usrp-users@lists.ettus.com
Discussion and technical support related to USRP, UHD, RFNoC
View all threadsRe: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I'll work on the script and post it when it's done.
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 03, 2012 3:29 AM
To: Marc Erickson; Mike McLernon
Cc: usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc and Mike,
Further to my last message, could you please provide an equivalent example of the Simulink model that you posted earlier but this time written in MATLAB.
Thanks,
Stan
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: 27 July 2012 15:43
To: 'Mike McLernon'
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Mike,
Thanks for the responses to my questions and thanks too for the Simulink model. The latter has proved to be quite useful. To experiment with the model, I connected together the RF1 and RF2 ports of my N210 (+SBX) via a 30 dB fixed attenuator and a steppable attenuator (0-10 + 0/10/20/30/40 dB). I also modified the model and replaced the switch with a slider that controls the center frequency of both the Tx and Rx blocks, simultaneously.
The model and hardware configuration described above allows me to set fc and the attenuation in the RF path. However, the behaviour of the output (Rx) spectrum is not quite as I would have expected--instead of the amplitude of the peak changing when the RF attenuator is adjusted, it appears that the level of what could be called the noise floor of the output changes instead. Could this be due to a numerical effect in the Simulink FFT block that holds the "peak" but instead changes the "floor"? Or has it something to do with the Ettus radio--perhaps the result of some automatic gain control (AGC) circuitry?
Regardless of all of the above, I would be very happy to experiment and to study the complementary MATLAB scripts that are used to construct a system similar to the Simulink model that you so kindly posted to this forum.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 20 July 2012 18:32
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
Responses interleaved below . . .
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, July 20, 2012 6:57 AM
To: Mike McLernon
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Mike,
With reference to the schematic of the Simulink model "mUSRPFullDuplex" that you posted earlier, could you please reveal what is in (or under) the block labeled "Receiver Processing"?
Hopefully you received my mail from earlier today with several models attached. If not, let me know.
Referring to item 5 in the list "experiment setup" (in which the SDRu Tx and SDRu Rx blocks are in the same model) would the system not perform full-duplex operation if two models were used?
If two models are used, and they are both using the same IP, they should still be performing full duplex. I have not run that case.
As a general question, could you please comment on how to configure the RF ports of the SBX board in which an RF switch is used to configure RF1 for either Tx or Rx.
Our R2011a and R2011b release notes state that the default antenna configurations cannot be changed. For my particular application, I did not need to perform any antenna configuration.
Finally, it would be very much appreciated if the sample Simulink model(s) could be shared in this community and it would be even more appreciated if they were to be complemented with equivalent MATLAB files. Such a system would be a useful example to include in future releases from the MathWorks.
Let me know if the models I sent earlier are helpful to you. We would want to include such models (and complementary MATLAB scripts) in future releases.
Hth,
Mike
Thanks in advance,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 19 July 2012 17:06
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I did the following experiment to ascertain our level of full duplex support:
Experiment Setup
- N210/SBX
- VERT900 antennas for both tx and rx
- Win64 on a 12-core development machine
- One MATLAB session, one Simulink model
- SDRu Tx block and SDRu Rx block in same model
- Model runs in accelerator mode
- Spectrum Scope blocks for tx and rx waveforms, a Simulink Scope to monitor rx overruns
- Interpolation/decimation factor of 512 (maximum allowed)
- Tx center frequency = 915 MHz
- Rx center frequency = 915 or 915.04 MHz (controlled with a manual switch)
- Optimizations:
* Use singles rather than doubles
* Use accelerator rather than normal mode
* Turned on compiler optimizations
* Turned off all data logging
Results
- Rx Spectrum Scope shows tones where they should be
- Simulink Scope shows no overruns
- When the SDRu Receiver block's center frequency is changed with the manual switch shortly after the simulation starts, the rx Spectrum Scope shows the change quickly.
- As the simulation continues, it takes progressively longer for the change to show in the Spectrum Scope.
* With one scope removed, the change is reflected more quickly, but it still takes progressively longer as the simulation continues.
* Conclusion is that a data buffer is slowly filling up and taking more time to empty out.
* If the tx Spectrum Scope were removed, the model would run faster, and the frequency changes would show more quickly.
I've attached images of the model, the block dialogs, the tx and rx spectra, and the time plot of the overrun signal with this message. Please let us know if you have additional questions.
Best,
Mike
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Sunday, July 15, 2012 6:14 AM
To: Marc Erickson; usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc,
Thanks for the kind offer to provide support.
In order to continue this thread within the community, I would like to respond by repeating my previous request:
Could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
With many thanks,
Stan
From: Marc Erickson [mailto:Marc.Erickson@mathworks.com]
Sent: 05 July 2012 19:19
To: Stan Gamla; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Stan,
Sorry for the confusion about our support. We have stated some limitations of our support in the product release notes which are available in the download package. Type 'help sdru' and follow the 'Help and Examples' link. Then click on the link under 'Release Notes'. The relevant limitation is stated as follows:
"Communications with USRP(tm) only supports half-duplex applications. MATLAB and Simulink applications must be receive-only or transmit-only. Full-duplex is not supported."
Our package has separate Rx and Tx blocks (for Simulink) and separate Rx and Tx System objects (for MATLAB) with their own configuration parameters. We allow binding the Rx and Tx to the same board in order to support some duplex operations. Full duplex operation might be possible when adhering to some modeling constraints.
Please feel free to contact us directly to discuss any use-cases and specific applications you might be interested in pursuing and we'd be happy to work with you.
