Hi Isaac! Your questions are really general signal processing questions, or at least GNURadio questions, not unique to a USRP. I was confused on the same subject when I first started using SDRs, though.
Nyquist really just means that, if you have a real signal with no power outside a given bandwidth (let's say 20MHz), then you need 40 million independent samples per second to represent it. You can get that by real-valued sampling using a regular clock at 40Msps, or you can simultaneously sample in-phase and quadrature components at 20Msps yielding complex valued samples. The latter approach is what you'll generally find with SDRs. However, if you want to accurately represent your 20MHz signal digitally, you need a bandpass filter in front of the analog to digital converters with a 20MHz pass band. Any real filter will have some rolloff at the band edges, which means that your signal may have power beyond the 20MHz that you care about, and you need a little higher sampling rate to avoid aliasing of that extra power. Secondly, the sampling has a limited number of bits to represent the analog signal, and you need a slightly higher sampling rate to guarantee an accurate representation of your signal. So, with complex sampling, you generally need to sample at roughly the same rate as the bandwidth you care about plus maybe 25%. -- Clint On May 11, 2021 6:15:26 PM EDT, isaac mario tupac davila <isacct...@gmail.com> wrote: >Hello community > >I'm Isaac. I'm dealing with some questions about the interpretation of >sampling rate and bandwidth in a USRP source block. > >What I understand is if I work with an USRP N210, my ADC works with a >100MS/s. If I use a Gigabit Ethernet and a data type of 16bits, I could >receive in my host up to 25MS/s with a bandwidth of 20MHz. >https://kb.ettus.com/About_USRP_Bandwidths_and_Sampling_Rates > >My questions are: > >1. If I can receive up to 25MS/s on my host, why my bandwidth is 20MHz? >I >think It is up to 12.5MHz according to Nyquist. > >2. Why is the sample rate value in the usrp source block equal to the >bandwidth I observe? I think this bandwidth should be the half >according to >Nyquist too. https://wiki.gnuradio.org/index.php/USRP_Source > >I appreciate any help to clarify this concepts > >Regards >Isaac T.
_______________________________________________ USRP-users mailing list -- usrp-users@lists.ettus.com To unsubscribe send an email to usrp-users-le...@lists.ettus.com
