Re: [Discuss-gnuradio] working with WFM sample data

[emilio gonzalez]

hi Marcus,

thanks for this tip, it pushed me the the right direction.  i guess 
this capture file isn't the standard "complex" datatype, and a 
conversion was the answer.  i found the answer on the gnuradio.org 
Octave page that said the magic word, short.  i use this graph to render 
on an FFT scope:

[ File Source (short) ]
  -> [ Throttle (complex, 25M) ]
    -> [ IShort to Complex ]
      -> [ FFT Scope (complex, rate 25M) ]

using IShort to Complex was just a result of experimentation, as Short 
to Complex seemed to output only half the captured band.  now i get to 
delve in to the world of filters, oof!

i'm surprised this isn't listed on the gnuradio.org Sample Data page ( 
http://gnuradio.org/redmine/projects/gnuradio/wiki/SampleData ), as i've 
come across messages from other folks confused about the data types and 
semantics of these USRP capture files.  i suggest an addition to that 
page that mentions the necessity of an IShort conversion.

thanks again!
- emilio
  KI6NVO


> It's very likely that your on-disk sample format isn't what you've
> declared to Gnu Radio, and it's seeing some garbage numbers.
>
> Is it
> possible that it's stored as 16-bit I 16-bit Q in the file? If so,
> you'll need to convert it before any of your complex-float blocks do
> anything with it.