Regards,
Marc
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Friday, June 29, 2012 7:03 AM
To: usrp-users@lists.ettus.com
Subject: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
As far as I understand, the Ettus SBX RF board comprises a full-duplex transceiver with separate receive and transmit chains, each having its own local oscillator.
https://www.ettus.com/product#tabs-3 states that:
"The SBX is a wide bandwidth transceiver that provides up to 100 mW of output power, and a typical noise figure of 5 dB. The local oscillators for the receive and transmit chains operate independently, which allows dual-band operation. The SBX is MIMO capable, and provides 40 MHz of bandwidth. The SBX is ideal for applications requiring access to a variety of bands in the 400 MHz-4400 MHz range. Example application areas include WiFi, WiMax, S-band transceivers and 2.4 GHz ISM band transceivers."
Further, according to their website "The MathWorks USRP(r) Hardware Support from [sic] MATLAB and Simulink is compatible with UHD(tm)-based Ettus radios".
http://www.mathworks.com/discovery/sdr/usrp.html viz:
"MATLAB and Simulink support for USRP(r) devices has been tested on N210 and USRP2(r) devices with WBX, SBX, XCVR2450, TVRX, TVRX2, LFRX, LFTX, DBSRX, and DBSRX2 daughterboards. Support may also extend to other UHD(tm)-based radios and daughterboards from National Instruments(r)."
Are both of these statements correct, I wonder?
If The MathWorks is tuned in to this channel, could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
On the other hand, if I have misunderstood what is meant by the word "supported" could either Ettus or The MathWorks comment please.
Thanks,
Stan
Hi Mike,
Sorry to ask again... but I would very much appreciate the MATLAB example
file.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 03 August 2012 14:24
To: Stan Gamla
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
I'll work on the script and post it when it's done.
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 03, 2012 3:29 AM
To: Marc Erickson; Mike McLernon
Cc: usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Marc and Mike,
Further to my last message, could you please provide an equivalent example
of the Simulink model that you posted earlier but this time written in
MATLAB.
Thanks,
Stan
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: 27 July 2012 15:43
To: 'Mike McLernon'
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Mike,
Thanks for the responses to my questions and thanks too for the Simulink
model. The latter has proved to be quite useful. To experiment with the
model, I connected together the RF1 and RF2 ports of my N210 (+SBX) via a 30
dB fixed attenuator and a steppable attenuator (0-10 + 0/10/20/30/40 dB). I
also modified the model and replaced the switch with a slider that controls
the center frequency of both the Tx and Rx blocks, simultaneously.
The model and hardware configuration described above allows me to set fc and
the attenuation in the RF path. However, the behaviour of the output (Rx)
spectrum is not quite as I would have expected--instead of the amplitude of
the peak changing when the RF attenuator is adjusted, it appears that the
level of what could be called the noise floor of the output changes instead.
Could this be due to a numerical effect in the Simulink FFT block that holds
the "peak" but instead changes the "floor"? Or has it something to do with
the Ettus radio--perhaps the result of some automatic gain control (AGC)
circuitry?
Regardless of all of the above, I would be very happy to experiment and to
study the complementary MATLAB scripts that are used to construct a system
similar to the Simulink model that you so kindly posted to this forum.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 20 July 2012 18:32
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
Responses interleaved below . . .
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, July 20, 2012 6:57 AM
To: Mike McLernon
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Mike,
With reference to the schematic of the Simulink model "mUSRPFullDuplex" that
you posted earlier, could you please reveal what is in (or under) the block
labeled "Receiver Processing"?
Hopefully you received my mail from earlier today with several models
attached. If not, let me know.
Referring to item 5 in the list "experiment setup" (in which the SDRu Tx and
SDRu Rx blocks are in the same model) would the system not perform
full-duplex operation if two models were used?
If two models are used, and they are both using the same IP, they should
still be performing full duplex. I have not run that case.
As a general question, could you please comment on how to configure the RF
ports of the SBX board in which an RF switch is used to configure RF1 for
either Tx or Rx.
Our R2011a and R2011b release notes state that the default antenna
configurations cannot be changed. For my particular application, I did not
need to perform any antenna configuration.
Finally, it would be very much appreciated if the sample Simulink model(s)
could be shared in this community and it would be even more appreciated if
they were to be complemented with equivalent MATLAB files. Such a system
would be a useful example to include in future releases from the MathWorks.
Let me know if the models I sent earlier are helpful to you. We would want
to include such models (and complementary MATLAB scripts) in future
releases.
Hth,
Mike
Thanks in advance,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 19 July 2012 17:06
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
I did the following experiment to ascertain our level of full duplex
support:
Experiment Setup
-
N210/SBX
-
VERT900 antennas for both tx and rx
-
Win64 on a 12-core development machine
-
One MATLAB session, one Simulink model
-
SDRu Tx block and SDRu Rx block in same model
-
Model runs in accelerator mode
-
Spectrum Scope blocks for tx and rx waveforms, a Simulink Scope to
monitor rx overruns -
Interpolation/decimation factor of 512 (maximum allowed)
-
Tx center frequency = 915 MHz
-
Rx center frequency = 915 or 915.04 MHz (controlled with a manual
switch) -
Optimizations:
-
Use singles rather than doubles
-
Use accelerator rather than normal mode
-
Turned on compiler optimizations
-
Turned off all data logging
Results
-
Rx Spectrum Scope shows tones where they should be
-
Simulink Scope shows no overruns
-
When the SDRu Receiver block's center frequency is changed with the
manual switch shortly after the simulation starts, the rx Spectrum Scope
shows the change quickly. -
As the simulation continues, it takes progressively longer for the
change to show in the Spectrum Scope.
-
With one scope removed, the change is reflected more quickly, but it
still takes progressively longer as the simulation continues. -
Conclusion is that a data buffer is slowly filling up and taking
more time to empty out. -
If the tx Spectrum Scope were removed, the model would run faster,
and the frequency changes would show more quickly.
