[Fwd: Re: [Discuss-gnuradio] USRP data rates]

[Jeff Brower]

Lisa-

> 32 Mbyte/sec --> 16 Msample/sec --> 16 MHz bandwidth.

After reading Richard's reply and thinking about Nyquist, I must be wrong.  16
Msample/sec, 8 MHz bandwidth, somewhat less usable bandwidth.

-Jeff

-------- Original Message --------
Subject: Re: [Discuss-gnuradio] USRP data rates
Date: Wed, 08 Aug 2007 23:07:14 -0500
From: Jeff Brower <address@hidden>
Organization: Signalogic, Inc
To: Lisa Creer <address@hidden>
CC: address@hidden

Lisa-

> The USRP sales brochure lists as a feature "Capable of processing signals up 
> to 16
> MHz wide". I can only figure out how it can process signals up to 8 MHz wide. 
> What
> am I missing?
> 
> Since samples sent across the USB are 16-bit signed integers (16-bit I and 
> 16-bit Q
> for complex samples) that makes each complex sample size 4 bytes. Since the 
> USB max
> is 32MB/s, we have a limit of 8MS/s which translates to an effective 
> bandwidth of
> 8MHz wide.

32 Mbyte/sec --> 16 Msample/sec --> 16 MHz bandwidth.

If you want to consider complex sampling, then divide into 8 MHz positive
frequencies, 8 MHz negative.

-Jeff