On 11/04/2010 01:09 AM, Steve M. wrote:
Hello:
I am running GNU Radio 3.3.0 under openSUSE 11.2, and I have a USRP2 board with
a WBX daughterboard. I’m still somewhat new to GNU Radio, so please bear with me.
What is the minimum and maximum frequency at which I can transmit and receive? I
thought it was between 50 MHz and 2.2 GHz, but I thought I have seen other ranges
specified such as 50 MHz and 2.3 GHz.
Any frequency between 50Mhz and 2.2GHz.
Also, can I select any frequency in that range to transmit or to receive? Can I
set the “frequency” parameter of a USRP2 sink block to any value between 50 MHz
and 2.2 GHz, such 51.4 MHz, or 773.66 MHz, or 1855.2 MHz, etc.? Must it be an
integer number of MHz? Can I specify the value down to a single Hz? Are there
limitations/rules as to what frequencies I can set a USRP2 sink block to be?
Thanks a lot for your help.
Steve McMahon
The ultimate frequency resolution is under 1Hz, thanks to the
digital-upconverter function in
the USRP2 FPGA (and similar in the USRP1).
It’s often the case that synthesized LOs (particularly for UHF and
higher frequencies) have
rather-coarse frequency steps (several 10s to several 100s of KHz is
not uncommon), so the
FPGAs can do digital up-conversion to make sure that your signal
appears exactly
where you want it (modulo any uncertainty in the reference clock).
The frequency specified in a USRP2 sink block has to match the range
provided by the
daughter-card you’re using, otherwise an error is provoked. That’s
it.
Now, understand that if you specify the center frequency down to 1Hz
levels, there’s non-zero
uncertainty in the reference clock, usually. On the order of
5-10PPM. So when that
error is effectively multiplied by the synthesizer, the error in
actual frequency can be
somewhat significant. Let’s take a +/-5PPM error as an example, and
you’ve set your
center frequency to 900.000MHz. That nominal 900MHz signal could
actually have a center
frequency that’s +/- 4.5Khz from the nominal 900.000Mhz you’ve asked
for.
If absolute frequency precision is important, you’ll need to use an
external reference clock for
the USRP2. People generally use 10MHz GPS timing receivers for this
purpose, but a “lab standard”
precision 10Mhz reference source (perhaps based on a DOCXO, for
example, and calibrated) could
also be used.
It’s pretty-normal for commercial and consumer equipment to have small
frequency errors, it’s
a natural consequence, ultimately, of the physics of crystal
oscillators, and the manufacturing
techniques that are used to make them. Crystal oscillator precision,
stability, and phase-noise,
generally scales non-linearly with price 
Here, for example, is a pretty-good 10MHz OCXO you can buy at Digikey:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=535-10099-ND
With a claimed +/-5ppb frequency stability.
Compared with a garden-variety TXCO:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=631-1067-2-ND
With a claimed +/-2.5ppm frequency stability.
That’s a frequency stability ratio of roughly 500:1
You’re much more likely to find that $150.00 OCXO inside a $20K lab
instrument than in a piece
of commercial or retail RF gear.
–
Principal Investigator
Shirleys Bay Radio Astronomy Consortium