I've attached images of the model, the block dialogs, the tx and rx spectra,
and the time plot of the overrun signal with this message. Please let us
know if you have additional questions.
Best,
Mike
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Sunday, July 15, 2012 6:14 AM
To: Marc Erickson; usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Marc,
Thanks for the kind offer to provide support.
In order to continue this thread within the community, I would like to
respond by repeating my previous request:
Could you please comment and perhaps provide an illustration of how to
configure the N210/SBX using both MATLAB and Simulink so that a signal is
transmitted from the RF1 port and received by the RF2. As a suggestion,
slider controls could be used in Simulink to independently adjust the RF
settings of both the transmitter and receiver to include at least gain and
center frequency. A spectrum of the received signal should be displayed.
With many thanks,
Stan
From: Marc Erickson [mailto:Marc.Erickson@mathworks.com]
Sent: 05 July 2012 19:19
To: Stan Gamla; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Stan,
Sorry for the confusion about our support. We have stated some limitations
of our support in the product release notes which are available in the
download package. Type 'help sdru' and follow the 'Help and Examples' link.
Then click on the link under 'Release Notes'. The relevant limitation is
stated as follows:
"Communications with USRPT only supports half-duplex applications. MATLAB
and Simulink applications must be receive-only or transmit-only. Full-duplex
is not supported."
Our package has separate Rx and Tx blocks (for Simulink) and separate Rx and
Tx System objects (for MATLAB) with their own configuration parameters. We
allow binding the Rx and Tx to the same board in order to support some
duplex operations. Full duplex operation might be possible when adhering to
some modeling constraints.
Please feel free to contact us directly to discuss any use-cases and
specific applications you might be interested in pursuing and we'd be happy
to work with you.
Regards,
Marc
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Friday, June 29, 2012 7:03 AM
To: usrp-users@lists.ettus.com
Subject: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
As far as I understand, the Ettus SBX RF board comprises a full-duplex
transceiver with separate receive and transmit chains, each having its own
local oscillator.
https://www.ettus.com/product#tabs-3 states that:
"The SBX is a wide bandwidth transceiver that provides up to 100 mW of
output power, and a typical noise figure of 5 dB. The local oscillators for
the receive and transmit chains operate independently, which allows
dual-band operation. The SBX is MIMO capable, and provides 40 MHz of
bandwidth. The SBX is ideal for applications requiring access to a variety
of bands in the 400 MHz-4400 MHz range. Example application areas include
WiFi, WiMax, S-band transceivers and 2.4 GHz ISM band transceivers."
Further, according to their website "The MathWorks USRPR Hardware Support
from [sic] MATLAB and Simulink is compatible with UHDT-based Ettus radios".
http://www.mathworks.com/discovery/sdr/usrp.html viz:
"MATLAB and Simulink support for USRPR devices has been tested on N210 and
USRP2R devices with WBX, SBX, XCVR2450, TVRX, TVRX2, LFRX, LFTX, DBSRX, and
DBSRX2 daughterboards. Support may also extend to other UHDT-based radios
and daughterboards from National InstrumentsR."
Are both of these statements correct, I wonder?
If The MathWorks is tuned in to this channel, could you please comment and
perhaps provide an illustration of how to configure the N210/SBX using both
MATLAB and Simulink so that a signal is transmitted from the RF1 port and
received by the RF2. As a suggestion, slider controls could be used in
Simulink to independently adjust the RF settings of both the transmitter and
receiver to include at least gain and center frequency. A spectrum of the
received signal should be displayed.
On the other hand, if I have misunderstood what is meant by the word
"supported" could either Ettus or The MathWorks comment please.
Thanks,
Stan
Hi Stan,
Still working on it - I have something going for a one-radio setup, but I'd like to prove out a two-radio configuration.
Thanks for your patience,
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 10, 2012 8:38 AM
To: Mike McLernon
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Mike,
Sorry to ask again... but I would very much appreciate the MATLAB example file.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 03 August 2012 14:24
To: Stan Gamla
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I'll work on the script and post it when it's done.
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 03, 2012 3:29 AM
To: Marc Erickson; Mike McLernon
Cc: usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc and Mike,
Further to my last message, could you please provide an equivalent example of the Simulink model that you posted earlier but this time written in MATLAB.
Thanks,
Stan
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: 27 July 2012 15:43
To: 'Mike McLernon'
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Mike,
Thanks for the responses to my questions and thanks too for the Simulink model. The latter has proved to be quite useful. To experiment with the model, I connected together the RF1 and RF2 ports of my N210 (+SBX) via a 30 dB fixed attenuator and a steppable attenuator (0-10 + 0/10/20/30/40 dB). I also modified the model and replaced the switch with a slider that controls the center frequency of both the Tx and Rx blocks, simultaneously.
The model and hardware configuration described above allows me to set fc and the attenuation in the RF path. However, the behaviour of the output (Rx) spectrum is not quite as I would have expected--instead of the amplitude of the peak changing when the RF attenuator is adjusted, it appears that the level of what could be called the noise floor of the output changes instead. Could this be due to a numerical effect in the Simulink FFT block that holds the "peak" but instead changes the "floor"? Or has it something to do with the Ettus radio--perhaps the result of some automatic gain control (AGC) circuitry?
Regardless of all of the above, I would be very happy to experiment and to study the complementary MATLAB scripts that are used to construct a system similar to the Simulink model that you so kindly posted to this forum.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 20 July 2012 18:32
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
Responses interleaved below . . .
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, July 20, 2012 6:57 AM
To: Mike McLernon
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Mike,
With reference to the schematic of the Simulink model "mUSRPFullDuplex" that you posted earlier, could you please reveal what is in (or under) the block labeled "Receiver Processing"?
