Re: [Discuss-gnuradio] Question about sampling

[Eric Blossom]

On Thu, Mar 17, 2005 at 03:28:26PM -0500, Dinkar Bhat wrote:

> Thanks for the info. One of my issues is the detection of the pilot
> frequency (which is a specific frequency relative to the
> channel). Now, using FFT on the samples that I captured using the
> ADC, I geta certain frequency resolution (bin size) that is
> inversely proportional to the the length of the ADC.

> But I might miss the pilot and hence I was wondering if I should
> sample at much higher rate than 2* max frequency. Even then I might
> miss the pilot, and I am wondering what is approach?
> 
> Dinkar
> 

No need to sample at much higher than 2x the bandwidth of interest.
The bandwidth of interest is 6 MHz, thus, assuming your IF frequency
aliases down to something convenient, you should be able to sample at
something on the order of 2.5 * 6 = 15 MS/sec (real).


The pilot tone is easy to see, particularly if you use an averaging
FFT.  See last night's new code ;-)

In this screenshot, the pilot is clearly visible at -2.7 MHz:

  http://comsec.com/images/atsc-ch44.png

It's RF channel 44 (653 MHz), KRXI-DT 11.1, the Reno FOX affiliate.

I captured this using a 4937 eval board with an IF of 5.75 MHz,
connected to a USRP with a "Basic Rx" daughterboard.  The antenna was
a pair of rabbit ears sitting on my kitchen table.  I was running the
usrp_fft.py example code.

The sample rate across the USB is 8MS/sec, complex.  
The screen shot is of a 512 point FFT.

Eric