Re: [Discuss-gnuradio] IF sampling in ISM band

[Eric Blossom]

On Tue, Jul 08, 2008 at 02:54:38PM +1000, Phaysal Khan wrote:
> 
> > In the trunk (sorry, I didn't take time to build 3.1.2), when I do> this, I 
> > see 16MHz of spectrum:> 

> address@hidden ~]$ usrp_fft.py -8 -f 92.1 -d 4
> > How do you _know_ you're not getting +/- 8MHz?

> I am looking at the signal spectrum in Matlab. The window shows me, I
> belive, -Fs/2 to Fs/2. i.e. +/-8MHz, and the main lobe is covering
> just half of this range. I am using matlab command

> plot(fftshift(abs(fft(data)))
> where 'data' contains the IF samples.  I can't use usrp_fft at -d 4
> coz I am using a laptop right now. It gives me over run errors
> (uOuO). I might need to try that on a desktop. 

If you're not using -d 4 you're not getting 16MHz worth of data,
you're getting 8MHz worth of data, thus the +/- 4MHz.

> > How wide is the signal that you want offset by 10MHz?

> My signal is 20MHz wide and I do use '-d 4' when I use usrp_rx_cfile
> command. (Mistakenly, I forgot to put it in mail).

Phaysal, there's no way that you can fit a 20MHz wide signal down a
16MS/s complex channel.  If you try to shift it up to 10MHz it's even
worse.

You best shot at getting the data is to put the center of the band of
interest at DC in the complex baseband.  That would give you +/- 8MHz
of relevant data.  You'd still be missing the outer 4 MHz (2MHz on
each side), but given the USB bandwidth, that's the best you can do
with the USRP.

Eric