Samples alignment of two USRP2s with seperate ethernet ports

GNU Radio
1

Hi GRers,

I’m quite new to USRP2. We are building a wideband MIMO receiver with
two
USRP2s. In order to archieve high bandwidth (20MHz) we have to use two
independent ethernet ports to host computer.
The question is how to get the the raw samples from different ethernet
ports
aligned. We plan to use a common 10MHz ref clock and 1 PPS source
synchronizing the two USRP2s. However, as the softwares for the two
boards
are not starting exactly in the same time, there should be a fixed delay
between the two channels of samples. How to get over this?
I read from this thread
http://lists.gnu.org/archive/html/discuss-gnuradio/2009-06/msg00075.html
regarding maybe the same issue. Has it already been solved?

Thank you in advance.

Hanwen

2

Hi,

Yes we have solved it but it is not perfect yet, probably due to jitter
in clocks, the threshold level at the 1PPS input and the quality of the
1PPS and 10MHz signals. The topic have also been touched here:
[Discuss-gnuradio] Re: delay through dsp pipelineblocked::http://lists.gnu.org/archive/html/discuss-gnuradio/2009-08/msg00173.html
blocked::http://lists.gnu.org/archive/html/discuss-gnuradio/2009-08/msg00162.html
-Patrik

Från: discuss-gnuradio-bounces+patrik.eliardsson=removed_email_address@domain.invalid
[mailto:discuss-gnuradio-bounces+patrik.eliardsson=removed_email_address@domain.invalid] För
hanwen
Skickat: den 8 september 2009 15:19
Till: [email protected]; Ulrika U.
Ämne: [Discuss-gnuradio] Samples alignment of two USRP2s with seperate
ethernet ports

Hi GRers,

I’m quite new to USRP2. We are building a wideband MIMO receiver with
two USRP2s. In order to archieve high bandwidth (20MHz) we have to use
two independent ethernet ports to host computer.
The question is how to get the the raw samples from different ethernet
ports aligned. We plan to use a common 10MHz ref clock and 1 PPS source
synchronizing the two USRP2s. However, as the softwares for the two
boards are not starting exactly in the same time, there should be a
fixed delay between the two channels of samples. How to get over this?
I read from this thread
http://lists.gnu.org/archive/html/discuss-gnuradio/2009-06/msg00075.html
regarding maybe the same issue. Has it already been solved?

Thank you in advance.

Hanwen