Hopefully you received my mail from earlier today with several models attached. If not, let me know.
Referring to item 5 in the list "experiment setup" (in which the SDRu Tx and SDRu Rx blocks are in the same model) would the system not perform full-duplex operation if two models were used?
If two models are used, and they are both using the same IP, they should still be performing full duplex. I have not run that case.
As a general question, could you please comment on how to configure the RF ports of the SBX board in which an RF switch is used to configure RF1 for either Tx or Rx.
Our R2011a and R2011b release notes state that the default antenna configurations cannot be changed. For my particular application, I did not need to perform any antenna configuration.
Finally, it would be very much appreciated if the sample Simulink model(s) could be shared in this community and it would be even more appreciated if they were to be complemented with equivalent MATLAB files. Such a system would be a useful example to include in future releases from the MathWorks.
Let me know if the models I sent earlier are helpful to you. We would want to include such models (and complementary MATLAB scripts) in future releases.
Hth,
Mike
Thanks in advance,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 19 July 2012 17:06
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I did the following experiment to ascertain our level of full duplex support:
Experiment Setup
- N210/SBX
- VERT900 antennas for both tx and rx
- Win64 on a 12-core development machine
- One MATLAB session, one Simulink model
- SDRu Tx block and SDRu Rx block in same model
- Model runs in accelerator mode
- Spectrum Scope blocks for tx and rx waveforms, a Simulink Scope to monitor rx overruns
- Interpolation/decimation factor of 512 (maximum allowed)
- Tx center frequency = 915 MHz
- Rx center frequency = 915 or 915.04 MHz (controlled with a manual switch)
- Optimizations:
* Use singles rather than doubles
* Use accelerator rather than normal mode
* Turned on compiler optimizations
* Turned off all data logging
Results
- Rx Spectrum Scope shows tones where they should be
- Simulink Scope shows no overruns
- When the SDRu Receiver block's center frequency is changed with the manual switch shortly after the simulation starts, the rx Spectrum Scope shows the change quickly.
- As the simulation continues, it takes progressively longer for the change to show in the Spectrum Scope.
* With one scope removed, the change is reflected more quickly, but it still takes progressively longer as the simulation continues.
* Conclusion is that a data buffer is slowly filling up and taking more time to empty out.
* If the tx Spectrum Scope were removed, the model would run faster, and the frequency changes would show more quickly.
I've attached images of the model, the block dialogs, the tx and rx spectra, and the time plot of the overrun signal with this message. Please let us know if you have additional questions.
Best,
Mike
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Sunday, July 15, 2012 6:14 AM
To: Marc Erickson; usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc,
Thanks for the kind offer to provide support.
In order to continue this thread within the community, I would like to respond by repeating my previous request:
Could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
With many thanks,
Stan
From: Marc Erickson [mailto:Marc.Erickson@mathworks.com]
Sent: 05 July 2012 19:19
To: Stan Gamla; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Stan,
Sorry for the confusion about our support. We have stated some limitations of our support in the product release notes which are available in the download package. Type 'help sdru' and follow the 'Help and Examples' link. Then click on the link under 'Release Notes'. The relevant limitation is stated as follows:
"Communications with USRP(tm) only supports half-duplex applications. MATLAB and Simulink applications must be receive-only or transmit-only. Full-duplex is not supported."
Our package has separate Rx and Tx blocks (for Simulink) and separate Rx and Tx System objects (for MATLAB) with their own configuration parameters. We allow binding the Rx and Tx to the same board in order to support some duplex operations. Full duplex operation might be possible when adhering to some modeling constraints.
Please feel free to contact us directly to discuss any use-cases and specific applications you might be interested in pursuing and we'd be happy to work with you.
Regards,
Marc
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Friday, June 29, 2012 7:03 AM
To: usrp-users@lists.ettus.com
Subject: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
As far as I understand, the Ettus SBX RF board comprises a full-duplex transceiver with separate receive and transmit chains, each having its own local oscillator.
https://www.ettus.com/product#tabs-3 states that:
"The SBX is a wide bandwidth transceiver that provides up to 100 mW of output power, and a typical noise figure of 5 dB. The local oscillators for the receive and transmit chains operate independently, which allows dual-band operation. The SBX is MIMO capable, and provides 40 MHz of bandwidth. The SBX is ideal for applications requiring access to a variety of bands in the 400 MHz-4400 MHz range. Example application areas include WiFi, WiMax, S-band transceivers and 2.4 GHz ISM band transceivers."
Further, according to their website "The MathWorks USRP(r) Hardware Support from [sic] MATLAB and Simulink is compatible with UHD(tm)-based Ettus radios".
http://www.mathworks.com/discovery/sdr/usrp.html viz:
"MATLAB and Simulink support for USRP(r) devices has been tested on N210 and USRP2(r) devices with WBX, SBX, XCVR2450, TVRX, TVRX2, LFRX, LFTX, DBSRX, and DBSRX2 daughterboards. Support may also extend to other UHD(tm)-based radios and daughterboards from National Instruments(r)."
Are both of these statements correct, I wonder?
If The MathWorks is tuned in to this channel, could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
On the other hand, if I have misunderstood what is meant by the word "supported" could either Ettus or The MathWorks comment please.
Thanks,
Stan
Hello all,
Could someone please point me in the right direction or perhaps provide some
guidance regarding the setting of signal levels using Simulink's SDRu
Transmitter and SDRu Receiver blocks, ideally in connexion with the Ettus
SBX RF board.
