Hi all!
I’ve been investigating
the use of Ettus SDRs for frequency metrology.
I noticed a strange
phase noise problem while testing the N210 with a BasicRX
daughterboard. The experiment is a simple dual channel
measurement, where 10 MHz signal from a DDS generator is
split into both input channels. Phase difference of the two
(identical) signals should reveal the “phase noise floor” of
the N210.
If I use internal
reference on the N210, I see considerable “jitter” or
fluctuations in the phase difference. See linked Fig1 for
the PSD (in time units). Using external reference split
directly from the input signal removes the fluctuations
(Fig2). However using an external reference not locked to
the signal input (e.g. high quality OCXO) results in similar
fluctuations as shown in Fig1.
Phase difference was
calculated simply as the argument of ch1 * conjugate(ch2).
Measuring a signal
detuned significantly from 10 MHz (e.g. 9.2 MHz) seems to
remove the fluctuations.
Interestingly, repeating
the same experiment with a B210 with internal reference
gives results similar to Fig2 without any problems. Signal
input was 100 MHz in this case, since B210 can’t measure 10
MHz directly.
Measurements were done
with two recent versions of gnuradio. Repeating the
experiment with uhd’s python interface gives similar
results.
Any ideas what this
might be related to?
Links to figures:
Fig1:
https://drive.google.com/open?id=1ZwAyGtDUXseIkKFOIo0yrjccriSC9PGt
Fig2:
https://drive.google.com/open?id=1wrqdBMuVT5lH0v8FJrLc0t7I0G2PGuBA
- Kalle Hanhijärvi