usrp-users@lists.ettus.com
Discussion and technical support related to USRP, UHD, RFNoC
View all threadsRe: [USRP-users] Random errors while transmitting certain BPSK/QPSK types on x310s
Hi Marcus,
you are correct, reducing the magnitude did seem to fix the issue. I am
wondering though, the gnuradio constellation modulator hier block also
seems to suffer from the same issue and doesn't have any options for
reduced magnitude output - do most people just throw in a downstream block
to reduce the magnitude? Also, is there any reference that speaks a bit
about the scaling issues that can happen with the USRP's or is that the
just type of thing you learn on the fly?
Thanks,
Mike
Date: Thu, 06 Jul 2017 21:43:54 -0400
From: "Marcus D. Leech" mleech@ripnet.com
To: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] Random errors while transmitting certain
BPSK/QPSK types on x310s
Message-ID: 595EE75A.4070805@ripnet.com
Content-Type: text/plain; charset="windows-1252"; Format="flowed"
On 07/06/2017 09:36 PM, Michael Carosino via USRP-users wrote:
A quick update to this question with more info. I did some further
analysis by capturing the received I/Q data from the USRP Source block
when transmitting the BPSK that works without errors (symbols are
0.707+0.707j, -0.707-0.707j) and also when using the BPSK that gives
errors (symbols are +1/-1). You can see in the attached image there is
quite a bit of small magnitude anomalies for the second image (also
errors are probably occurring a bit more frequently than I had
previously estimated).
To me this points to the issue being either something to do with the
USRP or possibly with the TX chain blocks.
With a baseband magnitude near 1, scaling issues/filtering/interpolation
issues can combine to produce distortions.
I usually use a baseband magnitude no greater than 0.9 or so.
On Thu, Jul 6, 2017 at 4:37 PM, Michael Carosino <m.carosino@gmail.com
mailto:m.carosino@gmail.com> wrote:
Hi all,
running Gnuradio 3.7.10.2 and UHD 4.0.0 rfnoc-devel latest commit
(tried earlier versions too). I've got a simple tx/rx flowgraph
going on. The simple description is:
Random input data -> Pack 1 Bit->Chunks to Symbols->Interpolating
FIR Filter->USRP Sink
USRP Source-> Polyphase Clock Sync -> Costas Loop-> Constellation
Receiver->Unpack 1 Bit
I'm transmitting on RF-B TX/RX and receiving on RF-B RX. The
system works almost perfectly, except that there are single bit
errors occurring (not many, maybe every couple of seconds at 500k
samp rate).
Now here's the real strange thing, these errors are ONLY present
if running real BPSK (-1,+1), imaginary BPSK (+j,-j), or rotated
QPSK/QAM (+1j,-1j,+1,-1)
If I use a BPSK having symbols with real and complex parts like
(-.707-.707j, .707+.707j) or QPSK (+/- .707+/- .707j) the errors
are NOT present.
A couple more notes:
Happens if using different x310s or the same for tx/rx.
Happens even if I try to add a small complex value before sending
data to the USRP.
Error always happens as a single bit error (not bursty).
Attached are the constellation plots out of the polyphase clock
sync (PFB) and the costas loop. My guess is the issue is either at
the USRP or the Polyphase Clock Sync block.
Anyone seen something like this before? I'll probably start diving
in the polyphase clock sync code to figure what's going on.
Thanks,
Mike
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
Hi Michael,
point is that having a magnitude too close to 1 can lead to problems in
the FPGA side of DSP – not in GNU Radio.
Best regards,
Marcus
On 07/07/2017 07:20 PM, Michael Carosino via USRP-users wrote:
Hi Marcus,
you are correct, reducing the magnitude did seem to fix the issue. I
am wondering though, the gnuradio constellation modulator hier block
also seems to suffer from the same issue and doesn't have any options
for reduced magnitude output - do most people just throw in a
downstream block to reduce the magnitude? Also, is there any reference
that speaks a bit about the scaling issues that can happen with the
USRP's or is that the just type of thing you learn on the fly?
Thanks,
Mike
Date: Thu, 06 Jul 2017 21:43:54 -0400
From: "Marcus D. Leech" <mleech@ripnet.com <mailto:mleech@ripnet.com>>
To: usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>
Subject: Re: [USRP-users] Random errors while transmitting certain
BPSK/QPSK types on x310s
Message-ID: <595EE75A.4070805@ripnet.com
<mailto:595EE75A.4070805@ripnet.com>>
Content-Type: text/plain; charset="windows-1252"; Format="flowed"
On 07/06/2017 09:36 PM, Michael Carosino via USRP-users wrote:
A quick update to this question with more info. I did some further
analysis by capturing the received I/Q data from the USRP Source
block
when transmitting the BPSK that works without errors (symbols are
0.707+0.707j, -0.707-0.707j) and also when using the BPSK that gives
errors (symbols are +1/-1). You can see in the attached image
there is
quite a bit of small magnitude anomalies for the second image (also
errors are probably occurring a bit more frequently than I had
previously estimated).
