My application is currently using an rtl_sdr (FC0013).
I get the best results when I use the built-in AGC option. However, there are a couple of corner conditions where their AGC is not desirable. My signal levels very pretty wildly from barely above the noise floor to just about saturating the input with no gain.
When I set the gain on the rtl_sdr to nominal and put in an AGC2 block, I found that resulting SNR was significantly worse than the builtin AGC of the rtl_sdr. I limited the max gain, but I still seemed to lose most of my signal as the noise floor rises significantly.
To elaborate a little bit ... through trial and error, I found that the best gain I could get out on the rtl_sdr before the noise floor rose unacceptably was about a 12.4 db gain setting. However, a not insignificant number of signal sources are at the limits of the input stage of the RTL already (as in nearby transmitters vs over the horizon transmitters) and over-modulates/distorts with that gain. What I'm thinking I need is something that will measure the output of the rtl_sdr and feed back through a conditioning algorithm to set the gain on the rtl_sdr. A little more insight: My signal is a voice modulated NBFM signal, 25khz channel separation, in a 1 MSPS wide baseband -- I use xlating FIR's to select the specific channel.
Does anyone have a good example or suggestion of an AGC usage to place right after
the source?
Thank you guys! This forum, the gnuradio and its collateral project, and the community have been awesome!
Jim