Dear all, A little question about the GPSDO-module disciplined local oscillator. We have run an experiment where we connect a GPS antenna to the GPSDO module, and let the GPSDO acquire enough satellites to achieve a lock (this is to align the time of different USRPs). We then disconnect the GPS antenna, and let the local oscillator run by itself. After disconnecting the GPS antenna, will the local oscillator behave like any oscillator, or will there be some lingering effect from the GPSDO module GPS correction? Thanks a lot, Francois -- Francois Quitin, Ph.D. Research Fellow INFINITUS - Infocomm Centre of Excellence School of EEE, Nanyang Technological University 50 Nanyang Ave, S2-B4b-05 Singapore 639798 Phone: +65-8502-3690 Email: fquitin@ntu.edu.sg <mailto:fquitin@ece.ucsb.edu>

Hi Francois, I assume you not only use the GPSDO to get a time and a PPS to set that time, but also to discipline the oscillator of your USRP. Generally, when losing GPS fix, the GPSDO will continue to provide a clock. However, in my experience, that (not GPS coordinated) clock is worse than the USRP-built in oscillator, but not much. Is there a reason why you would want to remove the GPS antenna during use? Also, you should be able to use the PPS / time feature of the GPSDO without disciplining the clock using the GPSDO, and from my perspective that might be what you want to do. Greetings, Marcus On 06.05.2014 09:46, Francois Quitin via USRP-users wrote: > Dear all, > > > > A little question about the GPSDO-module disciplined local oscillator. We > have run an experiment where we connect a GPS antenna to the GPSDO module, > and let the GPSDO acquire enough satellites to achieve a lock (this is to > align the time of different USRPs). We then disconnect the GPS antenna, and > let the local oscillator run by itself. > > > > After disconnecting the GPS antenna, will the local oscillator behave like > any oscillator, or will there be some lingering effect from the GPSDO module > GPS correction? > > > > Thanks a lot, > > Francois > > > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Hi Marcus, Thanks for your answer! The reason we want to remove the GPS antenna is the following: we need the time of the USRPs to be synchronized (with precision that can be as bad as microseconds), but we want the oscillators to be free-running (we're building a TDOA estimation architecture that also estimates the LO offsets and LO skew). So indeed, using the PPS/time feature of teh GPSDO without disciplining the clock seems the way to go. How should I do that? Should I just disconnect the 10MHz reference cable while leaving the PPS cable connected? Cheers, Francois On 06.05.2014 17:21, Marcus Müller via USRP-users wrote: > Hi Francois, > > I assume you not only use the GPSDO to get a time and a PPS to set > that time, but also to discipline the oscillator of your USRP. > > Generally, when losing GPS fix, the GPSDO will continue to provide a > clock. > However, in my experience, that (not GPS coordinated) clock is worse > than the USRP-built in oscillator, but not much. Is there a reason why > you would want to remove the GPS antenna during use? > > Also, you should be able to use the PPS / time feature of the GPSDO > without disciplining the clock using the GPSDO, and from my > perspective that might be what you want to do. > > Greetings, > Marcus > > On 06.05.2014 09:46, Francois Quitin via USRP-users wrote: > >> Dear all, >> >> A little question about the GPSDO-module disciplined local >> oscillator. We >> have run an experiment where we connect a GPS antenna to the GPSDO >> module, >> and let the GPSDO acquire enough satellites to achieve a lock (this >> is to >> align the time of different USRPs). We then disconnect the GPS >> antenna, and >> let the local oscillator run by itself. >> >> After disconnecting the GPS antenna, will the local oscillator behave >> like >> any oscillator, or will there be some lingering effect from the GPSDO >> module >> GPS correction? >> >> Thanks a lot, >> >> Francois >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> [1] > > > > Links: > ------ > [1] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

