the recovered data, since the input data
> isn't a continuous function.  Likewise, without an overlap-add operation,
> you can't work with a series of successive buffers./

When I do convolutional filtering in my software, using the overlap and save method (IMHO
more computationally efficient than the overlap and add), I do not apply any window to the
digitized data. If I do, then there are artifacts on the audio, like periodic variations in
amplitude at the rhythm of the incoming buffers. And not applying windowing does not cause
any ill effects, at least in this case.

Instead I saw that windowing is absolutely necessary when you use the FFT to compute the
spectrum and the waterfall, otherwise, as theory predicts, the smearing on the adjacent
bins is too high.

-- 
/*73 Alberto I2PHD*
Credo Ut Intelligam/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20140910/7148bda5/attachment-0003.htm>