Through experimentation with a closed-loop test bench (in which the RF1 and
RF2 ports of the N210 were connected together via a variable RF attenuator
and a CW tone was transmitted from one and received by the other) I have
observed that for similar gain settings of the Simulink Tx and Rx blocks and
for similar settings of the RF attenuator, the signal level observed in the
baseband spectrum varies with frequency. This might suggest that either the
Tx or the Rx sections of the radio board (or both) exhibit some form of
frequency dependent behaviour.
In the following set of plots, I have used the Simulink model
"mUSRPFullDuplex", kindly provided in an earlier post by The MathWorks, in
which the Tx and Rx blocks were configured as follows:
SDRu transmitter
LO offset (Hz) = 0
Gain (dB0 = 30
Interpolation = 512
Burst mode disabled
SDRu receiver
LO offset (Hz) = 0
Gain (dB) = 32
Decimation = 512
Sample time = 1/(10036/512)
Output data type = single
Frame length = 1024
Burst mode disabled
Figures 1-4 show the baseband output spectrum in which a Tx and Rx centre
frequency of 360 MHz was used and the RF attenuation was set to 36, 46, 56
and 66 dB respectively.
Figure 1: fc = 360 MHz, attenuation = 36 dB
Figure 2: fc = 360 MHz, attenuation = 46 dB
Figure 3: fc = 360 MHz, attenuation = 56 dB
Figure 4: fc = 360 MHz, attenuation = 66 dB
Similarly, figures 5-8 show the baseband output spectrum in which a Tx and
Rx centre frequency of 370 MHz was used and the RF attenuation was set to
36, 46, 56 and 66 dB respectively.
Figure 5: fc = 370 MHz, attenuation = 36 dB
Figure 6: fc = 370 MHz, attenuation = 46 dB
Figure 7: fc = 370 MHz, attenuation = 56 dB
Figure 8: fc = 370 MHz, attenuation = 66 dB
Figures 1 thru 4 show that the output level is reduced by approximately 10
dB with each 10 dB increment of the RF attenuation with the exception of the
last 10 dB step (Figures 3 and 4) in which the output level changes by 25
dB. Why?
Figures 5 thru 8 show that with a slightly increased carrier frequency of 10
MHz, the output level changes by approximately 10 dB for each 10 dB
increment of the RF attenuator.
However, whereas the noise floor in Figures 1 thru 4 is maintained at a
level of approximately -140 dB, it varies in Figures 5 thru 8 from -120 to
-145 dB. Why?
Further measurements were made in which the SDRu transmitter gain was
reduced to 0 dB and all other settings were kept constant. The results are
recorded in the table below in which each column relates to a given value of
RF attenuation:
40 dB 50 dB 60 dB 70 dB
f_c [MHz] Output level [dB]
400 -32 -42 -54 -56
600 -34 -45 -48 -47
800 -33 -39 -41 -42
1000 -38 -48 -46 -46
1200 -37 -43 -46 -47
1400 -46 -51 -47 -47
1600 -41 -46 -48 -48
1800 -42 -46 -47 -48
2000 -62 -50 -47 -47
2200 -43 -46 -46 -47
2400 -48 -46 -46 -46
The value observed at 200 MHz (40 dB RF attenuation) was repeatable.
A JFW 50DR-0446 DC - 2500 MHz dual rotary attenuator was used for the
experiments. Its performance against specification was checked using a
vector network analyzer and was found to conform.
In addition to the questions presented above, as a summary I wish to ask:
-
What is the maximum RF output level of the Tx section of the SBX RF
board? -
What is the minimum RF output level of the Tx section of the SBX RF
board? -
How does the above relate to SDRu transmitter gain and to the
amplitude of the input vector? -
What is the maximum RF input level of the Rx section of the SBX RF
board? -
What is the minimum RF input level of the Rx section of the SBX RF
board? -
How does the above relate to SDRu receiver gain and to the value of
output vector returned by the block?
Thanks,
Stan
On 08/13/2012 11:00 AM, Stan Gamla wrote:
Hello all,
Could someone please point me in the right direction or perhaps provide some
guidance regarding the setting of signal levels using Simulink's SDRu
Transmitter and SDRu Receiver blocks, ideally in connexion with the Ettus
SBX RF board.
Careful, the SBX is not spec'd to tune below 400 Mhz. The software will
not let the LO be tuned below that.
The units of the TX and RX samples are nothing more than counts where
+/-1.0 is equivalent to the ADC/DAC fullscale value.
The gain adjustments are in dB and should be very linear for SBX.
There isnt a calibrated power level, so it would be worth measuring a
test tone for a given frontend configuration.
A future UHD feature will be roughly calibrated power levels so samples
can be in volts.
-josh
Hi Mike and Marc,
Using the MathWorks example file "sdruFRSGMRSTransmitter", I have attempted
to create the example that I described below but have been far from
successful. Therefore, without wishing to become a nuisance, could you
please let me know when I can look forward to receiving a MATLAB version of
the Simulink model that you sent earlier?
With many thanks,
Stan
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Mike McLernon
Sent: 10 August 2012 14:41
To: Stan Gamla
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
Still working on it - I have something going for a one-radio setup, but I'd
like to prove out a two-radio configuration.
Thanks for your patience,
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 10, 2012 8:38 AM
To: Mike McLernon
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Mike,
Sorry to ask again... but I would very much appreciate the MATLAB example
file.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 03 August 2012 14:24
To: Stan Gamla
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
I'll work on the script and post it when it's done.
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 03, 2012 3:29 AM
To: Marc Erickson; Mike McLernon
Cc: usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Marc and Mike,
Further to my last message, could you please provide an equivalent example
of the Simulink model that you posted earlier but this time written in
MATLAB.
