On 2022-09-10 06:06, Aiden Morrison wrote:
Hi Marcus,
The timestamps returning from the radios are always in perfect
agreement in the metadata between the two USRPs. We are checking the
synchronization by doing a correlation on the captured sample blocks
and observing the peak shifting by 0-3 samples. If the front-ends were
actually synchronized at the ~100 ns level, this would be +/-1.
To be clear, this is with your laboratory setup, with an Rb refclock,
and some kind of shared 1PPS signal?
What are the characteristics of your 1PPS signal in this case? Rise
time, pulse duration, voltage, etc.
------------------------------------------------------------------------
*From:* Marcus D. Leech <[email protected]>
*Sent:* 08 September 2022 16:47
*To:* [email protected] <[email protected]>
*Subject:* [USRP-users] Re: N210 USRP synchronization performance 4x
worse than expected - suspect configuration problem
On 2022-09-08 04:54, Aiden Morrison via USRP-users wrote:
Hello,
I am attempting to do synchronized multi-site reception using USRP
N210 radios. Our first attempt was to make use of 2x radios with
ettus GPSDO modules installed, feeding the radios with a common PRN
sequence and correlating the outputs to evaluate timing offset. The
Ettus GPSDO radios say they provide +/-60ns timing accuracy, so even
in the worst case where one of the radios goes to each side of the
limit, at 10 MHz we would have slightly more than 1 sample of offset
between them, but in testing we observed +0 to +3 samples of offset.
To isolate the problem we repeated the test with two different N210
radios fed by a common external 10 MHz and PPS signal from a GNSS
disciplined Rubidium source, and observed exactly the same +0 to +3
samples of shift between the captured sequences. Seeing the same
error when using a common clock indicates the problem is very likely
to be in our test software.
In our software we follow
https://files.ettus.com/manual/page_sync.html
<https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Ffiles.ettus.com%2Fmanual%2Fpage_sync.html&data=05%7C01%7Caiden.morrison%40sintef.no%7Cfa4f6deef65f423e787c08da91a91677%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637982452653631828%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000%7C%7C%7C&sdata=O88MQv%2FF3wgfA8GurCme6mIlKvRLi8KwgK5WixFuba0%3D&reserved=0>
for synchronization:
1. We create two USRP objects, one for each radio.
I would repeat this experiment, but using a single multi-usrp object.
UHD will time-align samples across radios in the
same multi-usrp container object. But not across objects.
1.
2. We command the USRPs to use the external synchronization (radios
report back that they are synchronized to the GPSDO in the former
configuration)
3. We use a common time in the time spec commands for each
configuration change (e.g. setting LO frequencies)
4. For each burst capture of 50,000 samples we set a time_spec for
each USRP a long time in the future (1.0 seconds since the last
capture)
To be clear you are requesting streaming start at the same time?
1.
2. Step 4 is repeated 1000 times giving us 1000 bursts for
evaluating the synchronization
The only factor I can see is that we are using WBX radio modules, and
we are uncertain if this is somehow a factor in what we are seeing.
Thanks in advance for feedback.
-Aiden
Also, how are you checking time alignment? Are you actually looking
at the timestamps on the sample frames?
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
