On 22/04/2024 10:56, David Raeman via USRP-users wrote:
Hello,
I am also using multiple distributed E320s for an application that
uses the integrated GPSDO to have a common sense of time for precise
TOA estimation. I don’t have the measurements at my fingertips at the
moment, but I found that synchronization was substantially better
using a high-quality external clock/time source (e.g. Octoclock or a
good OCXO), compared to using the integrated GPSDO.
The synchronization performance of the integrated GPSDO improved if I
let the radios sit for quite a while. For example, if two radios had
GPS lock for 5 minutes before running an experiment, relative carrier
drift would be measurably worse than if they were allowed to sit with
GPS lock for an hour. I believe that once the GPS receiver acquires
time, the 10 MHz reference convergence occurs very slowly.
In the case of a using an external clock, to get good phase
performance I also had to power down the internal GPSDO. The RF switch
used to select between the two sources seems to introduce some
coupling/interference when both sources are present at the switch
inputs. My old mailing list inquiry about the problem is here
<https://lists.ettus.com/empathy/thread/XK7DJMNXTZVTKHGVAPZFJL3NZEF73XME>.
I pushed a patch to Ettus, incorporated into UHD 4.4, which allows you
to power off the integrated GPSDO by passing “enable_gps=0” in your
UHD session args.
Interestingly, in that thread somebody mentions having a similar issue
on a B210 device if the GPSDO add-on module is installed, which
doesn’t seem to have been an issue with your prior results.
I can try to follow-up with some measurements from our experiments,
and I’d appreciate any additional insights you might find as well
since this is still an active project for me.
-David
In my experience, the 1PPS from a GPSDO will vary in phase compared to
another GPSDO on the same antenna by
amounts up to 10s of ns. If two radios are setting their timestamp
clocks to "next PPS", there's absolutely ZERO
guarantee that they'll agree, on scales of 10s of nanoseconds, on
what time it is.
Further, phase-drift between two GPSDO on the 10MHz output, even when
connected to the same antenna, can be
quite noticeable, depending on what your observation interval is.
If you want really-good phase/frequency/time alignment of multiple
radios, you need a shared, common, reference.
Distributed implementations that are assuming that the GPSDO will
"solve everything" are "in a state of sin".
There are many articles "out there" on what to expect from a GPSDO.
Here's one I just picked at random:
https://www.febo.com/pages/stability/
*From:* edenmclaughlin...@gmail.com <edenmclaughlin...@gmail.com>
*Sent:* Monday, April 22, 2024 4:24 AM
*To:* usrp-users@lists.ettus.com
*Subject:* [USRP-users] About Two E320's Synchronization using GPSDO's
We have a setup with two USRP devices, attempting to acquire signals'
time of arrival (TOA) with matched filters. We can observe stable and
accurate TOA using USB-driven devices (i.e., B210's) with a single
external PPS and 10MHz feed.
However, when testing the design with two E320's using their
integrated GPSDO as the clock and timing source, we could not achieve
a steady TOA anywhere close to the clock performance specified here
<https://www.ettus.com/wp-content/uploads/2019/01/USRP_E320_Datasheet.pdf>.
A constant drift over time (i.e., +50 ns over ~120 seconds) was
observed on one E320 device with respect to the other for each run. We
are almost certain that neither of the devices is damaged (they are
fresh out of the box), also we are aware of this
<https://lists.ettus.com/empathy/thread/EDXCIKFY5LW7IEQN2JZOGXGWNCBELU3F?hash=4KGAPCNGJL6WXKC4IBZBITVNSFTDXXRK#4KGAPCNGJL6WXKC4IBZBITVNSFTDXXRK>
issue.
The setup includes one source signal symmetrically fed to the devices
from the same RF channels. The E320 devices share the same GPS
antenna, which is an externally fed active GPS antenna while the E320
GPSDO's ports are DC blocked.
We are yet to try:
1. Observing the behavior with another pair of E320 devices.
2. Feeding the source signal from the other RF ends.
3. Synchronizing the E320's with an external clock and timer, as we
have done with the B210 devices.
We are wondering if there are any software tricks applicable specific
to the E320 devices that we should be employing. Any help will be
appreciated!
_______________________________________________
USRP-users mailing list --usrp-users@lists.ettus.com
To unsubscribe send an email tousrp-users-le...@lists.ettus.com
_______________________________________________
USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-le...@lists.ettus.com