If you allow the GPSDO to lock and sit for a while, and then remove the antenna, it should go into "holdover mode". This will maintain more accuracy than if you had not had GPS to start with, but obviously not as good as if you had kept the antenna attached. If you remove power and power up again you lose the holdover performance and go into normal free-run. Matt On Tue, May 6, 2014 at 9:46 AM, Francois Quitin via USRP-users < usrp-users@lists.ettus.com> wrote: > Dear all, > > > > A little question about the GPSDO-module disciplined local oscillator. We > have run an experiment where we connect a GPS antenna to the GPSDO module, > and let the GPSDO acquire enough satellites to achieve a lock (this is to > align the time of different USRPs). We then disconnect the GPS antenna, and > let the local oscillator run by itself. > > > > After disconnecting the GPS antenna, will the local oscillator behave like > any oscillator, or will there be some lingering effect from the GPSDO > module GPS correction? > > > > Thanks a lot, > > Francois > > > > -- > > Francois Quitin, Ph.D. > > > > Research Fellow > > INFINITUS - Infocomm Centre of Excellence > > School of EEE, Nanyang Technological University > > 50 Nanyang Ave, S2-B4b-05 > Singapore 639798 > > Phone: +65-8502-3690 > > Email: fquitin@ntu.edu.sg <fquitin@ece.ucsb.edu> > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > >

Hi Francois, I'd suggest you'd just use the uhd::multi_usrp::set_clock_source[1] command to select the "internal" clock, instead of the GPSDO one, that should do the trick. Greetings, Marcus [1]http://files.ettus.com/uhd_docs/doxygen/html/classuhd_1_1usrp_1_1multi__usrp.html#a73ed40009d0d3787c183d42423d25026 On 07.05.2014 06:00, fquitin wrote: > Hi Marcus, > > Thanks for your answer! The reason we want to remove the GPS antenna > is the following: we need the time of the USRPs to be synchronized > (with precision that can be as bad as microseconds), but we want the > oscillators to be free-running (we're building a TDOA estimation > architecture that also estimates the LO offsets and LO skew). > > So indeed, using the PPS/time feature of teh GPSDO without > disciplining the clock seems the way to go. How should I do that? > Should I just disconnect the 10MHz reference cable while leaving the > PPS cable connected? > > Cheers, > Francois > > > On 06.05.2014 17:21, Marcus Müller via USRP-users wrote: >> Hi Francois, >> >> I assume you not only use the GPSDO to get a time and a PPS to set >> that time, but also to discipline the oscillator of your USRP. >> >> Generally, when losing GPS fix, the GPSDO will continue to provide a >> clock. >> However, in my experience, that (not GPS coordinated) clock is worse >> than the USRP-built in oscillator, but not much. Is there a reason why >> you would want to remove the GPS antenna during use? >> >> Also, you should be able to use the PPS / time feature of the GPSDO >> without disciplining the clock using the GPSDO, and from my >> perspective that might be what you want to do. >> >> Greetings, >> Marcus >> >> On 06.05.2014 09:46, Francois Quitin via USRP-users wrote: >> >>> Dear all, >>> >>> A little question about the GPSDO-module disciplined local >>> oscillator. We >>> have run an experiment where we connect a GPS antenna to the GPSDO >>> module, >>> and let the GPSDO acquire enough satellites to achieve a lock (this >>> is to >>> align the time of different USRPs). We then disconnect the GPS >>> antenna, and >>> let the local oscillator run by itself. >>> >>> After disconnecting the GPS antenna, will the local oscillator >>> behave like >>> any oscillator, or will there be some lingering effect from the >>> GPSDO module >>> GPS correction? >>> >>> Thanks a lot, >>> >>> Francois >>> >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com [1] >> >> >> >> Links: >> ------ >> [1] 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

Hi Matt, Thanks for the answer. And in this "holdover mode", there is no more correction of the clock, right? (I need the oscillator to have predictable behavior, e.g. more or less linear drift without any digital corrections) Francois On 07.05.2014 09:28, Matt Ettus wrote: > If you allow the GPSDO to lock and sit for a while, and then remove > the antenna, it should go into "holdover mode".  This will maintain > more accuracy than if you had not had GPS to start with, but obviously > not as good as if you had kept the antenna attached.  