[hpsdr] Frequency offset - not seen with gnuradio

[Tom McDermott]

Just an observation - ran some tests using an HP Z3801A GPSDO
connected to Hermes V2.0 as the 10 MHz ref clock, and using
gnuradio / linux as the software.

Listening to WWV on 5.0 MHz, the IQ baseband samples from
Hermes (at 48k sample rate) exactly line up at 0 hertz offset.
If I then dial in 5,000,005 Hz as the Hermes RX NCO frequency,
I see a -5 Hz offset in the baseband spectrum (as expected).
The gnuradio software is not adjusting any timing of the samples,
it is low pass filtering / decimating / plotting / the numeric value of
the digital samples from Hermes.


Sending IQ samples to a LR stereo speaker, it's pretty easy to
hear the binarual effect of the 5 Hz offset (it swishes left-right at
5 Hz). This also ceases at 0 Hz offset.

Filtering / decimating the spectrum down to 48 samples/sec and
integrating for 4096 sample FFT (quite slow) gives a very clear spectral
peak at the NCO offset frequency (WWV minus NCO). 


For Hermes + Gnuradio there does not seem to be a frequency offset
error.

-- Tom, N5EG
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20121209/5b92cbcd/attachment-0003.htm>