[USRP-users] Short-term accuracy and stability of the GPSDO
matt at ettus.com
Tue Apr 23 01:50:44 EDT 2013
The first thing to make sure is that both GPSDOs are locked, and that the
USRPs are locked to the GPSDOs. Also, the GPSDOs will take some time to
fully lock when first powered up. In general, these should be left on long
term for best results. You will also need to have self-surveyed the
On Mon, Apr 22, 2013 at 9:31 PM, Sivan Toledo <stoledo at tau.ac.il> wrote:
> I tried to measure the short term accuracy and stability of the GPSDO. My
> application requires high stability and the GPSDO seems to not perform very
> well in that respect. I would like to check if others had similar
> What I did was to send a carrier from one USRP using tx_waveform at
> 434.6Mhz (using a WBX) and to receive 10s chunks in a second USRP, both of
> which had a locked GPSDO. The FFTs of these samples were analyzed. I did
> this for 10 minutes.
> I discovered 3 interesting things. First, the signal has a bandwidth of
> about 20Hz, which I think means that the entire TX and RX chains had
> significant phase noise, at least when multiplied up to 434MHz.
> Second, the signal was about 45Hz off from where it should have been,
> which I think mean that the two GPSDOs where off by around 1hz (given that
> the 10MHz reference is multiplied by about 43).
> Third, the signal wandered over about 40Hz within short time spans of
> about a minute or less, again indicating frequency instability of about
> 0.5Hz per GPSDO.
> This is less than a 1e-7 error, which is huge over the long term average,
> but perhaps not so large for the short term (say a few seconds or a minute).
> Are these consistent with what others have seen (or what Ettus has
> measured with these GPSDO units)?
> Thanks, Sivan
> USRP-users mailing list
> USRP-users at lists.ettus.com
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the USRP-users