If you remove > power and power up again you lose the holdover performance and go into > normal free-run. > > Matt > > On Tue, May 6, 2014 at 9:46 AM, Francois Quitin via USRP-users > <usrp-users@lists.ettus.com> wrote: > >> Dear all, >> >>   >> >> A little question about the GPSDO-module disciplined local >> oscillator. We have run an experiment where we connect a GPS antenna >> to the GPSDO module, and let the GPSDO acquire enough satellites to >> achieve a lock (this is to align the time of different USRPs). We then >> disconnect the GPS antenna, and let the local oscillator run by >> itself. >> >>   >> >> After disconnecting the GPS antenna, will the local oscillator behave >> like any oscillator, or will there be some lingering effect from the >> GPSDO module GPS correction? >> >>   >> >> Thanks a lot, >> >> Francois >> >>   >> >> -- >> >> Francois Quitin, Ph.D. >> >>   >> >> Research Fellow >> >> INFINITUS - Infocomm Centre of Excellence >> >> School of EEE, Nanyang Technological University >> >> 50 Nanyang Ave, S2-B4b-05 >> Singapore 639798 >> >> Phone: +65-8502-3690 [1] >> >> Email: fquitin@ntu.edu.sg >> >>   >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> [2] > > > > Links: > ------ > [1] tel:%2B65-8502-3690 > [2] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Linear drift in frequency or linear drift in time? Linear drift in time is given by a roughly constant frequency error. Frequency error will change slowly over time, but it will be random. Also, there might be digital correction during the holdover period, but I am not certain. I will try to find out. Matt On Wed, May 7, 2014 at 9:42 AM, fquitin <fquitin@ulb.ac.be> wrote: > Hi Matt, > > Thanks for the answer. And in this "holdover mode", there is no more > correction of the clock, right? (I need the oscillator to have predictable > behavior, e.g. more or less linear drift without any digital corrections) > > Francois > > > On 07.05.2014 09:28, Matt Ettus wrote: > >> If you allow the GPSDO to lock and sit for a while, and then remove >> the antenna, it should go into "holdover mode". This will maintain >> more accuracy than if you had not had GPS to start with, but obviously >> not as good as if you had kept the antenna attached. If you remove >> power and power up again you lose the holdover performance and go into >> normal free-run. >> >> Matt >> >> On Tue, May 6, 2014 at 9:46 AM, Francois Quitin via USRP-users >> <usrp-users@lists.ettus.com> wrote: >> >> Dear all, >>> >>> >>> >>> A little question about the GPSDO-module disciplined local oscillator. >>> We have run an experiment where we connect a GPS antenna to the GPSDO >>> module, and let the GPSDO acquire enough satellites to achieve a lock (this >>> is to align the time of different USRPs). We then disconnect the GPS >>> antenna, and let the local oscillator run by itself. >>> >>> >>> >>> After disconnecting the GPS antenna, will the local oscillator behave >>> like any oscillator, or will there be some lingering effect from the GPSDO >>> module GPS correction? >>> >>> >>> >>> Thanks a lot, >>> >>> Francois >>> >>> >>> >>> -- >>> >>> Francois Quitin, Ph.D. >>> >>> >>> >>> Research Fellow >>> >>> INFINITUS - Infocomm Centre of Excellence >>> >>> School of EEE, Nanyang Technological University >>> >>> 50 Nanyang Ave, S2-B4b-05 >>> Singapore 639798 >>> >>> Phone: +65-8502-3690 [1] >>> >>> >>> Email: fquitin@ntu.edu.sg >>> >>> >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com [2] >>> >> >> >> >> Links: >> ------ >> [1] tel:%2B65-8502-3690 >> [2] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >

Linear drift in time (with of course a slowly varying frequency error). The main thing I want to avoid is digital corrections in the holdover mode, as this cannot be predicted by any type of LO model. Francois On 07.05.2014 09:49, Matt Ettus wrote: > Linear drift in frequency or linear drift in time?  Linear drift in > time is given by a roughly constant frequency error.  Frequency error > will change slowly over time, but it will be random.  Also, there > might be digital correction during the holdover period, but I am not > certain.  I will try to find out. > > Matt > > On Wed, May 7, 2014 at 9:42 AM, fquitin <fquitin@ulb.ac.be> wrote: > >> Hi Matt, >> >> Thanks for the answer. And in this "holdover mode", there is no more >> correction of the clock, right? (I need the oscillator to have >> predictable behavior, e.g. more or less linear drift without any >> digital corrections) >> >> Francois >> >> On 07.05.2014 09:28, Matt Ettus wrote: >> >>> If you allow the GPSDO to lock and sit for a while, and then remove >>> the antenna, it should go into "holdover mode".  This will maintain >>> more accuracy than if you had not had GPS to start with, but >>> obviously >>> not as good as if you had kept the antenna attached.  