[USRP-users] RFNoC timing clarification
jonathon.pendlum at ettus.com
Thu Jul 13 00:25:03 EDT 2017
The packetization happens in the radio_core (which is in
noc_block_radio_core) in a block called rx_control_gen3. That is where the
timestamp is applied. The output of that module goes to AXI Wrapper, which
doesn't do much since rx_control_gen3 generates the header too.
If you want to resample, consider using axi_rate_change. That is a module I
wrote to make N/M resampling easier. It handles things like EOB and
timestamps automatically. We use it in the DDC and DUC blocks.
On Wed, Jul 12, 2017 at 10:03 AM, Jason Matusiak <
jason at gardettoengineering.com> wrote:
> Howdy Jonathon!
> Is a "packet" determined before or after the RFNoC stage of the chain (I
> assume before since it is in the radio core)?
> Is there anyway to modify things if I decide to resample in an RFNoC
> block? It seems like it would throw off the timing since the sample rate
> into the block != the sample rate leaving the block in this case.
> On 07/12/2017 12:58 PM, Jonathon Pendlum wrote:
> Hey Jason,
> The timestamp of each packet describes the time the first sample in the
> packet was received. The timestamp is based on a counter in
> noc_block_radio_core (timekeeper.v.) that increments at the sample rate
> (i.e. radio_clk). The time is relative to that counter's placement in the
> processing chain. We do not do anything to back out the delays from
> elements in front of the counter, such as the delay through the front end
> filters, ADC, digital I/Q correction, etc.
> On Wed, Jul 12, 2017 at 9:20 AM, Jason Matusiak via USRP-users <
> usrp-users at lists.ettus.com> wrote:
>> I have a timing question that has me going in circles, I don't understand
>> how the USRP timestamps work when using RFNoC.
>> In a normal system, packets have timestamps with them (or at least the
>> first one does) and you can compute the time of any samples based on the
>> timestamp, sample rate, and number of samples since the timestamp, right?
>> Now, if I am using RFNoC, the timestamps aren't attached to the samples,
>> right? Where does that timestamp get associated: Further down the stream?
>> USRP-users mailing list
>> USRP-users at lists.ettus.com
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the USRP-users