Thanks,
Stan
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: 27 July 2012 15:43
To: 'Mike McLernon'
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Mike,
Thanks for the responses to my questions and thanks too for the Simulink
model. The latter has proved to be quite useful. To experiment with the
model, I connected together the RF1 and RF2 ports of my N210 (+SBX) via a 30
dB fixed attenuator and a steppable attenuator (0-10 + 0/10/20/30/40 dB). I
also modified the model and replaced the switch with a slider that controls
the center frequency of both the Tx and Rx blocks, simultaneously.
The model and hardware configuration described above allows me to set fc and
the attenuation in the RF path. However, the behaviour of the output (Rx)
spectrum is not quite as I would have expected--instead of the amplitude of
the peak changing when the RF attenuator is adjusted, it appears that the
level of what could be called the noise floor of the output changes instead.
Could this be due to a numerical effect in the Simulink FFT block that holds
the "peak" but instead changes the "floor"? Or has it something to do with
the Ettus radio--perhaps the result of some automatic gain control (AGC)
circuitry?
Regardless of all of the above, I would be very happy to experiment and to
study the complementary MATLAB scripts that are used to construct a system
similar to the Simulink model that you so kindly posted to this forum.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 20 July 2012 18:32
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
Responses interleaved below . . .
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, July 20, 2012 6:57 AM
To: Mike McLernon
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Mike,
With reference to the schematic of the Simulink model "mUSRPFullDuplex" that
you posted earlier, could you please reveal what is in (or under) the block
labeled "Receiver Processing"?
Hopefully you received my mail from earlier today with several models
attached. If not, let me know.
Referring to item 5 in the list "experiment setup" (in which the SDRu Tx and
SDRu Rx blocks are in the same model) would the system not perform
full-duplex operation if two models were used?
If two models are used, and they are both using the same IP, they should
still be performing full duplex. I have not run that case.
As a general question, could you please comment on how to configure the RF
ports of the SBX board in which an RF switch is used to configure RF1 for
either Tx or Rx.
Our R2011a and R2011b release notes state that the default antenna
configurations cannot be changed. For my particular application, I did not
need to perform any antenna configuration.
Finally, it would be very much appreciated if the sample Simulink model(s)
could be shared in this community and it would be even more appreciated if
they were to be complemented with equivalent MATLAB files. Such a system
would be a useful example to include in future releases from the MathWorks.
Let me know if the models I sent earlier are helpful to you. We would want
to include such models (and complementary MATLAB scripts) in future
releases.
Hth,
Mike
Thanks in advance,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 19 July 2012 17:06
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
Hi Stan,
I did the following experiment to ascertain our level of full duplex
support:
Experiment Setup
-
N210/SBX
-
VERT900 antennas for both tx and rx
-
Win64 on a 12-core development machine
-
One MATLAB session, one Simulink model
-
SDRu Tx block and SDRu Rx block in same model
-
Model runs in accelerator mode
-
Spectrum Scope blocks for tx and rx waveforms, a Simulink Scope to
monitor rx overruns -
Interpolation/decimation factor of 512 (maximum allowed)
-
Tx center frequency = 915 MHz
-
Rx center frequency = 915 or 915.04 MHz (controlled with a manual
switch) -
Optimizations:
-
Use singles rather than doubles
-
Use accelerator rather than normal mode
-
Turned on compiler optimizations
-
Turned off all data logging
Results
-
Rx Spectrum Scope shows tones where they should be
-
Simulink Scope shows no overruns
-
When the SDRu Receiver block's center frequency is changed with the
manual switch shortly after the simulation starts, the rx Spectrum Scope
shows the change quickly. -
As the simulation continues, it takes progressively longer for the
change to show in the Spectrum Scope.
-
With one scope removed, the change is reflected more quickly, but it
still takes progressively longer as the simulation continues. -
Conclusion is that a data buffer is slowly filling up and taking
more time to empty out. -
If the tx Spectrum Scope were removed, the model would run faster,
and the frequency changes would show more quickly.
I've attached images of the model, the block dialogs, the tx and rx spectra,
and the time plot of the overrun signal with this message. Please let us
know if you have additional questions.
Best,
Mike
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Sunday, July 15, 2012 6:14 AM
To: Marc Erickson; usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Marc,
Thanks for the kind offer to provide support.
In order to continue this thread within the community, I would like to
respond by repeating my previous request:
Could you please comment and perhaps provide an illustration of how to
configure the N210/SBX using both MATLAB and Simulink so that a signal is
transmitted from the RF1 port and received by the RF2. As a suggestion,
slider controls could be used in Simulink to independently adjust the RF
settings of both the transmitter and receiver to include at least gain and
center frequency. A spectrum of the received signal should be displayed.
With many thanks,
Stan
From: Marc Erickson [mailto:Marc.Erickson@mathworks.com]
Sent: 05 July 2012 19:19
To: Stan Gamla; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP
HardwareSupport for MATLAB
Hi Stan,
Sorry for the confusion about our support. We have stated some limitations
of our support in the product release notes which are available in the
download package. Type 'help sdru' and follow the 'Help and Examples' link.
Then click on the link under 'Release Notes'. The relevant limitation is
stated as follows:
"Communications with USRPT only supports half-duplex applications. MATLAB
and Simulink applications must be receive-only or transmit-only. Full-duplex
is not supported."
Our package has separate Rx and Tx blocks (for Simulink) and separate Rx and
Tx System objects (for MATLAB) with their own configuration parameters. We
allow binding the Rx and Tx to the same board in order to support some
duplex operations. Full duplex operation might be possible when adhering to
some modeling constraints.
Please feel free to contact us directly to discuss any use-cases and
specific applications you might be interested in pursuing and we'd be happy
to work with you.