If you remove >>> power and power up again you lose the holdover performance and go >>> into >>> normal free-run. >>> >>> Matt >>> >>> On Tue, May 6, 2014 at 9:46 AM, Francois Quitin via USRP-users >>> <usrp-users@lists.ettus.com> wrote: >>> >>>> Dear all, >>>> >>>>   >>>> >>>> A little question about the GPSDO-module disciplined local >>>> oscillator. We have run an experiment where we connect a GPS antenna >>>> to the GPSDO module, and let the GPSDO acquire enough satellites to >>>> achieve a lock (this is to align the time of different USRPs). We >>>> then disconnect the GPS antenna, and let the local oscillator run by >>>> itself. >>>> >>>>   >>>> >>>> After disconnecting the GPS antenna, will the local oscillator >>>> behave like any oscillator, or will there be some lingering effect >>>> from the GPSDO module GPS correction? >>>> >>>>   >>>> >>>> Thanks a lot, >>>> >>>> Francois >>>> >>>>   >>>> >>>> -- >>>> >>>> Francois Quitin, Ph.D. >>>> >>>>   >>>> >>>> Research Fellow >>>> >>>> INFINITUS - Infocomm Centre of Excellence >>>> >>>> School of EEE, Nanyang Technological University >>>> >>>> 50 Nanyang Ave, S2-B4b-05 >>>> Singapore 639798 >>>> >>>> Phone: +65-8502-3690 [1] [1] >>>> >>>> Email: fquitin@ntu.edu.sg >>>> >>>>   >>>> _______________________________________________ >>>> USRP-users mailing list >>>> USRP-users@lists.ettus.com >>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>>> [2] [2] >>> >>> Links: >>> ------ >>> [1] tel:%2B65-8502-3690 >>> [2] >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>> [2] > > > > Links: > ------ > [1] tel:%2B65-8502-3690 > [2] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Francois, The module will adjust its control voltage over time based on temperature and according to its model for crystal aging. So it will try to remain as accurate as possible. Matt On Wed, May 7, 2014 at 9:53 AM, fquitin <fquitin@ulb.ac.be> wrote: > Linear drift in time (with of course a slowly varying frequency error). > The main thing I want to avoid is digital corrections in the holdover mode, > as this cannot be predicted by any type of LO model. > > Francois > > > On 07.05.2014 09:49, Matt Ettus wrote: > >> Linear drift in frequency or linear drift in time? Linear drift in >> time is given by a roughly constant frequency error. Frequency error >> will change slowly over time, but it will be random. Also, there >> might be digital correction during the holdover period, but I am not >> certain. I will try to find out. >> >> Matt >> >> On Wed, May 7, 2014 at 9:42 AM, fquitin <fquitin@ulb.ac.be> wrote: >> >> Hi Matt, >>> >>> Thanks for the answer. And in this "holdover mode", there is no more >>> correction of the clock, right? (I need the oscillator to have predictable >>> behavior, e.g. more or less linear drift without any digital corrections) >>> >>> Francois >>> >>> On 07.05.2014 09:28, Matt Ettus wrote: >>> >>> If you allow the GPSDO to lock and sit for a while, and then remove >>>> the antenna, it should go into "holdover mode". This will maintain >>>> more accuracy than if you had not had GPS to start with, but obviously >>>> not as good as if you had kept the antenna attached. If you remove >>>> power and power up again you lose the holdover performance and go into >>>> normal free-run. >>>> >>>> Matt >>>> >>>> On Tue, May 6, 2014 at 9:46 AM, Francois Quitin via USRP-users >>>> <usrp-users@lists.ettus.com> wrote: >>>> >>>> Dear all, >>>>> >>>>> >>>>> >>>>> A little question about the GPSDO-module disciplined local oscillator. >>>>> We have run an experiment where we connect a GPS antenna to the GPSDO >>>>> module, and let the GPSDO acquire enough satellites to achieve a lock (this >>>>> is to align the time of different USRPs). We then disconnect the GPS >>>>> antenna, and let the local oscillator run by itself. >>>>> >>>>> >>>>> >>>>> After disconnecting the GPS antenna, will the local oscillator behave >>>>> like any oscillator, or will there be some lingering effect from the GPSDO >>>>> module GPS correction? >>>>> >>>>> >>>>> >>>>> Thanks a lot, >>>>> >>>>> Francois >>>>> >>>>> >>>>> >>>>> -- >>>>> >>>>> Francois Quitin, Ph.D. >>>>> >>>>> >>>>> >>>>> Research Fellow >>>>> >>>>> INFINITUS - Infocomm Centre of Excellence >>>>> >>>>> School of EEE, Nanyang Technological University >>>>> >>>>> 50 Nanyang Ave, S2-B4b-05 >>>>> Singapore 639798 >>>>> >>>>> Phone: +65-8502-3690 [1] [1] >>>>> >>>>> >>>>> Email: fquitin@ntu.edu.sg >>>>> >>>>> >>>>> _______________________________________________ >>>>> USRP-users mailing list >>>>> USRP-users@lists.ettus.com >>>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com[2] [2] >>>>> >>>> >>>> Links: >>>> ------ >>>> [1] tel:%2B65-8502-3690 >>>> [2] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com[2] >>>> >>> >> >> >> Links: >> ------ >> [1] tel:%2B65-8502-3690 >> [2] http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >