Hello, My team and I are doing some experiments with the USRP B210 and NB-IoT waveforms. During tests we have observed a few issues with the RF output, that we would like to understand better, and hopefully​ fix or mitigate. I must admit that we are not yet very familiar with USRP SDRs, so I apologize in advance for any stupid question! There are so many options to tweak the USRP :-) ! This is what we have observed: 1. Apparent RF leakage from RX to TX when idle (i.e. not transmitting the 180 kHz NB-IoT carrier). Our B210 was connected to an RF matrix where other signals in other channels not immediately adjacent to our center frequency also exist, and, every time we were not transmitting, some of those signals (a MHz away) were picked up from the RX and “leaked” into the TX – this could be observed from a spectrum analyzer connected to the B210 TX and it appears to be picking up quite a wide bandwidth (more than 5-10 MHz). * Is this a known issue with the B210 RF frontend? * Does anyone have any suggestions on how to mitigate it (if even possible)? * We tried limiting RX bandwidth as described here https://files.ettus.com/manual/page_usrp_b200.html#b200_fe_bw , but it seems not change anything at least when idle. Any ideas? 2. Out-of-band emissions and spectral regrowth: The output NB-IoT carrier shows quite large out-of-band emissions (~-30 dBc @200 kHz from center frequency, -50 dBc @400 kHz). * Could this be due to non-linear distortion caused by the B210 frontend (AD9361) ? * Would limiting TX analog bandwidth with set_tx_bandwidth(bw) help here? * Any other suggestions? I also have some other general questions...: 1. Even after reading this (https://files.ettus.com/manual/page_power.html ) I am still not super clear about if and how the set_tx_power_reference() (and rx) affect the gain level. I think understand TX gain setting will obviously affect the output power though. 2. When talking about multiple "gain stages", how does this translate to the B210? Different stages of the AD9361 ? 3. How are set_tx_bandwidth and set_rx_bandwidth implemented in practice in the B210? Many thanks Luca ________________________________ This communication is private and confidential and may contain information that is proprietary, privileged or otherwise legally exempt from disclosure. If you have received this message in error, please notify the sender immediately by e-mail and delete all copies of the message. In accordance with our guidelines, emails sent or received may be monitored. Inmarsat Global Limited, Registered No. 3675885. Registered in England and Wales with Registered Office at 99 City Road, London EC1Y 1AX INTERNAL | © INMARSAT

 1.

  1. Any other suggestions?

On 02/09/2021 03:48 PM, Luca Lodigiani via USRP-users wrote: > Hello, > > My team and I are doing some experiments with the USRP B210 and NB-IoT > waveforms. > During tests we have observed a few issues with the RF output, that we > would like to understand better, and /hopefully/​ fix or mitigate. I > must admit that we are not yet very familiar with USRP SDRs, so I > apologize in advance for any stupid question! There are so many > options to tweak the USRP :-) ! > > This is what we have observed: > > 1. Apparent RF leakage from RX to TX when idle (i.e. not transmitting > the 180 kHz NB-IoT carrier). Our B210 was connected to an RF > matrix where other signals in other channels not immediately > adjacent to our center frequency also exist, and, every time we > were not transmitting, some of those signals (a MHz away) were > picked up from the RX and “leaked” into the TX – this could be > observed from a spectrum analyzer connected to the B210 TX and it > appears to be picking up quite a wide bandwidth (more than 5-10 MHz). > 1. Is this a known issue with the B210 RF frontend? > 2. Does anyone have any suggestions on how to mitigate it (if > even possible)? > 3. We tried limiting RX bandwidth as described here > https://files.ettus.com/manual/page_usrp_b200.html#b200_fe_bw , but > it seems not change anything at least when idle. Any ideas? > Given that the leakage path that you're (I think) talking about would be VERY VERY early in the RF chain inside the AD9361 chip, setting analog bandwidth would have ZERO effect, since that sets the *IF* bandwidth--well after the bits and pieces you're likely dealing with. I haven't looked at the TX/RX isolation spec in the AD9361 data-sheet, but it would surprise me very much if it were better than 50dB. It's pretty hard inside such a tiny tiny piece of silicon to isolate RF signal pathways more than that. What power levels is the RX2 port seeing in your application--your "RF matrix" really needs to not be presenting more than about -15dBm to the RX2 port(s), or there's potential for damage. Also, when you talk about 180kHz carrier, you're talking *bandwidth* or center frequency? The center frequency on the B210 cannot go below about 50MHz. > 1. > > > > 1. Out-of-band emissions and spectral regrowth: The output NB-IoT > carrier shows quite large out-of-band emissions (~-30 dBc @200 kHz > from center frequency, -50 dBc @400 kHz). > 1. Could this be due to non-linear distortion caused by the B210 > frontend (AD9361) ? > 2. Would limiting TX analog bandwidth with set_tx_bandwidth(bw) > help here? > It might Help to lower TX analog bandwidth, but UHD automatically sets it based on your sample rate. Also try reducing baseband magnitude a little bit, as well as reducing TX gain a little bit. > 1. Any other suggestions? > > I also have some other general questions...: > > 1. Even after reading this > (https://files.ettus.com/manual/page_power.html ) I am still not > super clear about if and how the set_tx_power_reference() (and > rx) affect the gain level. I think understand TX gain setting > will obviously affect the output power though. > The tx_power_reference() API is new, so I cannot comment on how it works out in practice. The B2xx have about 80dB of TX gain setting range, but with peak power output of about +7dBm over most of its frequency range, I'd say only the top 30dB is really useful. > > 1. When talking about multiple "gain stages", how does this translate > to the B210? Different stages of the AD9361 ? > Yes, different stages within the AD9361 in the case of B2xx. > 1. How are set_tx_bandwidth and set_rx_bandwidth implemented in > practice in the B210? > By using features inside the AD9361 to set the effective analog bandwidth in front of/behind the ADCs/DACs. I'm not sure whether the implementation inside the AD9361 is purely-analog (like switched-capacitor filters, and changing clocks), or a hybrid approach that uses the DSP logic inside the AD9361. The ADCs and DACs in the AD936x family operate at very high sample rates internally, and use DSP logic to provide the actual delivered-to-digital-interface sample rates, so there's room to do both a digital and analog approach in providing the necessary effective analog bandwidth. But UHD normally picks an analog bandwidth setting that is appropriate to the desired sample rate. The only time this may not be appropriate is if your actual signal is very much narrower than is implied by the selected sample rate. > > Many thanks > Luca > > ------------------------------------------------------------------------ > > This communication is private and confidential and may contain > information that is proprietary, privileged or otherwise legally > exempt from disclosure. If you have received this message in error, > please notify the sender immediately by e-mail and delete all copies > of the message. In accordance with our guidelines, emails sent or > received may be monitored. Inmarsat Global Limited, Registered No. > 3675885. Registered in England and Wales with Registered Office at 99 > City Road, London EC1Y 1AX > > INTERNAL | © INMARSAT > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

