On 02/02/2023 16:45, Raúl Nieto Freire wrote:
Dear Marcus,
Thank you for your comment. For the moment we have not used RFNoC, but
if it helps measuring the replay latency, of course we could look at
it. Would that work for measuring the recording latency as well? Do
you recommend some configuration or strategy to find actual latency
for a record and replay of IQ samples? In any case I should check the
RFNoC documentation.
Regards
Raul Nieto
The 1PPS input signal isn't used (without FPGA mods) as a "sampling
trigger". It's not really exposed that way in the API.
Being able to measure the latency between its arrival on some other
useful event within the FPGA requires FPGA programming--
probably via the very-useful RFNoC framework, but FPGA programming
will be required, nonetheless.
Samples arrive from USRPs with timestamps, and in the X310, those
timestamps are applied as close to the output of
the "ADC output plane" as possible, but group delays and so on ahead
of the ADC aren't characterized, and will necessarily
be somewhat unpredictable from daughtercard-type to daughtercard-type
and even batch-to-batch in manufacturing and
across hardware rev numbers.
In a "burst type" replay environment, you can precisely control the
burst output timing using timed streaming, since the
timestamp/Time-of-Day clock on the USRP maintains time quite well if
you use a precise 10MHz reference and use the 1PPS
to synchronize to a particular time on session startup.
But for *continuous* replay (as in a linear translator application and
the like) this cannot be made to work well because
latency in the computer is variable and unpredictable.
*From:*Marcus D. Leech <[email protected]>
*Sent:* jueves, 2 de febrero de 2023 16:40
*To:* [email protected]
*Subject:* [USRP-users] Re: USRP x300 timing calibration
On 02/02/2023 04:56, Raúl Nieto Freire wrote:
Dear URSP community,
I'd like to launch some questions about the USRP x300 hardware
(timing) calibration for a RF Record and Replay experiment using
multifrequency GNSS, therefore using the two available RF
frontends. I appreciate any help or hint you may provide.
The samples recording and replaying starts when a 1 PPS edge is
detected.Also a very stable 10 MHz source is used. Would it be
possible to measure the actual latency between the time a 1 PPS
comes to the sma input port and the actual moment of the samples
recording/replaying? If I'm not wrong this would give us, after a
GNSS software demodulation and the latency removing, the total
propagation delay from the GNSS antenna phase center to the
internal USRP DAC.
1. Record setup:
USRP x300 is connected to a RF splitter + DC block + and to a GNSS
antenna for the recording setup.
2. Replay setup:
USRP x300 is connected to the splitter as a combiner to join the
two RF channels + DC block + 20dB atenuattor and its output to a
GNSS receiver.
To perform our tests we have been using the UHD software and we
are able to aligned our 1 PPS to a GNSSt timescale.
Thank you very much.
Best regards,
Raul Nieto
You're presumably using RFNoC to do the replay? Because otherwise,
latency is wildly unpredictable.
P Please consider the environment before printing this e-mail.
_______________________________________________
USRP-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
