Hi All, There is no pressing question here, but rather I am wondering if the following matches other people’s expectations and experience. I have a TDMA-based GR design with B200minis. Time is sync’d to GPS time and packets are scheduled and delineated using tx_time and SOB/EOB tags. The attached scope snapshots show the 1 PPS edge and a packet scheduled to transmit at that time. The following observations seem obvious, but you be the judge. The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high. It may be that I should put a couple of zero-valued samples up front, so I will try that soon. However, it appears that the EOB Tx-to-Rx transition is delayed by about 10 us past the last sample, and still results in a transient. This may not bode well for reducing the Tx transient. I am confident the transient is delayed as I calculated exactly how long my “packet” needs, which is 134.4 us It appears that it takes about 80 us to reach full transmit power. Again, zero-padding or extending my sync sequence at the front can mitigate that. Any and all comments are welcome. This behaviour is, for my particular project, not a problem, but I suspect would be for someone needing to meet a spectrum mask. Thanks for your time and consideration, steven Steven Knudsen, Ph.D., P.Eng. www. techconficio.ca www.linkedin.com/in/knudstevenknudsen Der entscheidende Augenblick der menschlichen Entwicklung ist immerwährend. Darum sind die revolutionären geistigen Bewegungen, welche alles Frühere für nichtig erklären, im Recht, denn es ist noch nichts geschehen. - Franz Kafka

- The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high.
It may be that I should put a couple of zero-valued samples up front, so I
will try that soon.
- However, it appears that the EOB Tx-to-Rx transition is delayed by
about 10 us past the last sample, and still results in a transient. This
may not bode well for reducing the Tx transient.
   - I am confident the transient is delayed as I calculated exactly
   how long my “packet” needs, which is 134.4 us
- It appears that it takes about 80 us to reach full transmit power.
Again, zero-padding or extending my sync sequence at the front can mitigate
that.

Hi Steven, Excellent analysis. Thank you. I have filed an issue against UHD for investigation to be done. What center frequency and gain were you using? Regards, Michael On Tue, Jun 28, 2016 at 9:57 AM, Steven Knudsen via USRP-users < usrp-users@lists.ettus.com> wrote: > Hi All, > > There is no pressing question here, but rather I am wondering if the > following matches other people’s expectations and experience. > > I have a TDMA-based GR design with B200minis. Time is sync’d to GPS time > and packets are scheduled and delineated using tx_time and SOB/EOB tags. > > The attached scope snapshots show the 1 PPS edge and a packet scheduled to > transmit at that time. The following observations seem obvious, but you be > the judge. > > > - The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high. > It may be that I should put a couple of zero-valued samples up front, so I > will try that soon. > - However, it appears that the EOB Tx-to-Rx transition is delayed by > about 10 us past the last sample, and still results in a transient. This > may not bode well for reducing the Tx transient. > - I am confident the transient is delayed as I calculated exactly > how long my “packet” needs, which is 134.4 us > - It appears that it takes about 80 us to reach full transmit power. > Again, zero-padding or extending my sync sequence at the front can mitigate > that. > > > Any and all comments are welcome. This behaviour is, for my particular > project, not a problem, but I suspect would be for someone needing to meet > a spectrum mask. > > Thanks for your time and consideration, > > steven > > > Steven Knudsen, Ph.D., P.Eng. > www. techconficio.ca > www.linkedin.com/in/knudstevenknudsen > > *Der entscheidende Augenblick der menschlichen Entwicklung > ist immerwährend. Darum sind die revolutionären geistigen Bewegungen, > welche alles Frühere für nichtig erklären, im Recht, denn es ist > noch nichts geschehen. - Franz Kafka* > > > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > >

