[Discuss-gnuradio] Phase changes between two USRP-2. Why and how?

[Mamoru Yamamoto]

Dear Friends,

I am Mamoru Yamamoto, a professor in RISH, Kyoto Univ., who is
using USRP-1 for dual-band satellite beacon receiver.  We use
our receivers are used for real research, and are very productive.
Thank you.  Today I want to ask you a different thing.

We are running another project with USRP-2.  This is to
develop a radar receiver.  We plan to use several (about 7)
USRP-2-based receivers for the radar imaging.  Thus the
receivers must be working with following conditions.

(1) Synchronized in frequency and timing with a transmitter.
(2) Phase between receivers kept constant.

The condition (1) is successful with synchronizing the USRP-2
with outside REF 10MHz or 100MHz, and applying a transmitter pulse
to PPS port.  Now we face a problem for (2).  We can check this
by applying the same signal to two USRP-2 receivers.  Our
symptom is,

[a] Phase between output data from two USRP-2 is stable in a single
run for hours. (GOOD)

[b] But at different runs of the USRP-2, the phase difference
are different (= look random). (BAD)

Our URSP-2 radar receiver is at 130MHz.  Then we use Basic-RX
board.  One more finding is,

[c] If center frequency of the USRP-2 is zero (=0Hz),
then the phase difference is stable at different runs. (WHY?)

Result of [b] and [c] may show that NCO in the FPGA are oscillating
at the same frequency but at different (random) initial phases.  At the
"zero-frequency" setup, NCO output may be fixed at a certain value
like 1+0j, then [c] is realized.  I would like to know,

-- Is this story true?
-- How can we set NCO initial phase equalized at different USRP-2?

Wakisaka-kun, one of our students asked a question to this mailing
list on October 1st, but not yet get an answer.  I attach his E-mail
here.  The figure is found from,
http://www.rish.kyoto-u.ac.jp/~yamamoto/Observation_data.png

We are sorry that his writing was not very clear.  But we need your
suggestions.  Thank you very much for your help in advance.

Regards,

-- 
Mamoru Yamamoto / RISH, Kyoto University
address@hidden
Phone +81-774-38-3814, Cell +81-90-5653-7555


=== Original question (E-mail from Mr. Wakisaka) ===

Subject: About synchronize two USRP2s using external reference (SMA)

I tried to synchronize two USRP2s using external reference (SMA).
I input 10-MHz signal to REFCLOCK pin and a trigger in PPS IN.

I modified rx_streaming_samples.cc and wrote the following code
 u2->config_mimo(usrp2::MC_WE_LOCK_TO_SMA);
I use txrx_raw_eth_20100608.bin and u2_rev3-20100603.bin.
but I failed to synchronize two USRP2s.

I synchronized start timing using PPS but two USRP2's signal
have phase difference. This phase difference change every observation.
Some figures are attached.

-- Figure --
http://www.rish.kyoto-u.ac.jp/~yamamoto/Observation_data.png

In this case, I input RF signals of 130 MHz + 0.05 MHz to RF1 and set as
follows:
Center frequency: 130 MHz
Decimation: 4

When I don't set center frequency, two USRP2's signal don't have phase
difference.
So I suppose this is due to the phase difference between NCOs of two USRP2.

What should I do?

Many Thanks.

Youhei Wakisaka.

=========================