|
On 09/11/2017 04:50 AM, John Shields
wrote:
Thanks for the feedback but I am not
sure that I understand it. What I was hoping to do was step
through the configurations with increasing levels of
synchronisation and expecting to see same.
Marcus' comment is correct and I have not, yet, put in the code
which synchronises SBXs.
I guess my basic point, from looking at previous post from
others Marcus L included, was that UHD would somehow improve the
synchronisation between two USRPs in the same container versus
those two separately.
What the multi_usrp object will do is to *align timestamps* among
all the devices within the "container". That is distinct from
arranging for
synchronization in time and phase. For that, you must have a
shared 10MHz reference and 1PPS source, and you must ask the
UHD source to use them.
What I do for testing ongoing phase coherence is to conjugate
multiply the two halves, then low-pass filter and decimate. This
allows me to
observe any phase-drift between the two sides over time. This
assumes that both sides are tuned to some common test signal.
In the case where two devices aren't sharing a reference clock, one
can expect the above to produce a noise-like output, since there
will be random
mutual phase noise. In the case where the two sides have a slow
phase-drift with respect to one another, one can expect a slow
sinusoidal output.
In the case where both sides are in constant phase coherence, one
can expect an essentially-constant output.
When I executed the FG shown (separately) with the USRPs
individually and then within a UHD container the results in
terms of phase variation was the same. I had expected that,
based on my understanding, the containerised USRPs would have
behaved better.
So, either my FG does not measure what I thought it should or
there is little UHD-related benefit to having USRPs individually
or in the 'domain' as MarcusL has mentioned previously. From my
situation it doesn't hence the first question in the post:
Does my FG not measure what I claim to be wishing
to measure?
Kind Regards,
John
On 11/09/17 01:03, Marcus D. Leech wrote:
On 09/10/2017 08:58 PM, Dan CaJacob
wrote:
I could be wrong, but I thought the SBX was one
of the few daughter cards that starts with s known phase
offset?
Only if you ask it to do so, and only if it's sharing clock with
its buddies...
On Sun, Sep 10, 2017, 2:49 PM Fulcrum
Associates < address@hidden>
wrote:
Dear
All,
I have a couple of USRPs connected, through
a strong
attenuator to a signal generator (NWT4001). While the
units have a MIMO
option, I don't have that cable. (Option A) When I run the
GRC as
attached, I see too good a result to the extent that the
differential
Phi seems to range over +/- 5 degrees.
What I had hoped to prove to myself that two
N200 with SBX
would have a varying offset without MIMO cable, then I
would connect the
MIMO cable and move the USRPs into a multi-unit and enable
GPSD O/B on
the unit which has the feature and MIMO for one without
(Option B) and
that the phase differential would improve noticeably and
be a variable
constant, but it didn't.
If it had, but there still was a fixed
phase offset which
varied each time it was setup (which is what I would
expect under B)
then I would hand-code the SBX stream initialisation code
to remove the
offset.
Does my FG not measure what I claim to be
wishing to
measure?
If it does measure it correctly, why do my
expectations
of options A and B leading to a different (though
improved) situation
not eventuate?
Kind Regards,
John
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
--
Very Respectfully,
Dan CaJacob
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
|