[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Bursty transmissions with X310?

From: Michael West
Subject: Re: Bursty transmissions with X310?
Date: Fri, 17 May 2024 08:32:43 -0700

Hi Ben,

The SOB signal is essentially ignored by the USRP Sink block.  If data is sent, the SOB is assumed and data is transmitted even if the SOB is not set..  TX samples not intended to be transmitted must not be sent to the USRP Sink block.  The EOB, as Phillip pointed out, tells the USRP Sink not to expect any more data so underruns are not detected.

Michael E. West

On Thu, May 16, 2024 at 6:54 AM Ben Wiederhake <ben.wiederhake@timelords.saarland> wrote:
Dear Philipp,

thanks for your response!

> from my experience with bursty transmissions the EOB tags are more to
> tell the USRP to not expect any further samples (and not to scream
> UUUUUUUUUnderruns), rather than to switch of the transmission.

Switching off the transmission is precisely what I want to do, in fact.
If the TX path is still actively driving the antenna (to 0 V, to +1 V,
to -1 V, doesn't matter), the RX sensitivity of the antenna is
necessarily very bad. This RX/TX switch alone makes a difference of
(very roughly) 20-30 dB in RX signal strength.
So I really want to switch off that TX path.

> I once implemented a custom "burst copy" block for this purpose as I
> couldn't find one in the core lib [2] and this works well for my use
> case (meanwhile i am using RFNoC which does not handle tx_eob tags, but
> stopping the stream still works well).
> To time it, the sample with SOB should be delayed or have a "tx_time"
> tag too.

Thanks! I've tried to avoid the tx_time tag until now, so I'll
experiment a bit based on your code and the tx_time tag.

I was really hoping the USRP Sink would do all that "by itself". The
timing is already implied by each sample's index in the sample stream,
so the tx_time approach seemed like unnecessary effort to me, but
apparently I'm overlooking something. I'll try it now.

If you (or someone else) has any other ideas, or can tell me what I
might be missing or doing wrong, I'm all ears.

> If you don't rely on precise timing, you can also simply set the samples
> between EOB and SOB to zero to achieve what you want I guess.

I already tried that; this just leads to the TX path still being active,
leading to the 20-30 dB loss I mentioned earlier.

With regards,
Ben Wiederhake

reply via email to

[Prev in Thread] Current Thread [Next in Thread]