Hi Michael, Thanks for taking the time to respond. First, to answer your question, I was using 100 MHz and gains in the range of 75 to 80. I can bump up to ~500 MHz, but beyond that I don’t have a better scope (nor do I have the money to get one :-( ) Some more thoughts… In talking to “real RF/analogue engineers” (I am an engineer, but not an expert in any one thing), they pointed out that the transients are very slow relative to the signalling rates, so any frequency components will stay in-band and not be all that noticeable. Indeed, I don’t seem to see anything particularly ugly on my spectrum analyzer (SA). There might be some spurs from the initial Tx turn-on, and I can look for that by keeping the Tx gain very low. That is, the Tx transient is there no matter what Tx gain I set. I have added a parameter to my OOT block to allow padding and doing as I suggest below, stuffing 80 us worth of zeros, the resulting tx signal looks good. It’s a shame to lose 80 us, but I can live with it. If I was to investigate further, I’d be tempted to get an Analog Devices eval kit for the frontend and put it through its paces. I didn’t see anything about Rx to Tx time and transients/ramp times in the datasheet, and had a quick scan of the ref manual. Doesn’t mean it’s not mentioned in the ref manual, so I guess if I really cared I’d look there. The ramp time does not seem to be a function of the sample rate, at least from some limited testing. I look forward to hearing what you think or discover. For now, as I say, I’m okay with my bandaid solution… Thanks! Steven Knudsen, Ph.D., P.Eng. www. techconficio.ca www.linkedin.com/in/knudstevenknudsen So fest wie die Hand den Stein hält. Sie hält ihn aber fest, nur um ihn desto weiter zu verwerfen. Aber auch in jene Weite führt der Weg. - Franz Kafka > On Jul 28, 2016, at 20:53, Michael West <michael.west@ettus.com> wrote: > > Hi Steven, > > Excellent analysis. Thank you. I have filed an issue against UHD for investigation to be done. What center frequency and gain were you using? > > Regards, > Michael > > On Tue, Jun 28, 2016 at 9:57 AM, Steven Knudsen via USRP-users <usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>> wrote: > Hi All, > > There is no pressing question here, but rather I am wondering if the following matches other people’s expectations and experience. > > I have a TDMA-based GR design with B200minis. Time is sync’d to GPS time and packets are scheduled and delineated using tx_time and SOB/EOB tags. > > The attached scope snapshots show the 1 PPS edge and a packet scheduled to transmit at that time. The following observations seem obvious, but you be the judge. > > The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high. It may be that I should put a couple of zero-valued samples up front, so I will try that soon. > However, it appears that the EOB Tx-to-Rx transition is delayed by about 10 us past the last sample, and still results in a transient. This may not bode well for reducing the Tx transient. > I am confident the transient is delayed as I calculated exactly how long my “packet” needs, which is 134.4 us > It appears that it takes about 80 us to reach full transmit power. Again, zero-padding or extending my sync sequence at the front can mitigate that. > > Any and all comments are welcome. This behaviour is, for my particular project, not a problem, but I suspect would be for someone needing to meet a spectrum mask. > > Thanks for your time and consideration, > > steven > > > Steven Knudsen, Ph.D., P.Eng. > www. techconficio.ca <http://techconficio.ca/> > www.linkedin.com/in/knudstevenknudsen <http://www.linkedin.com/in/knudstevenknudsen> > > Der entscheidende Augenblick der menschlichen Entwicklung ist immerwährend. Darum sind die revolutionären geistigen Bewegungen, welche alles Frühere für nichtig erklären, im Recht, denn es ist noch nichts geschehen. - Franz Kafka > > > > > > _______________________________________________ > 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> > >

 Hi All,

 There is no pressing question here, but rather I am wondering if
 the following matches other people’s expectations and experience.

 I have a TDMA-based GR design with B200minis. Time is sync’d to
 GPS time and packets are scheduled and delineated using tx_time
 and SOB/EOB tags.

 The attached scope snapshots show the 1 PPS edge and a packet
 scheduled to transmit at that time. The following observations
 seem obvious, but you be the judge.

