Re: [Discuss-gnuradio] --width-8 in usrp_fft.py causes spectral artifact when using --waterfall

[Eric Blossom]

On Fri, Dec 30, 2005 at 08:44:38AM -0500, Marcus Leech wrote:
> I just upgraded to the latest tarballs, and happily noticed that 
> usrp_fft.py now includes a waterfall display.
> 
> I also noticed the --width-8 option to use 8-bit, rather than 16-bit 
> samples.  While width-8 seems to work
>  in plain spectral display, it leaves a *strong* artifact in the 
> waterfall display.  Perhaps the waterfall
>  sink block hasn't been updated to handle 8-bit data?

Not sure abou the waterfall code.  OTOH, the reason we haven't been
promoting this mode is that it does not currently perform proper
rounding.  Thus the strong DC artifact.

> Also, there used to be a parameter you could set to control spectral
> averaging in FFT display-- something-mumble-alpha.  I can't find it
> anymore, and it would be really nice if I could turn on spectral
> averaging.

Still there. It's called avg_alpha:


class fft_sink_base(object):
    def __init__(self, input_is_real=False, baseband_freq=0, y_per_div=10, 
ref_level=50,
                 sample_rate=1, fft_size=512, fft_rate=15,
                 average=False, avg_alpha=None, title='', peak_hold=False):



Eric