On 09/06/2017 07:32 PM, Cho, Daniel J (332C) wrote:
So you need to have the 1 PPS AND a reference clock (whether 10MHz
input or GPSDO)?
Absolutely. The USRPs don't, in general, have a PLL/FLL implementation
to steer their onboard clock to an external 1PPS input. This is what GPSDOs
do, and there's reams and reams of code to make that happen inside
GPSDOs.
Since GPSDOs already exist, and devices like the Octoclock-G exist for
distributing a single high-quality clock, there's little reason to have
such a
space-consuming PLL implementation in the FPGA. So the 1PPS input
is just used as a time-of-day-clock synchronizer.
*From:*Marcus D. Leech [mailto:mle...@ripnet.com]
*Sent:* Wednesday, September 6, 2017 4:27 PM
*To:* Cho, Daniel J (332C) <daniel.j....@jpl.nasa.gov>
*Cc:* usrp-users@lists.ettus.com
*Subject:* Re: [USRP-users] pps ref input B210
On 09/06/2017 06:09 PM, Cho, Daniel J (332C) wrote:
So they are seeing exactly the same PPS source. I got a waveform
generator connected to a power splitter which is then fed into
both “PPS in” inputs. Since they are seeing the same PPS from the
same source, shouldn’t they be referencing off of the external
source? Why are they choosing an internal clocking source when I
am feeding both of them the same external PPS and choosing the
option of using the external clocking source on GnuRadio? What am
I doing wrong and how can I fix it?
Thanks,
Daniel Cho
*
I think that perhaps you are confused about what the 1PPS input does,
and what the difference between the "reference clock" and
"time source" are.
1PPS is simply used as a triggering event, to allow two (or more) USRP
devices to agree on the time-of-day (for timestamping purposes)
to within a very small margin of error (a few nanoseconds).
The *reference clock* input is used to drive the clocking system for
the entire board, including ADCs, reference clocks for RF synthesizers,
etc, etc.
In order for two or more streams to be time-synchronous *AND*
phase-coherent, you need both a 1PPS input to use as a trigger to
reset the time-of-day clock on the two-or-more USRPs, and the
reference clocks must all come from the same source. In your case,
you're only supplying the 1PPS. Even with each board having an
onboard GPSDO (that is actually "watching the sky"), they won't
agree completely on reference clock phase, and they will drift in
phase relative to one another a little bit, and will have relative
phase-noise--it will be *better*, but since the phase noise of the
oscillators in two GPSDOs aren't precisely the same, even when
being "steered" by the GPS signals, it won't be nearly as good as a
shared reference clock and 1PPS.
*
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