   * The (SOB-mediated?) Rx-to-Tx transient exists and is kind of
     high. It may be that I should put a couple of zero-valued
     samples up front, so I will try that soon.
   * However, it appears that the EOB Tx-to-Rx transition is
     delayed by about 10 us past the last sample, and still
     results in a transient. This may not bode well for reducing
     the Tx transient.
       o I am confident the transient is delayed as I calculated
         exactly how long my “packet” needs, which is 134.4 us
   * It appears that it takes about 80 us to reach full transmit
     power. Again, zero-padding or extending my sync sequence at
     the front can mitigate that.


 Any and all comments are welcome. This behaviour is, for my
 particular project, not a problem, but I suspect would be for
 someone needing to meet a spectrum mask.

 Thanks for your time and consideration,

 steven


 Steven Knudsen, Ph.D., P.Eng.
 www.techconficio.ca <http://techconficio.ca/>
 www.linkedin.com/in/knudstevenknudsen
 <http://www.linkedin.com/in/knudstevenknudsen>

 /Der entscheidende Augenblick der menschlichen Entwicklung
 ist immerwährend. Darum sind die revolutionären
 geistigen Bewegungen, welche alles Frühere für nichtig
 erklären, im Recht, denn es ist noch nichts geschehen. - Franz Kafka/





 _______________________________________________
 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

On 07/28/2016 11:19 PM, Steven Knudsen via USRP-users wrote: > Hi Michael, > > Thanks for taking the time to respond. > > First, to answer your question, I was using 100 MHz and gains in the > range of 75 to 80. I can bump up to ~500 MHz, but beyond that I don’t > have a better scope (nor do I have the money to get one :-( ) > > Some more thoughts… > > In talking to “real RF/analogue engineers” (I am an engineer, but not > an expert in any one thing), they pointed out that the transients are > very slow relative to the signalling rates, so any frequency > components will stay in-band and not be all that noticeable. Indeed, I > don’t seem to see anything particularly ugly on my spectrum analyzer > (SA). There might be some spurs from the initial Tx turn-on, and I can > look for that by keeping the Tx gain very low. */That is, the Tx > transient is there no matter what Tx gain I set./* > > I have added a parameter to my OOT block to allow padding and doing as > I suggest below, stuffing 80 us worth of zeros, the resulting tx > signal looks good. It’s a shame to lose 80 us, but I can live with it. > > If I was to investigate further, I’d be tempted to get an Analog > Devices eval kit for the frontend and put it through its paces. I > didn’t see anything about Rx to Tx time and transients/ramp times in > the datasheet, and had a quick scan of the ref manual. Doesn’t mean > it’s not mentioned in the ref manual, so I guess if I really cared I’d > look there. > > The ramp time does not seem to be a function of the sample rate, at > least from some limited testing. > > I look forward to hearing what you think or discover. For now, as I > say, I’m okay with my bandaid solution… > > Thanks! > > Steven Knudsen, Ph.D., P.Eng. > www.techconficio.ca <http://techconficio.ca> > www.linkedin.com/in/knudstevenknudsen > <http://www.linkedin.com/in/knudstevenknudsen> It takes time for a RF TX stage to reach steady-state operations. In the case of EOB/SOB, various bits and pieces get shut-down between bursts to limit things like LO leakage onto the channel between bursts. 