Re: [Discuss-gnuradio] usrp tx documentation/confusion

[Matt Ettus]

On Fri, 2006-05-05 at 16:48 -0700, Erik Tollerud wrote:
> Ah - I missed the usrp_siggen.py... at any rate, the situation still
> arises that I don't get anything resembling sin(wt)+cos(2wt)... I
> guess the biggest aspects of my confusion come from the interpolation
> (specified in the usrp.sink_c constructor) - is that interpolating
> between 2 points on the curve that's input into the sink, or does it
> just sample the signal however often the interpolation is and fill in
> the rest?  And for that matter, I don't see why the u.set_tx_freq is
> controlling the output - shouldn't the frequency specified for the
> input sine wave change something (at least if its around the tx
> frequency)?

The USRP DACs sample at 128 MS/s.  You can't send samples that fast over
USB, so you send them at a lower rate.  When they get to the USRP, they
are interpolated to 128 MS/s.  If you have an interpolation rate of 16
(the minimum), you need to send 8 MS/S over the bus (16 * 8e6 = 128e6).
In essence, the USRP inserts 15 zeros between every sample you send it,
and then filters it so that the intermediate samples form a smooth
curve.  

After interpolation, this signal is still limited to +/- 4 MHz since the
original sample rate is 8 MS/s.  The digital upconverter (as controlled
by u.set_tx_freq) then upconverts it to you desired center frequency,
which is in the range of +/- 44 MHz IIRC.

What I think you are running into is that your cos(2wt) is at double the
BASEBAND frequency of sin(wt).  Both are then shifted up by the
u.set_freq.  If w corresponds to 100 kHz, and 2w corresponds to 200 kHz,
and u.set_freq is equal to 5 MHz then you will get tones at 4.8 MHz
(from cos(2wt)), 4.9 MHz (from sin(wt)), 5.1 MHz (from sin(wt)), and 5.2
MHz, from cos(2wt)).

Matt

> 
> Again, if there's some web site that you know of that will explain
> this all, I'm fine with figuring it out with a little guidance... I
> just am getting somewhat buried in the terminology/conventions for
> something that seems it should be fairly simple...
> 
> On 5/5/06, Matt Ettus <address@hidden> wrote:
> > On Fri, 2006-05-05 at 15:45 -0700, Erik Tollerud wrote:
> > > I've been trying to work out how to do some things on the transmit
> > > side of gnuradio, and I've noticed there's something of a dearth of
> > > documentation.  Most of it amounts to "its just like receive, only
> > > backwards."  That's probably fine for people with signal processing
> > > background (which may be the majority of this community...), but I
> > > find myself getting confused... by looking at some of the examples, I
> > > get a sense of how to initialize all the components in python, but if
> > > I do anything beyond them, I get unexpected results.  For example, I
> > > took siggen_min2.py and modified it so that instead of outputting a
> > > pure sine wave, it output sin(wt)+cos(2wt).  This did not put anything
> > > resembling sin(wt)+cos(2wt) on the output of the tx board.
> > > Furthermore, chaging w does not some to alter the waveform at all -
> > > only calling u.set_tx_freq changed the output freqency.
> >
> > I agree that there is definitely a dearth of documentation, especially
> > for TX.
> >
> > The [unstated] purpose of siggen_min2 is to transmit different signals
> > to TX daughterboard A and TX daughterboard B.
> >
> > In general, without really changing the firmware, any signal sent to
> > DB-A will output the I signal on one output and the Q on the other.  The
> > same goes for DB-B.
> >
> > For what you are trying, I would suggest working with usrp_siggen.py
> > unless you want to use multiple TX daughterboards.
> >
> > Matt
> >
> >
> >
> >
> >
> 
> 
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
> 
-- 
Matt Ettus <address@hidden>