To me this points to the issue being either something to do with the
USRP or possibly with the TX chain blocks.
With a baseband magnitude near 1, scaling
issues/filtering/interpolation
issues can combine to produce distortions.
I usually use a baseband magnitude no greater than 0.9 or so.
On Thu, Jul 6, 2017 at 4:37 PM, Michael Carosino
<m.carosino@gmail.com <mailto:m.carosino@gmail.com>
<mailto:m.carosino@gmail.com mailto:m.carosino@gmail.com>> wrote:
Hi all,
running Gnuradio 3.7.10.2 and UHD 4.0.0 rfnoc-devel latest
commit
(tried earlier versions too). I've got a simple tx/rx flowgraph
going on. The simple description is:
Random input data -> Pack 1 Bit->Chunks to
Symbols->Interpolating
FIR Filter->USRP Sink
USRP Source-> Polyphase Clock Sync -> Costas Loop->
Constellation
Receiver->Unpack 1 Bit
I'm transmitting on RF-B TX/RX and receiving on RF-B RX. The
system works almost perfectly, except that there are single bit
errors occurring (not many, maybe every couple of seconds at
500k
samp rate).
Now here's the real strange thing, these errors are ONLY present
if running real BPSK (-1,+1), imaginary BPSK (+j,-j), or rotated
QPSK/QAM (+1j,-1j,+1,-1)
If I use a BPSK having symbols with real and complex parts like
(-.707-.707j, .707+.707j) or QPSK (+/- .707+/- .707j) the errors
are NOT present.
A couple more notes:
Happens if using different x310s or the same for tx/rx.
Happens even if I try to add a small complex value before
sending
data to the USRP.
Error always happens as a single bit error (not bursty).
Attached are the constellation plots out of the polyphase clock
sync (PFB) and the costas loop. My guess is the issue is
either at
the USRP or the Polyphase Clock Sync block.
Anyone seen something like this before? I'll probably start
diving
in the polyphase clock sync code to figure what's going on.
Thanks,
Mike
USRP-users mailing list
USRP-users@lists.ettus.com mailto:USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/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
If the modulator blocks always produce in {-1,+1}, then a downstream
multipier would be required to reduce the baseband magnitude.
It will happen with any radio where there is a translation from floating
point {-1.0,+1.0} into some integer representation on the wire, followed
by interpolation/upconversion in the digital domain on the radio, and
then out the DAC, where it may be driving the mixer a little hard, or
the subsequent RF gain stages.
On 2017-07-07 13:20, Michael Carosino via USRP-users wrote:
Hi Marcus,
you are correct, reducing the magnitude did seem to fix the issue. I am wondering though, the gnuradio constellation modulator hier block also seems to suffer from the same issue and doesn't have any options for reduced magnitude output - do most people just throw in a downstream block to reduce the magnitude? Also, is there any reference that speaks a bit about the scaling issues that can happen with the USRP's or is that the just type of thing you learn on the fly?
Thanks,
Mike
Date: Thu, 06 Jul 2017 21:43:54 -0400
From: "Marcus D. Leech" mleech@ripnet.com
To: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] Random errors while transmitting certain
BPSK/QPSK types on x310s
Message-ID: 595EE75A.4070805@ripnet.com
Content-Type: text/plain; charset="windows-1252"; Format="flowed"
On 07/06/2017 09:36 PM, Michael Carosino via USRP-users wrote:
A quick update to this question with more info. I did some further
analysis by capturing the received I/Q data from the USRP Source block
when transmitting the BPSK that works without errors (symbols are
0.707+0.707j, -0.707-0.707j) and also when using the BPSK that gives
errors (symbols are +1/-1). You can see in the attached image there is
quite a bit of small magnitude anomalies for the second image (also
errors are probably occurring a bit more frequently than I had
previously estimated).
To me this points to the issue being either something to do with the
USRP or possibly with the TX chain blocks.
With a baseband magnitude near 1, scaling issues/filtering/interpolation
issues can combine to produce distortions.
I usually use a baseband magnitude no greater than 0.9 or so.