80usec seems a tad long for this but not outrageous. A thing you could try is to keep TX running all the time, and crank down the gain and baseband magnitude between notional bursts--just as a laboratory experiment. If you channel requires hard TDMA-like timing, then obviously, this won't work that well, but it will give you a feel for the analog latencies involved in bring an RF power chain up to full-power and steady-state operation. > > So fest wie die Hand den Stein hält. Sie hält ihn aber fest, nur um > ihn desto weiter zu verwerfen. Aber auch in jene Weite führt der > Weg./- Franz Kafka/ > >> On Jul 28, 2016, at 20:53, Michael West <michael.west@ettus.com >> <mailto:michael.west@ettus.com>> wrote: >> >> Hi Steven, >> >> Excellent analysis. Thank you. I have filed an issue against UHD >> for investigation to be done. What center frequency and gain were >> you using? >> >> Regards, >> Michael >> >> On Tue, Jun 28, 2016 at 9:57 AM, Steven Knudsen via USRP-users >> <usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>> wrote: >> >> Hi All, >> >> There is no pressing question here, but rather I am wondering if >> the following matches other people’s expectations and experience. >> >> I have a TDMA-based GR design with B200minis. Time is sync’d to >> GPS time and packets are scheduled and delineated using tx_time >> and SOB/EOB tags. >> >> The attached scope snapshots show the 1 PPS edge and a packet >> scheduled to transmit at that time. The following observations >> seem obvious, but you be the judge. >> >> * The (SOB-mediated?) Rx-to-Tx transient exists and is kind of >> high. It may be that I should put a couple of zero-valued >> samples up front, so I will try that soon. >> * However, it appears that the EOB Tx-to-Rx transition is >> delayed by about 10 us past the last sample, and still >> results in a transient. This may not bode well for reducing >> the Tx transient. >> o I am confident the transient is delayed as I calculated >> exactly how long my “packet” needs, which is 134.4 us >> * It appears that it takes about 80 us to reach full transmit >> power. Again, zero-padding or extending my sync sequence at >> the front can mitigate that. >> >> >> Any and all comments are welcome. This behaviour is, for my >> particular project, not a problem, but I suspect would be for >> someone needing to meet a spectrum mask. >> >> Thanks for your time and consideration, >> >> steven >> >> >> Steven Knudsen, Ph.D., P.Eng. >> www.techconficio.ca <http://techconficio.ca/> >> www.linkedin.com/in/knudstevenknudsen >> <http://www.linkedin.com/in/knudstevenknudsen> >> >> /Der entscheidende Augenblick der menschlichen Entwicklung >> ist immerwährend. Darum sind die revolutionären >> geistigen Bewegungen, welche alles Frühere für nichtig >> erklären, im Recht, denn es ist noch nichts geschehen. - Franz Kafka/ >> >> >> >> >> >> _______________________________________________ >> 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 >> >> > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

