Hi,
After several days test, I got the conclusion that USRP2 could be
synchronized with UHD accurately. The lastest UHD version fixed stream
bug with PPS and make it possible to sampling with delay less than
0.5ns. I attached histogram of delay distribution. The test is run on
two USRP2 rev4. PPS signal and reference clock should be from same
device.
I also try synchronization between USRP2 rev3 and rev4. The mean of
delay is about 57ns… I guess it is because the USRP2 impedence are
different. The PPS difference caused delay should be close to integer
times of 10ns…
And I want to ask about the reason of delay distribution and the reason
of predictable mean of delay. Is there any way to decrease the mean and
variance of delay?
Thanks!
Liang