I can't comment on issue #1, but I do have experience with issue #2. To get a clean output spectrum, you need to do two things. First, the IQ values generated by your application can never exceed +1.0/-1.0 in value. You have to account for the PAPR of your waveform. If you're using GNU Radio, I have a block to dump peak IQ levels. https://github.com/drmpeg/gr-iqlevels Second, you can't set the TX gain to maximum. Something less than or equal to 78 will typically provide the best performance (although the exact maximum value will depend on the PAPR of your waveform). Here's a B210 transmitted signal at a gain of 78. Ron On 2/9/21 12:48, Luca Lodigiani via USRP-users wrote: > Hello, > > My team and I are doing some experiments with the USRP B210 and NB-IoT > waveforms. > During tests we have observed a few issues with the RF output, that we > would like to understand better, and /hopefully/​ fix or mitigate.   I > must admit that we are not yet very familiar with USRP SDRs, so I > apologize in advance for any stupid question!  There are so many > options to tweak the USRP :-) ! > > This is what we have observed: > > 1. Apparent RF leakage from RX to TX when idle (i.e. not transmitting > the 180 kHz NB-IoT carrier).  Our B210 was connected to an RF > matrix where other signals in other channels not immediately > adjacent to our center frequency also exist, and, every time we > were not transmitting, some of those signals (a MHz away) were > picked up from the RX and “leaked” into the TX – this could be > observed from a spectrum analyzer connected to the B210 TX and it > appears to be picking up quite a wide bandwidth (more than 5-10 MHz). > 1. Is this a known issue with the B210 RF frontend? > 2. Does anyone have any suggestions on how to mitigate it (if > even possible)? > 3. We tried limiting RX bandwidth as described here > https://files.ettus.com/manual/page_usrp_b200.html#b200_fe_bw , > but it seems not change anything at least when idle. Any ideas? > > 2. Out-of-band emissions and spectral regrowth:  The output NB-IoT > carrier shows quite large out-of-band emissions (~-30 dBc @200 kHz > from center frequency, -50 dBc @400 kHz). > 1. Could this be due to non-linear distortion caused by the B210 > frontend (AD9361) ? > 2. Would limiting TX analog bandwidth with set_tx_bandwidth(bw) > help here? > 3. Any other suggestions? > > I also have some other general questions...: > > 1. Even after reading this > (https://files.ettus.com/manual/page_power.html ) I am still not > super clear about if and how the set_tx_power_reference() (and > rx)  affect the gain level.  I think understand TX gain setting > will obviously affect the output power though. > 2. When talking about multiple "gain stages", how does this translate > to the B210?  Different stages of the AD9361 ? > 3. How are set_tx_bandwidth and set_rx_bandwidth implemented in > practice in the B210? > > > Many thanks > Luca > > ------------------------------------------------------------------------ > > This communication is private and confidential and may contain > information that is proprietary, privileged or otherwise legally > exempt from disclosure. If you have received this message in error, > please notify the sender immediately by e-mail and delete all copies > of the message. In accordance with our guidelines, emails sent or > received may be monitored. Inmarsat Global Limited, Registered No. > 3675885. Registered in England and Wales with Registered Office at 99 > City Road, London EC1Y 1AX > > INTERNAL | © INMARSAT > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com