- The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high.
It may be that I should put a couple of zero-valued samples up front, so I
will try that soon.
- However, it appears that the EOB Tx-to-Rx transition is delayed by
about 10 us past the last sample, and still results in a transient. This
may not bode well for reducing the Tx transient.
   - I am confident the transient is delayed as I calculated exactly
   how long my “packet” needs, which is 134.4 us
- It appears that it takes about 80 us to reach full transmit power.
Again, zero-padding or extending my sync sequence at the front can mitigate
that.

Hi Steven, Thanks for the additional information. Yes, even though the amplitude of the transient is large, it may not be noticeable in the frequency domain. As Marcus points out, the settling time of the TX power is "a tad long but not outrageous." I would not expect sample rate to affect it. It could be something with the AD9361 settings or just the TX PA taking a long time to turn on. You can force the TX PA on all the time by changing the ATR settings in UHD and see if it improves, but you will see more noise on the RX side. Regards, Michael On Fri, Jul 29, 2016 at 4:39 PM, Marcus D. Leech via USRP-users < usrp-users@lists.ettus.com> wrote: > On 07/28/2016 11:19 PM, Steven Knudsen via USRP-users wrote: > > Hi Michael, > > Thanks for taking the time to respond. > > First, to answer your question, I was using 100 MHz and gains in the range > of 75 to 80. I can bump up to ~500 MHz, but beyond that I don’t have a > better scope (nor do I have the money to get one :-( ) > > Some more thoughts… > > In talking to “real RF/analogue engineers” (I am an engineer, but not an > expert in any one thing), they pointed out that the transients are very > slow relative to the signalling rates, so any frequency components will > stay in-band and not be all that noticeable. Indeed, I don’t seem to see > anything particularly ugly on my spectrum analyzer (SA). There might be > some spurs from the initial Tx turn-on, and I can look for that by keeping > the Tx gain very low. *That is, the Tx transient is there no matter what > Tx gain I set.* > > I have added a parameter to my OOT block to allow padding and doing as I > suggest below, stuffing 80 us worth of zeros, the resulting tx signal looks > good. It’s a shame to lose 80 us, but I can live with it. > > If I was to investigate further, I’d be tempted to get an Analog Devices > eval kit for the frontend and put it through its paces. I didn’t see > anything about Rx to Tx time and transients/ramp times in the datasheet, > and had a quick scan of the ref manual. Doesn’t mean it’s not mentioned in > the ref manual, so I guess if I really cared I’d look there. > > The ramp time does not seem to be a function of the sample rate, at least > from some limited testing. > > I look forward to hearing what you think or discover. For now, as I say, > I’m okay with my bandaid solution… > > Thanks! > > Steven Knudsen, Ph.D., P.Eng. > www. techconficio.ca > www.linkedin.com/in/knudstevenknudsen > > > It takes time for a RF TX stage to reach steady-state operations. In the > case of EOB/SOB, various bits and pieces get shut-down between bursts > to limit things like LO leakage onto the channel between bursts. 80usec > seems a tad long for this but not outrageous. > > A thing you could try is to keep TX running all the time, and crank down > the gain and baseband magnitude between notional bursts--just as > a laboratory experiment. If you channel requires hard TDMA-like timing, > then obviously, this won't work that well, but it will give you a > feel for the analog latencies involved in bring an RF power chain up to > full-power and steady-state operation. > > > > > So fest wie die Hand den Stein hält. Sie hält ihn aber fest, nur um ihn > desto weiter zu verwerfen. Aber auch in jene Weite führt der Weg. *- > Franz Kafka * > > On Jul 28, 2016, at 20:53, Michael West <michael.west@ettus.com> wrote: > > Hi Steven, > > Excellent analysis. Thank you. I have filed an issue against UHD for > investigation to be done. What center frequency and gain were you using? > > Regards, > Michael > > On Tue, Jun 28, 2016 at 9:57 AM, Steven Knudsen via USRP-users < > usrp-users@lists.ettus.com> wrote: > >> Hi All, >> >> There is no pressing question here, but rather I am wondering if the >> following matches other people’s expectations and experience. >> >> I have a TDMA-based GR design with B200minis. Time is sync’d to GPS time >> and packets are scheduled and delineated using tx_time and SOB/EOB tags. >> >> The attached scope snapshots show the 1 PPS edge and a packet scheduled >> to transmit at that time. The following observations seem obvious, but you >> be the judge. >> >> >> - The (SOB-mediated?) Rx-to-Tx transient exists and is kind of high. >> It may be that I should put a couple of zero-valued samples up front, so I >> will try that soon. >> - However, it appears that the EOB Tx-to-Rx transition is delayed by >> about 10 us past the last sample, and still results in a transient. This >> may not bode well for reducing the Tx transient. >> - I am confident the transient is delayed as I calculated exactly >> how long my “packet” needs, which is 134.4 us >> - It appears that it takes about 80 us to reach full transmit power. >> Again, zero-padding or extending my sync sequence at the front can mitigate >> that. >> >> >> Any and all comments are welcome. This behaviour is, for my particular >> project, not a problem, but I suspect would be for someone needing to meet >> a spectrum mask. >> >> Thanks for your time and consideration, >> >> steven >> >> >> Steven Knudsen, Ph.D., P.Eng. >> www. techconficio.ca >> www.linkedin.com/in/knudstevenknudsen >> >> *Der entscheidende Augenblick der menschlichen Entwicklung >> ist immerwährend. Darum sind die revolutionären geistigen Bewegungen, >> welche alles Frühere für nichtig erklären, im Recht, denn es ist >> noch nichts geschehen. - Franz Kafka* >> >> >> >> >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >> > > > > _______________________________________________ > USRP-users mailing listUSRP-users@lists.ettus.comhttp://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 > >