Hi Marcus,

Thanks for your help. Don't you think that 40-60ms is a bit long fot the amplifier to shut down ? I tried to increase the gain at the receiver and the sender sides but I still have the problem.


More important, I tried an other scenario by sending a burst of symbols whose value is 0. According to me, it's like if I do not send anything. However, if I take a look at the received signal over another USRP, we detect a signal which is not a "zero" signal. You may find this received signal attached and the flowgraph used (we nullify all the symbols coming into the URSP transmitter). This residual signal is very close to the signal I get in my first scenario at the end of each packet and I think the "error" is the same

May you reproduce this scenario and have the same result ? Does this result seem weird to you too? For me, it comes from an hardware problem. Or Can it come from the fact we are using the burst mode?

Thanks,

Raphaël


On Mon, 12 Feb 2018 23:13:45 +0100, Marcus Müller wrote:
Hi!

This might really be the fact that the amplifier takes some time to
shut down. This, paired with the possibility of TX/RX crosstalk might
be a contributing factor: That might actually change the DC offset on
the receiving side, and trigger the step response of the DC offset
cancellation (which is observable for about 40-60 ms, if I remember
correctly). You can try to disable the DC offset cancellation on the
USRP source and see if that helps (but probably introduces a DC offset,
which doesn't matter for practical OFDM).

If I read your flow graph correctly, you're using 0 gain on TX and RX -
can you try *a little more* gain on each end? This might change the
ratio of received signal to in-device crosstalk sufficiently that you
won't be able to spot this relative to your signal amplitude.

Regarding consecutive packets: if you look at that overlayed signal,
compared to your OFDM symbol, it's very slowly changing – almost
constant.

In other words, this will end up in the DC bin of the next OFDM frame
anyways, and those are usually not used at all. This would have no
adverse effects at all!

Best regards,
Marcus

On Mon, 2018-02-12 at 14:29 +0100, NAVES Raphael wrote:
Hi Marcus,

Thanks for your answer. I will try to be clearer.

You may find attached the screen shot of my Gnuradio flowgraph which
is
a classical OFDM transmission. First, I'm getting PDU on the
described
socket. PDU are sent on the socket every 25ms by another python
script
running in parallel. Then the message is transormed to a tagged
stream
and go through an classical OFDM modulator. I'm plotting the
received
symbols at the same USRP (reception antenna).

The problem is described on the screen shot of the received samples.
As
you may see, after each burst, the received samples take few time to
go
back to "the ambient noise level".
In that case it's not a problem because following packets are
separated
enough in time and have the same power. However, imagine the case
where
I want to receive directly after a packet with high power, a new
packet
with lower power : this packet reception will be disturbed by the
noise
coming after the first packet. How could I remove the noise following
a
packet burst ? Is that a problem of turning off the power amplifier
in
the transmission part ?

Do not hesitate if I'm not clear enough

Best,

Raphael


On Sat, 10 Feb 2018 17:56:11 +0100, Marcus Müller wrote:
> Hi Raphaël,
>
> not quite sure I get your problem, but this is rather hard to debug
> without knowing exactly what your transmitter does.
>
> For example, if you transmit something that isn't zero-mean, then
> the
> DC offset cancellation *in* the receiving end would start to cancel > out. As that cancellation is effectively a narrow high-pass filter,
> you'd see its (inverted) step response after you turn off the
> transmitter.
>
> So, if you could share both your full transmit block diagram as
> well
> as
> your receiving block diagram, we might be able to help you!
>
> Best regards,
> Marcus
> On Fri, 2018-02-09 at 09:28 +0100, NAVES Raphael via USRP-users
> wrote:
> > Hello,
> >
> > For more details, please find attached the received signal in the
> > time
> > domain. Clearly after the end of the packet, the signal takes a
> > certain
> > time to come back to "the ambient noise". If I zoom, I notice
> > there
> > is
> > still some noise 40ms after the packet. It can disturb the
> > reception
> > of
> > the following packet if its power is less than this additional
> > noise.
> >
> > What do you suggest to cancel it ? I'm using the USRP sink/source
> > blocks from gnuradio for the transmission/reception parts.
> >
> > Best,
> >
> > Raphaël
> >
> >
> > On Thu, 08 Feb 2018 21:48:10 +0100, NAVES Raphael via USRP-users
> > wrote:
> > > Hello Dan,
> > >
> > > Thanks for your answer. I'm using for transmitting the
> > > traditional
> > > USRP sink block provided by Gnuradio Companion. Each packet
> > > coming
> > > to
> > > this block is tagged with its length at the first sample. For
> > > the
> > > receiving part, I'm using the USRP source block. Both are used
> >
> > with
> > > basic parameters.
> > >
> > > Do you think I should modify/use different parameters for these
> > > blocks ?
> > >
> > > Raphaël
> > >
> > > On Thu, 8 Feb 2018 11:54:07 -0500, Dan Veeneman wrote:
> > > > Hello Rafael,
> > > >
> > > > Are you sure the transmitter has stopped radiating
> > > > immediately
> > > > after
> > > > the
> > > > end of a packet?  The power amplifier on the transmitter may
> >
> > take
> > > > a
> > > > small amount of time to go from powered up to powered down,
> > > > although
> > > > 40
> > > > milliseconds may be excessive.
> > > >
> > > > Do you have a writeup and/or code for your burst transmission
> > > > system
> > > > (transmitter and receiver), that perhaps others may be able
> > > > to
> > > > duplicate
> > > > the issue?
> > > >
> > > >
> > > > Regards,
> > > > Dan
> > > >
> > > >
> > > > On 2/8/2018 11:33 AM, NAVES Raphael via USRP-users wrote:
> > > > > Hello everyone,
> > > > >
> > > > > I implemented a burst transmission between two USRP N210.
> > > > > The
> > > > > transmission and the packet receiving works well, however
> > > > > I'm
> > > > > facing the
> > > > > following problem. From the reception side, I notice that
> >
> > after
> > > > > the
> > > > > packet receiving, the sampling signal does not come back to
> > > > > the
> > > > > ambient
> > > > > noise immediatly. It takes few milliseconds (about 40 ms)
> > > > > to
> > > > > come
> > > > > back
> > > > > to "0". It's like if there is still a signal transmitted
> > > > > after
> > > > > the
> > > > > real
> > > > > packet transmission. It may be a problem when you want to
> > > > > receive
> > > > > many
> > > > > consecutive packets with few space between them
> > > > >
> > > > >  I suppose that it comes from an hardware problem when the
> > > > > signal
> > > > > is put
> > > > > on the baseband ? Does that deal with the LO leakage ?
> > > > > What
> >
> > can
> > > > > we
> > > > > do to
> > > > > avoid this additional signal ?
> > > > >
> > > > > Thanks,
> > > > >
> > > > >
> > > > > Raphaël
> > > > >
> > > > > _______________________________________________
> > > > > USRP-users mailing list
> > > > > USRP-users@lists.ettus.com
> > > > >
> >
> > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.
> > > > > com
> > >
> > >
> > > _______________________________________________
> > > USRP-users mailing list
> > > USRP-users@lists.ettus.com
> > > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.
> > > com
> >
> > _______________________________________________
> > USRP-users mailing list
> > USRP-users@lists.ettus.com
> > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.co
> > m
_______________________________________________
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Reply via email to