On Thu, Jul 6, 2017 at 4:37 PM, Michael Carosino <m.carosino@gmail.com
mailto:m.carosino@gmail.com> wrote:
Hi all,
running Gnuradio 3.7.10.2 and UHD 4.0.0 rfnoc-devel latest commit
(tried earlier versions too). I've got a simple tx/rx flowgraph
going on. The simple description is:
Random input data -> Pack 1 Bit->Chunks to Symbols->Interpolating
FIR Filter->USRP Sink
USRP Source-> Polyphase Clock Sync -> Costas Loop-> Constellation
Receiver->Unpack 1 Bit
I'm transmitting on RF-B TX/RX and receiving on RF-B RX. The
system works almost perfectly, except that there are single bit
errors occurring (not many, maybe every couple of seconds at 500k
samp rate).
Now here's the real strange thing, these errors are ONLY present
if running real BPSK (-1,+1), imaginary BPSK (+j,-j), or rotated
QPSK/QAM (+1j,-1j,+1,-1)
If I use a BPSK having symbols with real and complex parts like
(-.707-.707j, .707+.707j) or QPSK (+/- .707+/- .707j) the errors
are NOT present.
A couple more notes:
Happens if using different x310s or the same for tx/rx.
Happens even if I try to add a small complex value before sending
data to the USRP.
Error always happens as a single bit error (not bursty).
Attached are the constellation plots out of the polyphase clock
sync (PFB) and the costas loop. My guess is the issue is either at
the USRP or the Polyphase Clock Sync block.
Anyone seen something like this before? I'll probably start diving
in the polyphase clock sync code to figure what's going on.
Thanks,
Mike
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com [1]
USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
You can change the gain setting on the interpolating FIR filter to
accomplish this.
The QT GUI Histogram Sink can be used to set a gain value that limits
the IQ values to +1/-1.
Ron
On 07/07/2017 10:20 AM, Michael Carosino via USRP-users wrote:
Hi Marcus,
you are correct, reducing the magnitude did seem to fix the issue. I
am wondering though, the gnuradio constellation modulator hier block
also seems to suffer from the same issue and doesn't have any options
for reduced magnitude output - do most people just throw in a
downstream block to reduce the magnitude? Also, is there any reference
that speaks a bit about the scaling issues that can happen with the
USRP's or is that the just type of thing you learn on the fly?
Thanks,
Mike
Date: Thu, 06 Jul 2017 21:43:54 -0400
From: "Marcus D. Leech" <mleech@ripnet.com <mailto:mleech@ripnet.com>>
To: usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>
Subject: Re: [USRP-users] Random errors while transmitting certain
BPSK/QPSK types on x310s
Message-ID: <595EE75A.4070805@ripnet.com
<mailto:595EE75A.4070805@ripnet.com>>
Content-Type: text/plain; charset="windows-1252"; Format="flowed"
On 07/06/2017 09:36 PM, Michael Carosino via USRP-users wrote:
A quick update to this question with more info. I did some further
analysis by capturing the received I/Q data from the USRP Source
block
when transmitting the BPSK that works without errors (symbols are
0.707+0.707j, -0.707-0.707j) and also when using the BPSK that gives
errors (symbols are +1/-1). You can see in the attached image
there is
quite a bit of small magnitude anomalies for the second image (also
errors are probably occurring a bit more frequently than I had
previously estimated).
To me this points to the issue being either something to do with the
USRP or possibly with the TX chain blocks.
With a baseband magnitude near 1, scaling
issues/filtering/interpolation
issues can combine to produce distortions.
I usually use a baseband magnitude no greater than 0.9 or so.
On Thu, Jul 6, 2017 at 4:37 PM, Michael Carosino
<m.carosino@gmail.com <mailto:m.carosino@gmail.com>
<mailto:m.carosino@gmail.com mailto:m.carosino@gmail.com>> wrote:
Hi all,
running Gnuradio 3.7.10.2 and UHD 4.0.0 rfnoc-devel latest
commit
(tried earlier versions too). I've got a simple tx/rx flowgraph
going on. The simple description is:
Random input data -> Pack 1 Bit->Chunks to
Symbols->Interpolating
FIR Filter->USRP Sink
USRP Source-> Polyphase Clock Sync -> Costas Loop->
Constellation
Receiver->Unpack 1 Bit
I'm transmitting on RF-B TX/RX and receiving on RF-B RX. The
system works almost perfectly, except that there are single bit
errors occurring (not many, maybe every couple of seconds at
500k
samp rate).
Now here's the real strange thing, these errors are ONLY present
if running real BPSK (-1,+1), imaginary BPSK (+j,-j), or rotated
QPSK/QAM (+1j,-1j,+1,-1)
If I use a BPSK having symbols with real and complex parts like
(-.707-.707j, .707+.707j) or QPSK (+/- .707+/- .707j) the errors
are NOT present.
