[USRP-users] Frequency offset between two USRPs B210 synchronized with use of Octoclock-G

Piotr Krysik perper at o2.pl
Wed Oct 11 15:30:33 EDT 2017

Hi all,

After discussion on IRC and few questions (by Brian Padalino and Marcus
D. Leech), that made me not so sure about how I made the measurement, I
found the reason for frequency offset.

It wasn't fault of USRP B210 or broken cable providing 10MHz reference
signal. The issue was a bit more complex:
-two B210s were not time synchronized,
-signal generator wasn't producing ideal sine wave but with slightly
drifting frequency (connecting the generator's Ref input to Octoclock
didn't help in case of that particular generato, which was FLUKE 6061A).

If at two points in time frequency of the generated signal is different,
then by looking at the phase difference between two USRPs, that are time
shifted with respect to each other, the effects of that frequency
difference will be observed.

I made the following experiment to show that it was generator's fault:
-phase difference between inputs of two USRPs was constantly measured,
-first I connected FLUKE 6061A signal generator to USRPs (through splitter),
-then I've disconnected the splitter from that generator and connected
another one(Agilent N5182A),
-both generators' Ref inputs were connected to Octoclock-G.

The effect can be seen on the attached image. For FLUKE 6061A you can
observe frequency offset while for Agilent N5182A in the measured period
no frequency offset effect can be observed.

So the conclusion is to synchronize USRPs B210 in time and to not trust
our FLUKE 6061A signal generator as a laboratory equipment.

Best Regards,
Piotr Krysik
-------------- next part --------------
A non-text attachment was scrubbed...
Name: two_generators_two_b210.png
Type: image/png
Size: 54043 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20171011/8646a652/attachment.png>

More information about the USRP-users mailing list