Re: [Discuss-gnuradio] N210 MIMO packet TX time alignment when using message strobe

[Martin Braun]

On 11/18/2015 12:46 AM, Ludwig Stephan (CR/AEH4) wrote:
> Hi Martin,
> 
> thank you for your quick reply. I will try your suggestions.
> 
> As I wrote, we do not have a GPSDO for the PPS. Just to be clear: Do
> you want me to connect an external PPS to both USRPs and then change
> the python code of the model to set_time_unknown_pps() before the
> flow graph is executed?
> 
> Can I use the PPS only without the 10MHz reference or do I need both?
> 
No, that's not what my suggestion was: Set the clock and time sources to
MIMO, that'll share the PPS + 10 MHz. Then, sync to PPS explicitly.

> Doesn't the USRP block do this by setting the sync parameter to
> "unknown PPS/sync" (cp. uhd_usrp_sink.xml)?

Yes, it does. If you're already doing that, you're good!

> Ad 1) That would be perfect ;-) Ad 2) No, we do not see any 'L's.
> There are two 'U's when starting, but that's OK.
> 
> I will also test the tx_time injection. But remember: All works fine
> with plain tagged streams, but not if the tagged stream is
> "triggered" by a message strobe block (see attached model or image).
> Do you have an idea what could interfere there at all?

In that case, I misunderstood your original post. There should be no
difference between 'plain tagged streams' and those coming out of the
"PDU to Tagged Stream" block.

One thing you could be doing differently is use a 'repack' block instead
of 'packed to unpacked'. The former will update the length tags
correctly; no more use for the tagged stream multiply length tag block.

Does the QT GUI Time sink show all the tags you were expecting?

Cheers,
Martin