Regards,
Marc
From: usrp-users-bounces@lists.ettus.com
[mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Friday, June 29, 2012 7:03 AM
To: usrp-users@lists.ettus.com
Subject: [USRP-users] SBX RF board simultaneous mode with USRP Hardware
Support for MATLAB
As far as I understand, the Ettus SBX RF board comprises a full-duplex
transceiver with separate receive and transmit chains, each having its own
local oscillator.
https://www.ettus.com/product#tabs-3 states that:
"The SBX is a wide bandwidth transceiver that provides up to 100 mW of
output power, and a typical noise figure of 5 dB. The local oscillators for
the receive and transmit chains operate independently, which allows
dual-band operation. The SBX is MIMO capable, and provides 40 MHz of
bandwidth. The SBX is ideal for applications requiring access to a variety
of bands in the 400 MHz-4400 MHz range. Example application areas include
WiFi, WiMax, S-band transceivers and 2.4 GHz ISM band transceivers."
Further, according to their website "The MathWorks USRPR Hardware Support
from [sic] MATLAB and Simulink is compatible with UHDT-based Ettus radios".
http://www.mathworks.com/discovery/sdr/usrp.html viz:
"MATLAB and Simulink support for USRPR devices has been tested on N210 and
USRP2R devices with WBX, SBX, XCVR2450, TVRX, TVRX2, LFRX, LFTX, DBSRX, and
DBSRX2 daughterboards. Support may also extend to other UHDT-based radios
and daughterboards from National InstrumentsR."
Are both of these statements correct, I wonder?
If The MathWorks is tuned in to this channel, could you please comment and
perhaps provide an illustration of how to configure the N210/SBX using both
MATLAB and Simulink so that a signal is transmitted from the RF1 port and
received by the RF2. As a suggestion, slider controls could be used in
Simulink to independently adjust the RF settings of both the transmitter and
receiver to include at least gain and center frequency. A spectrum of the
received signal should be displayed.
On the other hand, if I have misunderstood what is meant by the word
"supported" could either Ettus or The MathWorks comment please.
Thanks,
Stan
Hi Stan,
Sorry for the delay in responding - I was out of town on vacation most of last week.
I am still working on the MATLAB script, and it is a high priority this week. To help me with my development, I have a few questions:
-
Can you send me the Simulink model that you created, in which the noise floor appeared to change? I'd like to reproduce the behavior you're seeing. -
Is your objective to use one radio, where bidirectional transmission occurs over the GigE cable? (As we've demonstrated, such a configuration can be implemented with one Simulink model.) Or do you want to use two host computers and two radios, where USRP #1 transmits to USRP #2, and vice-versa?
Best,
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Monday, August 20, 2012 9:37 AM
To: usrp-users@lists.ettus.com
Cc: Mike McLernon; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Importance: High
Hi Mike and Marc,
Using the MathWorks example file "sdruFRSGMRSTransmitter", I have attempted to create the example that I described below but have been far from successful. Therefore, without wishing to become a nuisance, could you please let me know when I can look forward to receiving a MATLAB version of the Simulink model that you sent earlier?
With many thanks,
Stan
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Mike McLernon
Sent: 10 August 2012 14:41
To: Stan Gamla
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
Still working on it - I have something going for a one-radio setup, but I'd like to prove out a two-radio configuration.
Thanks for your patience,
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 10, 2012 8:38 AM
To: Mike McLernon
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Mike,
Sorry to ask again... but I would very much appreciate the MATLAB example file.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 03 August 2012 14:24
To: Stan Gamla
Cc: usrp-users@lists.ettus.com; Marc Erickson
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I'll work on the script and post it when it's done.
Mike
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, August 03, 2012 3:29 AM
To: Marc Erickson; Mike McLernon
Cc: usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc and Mike,
Further to my last message, could you please provide an equivalent example of the Simulink model that you posted earlier but this time written in MATLAB.
Thanks,
Stan
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: 27 July 2012 15:43
To: 'Mike McLernon'
Cc: usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Mike,
Thanks for the responses to my questions and thanks too for the Simulink model. The latter has proved to be quite useful. To experiment with the model, I connected together the RF1 and RF2 ports of my N210 (+SBX) via a 30 dB fixed attenuator and a steppable attenuator (0-10 + 0/10/20/30/40 dB). I also modified the model and replaced the switch with a slider that controls the center frequency of both the Tx and Rx blocks, simultaneously.
The model and hardware configuration described above allows me to set fc and the attenuation in the RF path. However, the behaviour of the output (Rx) spectrum is not quite as I would have expected--instead of the amplitude of the peak changing when the RF attenuator is adjusted, it appears that the level of what could be called the noise floor of the output changes instead. Could this be due to a numerical effect in the Simulink FFT block that holds the "peak" but instead changes the "floor"? Or has it something to do with the Ettus radio--perhaps the result of some automatic gain control (AGC) circuitry?
Regardless of all of the above, I would be very happy to experiment and to study the complementary MATLAB scripts that are used to construct a system similar to the Simulink model that you so kindly posted to this forum.
Thanks,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 20 July 2012 18:32
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
Responses interleaved below . . .
From: Stan Gamla [mailto:cystrin@hotmail.com]
Sent: Friday, July 20, 2012 6:57 AM
To: Mike McLernon
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Mike,
With reference to the schematic of the Simulink model "mUSRPFullDuplex" that you posted earlier, could you please reveal what is in (or under) the block labeled "Receiver Processing"?
Hopefully you received my mail from earlier today with several models attached. If not, let me know.
Referring to item 5 in the list "experiment setup" (in which the SDRu Tx and SDRu Rx blocks are in the same model) would the system not perform full-duplex operation if two models were used?
If two models are used, and they are both using the same IP, they should still be performing full duplex. I have not run that case.
As a general question, could you please comment on how to configure the RF ports of the SBX board in which an RF switch is used to configure RF1 for either Tx or Rx.
