Re: [Discuss-gnuradio] TVRX flatness + USRP gain setting?

[Eric Blossom]

On Fri, Oct 21, 2005 at 05:33:22PM -0400, Weber, Michael J. (US SSA) wrote:
> What is the recommended position of the USRP gain setting when using the
> TVRX daughtercard?

Hi Mike,

First off, I strongly suggest you use CVS.  Things are a lot different
with that code.  Sooo, for the code currently in CVS, when using a
simple dipole, I typically set the gain to about 35.

> Looking at the FFT plots, if the gain is maxed out it appears that the
> signal drops about 13dB from center to edge, regardless of the bandwidth
> of the decimated signal. I.E, if the signal is 640KHz wide, the edges
> are 13dB down from the center, and if the signal is 4MHz wide, the edges
> are still 13dB down from the center. At minimum gain, the edges seems to
> drop about 6dB, but the input from the TVRX goes deaf at that setting.

The rolloff is caused by the 4 stage CIC filter in the FPGA.  There's a
half-band filter in progress that will flatten this out.  We'll use the
half-band as the last stage decimator.

> Is there a way to have a flatter signal response across the bandwidth
> and still have gain?

Gain and bandwidth are pretty much orthogonal.
In the meanwhile the quick fix is to sample at twice the rate you
really need, then filter and decimate by 2 in software.  Use a
combination of optfir.low_pass and gr.fir_filter_ccf

We do this in usrp_wfm_rcv.py that's in CVS.  

Eric