On 07/07/2021 08:49 AM, Cédric Hannotier via USRP-users wrote:
Hi Jean-Michel,
On 07/07/21 12:07, frie...@free.fr wrote:
why is the multichannel behaviour in a single receiver such as the B210
different from the single
channel streaming? I can imagine the different behaviour for networked
multichannels USRP (e.g.
X310) but how about the B210?
I think it is related to this topic:
https://lists.ettus.com/empathy/thread/YHNX5U6XFKBCV3XPC34LT3JFOG2ZGA2W?hash=YHNX5U6XFKBCV3XPC34LT3JFOG2ZGA2W#YHNX5U6XFKBCV3XPC34LT3JFOG2ZGA2W
Regards
When you use set_time_now(), it's an non-synchronized operation. it sets
the clock on the target USRP to whatever value you specify.
But if you have more than one USRP clock to set, that *necessarily*
involves multiple transactions between your host and the
target USRPs, and those transactions take a finite amount of time to
complete. The clock for the Nth USRP isn't "stopped" while
the setting takes place on the N+1th USRP, etc. Which is precisely
why the whole 1PPS infrastructure exists--to allow a single
synchronization event across multiple devices.
Now, how does this apply to the B210 for multiple channels? Well
because of the way the FPGA is structured, there are effectively
TWO time-of-day clocks on the B210--one for each channel. So, you
have to use "set_time_next_pps()" on it to force both
time-of-day clocks to be set to the same value. Using set_time_now()
will cause two internal transactions--one for each clock,
and there will necessarily be some amount (several clock cycles) of
skew between the two clocks.
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