A couple more notes:
Happens if using different x310s or the same for tx/rx.
Happens even if I try to add a small complex value before
sending
data to the USRP.
Error always happens as a single bit error (not bursty).
Attached are the constellation plots out of the polyphase clock
sync (PFB) and the costas loop. My guess is the issue is
either at
the USRP or the Polyphase Clock Sync block.
Anyone seen something like this before? I'll probably start
diving
in the polyphase clock sync code to figure what's going on.
Thanks,
Mike
USRP-users mailing list
USRP-users@lists.ettus.com mailto:USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/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
I've created an OOT module with a little utility I use to print peak IQ
levels. This allows for precise setting of the gain to insure IQ levels
are within +1/-1.
https://github.com/drmpeg/gr-iqlevels
Ron
On 07/07/2017 11:28 AM, Ron Economos via USRP-users wrote:
You can change the gain setting on the interpolating FIR filter to
accomplish this.
The QT GUI Histogram Sink can be used to set a gain value that limits
the IQ values to +1/-1.
Ron
On 07/07/2017 10:20 AM, Michael Carosino via USRP-users wrote:
Hi Marcus,
you are correct, reducing the magnitude did seem to fix the issue. I
am wondering though, the gnuradio constellation modulator hier block
also seems to suffer from the same issue and doesn't have any options
for reduced magnitude output - do most people just throw in a
downstream block to reduce the magnitude? Also, is there any
reference that speaks a bit about the scaling issues that can happen
with the USRP's or is that the just type of thing you learn on the fly?
Thanks,
Mike
Date: Thu, 06 Jul 2017 21:43:54 -0400
From: "Marcus D. Leech" <mleech@ripnet.com
<mailto:mleech@ripnet.com>>
To: usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>
Subject: Re: [USRP-users] Random errors while transmitting certain
BPSK/QPSK types on x310s
Message-ID: <595EE75A.4070805@ripnet.com
<mailto:595EE75A.4070805@ripnet.com>>
Content-Type: text/plain; charset="windows-1252"; Format="flowed"
On 07/06/2017 09:36 PM, Michael Carosino via USRP-users wrote:
A quick update to this question with more info. I did some further
analysis by capturing the received I/Q data from the USRP
Source block
when transmitting the BPSK that works without errors (symbols are
0.707+0.707j, -0.707-0.707j) and also when using the BPSK that
gives
errors (symbols are +1/-1). You can see in the attached image
there is
quite a bit of small magnitude anomalies for the second image (also
errors are probably occurring a bit more frequently than I had
previously estimated).
To me this points to the issue being either something to do
with the
USRP or possibly with the TX chain blocks.
With a baseband magnitude near 1, scaling
issues/filtering/interpolation
issues can combine to produce distortions.
I usually use a baseband magnitude no greater than 0.9 or so.
On Thu, Jul 6, 2017 at 4:37 PM, Michael Carosino
<m.carosino@gmail.com <mailto:m.carosino@gmail.com>
<mailto:m.carosino@gmail.com mailto:m.carosino@gmail.com>> wrote:
Hi all,
running Gnuradio 3.7.10.2 and UHD 4.0.0 rfnoc-devel latest
commit
(tried earlier versions too). I've got a simple tx/rx flowgraph
going on. The simple description is:
Random input data -> Pack 1 Bit->Chunks to
Symbols->Interpolating
FIR Filter->USRP Sink
USRP Source-> Polyphase Clock Sync -> Costas Loop->
Constellation
Receiver->Unpack 1 Bit
I'm transmitting on RF-B TX/RX and receiving on RF-B RX. The
system works almost perfectly, except that there are single bit
errors occurring (not many, maybe every couple of seconds
at 500k
samp rate).
Now here's the real strange thing, these errors are ONLY
present
if running real BPSK (-1,+1), imaginary BPSK (+j,-j), or
rotated
QPSK/QAM (+1j,-1j,+1,-1)
If I use a BPSK having symbols with real and complex parts like
(-.707-.707j, .707+.707j) or QPSK (+/- .707+/- .707j) the
errors
are NOT present.
A couple more notes:
Happens if using different x310s or the same for tx/rx.
Happens even if I try to add a small complex value before
sending
data to the USRP.
Error always happens as a single bit error (not bursty).
Attached are the constellation plots out of the polyphase clock
sync (PFB) and the costas loop. My guess is the issue is
either at
the USRP or the Polyphase Clock Sync block.
Anyone seen something like this before? I'll probably start
diving
in the polyphase clock sync code to figure what's going on.
Thanks,
Mike
USRP-users mailing list
USRP-users@lists.ettus.com mailto:USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/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.com