Our R2011a and R2011b release notes state that the default antenna configurations cannot be changed. For my particular application, I did not need to perform any antenna configuration.
Finally, it would be very much appreciated if the sample Simulink model(s) could be shared in this community and it would be even more appreciated if they were to be complemented with equivalent MATLAB files. Such a system would be a useful example to include in future releases from the MathWorks.
Let me know if the models I sent earlier are helpful to you. We would want to include such models (and complementary MATLAB scripts) in future releases.
Hth,
Mike
Thanks in advance,
Stan
From: Mike McLernon [mailto:Mike.McLernon@mathworks.com]
Sent: 19 July 2012 17:06
To: Stan Gamla
Cc: Marc Erickson; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
Hi Stan,
I did the following experiment to ascertain our level of full duplex support:
Experiment Setup
- N210/SBX
- VERT900 antennas for both tx and rx
- Win64 on a 12-core development machine
- One MATLAB session, one Simulink model
- SDRu Tx block and SDRu Rx block in same model
- Model runs in accelerator mode
- Spectrum Scope blocks for tx and rx waveforms, a Simulink Scope to monitor rx overruns
- Interpolation/decimation factor of 512 (maximum allowed)
- Tx center frequency = 915 MHz
- Rx center frequency = 915 or 915.04 MHz (controlled with a manual switch)
- Optimizations:
* Use singles rather than doubles
* Use accelerator rather than normal mode
* Turned on compiler optimizations
* Turned off all data logging
Results
- Rx Spectrum Scope shows tones where they should be
- Simulink Scope shows no overruns
- When the SDRu Receiver block's center frequency is changed with the manual switch shortly after the simulation starts, the rx Spectrum Scope shows the change quickly.
- As the simulation continues, it takes progressively longer for the change to show in the Spectrum Scope.
* With one scope removed, the change is reflected more quickly, but it still takes progressively longer as the simulation continues.
* Conclusion is that a data buffer is slowly filling up and taking more time to empty out.
* If the tx Spectrum Scope were removed, the model would run faster, and the frequency changes would show more quickly.
I've attached images of the model, the block dialogs, the tx and rx spectra, and the time plot of the overrun signal with this message. Please let us know if you have additional questions.
Best,
Mike
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Sunday, July 15, 2012 6:14 AM
To: Marc Erickson; usrp-users@lists.ettus.com
Subject: Re: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Marc,
Thanks for the kind offer to provide support.
In order to continue this thread within the community, I would like to respond by repeating my previous request:
Could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
With many thanks,
Stan
From: Marc Erickson [mailto:Marc.Erickson@mathworks.com]
Sent: 05 July 2012 19:19
To: Stan Gamla; usrp-users@lists.ettus.com
Subject: RE: [USRP-users] SBX RF board simultaneous mode with USRP HardwareSupport for MATLAB
Hi Stan,
Sorry for the confusion about our support. We have stated some limitations of our support in the product release notes which are available in the download package. Type 'help sdru' and follow the 'Help and Examples' link. Then click on the link under 'Release Notes'. The relevant limitation is stated as follows:
"Communications with USRP(tm) only supports half-duplex applications. MATLAB and Simulink applications must be receive-only or transmit-only. Full-duplex is not supported."
Our package has separate Rx and Tx blocks (for Simulink) and separate Rx and Tx System objects (for MATLAB) with their own configuration parameters. We allow binding the Rx and Tx to the same board in order to support some duplex operations. Full duplex operation might be possible when adhering to some modeling constraints.
Please feel free to contact us directly to discuss any use-cases and specific applications you might be interested in pursuing and we'd be happy to work with you.
Regards,
Marc
From: usrp-users-bounces@lists.ettus.com [mailto:usrp-users-bounces@lists.ettus.com] On Behalf Of Stan Gamla
Sent: Friday, June 29, 2012 7:03 AM
To: usrp-users@lists.ettus.com
Subject: [USRP-users] SBX RF board simultaneous mode with USRP Hardware Support for MATLAB
As far as I understand, the Ettus SBX RF board comprises a full-duplex transceiver with separate receive and transmit chains, each having its own local oscillator.
https://www.ettus.com/product#tabs-3 states that:
"The SBX is a wide bandwidth transceiver that provides up to 100 mW of output power, and a typical noise figure of 5 dB. The local oscillators for the receive and transmit chains operate independently, which allows dual-band operation. The SBX is MIMO capable, and provides 40 MHz of bandwidth. The SBX is ideal for applications requiring access to a variety of bands in the 400 MHz-4400 MHz range. Example application areas include WiFi, WiMax, S-band transceivers and 2.4 GHz ISM band transceivers."
Further, according to their website "The MathWorks USRP(r) Hardware Support from [sic] MATLAB and Simulink is compatible with UHD(tm)-based Ettus radios".
http://www.mathworks.com/discovery/sdr/usrp.html viz:
"MATLAB and Simulink support for USRP(r) devices has been tested on N210 and USRP2(r) devices with WBX, SBX, XCVR2450, TVRX, TVRX2, LFRX, LFTX, DBSRX, and DBSRX2 daughterboards. Support may also extend to other UHD(tm)-based radios and daughterboards from National Instruments(r)."
Are both of these statements correct, I wonder?
If The MathWorks is tuned in to this channel, could you please comment and perhaps provide an illustration of how to configure the N210/SBX using both MATLAB and Simulink so that a signal is transmitted from the RF1 port and received by the RF2. As a suggestion, slider controls could be used in Simulink to independently adjust the RF settings of both the transmitter and receiver to include at least gain and center frequency. A spectrum of the received signal should be displayed.
On the other hand, if I have misunderstood what is meant by the word "supported" could either Ettus or The MathWorks comment please.
Thanks,
Stan